the association between diet and working hours with

THE ASSOCIATION BETWEEN DIET AND WORKING

HOURS WITH MARKERS OF CARDIOMETABOLIC

HEALTH IN THE BRITISH POLICE FORCE

Thesis submitted for the degree of

Doctor of Philosophy

Imperial College London

Rachel Gibson

Division of Diabetes, Endocrinology and Metabolism

Section of Investigative Medicine

Department of Medicine

2016

! ! ! ! "!

PUBLICATIONS

Publications and conference presentations based on work presented in or related to this

thesis are as follows:

Conference presentations

Nutrition Society Winter Meeting 8th December 2015: Sex differences in the relationship

between work patterns and diet in British police force employees: a nested cross-sectional

study

The Rank Prize Funds Mini-symposium on Metabolism and Circadian Rhythms: 24th -27th

February 2014: The impact of shift work on nutrient intake and metabolic risk:

methodological issues

Published abstract

Gibson R, Eriksen R, Chan Q, Vergnaud AC, Singh D, Heard A, Spear J, Aresu M,

McRobie D, Elliott P, Frost G. (2016) Sex differences in the relationship between work

patterns and diet in British police force employees: a nested cross-sectional study Proc

Nutr Soc, 75 OCE1: E20

! ! ! ! #!

THESIS ABSTRACT

Long hours and shift work have been associated with cardiometabolic disease risk.

Despite diet being an established modifiable risk factor, few studies have examined how

dietary behaviours vary in relation to working hours. The aim of this thesis was to

investigate the relationship between working hours and diet quality with markers of

cardiometabolic health.

Cross-sectional analyses were conducted using data from The Airwave Health Monitoring

Study - a British police occupational cohort (n=5,849). Number of weekly working hours

was determined from questionnaire data. The Dietary Approaches to Stop Hypertension

score was calculated using 7-day diet data to measure diet quality. Markers of

cardiometabolic health included: adiposity (body mass index, waist circumference and

body fat), blood pressure, cholesterol, HbA1c and C-reactive protein. Sub-group analyses

were conducted in participants with available shift work data (based on police radio

records) (n=2,323). As part of this thesis a revised food diary and shift work questionnaire

were developed and piloted for use in future studies.

Male employees (n=3,332) working >49hrs per week (vs. 35-40hrs) were more likely to

have a dietary pattern associated with elevated cardiometabolic risk. There was a positive

dose-response relationship across working hours (!35-40, 41-48, 49-54, !55hrs per week)

with markers of adiposity in male employees. Diet quality did not modify this association.

Based on limited shift work data: night workers (vs. day) were found to consume a higher

quantity of sugar-sweetened beverages. However shift work was not associated with

increased cardiometabolic risk.

This thesis suggests a sex specific positive association between weekly working hours and

adiposity that is independent of established risk factors. Temporal eating pattern and

previous shift work data collected using the revised food diary and shift questionnaire will

be important to future studies exploring the relationship between diet, work hours and

health.

! ! ! ! $!

TABLE OF CONTENTS

Abstract 3

List of tables 12

List of figures 15

Statement of personal contribution 16

Acknowledgements 17

Copyright declaration 18

Glossary of abbreviations 19

Chapter 1: Background 21

1.1 Chapter overview 19

1.1.1 The UK burden of cardiovascular disease and type 2 diabetes 21

1.1.2 Diet and cardiometabolic disease risk 21

1.1.4 The workplace and cardiometabolic disease risk 21

1.1.5 The workplace and dietary intake 22

1.1.6 Diet, cardiometabolic health and the police force 22

1.1.7 Background chapter aims 22

1.2 Cardiometabolic disease and risk markers 23

1.2.1 Definitions of cardiometabolic disease end points 23

1.2.2 Markers of cardiometabolic risk and cardiometabolic syndrome 23

1.2.3 Determinants of cardiometabolic disease risk markers 28

1.3 Diet and cardiometabolic disease risk 29

1.3.1 Energy and macronutrient intake and cardiometabolic disease risk 29

1.3.2 Dietary fibre, sodium and cardiometabolic disease risk 32

1.3.3 Food groups, dietary patterns and cardiometabolic disease risk 33

1.3.4 Eating patterns and cardiometabolic health 35

1.4 Factors that influence food choice and dietary behaviour 37

1.4.1 Determinants of food choice 37

1.4.2 Occupational factors and food choice 38

1.5 Non-dietary lifestyle factors and cardiometabolic disease risk 39

1.5.1 Physical activity and inactivity 39

1.5.2 Smoking 40

1.5.3 Sleep 40

1.5.4 Psychological stress 40

1.5.6 Diet and other lifestyle behaviours 41

1.6 Non-modifiable risk factors of cardiometabolic diseases 41

! ! ! ! %!

1.6.1 Age and sex 41

1.6.2 Ethnicity 41

1.6.3 Genetics 42

1.6.4 Socioeconomic status 42

1.5.6 Diet and non-modifiable risk factors 42

1.7 Working hours and cardiometabolic disease risk 43

1.7.1 Definition and prevalence of long working hours 43

1.7.2 Long hours and cardiometabolic disease risk 43

1.7.3 Shift work and cardiometabolic disease risk 44

1.7.4 Defining shift work 44

1.7.5 Potential pathways between work hours and cardiometabolic health 46

1.8 Working patterns and dietary intake: A literature review 48

1.8.1 Long working hours and diet 48

1.8.2 Shift work and diet 48

1.8.3 Food choice and working patterns 49

1.9 Limitations and gaps in current research 56

1.10 Chapter summary 56

1.11 Hypotheses, aims and thesis structure 57

1.11.1 Study hypotheses 57

1.11.2 Research objectives 57

1.11.3 Thesis structure 58

Chapter 2: Core methods (non-dietary) 59

2.1 Introduction to the Airwave Health Monitoring Study 59

2.1.1 Study design 59

2.1.2 Participant recruitment 59

2.1.3 Data collection 59

2.1.4 Sample selection for nutritional studies 61

2.2 Occupational variables 62

2.2.1 Rank 63

2.2.2 Working environment 63

2.2.3 Working hours 64

2.2.4 Job satisfaction, control and demand 66

2.2.5 Length of service and time in current job role 66

2.2.6 Employment region 66

2.3 Measures of cardiometabolic health 66

2.3.1 Anthropometric measures 66

2.3.2 Biochemical and clinical measurements 67

! ! ! ! &!

2.3.3 Medical and pharmacological information 68

2.3.4 Classification of cardiometabolic health 68

2.4 Additional covariates 70

2.4.1 Demographic and socio-economic 70

2.4.2 Physical activity 70

2.4.3 Sedentary time 70

2.4.4 Additional lifestyle behaviours 70

2.5 Missing data 71

2.6 Collinearity between explanatory variables 72

2.7 Core statistical methods 72

2.7.1 Descriptive statistics 73

2.7.2 Multivariable analyses 73

2.7.3 Sub group and sensitivity analyses 74

CHAPTER 3: Dietary protocol development 75

3.1 Introduction 75

3.1.1 Background and rationale 75

3.1.2 Standard operating procedure aims 75

3.2 Methods 76

3.2.1 Dietary data collection 76

3.2.2 Food record coding protocol 79

3.2.3 Coder training 80

3.2.4 Quality control 81

3.2.5 Dietary data cleaning 82

3.2.6 Dietary variable generation 83

3.2.7 Missing dietary data 86

3.3 Results 86

3.3.1 Codebook development 86

3.3.2 Summary coding statistics 87

3.3.3 Quality check of coding 89

3.3.4 Data cleaning 89

3.4 Discussion 89

3.4.1 Coding consistency and error rates 90

3.4.2 Nutritional database limitations 90

3.5 Conclusions 91

CHAPTER 4. STUDY 1: Dietary energy intake misreporting 92

4.1 Introduction 92

! ! ! ! '!

4.1.1 Background and study rationale 92

4.1.2 Study aims and objectives 92

4.2 Methods 93

4.2.1 Participants 93

4.2.2 Dietary and non-dietary variables 93

4.2.3 Classification of energy intake misreporting 93

4.2.4 Statistical methods 95

4.3 Results 96

4.3.1 Summary characteristics 96

4.3.2 Logistic regression 102

4.3.3 Sensitivity analyses 105

4.4 Discussion 105

4.4.1 Summary of key findings 105

4.4.2 Discussion of main findings 106

4.4.3 Study strengths and limitations 109

4.5 Study conclusions and relevance to further studies 109

CHAPTER 5: STUDY 2 - Dietary profile of British police force employees 111

5.1 Introduction 111



5.2 Methods 112


5.2.2 Dietary measurements 112

5.2.3 Group level variables 114


5.3 Results 115


5.3.2 Dietary profile across sex 119

5.3.3 Dietary profile across region of employment 121

5.3.4 Dietary profile across job role / rank 124

5.3.5 Dietary profile across working hours 127

5.3.6 Sub-group analysis: Dietary profile across shift work categories 133

5.4 Discussion 136





! ! ! ! (!

CHAPTER 6. STUDY 3: Diet quality and cardiometabolic risk 142




6.2 Methods 143


6.2.2 Dietary variables 143

6.2.3 Outcome measurements of cardiometabolic risk 143

6.2.4 Covariate measures 143


6.3 Results 146


6.3.3 Association between DASH score and markers of cardiometabolic health 146

6.3.4 Participant characteristics associated with a poor diet quality 154

6.4 Discussion 158





CHAPTER 7. STUDY 4: Working hours and cardiometabolic risk 164




7.2 Methods 165


7.2.2 Working hour exposure measures 165

7.2.3 Outcome measurements of cardiometabolic risk 165

7.2.4 Dietary variables 166

7.2.5 Covariate measures 166


7.3 Results 168


7.3.3 Characteristics across working hour groups 171

7.3.3 Working hours and markers of cardiometabolic health 171

7.4 Discussion 178


! ! ! ! )!




CHAPTER 8. Methodological development – mapping diet to working hours 184


8.2 Time of day food diary 185

8.2.1 Design and pilot 185

8.1.2 Exploratory cross-sectional study 185

8.3 Shift work questionnaire 194

8.3.1 Design phase 194

8.3.2 Testing phase 195

8.3.3 Preliminary analyses 196

8.3.4 Preliminary results 196

8.3.5 Discussion of future work 197

8.4 Chapter summary 197

CHAPTER 9. Synthesis of results and overall conclusions 198

9.1 Synthesis of results 198


9.2 Interpretation and implications of findings 201

9.3 Overall strengths and limitations 203

9.4 Future work 204

9.5 Overall conclusion 205

References 206

Appendices 229

A1.1 Working hours and diet literature search methods 231

A2.1 Key participant characteristics compared across

study samples 233

A2.2 Random selection criteria applied for sample selection 234

A2.3 Bland-Altman Plot: Weekly working hours measured by payroll data and self

reported values 235

A2.4 Police force regions by country 236

A2.5 Investigation of missing data by participant characteristics 237

A2.6 Tests of collinearity between categorical variables 238

A3.1 Standard operating procedure for coding of food diaries in the Airwave Health

Monitoring Study 240

! ! ! ! *+!

A3.2 Example of the food diary used to capture dietary intake in the Airwave Health

Monitoring Study 2007-2013 263

A3.3 Food groups and descriptions used in the dietary assessment of the Airwave

Health Monitoring Study 264

A4.1 Study 1: Included vs. excluded for energy intake misreporting 265

A4.2 Sensitivity analyses results tables: predictors of energy intake under reporting 266

A5.1 Quintile cut-offs for calculating DASH scores based on individuals from the

Airwave Health Monitoring Study 271

A5.2 Dietary profile by quintile cut off of DASH score for men and women in the


A5.3 Comparison of key characteristics across ranked and non-ranked police force

employees included in the Airwave Health Monitoring Study 276

A5.4 Comparison of key characteristics region of employment for Airwave Health

Monitoring Study participants 278

A 5.5 Comparison of key characteristics by working hours for Airwave Health

Monitoring Study participants 280

A5.6 Partial correlation coefficients for dietary macronutrients and food group intakes

of the Airwave Health Monitoring Study participants 287

A5.7 Sensitivity analyses: Dietary profiles across working hour groups amongst

mid-ranking police officers: men 288

A5.8 Sensitivity analyses: Dietary profiles across working hour groups amongst

ranked police officers: women 290

A5.9 Sub cohort profile: participant characteristics across shift work classification 292

A6.1 Airwave Health Monitoring Study participant characteristics across fifths of

DASH score 296

A6.2 Partial correlation coefficients for dietary variables, markers of cardiometabolic

risk, working hours and activity 307

A6.3 Sensitivity analyses: diet and cardiometabolic risk (energy intake under-reporters

excluded) 308

A7.1 Comparison of participants reporting part-time work and/or chronic disease

diagnosis with the rest those who work full time without chronic disease diagnosis 313

A7.2 Comparison of demographic, lifestyle and occupational characteristics (Study 4) 314

A7.3 Prevalence of cardiometabolic biomarkers and anthropometric measurements

against risk categories across men and women in the Airwave Health Monitoring Study

with dietary data 316

A7.4 Sample characteristics by group of weekly working hours 317

A7.5 Sensitivity analyses Study 4 (non-mid rank employees excluded) 326

A7.6 Sensitivity analyses Study 4 (energy intake under-reporters excluded) 329

! ! ! ! **!

A7.7 Sensitivity analyses Study 4 (include part time workers) 336

A7.8 Association between shift work and cardiometabolic health markers 337

A8.1 By time food diary – sample pages 339

A8.2 Amended food diary: pilot study 340

A8.3 Decision tree classification to determine common shift type reported in a sub-group

of participants from the Airwave Health Monitoring Study 344

A8.4 Summary of frequency and mean length of shift categories recorded included

for analyses 345

A8.5 New shift work questionnaire 346

Copyright permissions appendices 354

! ! ! ! *"!

LIST OF TABLES

Chapter 1: Background

Table 1.1 Components of the Metabolic Syndrome Definition: Consensus definition 24

Table 1.2 WHO body mass index cut-off points and associated health risks 25

Table 1.3 Examples of the types of shift work associated with cardiometabolic risk 45

Table 1.4 Summary of studies investigating working hour arrangements and diet 50

Chapter 2: Core methods (non-dietary)

Table 2.1 Summary of the data collected by the Airwave Health Monitoring 61

Table 2.2 Classification of working environment based on job description 63

Table 2.3 Intra-class correlation coefficients of biological and clinical measurements 68

Table 2.4 Classification of increased cardiometabolic risk across anthropometric and

biochemical measurement taken in the Airwave Health Monitoring Study 69

Table 2.5 Sub-group and sensitivity analyses conducted by study 74

CHAPTER 3: Dietary protocol development

Table 3.1 Key elements of the Airwave Health Monitoring Study standard

dietary coding protocol 77

Table 3.2 Example scenarios and possible coding solutions when exact code

matches between recorded foods and UKN database codes are not available 80

Table 3.3 Example of how composite dishes were disaggregated into each food

group of interest 85

Table 3.4 Food diaries coded and checked per coder, and overall mean error rate 87

Table 3.5 Mean food record coding error rates per coder per three-month period 88

Table 3.6 Examples of different errors found during quality checking procedures 89

CHAPTER 4. STUDY 1: Dietary energy intake misreporting

Table 4.1 Example of upper and lower cut off values for ratio of energy intake to basal

metabolic rate 94

Table 4.2 Characteristics of participants included in the study of energy intake reporting 97

Table 4.3 Comparison of demographic, anthropometric, lifestyle and occupational

characteristics of under- and plausible reporters of energy intake 98

Table 4.4 Predictors of under-reporting energy amongst men 103

Table 4.5 Predictors of under-reporting energy amongst women 104

! ! ! ! *#!

CHAPTER 5: STUDY 2 - Dietary profile of British police force employees

Table 5.1 Comparison of demographic, lifestyle and occupational characteristics across

men and women with dietary data from the Airwave Health Monitoring Study 117

Table 5.2 Comparison of dietary intakes across men and women in the Airwave Health

Monitoring Study and comparison against the UK National Diet and Nutrition Survey

and UK dietary guidelines 120

Table 5.3 Measurement of dietary differences across region of employment: men 122

Table 5.4 Measurement of dietary differences across region of employment: women 123

Table 5.5 Measurement of dietary differences across job role/rank: men 125

Table 5.6 Measurement of dietary differences across job role/rank: women 126

Table 5.7 Dietary profile across working hour groups: men 129

Table 5.8 Dietary profile across working hour groups: women 131

Table 5.9 Sub analyses: dietary profile across shift work classification: men 134

Table 5.10 Sub analyses: dietary profile across shift work classification: women 135

CHAPTER 6. STUDY 3: Diet quality and cardiometabolic risk

Table 6.1 Cut off values for cardiometabolic risk factor calculation 144

Table 6.2 Model variables for general linear models investigating the relationship

between diet quality (DASH score) and markers of cardiometabolic risk 145

Table 6.3 Association of DASH score with markers of cardiometabolic health in the

Airwave Health Monitoring Study: Men 149

Table 6.4 Association of DASH score with markers of cardiometabolic health in the

Airwave Health Monitoring Study: Women 151

Table 6.5 Odds ratio of having three or more markers of metabolic risk per quartile

of DASH score: men 153

Table 6.6 Odds ratio of having three or more markers of metabolic risk per quartile

of DASH score: women 153

Table 6.7 Odds ratio of being in the lowest DASH diet quality group 155

CHAPTER 7. STUDY 4: Working hours and cardiometabolic risk

Table 7.1 Model variables in general linear models investigating the association between

weekly number of working hours and markers of cardiometabolic risk 167

Table 7.2 Association between number of weekly working hours and markers of

cardiometabolic health: men 173

Table 7.3 Association between number of weekly working hours and markers of

cardiometabolic health: women 176

! ! ! ! *$!

CHAPTER 8. Methodological development – mapping diet to working hours

Table 8.1 Criteria for the classification of each shift recoded by participants in the


Table 8.2 Summary characteristics of participants enrolled from the London Metropolitan

police force in 2014 that completed the baseline food diary 188

Table 8.3 Differences in average nutrient, food group and eating occasions across

different shifts for men and women 190

Table 8.4 Percentage of participants recording energy intake by four-hourly periods

per shift worked 191

Table 8.5 Criteria for the Airwave Health Monitoring Health Monitoring Study shift work

questionnaire 194

Table 8.6 Selected occupational characteristics obtained using the new shift work

questionnaire during follow-up of the Airwave Health Monitoring Study cohort 196

CHAPTER 9. Synthesis of results and overall conclusions

Table 9.1 Summary of the research objectives achieved in this thesis and the outcome

of each hypothesis 199

! ! ! ! *%!

LIST OF FIGURES

Chapter 1: Background

Figure 1.1 Schematic of the hypothesized causal pathway between visceral adiposity,

type 2 diabetes and cardiovascular diseases 26

Figure 1.2 Proposed pathophysiology of the metabolic syndrome including

inflammatory pathways 27

Figure 1.3 Factors, and markers associated with cardiometabolic disease development 29

Figure 1.4 Framework for the determinants of eating behaviour 38

Chapter 2: Core methods (non-dietary)

Figure 2.1 Enrolment in the Airwave Health Monitoring Study per region 60

Figure 2.2 Schematic showing the sample selection for inclusion in cross-sectional

studies included in this thesis 62

CHAPTER 3: Dietary protocol development

Figure 3.1 Example of an algorithm from the standard coding protocol to aid portion

size estimation for cold and ambient beverages 78

Figure 3.2 Schematic of quality control cycle used in the Airwave Health Monitoring

Study to maintain food record coding consistency 81

Figure 3.3 Schematic of the gross quality check and data cleaning applied to the

Airwave Health Monitoring study dietary data 83

Figure 3.4 Schematic of food group allocation procedure applied to the UKN and

Airwave Health Monitoring Study dietary codes. 84

CHAPTER 7. STUDY 4: Working hours and cardiometabolic risk

Figure 7.1 Model of effect modification 167

Figure 7.2 Cardiometabolic phenotype of men and women classified with having three

or more cardiometabolic risk 170

CHAPTER 8. Methodological development – mapping diet to working hours

Figure 8.1 Schematic showing sample selection for inclusion in cross-sectional study

investigating the mapping of dietary intake to working hours 186

Figure 8.2 Percentage of participant’s energy intake by four-hourly periods per shift type 192

STATEMENT OF PERSONAL CONTRIBUTION

The author performed all of the work described in this thesis. Collaborations and assistance

are detailed below. The work of others is fully cited, referenced and/or acknowledged. The

author was responsible for the processing and handling of the nutritional data presented in this

thesis. All critical appraisal and interpretation presented in this thesis are the opinion of the

author. The Airwave Health Monitoring Study team at Imperial College London undertook

participant recruitment and primary data collection. Dietary data generation: protocol development (Chapter 3)

The author jointly developed the standard dietary coding protocol, trained the dietary coders,

audited the coding process, and cleaned the dietary data with Rebeca Eriksen (PhD student,

Imperial College). Yvonne McMeel (Research Technician, Imperial College) and Katie Lamb

(Research Technician, Imperial College) contributed to the testing of the standard dietary

coding protocol. Rebeca Eriksen, Jessica Ayling (MRes student, Imperial College), Andrea

Carames (MRes student, Imperial College), Yvonne McMeel and Katie Lamb assisted the

author in the composite food disaggregation. The author contributed to the coding of the food

diaries included in this thesis as part of the Airwave Health Monitoring Study nutritional

assessment team.

Methodological development – mapping diet data to working hours (Chapter 8)

The author amended the diet diary, designed and managed the piloting of the new food diaries

with the assistance of Dr Deepa Singh (Imperial College) who organised the recruitment of

participants for the pilot study. Dr Anne-Claire Vergnaud and Dennis McRobie (Imperial

College) assisted the author in the testing of the shift work questionnaire. Signature: Date:

Name of author of PhD thesis: Rachel Gibson

12/12/2016

We confirm that this is an accurate representation of this work:

Signature: Date: 13/12/2016

Name of supervisor: Professor Paul Elliott

Signature: Date:

Name of supervisor: Professor Gary Frost

Signature: Date:

Name of supervisor: Dr. Queenie Chan

16

13/12/2016

13/12/2016

! ! ! ! *'!

ACKNOWLEDGEMENTS

The Airwave Health Monitoring Study is funded by the Home Office (Grant number 780-

TETRA) with additional support from the National Institute for Health Research, Imperial

College Healthcare NHS Trust and Imperial College Biomedical Research Centre. The

diet coding was supported through discretionary departmental funds.

I am extremely grateful to my academic supervisors Professor Gary Frost, Professor Paul

Elliott and Dr Queenie Chan for their support and guidance throughout my PhD.

I would also like to thank - the participants that have taken part in the Airwave Health

Monitoring Study. Also, The Airwave Health Monitoring Study investigators at Imperial

College involved in data collection and data extraction of the non-dietary data: Anne-Claire

Vergnaud, Deepa Singh, Andy Heard, Jeanette Spear, Dennis McRobie and Maria Aresu.

The food diary coders: Jessica Ayling, Andrea Carames, Chujie Chen, Zhengyu Fan,

Kirsty Frost, Louise Hirichi, Zanna Hofstede, Niamh O’Sullivan, Kristina Petersen,

Fathimah Sigit, Claudia Schreuder, Manny Singh, Elizabeth Slack, Jill Twomey, Alan Wan,

Jessica Ware and Yiling Zhu. With a special thank you to Yvonne McMeel and Katie Lamb

the Research Technicians for three years of food diary coding which generated a large

proportion of the data used in this thesis and Rebeca Eriksen for being a great co-

researcher over the last three years. David Hughes at Forestfield Systems Ltd. for his

support in adapting the Dietplan nutritional software to meet the needs of our study and

Nima Khandan-Nia (Imperial College, London) for assisting with nutritional and statistical

software management. Dr Lesley Rushton (Imperial College), Dr Ruth Travis (Oxford

University) and Professor Simon Folkard (University Paris Descartes, France and

Swansea University) for advice regarding the development of the shift work questionnaire

(Chapter 8).

I would also like to thank family, friends and also colleagues in the Nutrition and Dietetic

Research Group at Imperial College for their advice, support and interest in my research.

Finally, thank you to my husband Jay for his unwavering support and encouragement.

! ! ! ! *(!

COPYRIGHT DECLARATION

The copyright of this thesis rests with the author and is made available under a Creative

Commons Attribution Non-Commercial No Derivatives licence. Researchers are free to

copy, distribute or transmit the thesis on the condition that they attribute it, that they do not

use it for commercial purposes and that they do not alter, transform or build upon it. For

any reuse or redistribution, researchers must make clear to others the licence terms of this

work.

! ! ! ! *)!

GLOSSARY OF ABBREVIATIONS

ACPO Association of Chief Police Officers Apo(B) Apolipoprotein B BCOPS Buffalo Cardio-metabolic Occupational Police Stress study BMI Body Mass Index BMR Basal metabolic rate CI Confidence Interval cm Centimetres COPD Chronic obstructive pulmonary disease CRP C-reactive protein CVD Cardiovascular Disease DASH Dietary Approaches to Stop Hypertension DW Day work EI Energy intake EPIC European Prospective Investigation into Cancer ERFC Emerging Risk Factors Collaboration FFQ Food frequency questionnaire g Gram GCSE General Certificate of Higher Education GI Glycaemic index GLM General Linear Model GWAS Genome-wide association studies HbA1c Glycated haemoglobin Hs-CRP High sensitivity C-reactive protein HDL High Density Lipoprotein HR Hazard Ratio HSE Health Survey for England IL-6 Interleukin-6 IDL Intermediate-density lipoprotein INTERMAP INTERnational collaborative of MAcronutrients, micronutrients

and blood Pressure IPAQ-SF International Physical Activity Questionnaire – Short form IQR Interquartile range Kg Kilogram Kcal Kilocalorie LDL Low Density Lipoprotein m Meters mmol/L Millimoles per Litre MedD Mediterranean Diet MET Metabolic equivalent MetS Metabolic syndrome mmHg Milligrams of mercury

! ! ! ! "+!

MUFA Mono Unsaturated Fatty Acids NCD Non communicable diseases NDNS National Diet and Nutrition Survey NHANES National Health and Examination Survey NME Non-milk extrinsic sugars NS Night shift NSP Non starch polysaccharides OGTT Oral glucose tolerance test OR Odds ratio PAL Physical activity level PUFA Poly unsaturated fatty Acids RCT Randomised Controlled Trial RSW Rotating shift work RR Relative Risk SD Standard deviation SE Standard error SACN Scientific Advisory Committee on Nutrition SFA Saturated Fatty Acids SAS Statistical Analysis Software SCFA Short chain fatty acids SSB Sugar sweetened beverages SNP Single nucleotide polymorphisms SUN Seguimento Universidad de Navarra cohort study SW Shift work T2DM Type 2 Diabetes Mellitus TEE Total energy expenditure TG Triacylglyceride TV Television UK United Kingdom UKN United Kingdom Nutritional dataset USA United States of America VAT Visceral adipose tissue VLDL Very low-density lipoprotein WC Waist circumference WHO World Health Organisation

! ! ! ! "*!

CHAPTER 1: BACKGROUND

1.1 Chapter overview

1.1.1 The UK burden of cardiovascular disease and type 2 diabetes

Cardiovascular disease (CVD) is the leading cause of death in the UK for men and the

second leading cause for women (1). Type 2 diabetes (T2DM) doubles the risk of

developing CVD (2). It is estimated that 10% of the UK adult population are diagnosed

with CVD and that 6% are diagnosed or living with undiagnosed T2DM (3). The annual

UK health care cost to treat these diseases is £8.6bn and £10bn respectively (3,4). T2DM

prevalence is forecast to increase to 9.5% in the UK population by 2030 (5) contributing to

the future UK burden of CVD.

1.1.2 Diet and cardiometabolic disease risk

The role of diet in the aetiology CVD and cardiometabolic disease is well established (6),

with dietary modification central to primary prevention and secondary management (7–9).

Understanding the dietary behaviours of different population groups provides an important

link between diet and cardiometabolic disease. However food choice is influenced by a

variety of physiological, environmental and psychosocial factors (10,11).

1.1.4 The workplace and cardiometabolic disease risk

Two types of work related disease have been described by the World Health Organization

(WHO): i) ‘occupational diseases’ which occur as a direct result of workplace specific

exposures and ii) ‘multifactorial diseases’ which describe diseases that occur in the

general population, but may be exacerbated or partially caused by workplace exposures

(12). There are an increasing number of studies suggesting that T2DM and CVD are

multifactorial work related diseases associated with long and atypical working hours (13–

15).

! ! ! ! ""!

1.1.5 The workplace and dietary intake

The number of years spent in training and employment can be substantial and it is

estimated that about a third of daily calorie intake is consumed in the workplace (16).

Consequently the workplace is likely to play a key role in shaping dietary behaviours that

impact on health outcomes. The workplace is now considered an important area for public

health interventions (17). It is therefore important to understand if dietary behaviours are

associated with long and atypical working hours, and if they modify the relationship

between working hours and cardiometabolic diseases.

1.1.6 Diet, cardiometabolic health and the police force

The British Police Force employs over 250,000 people (18) many of these job roles require

long and atypical working hours. A previous study in America observed front line police

officer duties to be associated with increased risk of sudden cardiac death (19). It has also

been suggested that police officers have a higher prevalence of traditional cardiometabolic

risk factors (20,21). The diet of the British police force is of particular interest as the Home

Office commissioned report ‘Managing sickness absence in the police service’ cited

concerns about the impact of closing staff canteens on employee health and wellbeing

(22). To date the diet of this large occupational group in the UK has not been studied.

1.1.7 Background chapter outline

This chapter explores the relevant literature relating to how diet influences cardiometabolic

disease risk. It provides an overview of the factors that affect food choice and presents a

narrative review of studies that have assessed dietary behaviours across various working

hour arrangements. The chapter also summarises the evidence relating working hours to

cardiometabolic disease risk, which supports the need for dietary research in this area.

! ! ! ! "#!

1.2 Cardiometabolic disease and risk markers

1.2.1 Definitions of cardiometabolic disease end points

1.2.1.1 Cardiovascular disease

CVD covers a range of diseases relating to the heart and cardiovascular system and

includes heart disease, stroke, heart failure, peripheral arterial disease and aortic disease

(4). Although CVD includes various diseases, a generalised pathological process is

initiated by lesions in the vascular endothelium causing atherosclerotic plaques and/or

reduced vascular compliance (23).

1.2.1.2 Type 2 diabetes

T2DM is a disease characterised by high concentrations of glucose in the blood. The

current criteria for T2DM diagnosis is a glycated haemoglobin (HbA1c) of >6.5%

(48mmol/mol) or a fasting blood glucose concentration of 7mmol/L with >11mmol/L two

hours after a 75g-glucose oral glucose load (24). T2DM follows a gradual pathology

involving the reduced sensitivity of cells to insulin and the gradual reduction of insulin

secretion from pancreatic beta-cells (24).

1.2.2 Markers of cardiometabolic risk and cardiometabolic syndrome

CVD and T2DM are also referred to as cardiometabolic diseases since metabolic

abnormalities including dyslipidaemia and insulin resistance are associated with an

increased risk of developing these diseases (25). Individual markers (insulin resistance,

blood lipids, blood pressure, inflammation and adiposity) are independently associated

with increased risk of CVD and or T2DM (26), however the co-occurrence of these

markers potentiates disease risk. This clustering of metabolic abnormalities is commonly

referred to as metabolic or cardiometabolic syndrome (MetS). Various criteria have been

used for MetS since the initial definition published by the WHO in 1998. The common

criterion across all definitions include measurements of triglycerides, HDL, blood pressure

and blood glucose (27). The 2009 consensus statement of the components that should be

included when screening for MetS is detailed in Table 1.1 (28). Observational studies

have estimated that MetS increases the risk of developing T2DM by almost five-fold (29)

and CVD two-fold (30). The prevalence of MetS in Northern Europe is estimated to be

between 20-30% depending on the MetS definition applied (31).

! ! ! ! "$!

Table 1.1 Components of the Metabolic Syndrome Definition: Consensus definition

Measure Categorical Cut-off Points Elevated waist circumference (Ethnic and gender specific)

European: !94cm (men), !80cm (women)

Elevated triglycerides (or on drug treatment for elevated triglycerides)

!150mg/dL (1.7mmmol/L)

Reduced high density lipoprotein (HDL) (or on drug treatment for reduced HDL)

< 40mg/dL (1.0mmol/L) Males < 50mg/dL (1.3mmol/L) Females

Elevated blood pressure (or on antihypertensive drug treatment)

Systolic !130 and/or diastolic !85 mm Hg

Elevated fasting glucose (or on drug treatment for elevated glucose)

!100mg/dL

Source of information: Alberti et al,. 2009, Table 1 (28).

1.2.2.1 Obesity and cardiometabolic disease risk

Obesity is a key risk factor for CVD and T2DM (32). The WHO defines obesity as “a

degree of fat storage associated with clearly elevated health risks” (33). Body mass index

(BMI) is a proxy measure of adiposity (due to its high correlation with adiposity) and is

calculated by the following equation: body-mass (kg) / height (m2). BMI ranges are

classified by the WHO based on associations between BMI with morbidity and mortality

(33), Table 1.2. The prevalence of adult obesity in the UK is ~25% and it is forecast to

increase to over 50% by 2050 (34). Despite obesity being a strong predictor of CVD and

T2DM development (32), evidence increasingly shows that cardiometabolic disease risk

can vary within BMI classifications (35,36). A key limitation of BMI in the measure of

obesity is that it cannot distinguish between fat and fat free mass (37). It is estimated that

>25% of those with a ‘healthy’ BMI may be at risk of developing chronic metabolic

diseases such as CVD and T2DM (38).

! ! ! ! "%!

Table 1.2 World Health Organization body mass index cut-off points and associated health risks

Classification BMI range (kg/m2) Health risk

Underweight <18.5 Increased

Normal weight !18.5 <25.0 Normal

Overweight !25.0 <30.0 Increased

Obese class I !30.0 <35.0 Moderate

Obese class II !35.0 <40.0 Severe

Obese class III !40.0 Very severe

Source of information: World Health Organization, 1995 (33)

1.2.2.2 Distribution of body fat and cardiometabolic disease risk

Research from National Health and Examination Survey III (NHANES) has shown that

those with a healthy BMI (WHO ‘normal’ weight) but with central obesity (defined as a

waist to hip circumference ratio >1.0 for men and >0.92 for women) were at higher risk of

cardiovascular mortality compared to a person with the same BMI without central obesity:

the hazard ratio (HR) for men was 1.87 with a 95% confidence interval (95%CI) of 1.53,

2.29 and HR 1.48 95%CI 1.35, 1.62 for women (35). These observations suggest that fat

distribution may be a more important predictor of cardiometabolic health than overall

measures of obesity. Imaging studies have shown excess visceral adipose tissue (VAT) to

be a candidate marker for the sequale of metabolic abnormalities associated with

cardiometabolic disease development (39) Figure 1.1. VAT is adipose tissue that

accumulates in the intra-abdominal cavity. Compared to subcutaneous adipose tissue

(located below the dermal layer) VAT has been shown to have a distinct morphology and

function. VAT secretes less adiponectin (anti-inflammatory cytokine) and more pro-

inflammatory cytokines such as interleukin-6 (IL-6) and tumour necrosis factor-! than

subcutaneous adipose tissue (40). Additionally it contains higher concentrations of

hormone sensitive lipase, facilitating hydrolysis of triglycerides to free fatty acids. This in

turn leads to increased low-density lipoprotein (LDL) cholesterol secretion from the liver.

The CardioMetabolic Health Alliance stated that excess visceral fat was an essential

component of the pathophysiology of cardiometabolic disease (41).

! ! ! ! "&!

Figure 1.1 Schematic of the hypothesized causal pathway between visceral adiposity, type 2 diabetes and cardiovascular diseases

Image adapted by permission from Macmillan Publishers Ltd Nat Clin Pract Cardiovasc Med

(Matsuzawa Y. 2006 Nat 3: 35–42). © 2006 Nature Publishing Group

1.2.2.3 Inflammation and cardiometabolic disease risk

Inflammation is a process that occurs as a reaction to trauma and infection to initiate the

repair of damaged tissues. C-reactive protein (CRP) is an acute phase protein produced

by the liver in response to IL-6. Various tissues in the body including immune response

cells and visceral adipocytes secrete IL-6. Research has shown CRP to be a strong

correlate with future CVD risk (42,43). CRP is recognised as an emerging risk marker that

may explain some of the variability in CVD risk against established markers of risk (42,43).

Inflammation has been linked to the progression of atherosclerosis. It promotes initial

vascular endothelial dysfunction, causing increased permeability to lipoproteins, and

attracts macrophages, thus facilitating the creation of foam cells (44). A meta-analyses

found an independent, positive association between CRP and CVD incidence in those with

a CRP of >3.0mg/L compared to <1mg/L (RR 1.58 95%CI 1.37, 1.83) (45). Research from

the Women’s Health Study (n =15,745) observed CRP to be associated with increased risk

of T2DM, relative risk (RR) 4.2 (95%CI 1.5, 12.0) after adjustment for lifestyle factors and

BMI (46). Figure 1.2 illustrates the proposed pathways between the elements of MetS

including the proinflammatory pathways.

!

Pro-inflammatory signaling proteins, free fatty acids

! !Visceral adiposity

!

Hyperglycemia ! Hypertension

!

Dyslipidemia

!

Type 2 diabetes

! Atherosclerosis -> cardiovascular disease

! ! ! ! "'!

Figure 1.2 Proposed pathophysiology of the metabolic syndrome including inflammatory pathways

Reprinted from The Lancet Volume 365, Eckel RH, Grundy SM, & Zimmet PH, The metabolic

syndrome, p.1415 (63). Copyright 2005, with permission from Elsevier.

1.2.2.4 Blood lipids and cardiometabolic disease risk

Lipoproteins transport triglycerides and cholesterol around the body. CVD risk is related to

the role that each lipoprotein performs in lipid metabolism. High-density lipoproteins (HDL)

have been shown to be protective against CVD (47). The precise mechanisms are yet to

be fully defined in vivo, however it is known that HDL particles promote cholesterol

removal from tissues and have a role in the prevention of oxidation of LDL particles (48).

Conversely LDL promotes delivery of cholesterol to tissues and is liable to oxidation,

leading to lesion formation in the arterial endothelium. High levels of LDL are associated

with an elevated risk of CVD, and therefore LDL reduction is the primary target for CVD

risk management (49). However, non-HDL cholesterol concentration (total cholesterol

concentration minus HDL concentration) measures a larger fraction of atherogenic lipids in

the blood as it also includes very low-density lipoprotein (VLDL) and intermediate-density

(IDL) particles. Meta-analyses have shown non-HDL measurements to be a better

predictor of CVD development than LDL (50).

! ! ! ! "(!

1.2.2.5 Impaired glucose metabolism and cardiometabolic disease risk

Insulin resistance and/or a reduced beta-cell secretion of insulin results in increased blood

glucose concentration (51). Cumulative exposure to elevated levels of blood glucose as a

result of T2DM is associated with tissue damage, in particular tissues of the vascular

system are susceptible, leading to long term disease complications which include

blindness, kidney disease and CVD. Adverse blood glucose concentrations that are

below T2DM diagnostic criteria are described by the WHO as ‘intermediate conditions in

the transition between normal blood glucose levels and diabetes’ (52). Diagnostic criteria

are a fasting blood glucose concentration of !100 to <126 mg/dl (impaired fasting

glucose), or a 2-hr post oral 75g glucose load blood concentration of !140 to <200 mg/dl

(impaired glucose tolerance) (53), or a HbA1c measurement of 5.7-6.4% (54). The Dutch

Hoorn prospective cohort study (n = 1,342) found that impaired fasting glucose and

impaired glucose tolerance corresponded to an odds ratio (OR) of 10.0 (95%CI 6.1, 16.5)

and 10.9 (95%CI 6.0, 19.9) respectively of developing T2DM (mean follow-up period of 6.4

years) compared to a healthy glucose level (55). Although T2DM has been shown to

increase CVD risk two-fold (2) no consistent linear association has been shown in

epidemiological research between glucose tolerance in non-diabetics and CVD risk (56).

1.2.2.6 Blood pressure and cardiometabolic disease risk

Hypertension is characterised by impaired endothelial function causing blood pressure,

either systolic (heart in contracting state) or diastolic (heart in relaxed state), to be elevated

to a level associated with increased CVD risk. Hypertension initiates arterial wall damage

as a result of sheer stress, making the endothelium susceptible to lesion formation.

Hypertension is diagnosed and clinical intervention indicated when systolic blood pressure

measures !140mmHg and/or diastolic blood pressure !90mmHg (57). However a

positive, independent and graded association between both systolic and diastolic blood

pressure with CVD has been observed in epidemiological studies (58). Significant

elevated CVD risk and CVD mortality has been observed at ‘pre-hypertensive’ blood

pressures (systolic 120-139mmHg or diastolic 80-89mmHg) (59).

1.2.3 Determinants of cardiometabolic disease risk markers

T2DM and CVD typically have long latency periods and develop in response to

environmental, behavioural, genetic and epigenetic factors making their aetiology complex

and multifactorial (60). Therefore risk of CVD and T2DM is commonly estimated based on

cardiometabolic risk markers and MetS criteria. Variations in these cardiometabolic health

! ! ! ! ")!

biomarkers are the result of interactions between modifiable and non-modifiable risk

factors (27), Figure 1.3. However it is acknowledged that the main contributors to the

development of CVD and T2DM are modifiable (6,61), with diet considered a major

determinant of cardiometabolic disease risk (62). Interventions to manage risk markers

can reduce and/or delay the development of T2DM and/or CVD disease endpoints (63).

Therefore understanding what determines dietary behaviours is an important aspect in

reducing the public health burden of T2DM and CVD.

Figure 1.3 Factors, and markers associated with cardiometabolic disease development

1.3 Diet and cardiometabolic disease risk

1.3.1 Energy and macronutrient intake and cardiometabolic disease risk

1.3.1.1 Energy intake, adiposity and cardiometabolic risk

In humans energy metabolism dictated by the first law of thermodynamics:

Energy stored = Energy intake - Energy expenditure Lipid Carbohydrate Basal Metabolic rate

Fat Physical activity Protein Adaptive thermogenesis

Alcohol

!

Modifiable risk factors: Diet, physical activity,

smoking, sleep, psychological stress!

!

Non-modifiable risk factors:

Age, gender, ethnicity, genetic, socio-economic

status!

!

Markers of risk: Obesity, dyslipidaemia,

insulin resistance, hypertension, inflammation!

!

Clinical disease end point: Cardiovascular disease Type 2 diabetes mellitus!

!

!

! ! ! ! #+!

If energy intake exceeds energy expenditure the human body will store the excess as lipid

in adipose tissue. An established component for the treatment or prevention of T2DM and

CVD is the maintenance of a healthy weight body weight and waist circumference - as

these measures correlate with adiposity (64). Even without achieving these targets, it has

been shown that a 5% reduction in body weight can improve metabolic function in obese

individuals (65). Therefore the cornerstone of weight management is the reduction of

dietary energy intake to be equal to, or less than, the energy requirement to maintain an

individual’s current weight. Dietary energy (kcal) is derived from four macronutrients: fat,

carbohydrate, protein and alcohol.

1.3.1.2 Fat intake and cardiometabolic risk

Dietary fat (mainly triacylglycerol) is composed of a heterogeneous group of fatty acids,

varying in their carbon chain length and saturation (the number of double bonds along the

carbon chain), attached to a glycerol molecule. The structural differences of fatty acids

confer different physiological properties. Generally saturated fats, particularly those with a

chain length of 14-20 carbons are positively associated with both LDL and HDL

concentrations. A Cochrane review of RCTs (15 studies) published in 2015, concluded

that reducing saturated fat intake had a positive impact on CVD risk reduction (66).

Evidence suggests that consuming polyunsaturated fatty acids (PUFA) in place of

saturated fat can be protective against CVD (67), however the evidence of beneficial

effects of monounsaturated fatty acids (MUFA) are inconclusive (66).

1.3.1.2 Carbohydrates and cardiometabolic risk

Two main sub-groups of glycaemic carbohydrate contribute to energy intake: sugars and

starches. In common with dietary lipids the various molecular structures of carbohydrates

exert different physiological effects. Research suggests that certain types of

carbohydrates, for example those with a low glycaemic index (GI), may reduce central

adiposity (68) and T2DM risk (69). GI refers to the extent a food raises blood sugar levels

after eating - the lower the GI, the slower digestion and/or absorption. Dietary sugars are

single monomers e.g. glucose (monosaccharaides) or formed from two units e.g. sucrose

(disaccharides). These ‘simple’ sugars are more rapidly absorbed and metabolised by the

body (therefore generally classified as high GI). Starches are formed from a large number

of monosaccharide units arranged in straight or branched chains with their structure

dictating how quickly they are digested and absorbed, hence their GI can vary. Although

the concept of GI and carbohydrate quality has been applied in epidemiological studies it

! ! ! ! #*!

currently has limited application in public health nutrition. This lack of translation is based

on numerous factors, firstly the financial cost of manufacturers to measure the GI of a

product for labelling (GI can only be measured using healthy human volunteers), secondly

the processing and the combination of nutrients consumed together (e.g. at one meal) can

influence the GI of that eating occasion (70). Lastly, not all foods that are classified as

having a low GI are in line with current healthy eating guidelines – for example ice cream

and croissants are classified as low GI due to their fat content (slowing carbohydrate

absorption).

Sugars consumed in the diet are obtained mainly from plant sources and a small amount

from dairy (lactose). Added sugar refers to sugar that is not intrinsic to a food in its natural

state, i.e. it is added through food processing and at home preparation. The 2014 SACN

Carbohydrates and Health report concluded that there was insufficient evidence to suggest

sugar intake was associated with CVD risk (71). The report suggested that sugar intake

was positively associated with overall energy intake and that sugar consumed from sugar-

sweetened beverages (SSBs) was positively associated with T2DM and BMI (71). High

intakes of fructose (from high fructose corn syrup which is used as a sweetener in SSBs)

are hypothesised to increase MetS (72) due to the different metabolic pathway used for

metabolism compared to glucose - with fructose being predominantly metabolised in the

liver and favouring lipogenesis (73,74). Moreover, SSBs are classified as a low nutrient

dense food, i.e. they contain a low number/quantity of micronutrients compared to the

amount calories they provide (75). Observational evidence from the Rotterdam study (n =

4,969) found that carbohydrate foods such as bread, vegetables, fruit and potatoes

contributed most to the nutrient density of the diet (76). A higher nutrient dense diet (high

in fruit and vegetables) compared to a low nutrient dense diet (low in fruit and vegetables,

high in added sugar and confectionary) was associated with lower all-cause mortality (76).

This suggests that the concentration of other nutrients contained within carbohydrate foods

may be important in determining associated health benefits.

1.3.1.3 Protein and cardiometabolic risk

Proteins are composed of various amino acids and are obtained from both animal (meat,

dairy, eggs) and plant (nuts, legumes, cereals) sources. Research suggests that the food

source of protein may infer different physiological effects on cardiometabolic health.

Cross-sectional analyses from the INTERMAP cohort observed vegetable protein to be

negatively related to blood pressure (-1.11mmHg, p <0.001 systolic and -0.71mmHg,

! ! ! ! #"!

diastolic per 2.8% kcal increase of animal protein) (77). Results from the OTISCAV-LUX

study found that high meat derived protein intakes to be positively associated with central

adiposity, with no attenuation of significance after adjustment for physical activity, age and

other dietary variables (78). These observations may relate to correlated intakes of micro

and macronutrients that are consumed with plant or animal protein rather than the protein

per se. For example, vegetables are also a source of nitrates which have been shown in

acute feeding studies to reduce blood pressure (79).

1.3.1.4 Alcohol and cardiometabolic risk

Alcohol (ethanol) is a nonessential nutrient that can contribute significantly to energy

intake. The evidence regarding the benefit of moderate intake of alcohol to

cardiometabolic health is contradictory. Evidence suggests that as part of a

Mediterranean diet (MedD) moderate alcohol intake may prove beneficial (80). However

excessive alcohol intake is associated with CVD intermediaries including hypertension and

atrial fibrillation (81).

1.3.2 Dietary fibre, sodium and cardiometabolic disease risk

1.3.2.1 Dietary fibre (non-starch polysaccharides)

Non-starch polysaccharides (NSP) are a class of carbohydrate that are mainly derived

from the cell walls of plants. NSP comprises of soluble and insoluble sub-types. The

protective benefits of fibre against cardiometabolic disease relate to numerous

physiological properties. Short chain fatty acids (SCFA) are produced by colon microbiota

from the fermentable fractions of dietary fibre. These SCFA have been shown to reduce

appetite and potentially modulate body weight (82,83). Soluble fibre, particularly beta-

glucans (found in oats and rye), bind with cholesterol in the colon and reduce absorption

therefore lowering total and LDL at doses of 3g per day (84). Additionally, high fibre foods

are absorbed more slowly from the intestine and are generally classified as low GI due to

their postprandial glucose response.

1.3.2.2 Sodium

Sodium in the diet is mainly derived from table salt (as sodium chloride) and salt added in

food manufacture. High levels of dietary sodium intakes have been consistently

associated with hypertension (85,86). A large systematic review of 37 RCTs found that a

decrease in sodium intake reduced systolic blood pressure by 3.39mmHg (95%CI 2.46,

4.31) in adults (87). Renal control of sodium excretion and water balance is thought to

! ! ! ! ##!

play a role in the aetiology of hypertension (88). However the exact mechanisms that

associate high sodium intake to increased blood pressure are yet to be fully characterised

(88).

1.3.3 Food groups, dietary patterns and cardiometabolic disease risk

The evidence that associates individual macronutrients with health outcomes in

epidemiological studies can be contradictory when a reductionist approach is applied. For

example separating out the association between added sugar and fat on health can be

problematic due to the collinearity of these two nutrients in the diet. In the UK diet ~50%

of added sugar is consumed from confectionary and baked goods (which are also

commonly high in SFA) and 11% from alcoholic drinks (89). Moreover, the food matrix is

highly complex containing various micronutrients and non-nutrient bioactive compounds in

addition to macronutrients. Therefore as people select foods rather than nutrients to eat,

there has been a move towards food-based recommendations for healthy eating

guidelines (90).

1.3.3.1 Food groups and cardiometabolic disease risk

An extensive appraisal of systematic reviews and meta-analysis conducted by Fardet and

Boirie reported the following food groups to be negatively associated with diet-related

cardiometabolic disease risk: fruits, vegetables, whole grains, fish, legumes, dairy, nuts

and seeds! (91). Conversely analyses showed red/processed meat and sweetened

beverages were consistently associated with increased risk for all diet-related chronic

diseases (91). However, it can be difficult to determine if an individual food group is

beneficial or detrimental to health, due to multi-collinearity between the foods eaten.

Research from the Malmo Diet and Cancer Cohort found that sugar sweetened beverage

intake was significantly associated with consuming a general poor quality diet (low fruit,

vegetable, cereal and fish consumption), highlighting the difficulty in separating the effect

of a single food group on health outcomes in observational studies (92).

1.3.3.2 Dietary patterns and cardiometabolic disease risk

The combinations of foods consumed may in part be responsible for health effects through

nutrient-nutrient interactions. A longitudinal study conducted in American men observed

that a diet that combined a low saturated fat intake with high vegetable and fruit intake to

provide greater protection against death from coronary heart disease than the individual

components (93). ‘Food synergy’ is the theory that the interaction between chemical

! ! ! ! #$!

compounds within the diet may have additive effects on physiology (94). For example in

vivo research has observed green leafy vegetables and olive oil when consumed together

may have a synergistic effect by forming nitro fatty acids that inhibit the enzyme soluble

epoxide hydrolase which lowers arterial blood pressure (95), the authors suggested this as

one of the potential mechanisms by which the MedD reduces cardiovascular risk.

Although the cardiometabolic benefits of certain dietary patterns (outlined below) have

been tested in RCTs evidence to recommend specific foods to consume at the same

eating occasion is currently limited (96).

1.3.3.3 Dietary patterns associated with reduced cardiometabolic risk

Two of the main food-based dietary patterns that have shown to be beneficial in the

management of cardiometabolic disease risk are the MedD and Dietary Approaches to

Stop Hypertension diet (DASH) (97). The MedD is based on a dietary pattern comparable

to traditional South Mediterranean dietary intakes identified in the 1970’s (a diet rich in

whole grains, olive oil, fish, fruit and vegetables, low in diary and red meat with moderate

alcohol intake) observed to be inversely linked with CVD risk (80). Further epidemiological

studies and clinical trials have shown MedD to be inversely associated with MetS (98).

The DASH diet was developed from a landmark American clinical trial testing a dietary

intervention to reduce blood pressure (99). DASH encourages intakes of positive food

groups: whole grains, low fat dairy, total fruit, vegetables, nuts legumes and seeds, and

limits intakes of SSBs, processed red meat and sodium (99). Meta-analyses of 20 RCTs

of DASH diet and metabolic risk markers found that in addition to a significant lowering of

blood pressure, DASH showed positive effects in the lowering of LDL cholesterol (100).

Additionally the Insulin Resistance Atherosclerosis Study (n = 862) found an inverse

association between DASH diet score and T2DM (101).

It has been suggested that, although both scores estimate a similar T2DM disease risk

reduction, DASH captures the dietary characteristics related to T2DM to a greater extent

than other diet quality scores (102). This relates to the evidence of high SSB intake and

adverse effects on metabolic health (103) and increasing evidence to suggest dairy intake

may be protective against CVD (104) and T2DM (105). Moreover, dietary salt intake has

been consistently and positively associated with blood pressure, a key risk factor for CVD

(85). Additionally the inclusion of alcohol, even at moderate levels, in the MedD score (up

to three units per day) contradicts Public Health England guidance of a safe limit for

alcohol consumption (106).

! ! ! ! #%!

1.3.3.4 Physiological benefits of dietary patterns

Various scores and indices exist to measure dietary intake against characteristics of the

DASH and MedD diets. However, the common variables included across the different

scores are: fruits, vegetables, legumes and whole grains (positive components) and red

meat intake (negative component). A predominantly plant based diet, as characterised by

either MedD and DASH, has been shown to be associated with a reduction in long-term

weight gain (107). The combined results from the Nurses’ Health Study and Health

Professionals Study longitudinal study (n = 331,505) found that a predominantly plant

based diet (also low in added and refined carbohydrates) to reduce the risk of developing

T2DM (HR 0.66, 95%CI 0.61, 0.72) post adjustment for BMI (108). The mechanisms that

explain the association between predominantly plant based diets and positive health

outcomes are hypothesised to be due to the high nutrient, antioxidant and dietary fibre

content (107). Bioactive compounds such as flavonoids are found in fruits, vegetables and

whole grains. Higher intakes of one of the classes of these bioactive compounds

(flavonoids) have been associated with long-term weight maintenance in longitudinal

studies (109).

1.3.4 Eating patterns and cardiometabolic health

1.3.4.1 Frequency of energy intake

Observational studies investigating the effect that the number of daily eating occasions

has on metabolic health have shown inconsistent results. Daily eating occasions have

been inversely associated with total blood cholesterol concentrations (0.15mmol/L

difference between 1-2 and >6 eating occasions/day) in the EPIC-Norfolk study (110). A

positive association was observed between eating occasions and hypertension in the

Korean Nation Health Survey (111). However, cross-sectional analyses of the NHANES

cohort (n = 18,696) observed a higher eating frequency to be associated with central

obesity; eating "3 vs. !5 times per day was associated with an OR for central obesity of

1.42 (95%CI 1.15, 1.75) and 1.29 (95%CI 1.15, 1.75) in men and women respectively after

adjusting for demographic factors and energy intake (112). A study in British adults

observed that snacking frequency, an important aspect of eating frequency, was

associated with lower diet quality, potentially mediating the influence of eating frequency

on measures of adiposity (113). Conversely, a cross-sectional study using data from the

INTERMAP cohort found that higher number of daily eating occasions were associated

with a healthier diet and lower BMI compared to those reporting a low number of eating

! ! ! ! #&!

occasions (114). Systematic reviews report limited evidence from clinical trials and animal

studies to support the influence of eating frequency on metabolic and anthropometric

measures (115,116). The conflicting observations relating eating occasions to health

outcomes may be due to the definition of an ‘eating occasion’ applied, as no standard

definition exists (117).

1.3.4.2 Regularity of energy intake

Regularity of dietary intake refers to the consumption of meals of similar energy

composition at a similar time each day. In a Swedish cohort (n = 3,607) regular eating

(defined as not skipping meals) was inversely associated with components of MetS and

positively associated with higher intake of fish, fruit and vegetables (118). The inverse

association remained after adjustment for diet quality (118). A novel scoring system

developed to measure daily variations in total and meal specific energy intakes, observed

that high daily variation in total energy intake was significantly associated with an

increased waist circumference (OR 1.34 95%CI 1.04, 1.72) in the 1946 British Birth Cohort

(119). Moreover, this study also observed variations in daily energy intake at specific meal

times, particularly at breakfast and between meals, to be associated with elevated risk of

MetS independent of energy intake (119). A prospective study of the Nurses’ Health

cohort found irregular breakfast consumers ("6 times per week) compared to daily

consumers to be at increased risk of T2DM (RR 1.28, 95%CI 1.14, 1.44) (120). This study

also found that daily breakfast eaters reported a significantly healthier diet (measured by

the Alternative Healthy Eating Index) - potentially a contributing factor to reduced T2DM

risk, however the authors adjusted for this in their analyses (120). A 14-day randomised

crossover trial in 10 healthy obese women reported improved fasting lipid profiles and

postprandial insulin response following a regular meal protocol versus an irregular meal

pattern (6 vs. 3-9 eating occasions/day) (121).

1.3.4.3 Timing of energy intake

Studies have found that, independent of calorie intake and physical activity, the timing of

food intake can influence weight gain (122). In particular, consuming a higher percentage

of daily energy intake during the evening has been associated with MetS and obesity

(123). Assessment of the 1946 British Birth Cohort observed a higher prevalence of

hypertension in those eating a greater percentage of their energy intake in the evening

compared to the morning (124). This finding supports the positive association found

between breakfast frequency and reduced metabolic risk (125). However breakfast

! ! ! ! #'!

skipping has been shown to be associated with higher evening energy intake (123). The

benefit of eating breakfast may be mediated by the macronutrient content of ‘breakfast’

with morning intake of carbohydrate being found to be protective against abdominal

obesity (126) and T2DM (127). The Bath Breakfast Project observed that breakfast eaters

demonstrated higher physical activity thermogenesis (442 kcal/day 95%CI 34, 841), and

although during six week RCT no changes in anthropometric measures were reported,

more stable afternoon and evening blood glucose concentrations were observed when

breakfast was consumed (128). Additionally a RCT demonstrated that the timing of food

intake can significantly influence post-prandial blood triglyceride and glucose

concentrations, with higher levels being recorded after identical snacks were consumed at

4a.m. compared to 4p.m. (129). These findings have been replicated in a similar

randomised crossover trial which provided healthy volunteers with identical meals at 8a.m.

and 8p.m. (130). This study also reported a significant increase in post-prandial resting

metabolic rate after the 8a.m. meal (36.7 95%CI 34.6, 38.8 kcal/kg fat free mass vs. 33.7

95%CI 31.4, 36.0 kcal/kg fat free mass, p <0.05) (130).

1.3.4.4 Eating patterns and circadian rhythms

The influence of eating patterns on metabolism is related to circadian rhythms. Humans

have evolved endogenous physiological rhythms to mirror a diurnal existence where

activities take place during daylight hours and sleep during darkness. These rhythms have

been conserved as advantageous as they anticipate regular changes in our environment

to maintain homeostasis. These endogenous molecular rhythms are entrained via

‘zeitgebers’ (cues). The main zeitgeber is daylight that entrains the ‘master clock’

(suprachiasmatic nucleus), which controls various neuroendocrine pathways, including

those involved in metabolism (131). Therefore eating out of ‘sync’ with our biological clock

has been hypothesised to contribute to adverse metabolic profiles (131).

1.4 Factors that influence food choice and dietary behaviour

1.4.1 Determinants of food choice

There are many factors that influence the dietary behaviours that affect health.

Interventions to modify dietary intakes have been traditionally person centred focusing on

behaviour change strategies. It has been suggested that the lack of efficacy of individual

nutritional interventions may be partly due to the paucity of interventions altering

environmental determinants of food choice (11). Figure 1.4 presents a framework of the

38

determinants of eating behaviour as proposed by Booth et al. The two outer layers in

Figure 1.4 determine the behavioural settings (micro-environments) that influence food

choice. For example employers determine working schedules (time of work and breaks),

while local government, food retailers and employers dictate the availability of foods

(product offer, opening times and workplace canteens/vending machines). The inner

layers (physiology, cultural and social) shape individual preferences for foods to interact

with the factors that facilitate food choice (enablers of choice). Booth et al. propose that

diet choices occur as enablers of food choice interact with the behavioural setting (11).

Figure 1.4 Framework for the determinants of eating behaviour

Figure from Booth et al.,Environmental and Societal Factors Affect Food Choice and Physical Activity:

Rationale, Influences, and Leverage Points 2001 Nutr Rev 59 S21-36. (11). © 2001 International Life

Sciences Institute, Oxford University Press.

1.4.2 Occupational factors and food choice

Workplaces are heterogeneous both within and between occupational groups. They vary

in the nature of work, the working hours, level of skill involved, environment, support

structure and remuneration. The Scottish Heart Health Study conducted a survey of

nutrient intakes across different socioeconomic groups of employees (132). The study

found that alcohol intake was higher in manual compared to non-manual male workers,

! ! ! ! #)!

and amongst men and women lower occupational groups were associated with a poorer

quality diet (132). Similarly the Whitehall II study conducted in London civil servants found

gender and employment grade to be associated with differences in dietary behaviours,

with men and those in a lower employment grade consuming a poorer diet (133). These

studies suggest that socio-economic factors may influence dietary choices in different

occupational classes. However, the determinants of food choice may be mediated by

factors related to occupation and/or occupational grade. Studies conducted in American

employees have found that a poorer diet quality was associated with ‘burnout’ (feelings of

negativity towards work) (134) and ‘high work stress’ (135). However the nutritional

choices associated with work stress may differ according to sex and levels of job strain

(136). A cross-sectional study in America found psychological job demands to be

positively associated with high fat food intakes in men, but not in women (136). While

qualitative research in a cohort of Australian paramedics found that food choices at work

were influenced by the organisational environment (e.g. time of work and meal break

structure), the physical working environment (e.g. station location and ambulance

environment), physiological factors (e.g. hunger) and psychosocial aspects (e.g. shift work

colleagues) (137).

1.5 Non-dietary lifestyle factors and cardiometabolic disease risk

1.5.1 Physical activity and inactivity

Physical activity comprises of aerobic (improves cardio-respiratory fitness) and anaerobic

(improves muscle tone via force exertion) activities. Physical activity has been shown to

be important for maintaining body weight (138). Regular aerobic exercise has been found

to reduce blood pressure and is also associated with increased HDL cholesterol (139).

Conversely being sedentary, measured through time sitting or TV viewing is considered

detrimental to cardiometabolic health. The Nurses’ Health Study found that independent

of physical exercise TV viewing and daily sitting were associated with increased risk of

T2DM and obesity (140). In fully adjusted analyses for every two hours increase in TV

viewing, there was an observed 23% increased risk of obesity (95%CI 17%, 30%), and

14% increased risk of T2DM (95%CI 5%, 23%) (140).

! ! ! ! $+!

1.5.2 Smoking

The causal relationship between smoking and CVD is well established and it is associated

with an up to fourfold increase in risk (141). Smoking is associated with lower HDL

cholesterol (142), and it is known to raises blood pressure acutely via the effect of nicotine.

The Framingham study observed that when combined with additional risk factors, such as

dyslipidaemia and glucose intolerance, smoking had a synergistic effect on future CVD risk

(143). Additionally, the 2014 report from the US Surgeon General stated that there was

sufficient causal evidence to link smoking to T2DM development, with current compared to

non-smokers having a 30-40% increased risk of T2DM (141).

1.5.3 Sleep

Short sleep duration and sleep disturbance have been shown to be independently

associated with obesity and hypertension (144,145). Prospective analyses conducted in

the Nurses’ Health Study cohort (n = 70,026) observed a U-shaped association between

hours of sleep and T2DM (146). This study observed that ‘short’ sleepers (5 or less hours

per night) and ‘long’ sleepers (9 hours or more) compared to ‘regular’ sleepers (8 hours)

had a RR of 1.57 (95%CI 1.28, 1.92) and RR 1.47 (95%CI 1.19, 1.80) respectively for

T2DM development (146). However, when the analyses were adjusted for BMI the

association was attenuated and only remained significant in long sleepers, suggesting that

increased body mass may be a mediating factor in the relationship between short sleep

duration and T2DM (146).

1.5.4 Psychological stress

Stress is used to describe the physiological response to a stressor (perceived or actual

threat). The steroid hormone cortisol is released in response psychological and physical

stress. The association between chronic (sustained) stress with CVD and T2DM is based

on the theory of ‘allostatic load’ (147). This theory centres on the physiological response to

environmental stress via the hypothalamic-pituitary-adrenal axis, and sustained stress

resulting in chronic over activity of adrenal corticoid secretion (147). A case-control study

conducted in the Whitehall II cohort found that those with MetS to have higher elevated

rates of cortisol secretion compared to matched controls (148). Behavioural responses to

increased physiological stress may include alcohol, smoking and changes in dietary

habits, which are all established risk factors of CVD. However, results from Whitehall II

suggest that in women the association may be independent of these potential mediators

(149).

! ! ! ! $*!

1.5.6 Diet and other lifestyle behaviours

It is important to note that the lifestyle factors that impact on cardiometabolic health do not

occur in isolation to dietary behaviours. It has been suggested that health behaviours

occur in ‘clusters’ (150). Previous studies have associated psychological stress to poorer

dietary choices (135,136). The Nurses’ Health Study observed that high rates of

sedentary behaviour (weekly TV viewing) to be associated with a diet high in energy,

saturated fats, snacks and sweets (140). Moreover, it has been suggested that

combinations of negative lifestyle behaviours have a synergistic relationship with

cardiometabolic disease risk (151,152).

1.6 Non-modifiable risk factors of cardiometabolic diseases

1.6.1 Age and sex

Ageing is a key determinant of cardiovascular health (153). Moreover, men tend to

develop CVD at younger age compared to women. The Framingham Study estimated that

by 60 years of age ~20% of men and ~6% of women have CVD (154). The ageing

process involves cellular degenerative changes that result in a functional decline in tissues

and organs (155). A large prospective study in Finland (n = 14,786) found that age related

increases in CVD incidence were explained by diabetes, blood pressure and body weight

in both men and women (156). The difference in risk between the sexes is thought to be

due to lifestyle and sex specific physiology. The Finnish cohort found that the difference in

CVD incidence between men and women was largely explained by differences in smoking

rate (with men more likely to smoke than women) and higher HDL/total cholesterol ratio

amongst women (156). Prior to the menopause women produce oestrogen which may

have cardio protective properties potentially explaining why a women’s CVD risk increases

post menopause as oestrogen production declines (157).

1.6.2 Ethnicity

Ethnicity is frequently used in epidemiology to establish factors and patterns relating to

disease risk. Ethnicity broadly refers to a population cohort with shared origins and

common traditions and culture (158). Observational studies have shown that specific

ethnic population groups are at disproportional risk of developing CVD and T2DM

compared to Caucasians. In the UK, when compared to Caucasians, South Asians are at

increased risk of CVD and T2DM and those of African Caribbean origin are at higher risk

of being hypertensive and/or developing T2DM (159). As these diseases are

! ! ! ! $"!

multifactorial, cultural lifestyle differences (diet, physical activity), and socioeconomic

factors are likely to contribute to health inequalities.

1.6.3 Genetics

Genome-wide association studies (GWAS) have enabled researchers to understand the

role of genetic background in CVD and T2DM risk variation. While twin studies have

suggested a strong genetic basis for T2DM (160), GWAS have shown that T2DM genetic

susceptibility is polygenetic. Although approximately 63 gene loci have been associated

with T2DM effect sizes are small and relate to the clustering of numerous single-nucleotide

polymorphisms (SNPs) (161). A systematic review of 23 studies concluded that adding

genetic markers of T2DM risk to established risk factor models (age, waist circumference,

BMI) did not improve predictive performance (162). Mutations in the genes that code for

key proteins involved in cholesterol metabolism (LDL receptor, Apo(B) and PcSK9

enzyme) are well characterised with homozygous carriers (i.e. two copies of the faulty

gene) resulting in atherogenic levels of cholesterol from a young age and high risk of

coronary disease before the age of 40. The most well characterised gene in obesity

research is the fat mass and obesity-associated (FTO) gene. Based on large cross-

sectional analyses (n = 38,759) it is estimated that ~16% of the population are

homozygous for the FTO risk allele; however versus the protective allele the risk allele is

only estimated to account for ~3kg increase in adult weight (163). Both diet and physical

activity have been shown to interact with FTO polymorphisms to modulate bodyweight and

waist circumference (164).

1.6.4 Socioeconomic status

The most common measures for socioeconomic status are occupation, education and/or

income. The Marmot report ‘Fair Society, Healthy Lives’ identified a social gradient in

health outcomes in the UK, with those in lower socioeconomic strata having worse

outcomes, particularly for CVD, than those in the higher strata (165). The reasons for this

gradient are multifactorial and include lifestyle choices, environmental (housing, access to

facilities), occupational factors, education, and health service provision (165).

1.5.6 Diet and non-modifiable risk factors

In common with modifiable lifestyle factors it is difficult to segregate non-modifiable risk

factors from diet and some health behaviours. Previous studies have shown that dietary

intakes differ between men and women, with women more likely make healthier choices

(higher fruit and vegetable intake with lower intake of high fat foods) (166). Findings from

! ! ! ! $#!

the UK National Diet and Nutrition Survey (NDNS) show that fruit and vegetable intakes

have a negative graded relationship with socioeconomic position (measured using income

or final educational level) (167). There is also evidence that genetic variation can

influence eating behaviours with gene variants being associated with taste preferences,

snacking and satiety (168).

1.7 Working hours and cardiometabolic disease risk

1.7.1 Definition and prevalence of long working hours

There is no definition of ‘standard’ weekly working hours however, working between 35

and 40-hours per week is commonly stated as the reference in occupational cohort studies

(15,169,170). The European Working Time Directive (2003/88/EC) aimed to limit working

hours to an average of 48 hours per week (calculated across a 17-week period). In the UK

emergency service employees are exempt from the directive, junior doctors have a 51-

hour restriction, and additional opt-out agreements are in existence (171). In 2014 the

proportion of UK employees working over 40-hours and 48-hours per week were 44% and

13% respectively (172).

1.7.2 Long hours and cardiometabolic disease risk

A large scale meta-analysis that controlled for sex, age and socioeconomic status (n =

603,838) and included 24 cohort studies across USA, Europe and Australia reported that

those working long hours (!55hrs per week) compared to standard hours (35-40hrs per

week) had an increased risk of stroke (RR 1.33; 95%CI 1.11, 1.61) and coronary heart

disease (RR 1.13; 95%CI 1.02, 1.26) (15). This study observed a dose-response

relationship across categories of weekly working hours and stroke (Ptrend <0.0001) (15).

A further meta-analysis investigating the association between number of working hours

and T2DM in prospective cohort studies (n = 222,120) found that the association was only

significant in employees classified as having a low socioeconomic status (RR 1.29; 95%CI

1.06, 1.57) (169). A study in female Australian employees (n = 9,276) followed-up over a

two year period reported working long (41-48hours per week) and very long hours

(!49hours per week) compared to part-time work was associated with greater weight gain,

with a dose-response relationship between length of weekly working hours and percentage

weight gain (173). There are limited studies published from UK cohorts; however, findings

from Whitehall II observed the number of overtime hours worked to be positively

! ! ! ! $$!

associated with increased risk of incident CHD independently of conventional risk factors

(174).

1.7.3 Shift work and cardiometabolic disease risk

Meta-analyses have reported shift workers to be at increased risk of developing T2DM

(OR 1.09 95%CI 1.05, 1.12) (13) and of experiencing a coronary event (RR 1.24 95%CI

1.10, 1.39) (14). The latter study estimated the population attributable risk of shift work for

coronary events to be 7.3% in Canada (14). However the Canadian population has over

twice the prevalence of shift work (~33%) compared to the UK (14). Research from the

Nurses’ Health Study in America observed that nurses working shifts for 15 years or more

were at elevated risk from CVD mortality compared to never shift workers (HR 1.23,

95%CI 1.09, 1.38) (175). Further cohort studies have shown shift work to be associated

with markers of cardiometabolic disease risk including obesity (176–180) and MetS (181–

184). There is limited research investigating UK shift working populations; however, the

Health Survey for England reported a higher prevalence of diabetes and a high waist

circumference among shift workers compared to non-shift workers (185). Approximately

15% (3.6 million) of the UK workforce undertake shift work (186).

1.7.4 Defining shift work

There is no definitive definition of shift work, however shift work is commonly defined as

‘working outside of the standard daytime hours’ which in the UK is taken as 7a.m. to 7p.m.

(187). This lack of a standard definition makes it difficult to draw consistent conclusions

about the association between shift work and cardiometabolic health. Shifts can occur at

various parts of the day (e.g. night, early morning and evening) and arranged in various

patterns (e.g. fixed, rotating or random), however these differences are rarely reported in

studies (188). The aspects of shift work which have been associated with cardiometabolic

health are summarised in Table 1.3. A further limitation of existing studies is the lack of

consideration of the association between long work hours and shift work (189). A small

cross sectional study (n = 480) in American police officers found long work hours

interacted with night shift work to increase the risk of obesity (190). Conversely a large

scale meta-analyses investigating the association between long-working hours and T2DM

risk reported no change in the association between these two variables in sensitivity

analyses that excluded shift workers (169).

! ! ! ! $%!

Table 1.3 Aspects of shift work previously found to be associated with cardiometabolic health outcomes Shift type

Typical definition

Cardiometabolic outcome

Cohort Associated risk for shift workers

Study reference

Night work

Typically includes working between midnight and 3.a.m.

MetS

T2DM

Various. Meta-analysis (up to 2009)

Denmark. Nurses (n =19,873)

Pooled OR1.57 (95%CI 1.24, 1.98)

HR 1.58 (95%CI 1.25, 1.99)

(191)

(188)

Rotating Various shifts usually includes night work and follows a fixed pattern that repeats

Stroke America Nurses’ Health Study (n = 8,008)

4% increase risk per 5yrs SW. Adjusted HR 1.04 (95%CI 1.01,1.07)

(192)

T2DM America Nurses’ Health Study. (n = 69,269)

Fully adjusted HR 1.05 (95%CI 1.04,1.06)

(193)

MetS Belgium BELTRESS Men (n = 1,529)

Taiwan. Factory workers. Female. (n = 387)

Argentina. Factory workers. Men. (n = 1,351)

OR 1.77 (95%CI 1.34, 2.32).

OR: 3.5 (95%CI 1.3, 9.0)

OR 1.51 (95%CI 1.01, 2.25), independent of age and physical activity

(181)

(194)

(182)

Insulin resistance

France. Chemical workers. Men (n = 192)

SW had sig lower insulin sensitivity

(195)

Raised body mass index (>25kg/m2)

Italy. Rail workers. Men (n = 339)

OR 1.93 (95%CI 1.01-3.71)

(180)

Retired shift workers

T2DM

Hypertension

China. Men and women. (n = 26,436)

Previous SW (>10years): T2DM OR: 1.10 (95%CI 1.03,1.17). Hypertension OR: 1.05 (95%CI 1.01,1.09)

(196)

Abbreviations: BELTRESS Belgium job stress project; CI confidence intervals; HR hazard ratio; MetS

metabolic syndrome; OR odds ratio; SW shift work; T2DM type two diabetes;

! ! ! ! $&!

1.7.5 Potential pathways between work hours and cardiometabolic health

1.7.5 1 Working hours and circadian disruption

The clash between peripheral and central rhythmicity that occurs when food intake is out

of sync with the master clock is a possible cause for increased metabolic risk in shift

workers (197). Working shifts requires employees to alter their eating, physical activity

and sleep behaviours to accommodate their work schedule. It has been proposed that

there are two interlinked aspects to the increase in metabolic risk observed in shift

workers: i) metabolic disturbances caused by circadian disruption and ii) the impact of shift

work on diet and other health behaviours (198). A high number of weekly working hours

may be related to extended shift length and/or more frequent shifts. ‘Social jet lag’ theory

hypothesises that electric light, occupational and social demands artificially lengthens our

biological day and is a potential contributing factor to the global epidemic of obesity

through circadian disruption (199).

1.7.5.2 Working hours and non-dietary risk factors

Physical activity

Establishing the relationship between physical activity and working hours is problematic

due to the different occupational groups that have been included in studies (e.g. office

based employees vs. front line nursing staff). The analyses of the SUN cohort found that

those working 50 hours or more per week compared to those working less that 24 hours

per week reported higher physical activity (22.6 SD 25.1 vs. 18.9 SD 20.8 metabolic

equivalent (MET)-hrs per week, p = 0.006), and less time watching TV (1.4 SD 1.2 vs. 1.8

SD 1.3 hrs per day, p <0.001) (200). Conversely, a study in radiographers found a

negative correlation between physical activity and length of working hours (201).

Smoking

A cohort study in Holland (n = 12,140) found that found that employees working shifts

compared to day work, independent of educational level, were more likely to take up

smoking (202). The Health Survey for England 2013 also found smoking prevalence to be

higher in shift workers compared to non-shift workers with 28% of men and 26% of women

working shifts reporting smoking compared to 23% and 15% of day working men and

women respectively (185).

! ! ! ! $'!

Sleep

Both shift work and long working hours are associated with adverse sleep patterns

(203,204). At the current time evidence is limited as to the contribution that sleep makes

to chronic disease risk in populations that work shifts and/or long working hours. A cross-

sectional study conducted in Canadian female hospital workers (n = 171) observed

rotating shift patterns, compared to day-only working, to be positively associated with self-

reported poor sleep and increased risk of MetS (OR 2.29, 95%CI 1.12, 4.70) (203).

Mediation analyses in this study suggested that poor sleep did not mediate the relationship

between rotating shift work and MetS (203).

Job strain

One of the largest studies investigating work stress is the Whitehall II study of UK civil

servants (n = 10,308) (205). Over a 14-years follow-up period a dose-response

relationship between work stressors (demand-control model) and MetS was observed OR

2.25 (95%CI 1.31, 3.85) after adjustment for employment grade and age (205). The

Nurses’ Health Study (n = 52,656), 4-year follow-up work-stress (high-demand with low-

control) associated with a greater increase in body weight than those reporting no work

stress (206) .

Socio-economic position

In socioeconomic status stratified meta-analysis (n = 1,502) ‘low’ socioeconomic class

strata had a RR 1.26 (95%CI 1.02, 1.57) of developing T2DM working long (!55hrs)

versus standard working hours after adjustment for BMI, smoking, alcohol, sex, age and

physical activity (PAL) (169). No associations were seen in analyses for intermediate or

high socioeconomic status, the authors hypothesised that the association observed in low

socioeconomic occupations may reflect financial constraints (i.e. having to work long

hours), reduced work recovery and leisure time (169). The Whitehall II study found that

the negative association between socio-economic status and heart disease observed to be

partly attributable to differences in job strain (207).

! ! ! ! $(!

1.8 Working patterns and dietary intake: A literature review

Table 1.4 summarises the results of an extensive literature search of observational studies

that have investigated diet across different working hour arrangements. Appendix A1.1

details the literature search criteria.

1.8.1 Length of working hours and diet

As shown in Table 1.4, seven studies were identified that investigated length of working

hours and diet. The majority of studies were conducted in men across a wide range of

occupational groups. The method of working hour measurement also varied across

studies, with one study measuring length of working day (208) rather than weekly working

hours. Despite the heterogeneity across population samples, generally longer working

hours were associated with differences in the distribution and/or regularity of food intake

(208–210) and/or more takeaway or fast food consumption (210–212). The literature

search did not find any published UK based studies that had investigated diet and length

of working hours. Results reported from Whitehall II (UK civil servants) investigating

overtime and coronary heart disease did report fruit and vegetable intake as a covariate

(174). This study found no difference in intake of fruit and vegetables across overtime

hours, however they did observe a higher alcohol intake with increased length of overtime

(174). As the literature outcome search criteria (Appendix A1.1) was focused on dietary

intake no studies were retrieved that reported solely alcohol intake. However, a large-

scale pooled meta-analysis (27 studies) investigating the association between length of

weekly working hours and alcohol intake reported that compared to standard working

hours (35-40hrs per week) working 55 hours or more per week was associated with OR

1.12 (95%CI 1.12, 1.25) for risky alcohol use (>14 alcohol drinks per week for women and

>21 for men) (170). The majority of studies identified in the literature search used

retrospective dietary data collection methods, with only one study using a prospective

dietary measurement tool (210).

1.8.2 Shift work and diet

Table 1.4 details 19 observational studies that have investigated shift work and diet. In the

studies reviewed a high level of heterogeneity was observed, with differences in industry,

shift work definition and dietary assessment methods used. A third of studies used

prospective dietary collection methods (213–219); however, the studies using these

methods had a small sample size (n <150). The largest study used data collected from

! ! ! ! $)!

the EPIC Netherlands cohort (220). However <10% of the sample were shift workers and

retrospective shift work exposure measurement methods were used (220).

The variation in shift work exposure and dietary measurements methods used in the

studies identified in this literature review make comparison of results problematic.

However, common themes in the literature are differences in meal and snacking frequency

(213,214,216,221–223). Disagreements with regard to whether these increase or

decrease as a result of shift work are likely to be due to the variable definitions of ‘meal’ or

‘snack’ used. Although four studies reported no significant differences between groups of

different shift workers in terms of nutrient intakes (180,218,224,225), observations in the

majority of studies suggested a poorer diet quality in shift workers, with higher sugar (217),

saturated fat (226,227), SSBs (217,228), alcohol (217,228,229), and lower vegetable

(227,228,230) intakes observed. There is limited research conducted in the UK

investigating the diet of shift workers. The small study by Waterhouse et al. used a survey

questionnaire to measure dietary variables and cases and controls were from different

employment sectors (221). The Health Survey for England, based on questionnaire data

of a randomly selected sample, reported on the dietary intake of shift workers in 2013. It

found that fruit and vegetable intake was lower in shift workers compared to non-shift

workers (3.3 vs. 3.6 portions for men, and 3.6 vs. 3.8 portions for women) (185).

1.8.3 Food choice and working patterns

The reason for the differences in food choices made by employees working different hours

is likely to be multifactorial, however published research in this area is limited. A

qualitative study in South African nurses found that the availability of food in the work

place in addition to unsocial working hours influenced health behaviours including over-

eating and unhealthy food choices (231). Research into possible physiological

mechanisms have found ghrelin, the hormone that stimulates appetite, to be higher in

night shift workers when compared to morning shift workers (232). This study also found a

lower self-reported post prandial satiety rating in the night-workers (232). This suggests a

possible causal mechanism to explain the larger evening meals eaten in a cohort of health

workers when they switched from day to night work (233).

!!

!!

"#!

Table 1.4 Sum

mary of observational studies investigating w

orking hour arrangements and diet

Reference

Study design

Country

Occupation /

industry n

Gender /

age range W

ork hours exposure / com

parison groups

Dietary

assessment

method

Dietary

outcomes

reported

Results

Length working hours

Maruyam

a and M

orimoto et al;

1996 (208)

Cross-

sectional Japan

Foremen and

managers

6,827 M

en

Age not

stated

Length of daily w

ork hours

9 or less hrs/day vs. 10 hrs/day or m

ore

Questionnaire

relating to breakfast, snacking, salt intake, eating patterns

Eating patterns,

salt intake N

o difference in salt intake or breakfast habits.

Irregular meals m

ore likely reported in longer w

orking hours

Nakam

ura et al; 1998 (209)

Cross-

sectional Japan

Com

puter m

anufacturing 248

Men

21-56 yrs

Mean m

onthly over tim

e Q

uestionnaire E

ating patterns (occasions and tim

e), alcohol, high fat foods

Overtim

e hours significant positive correlation w

ith later dinner tim

e

Devine et al;

2009 (210) C

ross-sectional

US

A

Low incom

e w

orking parents

50 M

en (50%)

Mean age

37yrs

Length of hours and shift w

ork 24-hr recall (x3)

Survey of food

choices

Food choice Long hours and/or shift w

ork associated with

more take-aw

ay foods, m

issed meals and

ready meals

Escoto et al;

2010 (211) C

ohort U

SA

Transport w

orkers 1,086

Men (78%

)

Mean 47.7

(SD

10.2) yrs

Length of weekly

working hours:

<40, 40-49, >50

FFQ and survey

Food choice and food group intake

>50hrs/week

associated with higher

intake of salty snacks and use of vending m

achines E

scoto et al; 2012 (234)

Cross-

sectional U

SA

V

arious (P

roject EA

T III cohort)

2,287 M

en (42%)

25-30yrs

>40 vs. <40 hrs per w

eek S

urvey and semi

quantitative FFQ

Food group intake

Wom

en working

>40hrs less likely to consum

e 5 or more

portions fruit and vegetables.

!!

!!

"$!

Reference

Study design

Country

Occupation /

industry n

Gender /

age range W


parison groups

Dietary

assessment

method

Dietary

outcomes

reported

Results

Fan et al; 2015 (212)

Cross-

sectional U

SA

Inform

ation technology

823 M

en (68%)

Mean age

46yrs

Length of weekly

work hours

Survey

Fast food consum

ption Fast food consum

ption higher in m

en working

>50hrs /week

Pim

enta et al; 2015 (200)

Cohort

Spain

Various

university graduates (S

UN

cohort)

6,845 M

en (35%)

Mean 36yrs

Length of weekly

working hours:

<24

25-39

40-49

>50

136 item sem

i-quantitative FFQ

D

iet quality (M

edD score),

energy intake

Higher m

ean daily energy intake in em

ployees working

>50 hours per week

compared to 25-35

hours per week.

No significant

difference in MedD

score across

Shift w

ork

Reinberg et al;

1979 (213) C

ross-sectional

UK

O

il refinery w

orkers 7

Men

21-36yrs

NS

= 5

DW

= 2

8 week food

diaries E

ating frequency, m

acro nutrient and energy intake

NS

= modified pattern

of carbohydrate intake w

ith increased snacking

Lennernäsa, et al. 1994. (214)

Cohort

Sw

eden M

anufacturing

96 M

en

25-55yrs

DW

= 37. M

ixed SW

= 34 3-m

ixed SW

= 25 (36-40hrs/w

eek)

24-hr recall:

work and rest

days (mean 6

per participant)

Micro and m

acro nutrient intake

NW

= redistribution of food intake.

No difference in

nutrient intakes across

!!

!!

"%!

Reference

Study design

Country

Occupation

/ industry n

Gender /

age range W


parison groups

Dietary

assessment

method

Dietary

outcomes

reported

Results

Sudo and

Ohtsuka; 2001

(215)

Cohort

Japan M

anufacturing

137 W

omen

DW

= 44

RS

W = 93

4-day dietary questionnaire (3 w

ork days / 1 rest day)

Energy, m

icro and m

acro-nutrients, and m

eal type

Late SW

= lower

energy and nutrient intakes due to low

er m

eal frequency and poor m

eal quality de A

ssis et al; 2003 (217)

Cohort

Brazil

Refuse

collectors 66

Men

SW

(morning)

SW

(afternoon)

NS

1 x 24hr recall + 2 x 24hr food records on consecutive w

ork days

Energy, m

acro nutrient and food group intake

NS

= higher carbohydrate, sugar sw

eetened beverage, and alcohol

de Assis et al;

2003 (216) C

ohort B

razil R

efuse collectors

66 M

en S

W (m

orning)

SW

(afternoon)

NS

As above

Num

ber of eating occasions

NS

= higher number

eating occasions

Waterhouse et

al. 2003 (221) C

ase control

UK

H

ospital vs. controls in academ

ia

93 W

omen

69%

Working

age

DW

= 50

NS

= 43

Survey

questionnaire over 7-days

Differences in

the type of meals

and influencing factors for m

eals betw

een day and night shift

NS

= more snacking,

less large hot meals

more cold food

Pasqua and

Moreno; 2004

(218)

Cross-

sectional B

razil Transport w

orkers 28

Men

Mean age

32.8 (SD

5.3) yrs

Fixed SW

3-day diet record (2 w

ork days / 1 rest day)

Energy intake /

temporal

distribution of energy intake across shifts an season

No significant

differences reported

!!

!!

"&!

Reference

Study design

Country

Occupation

/ industry n

Gender /

age range W


parison groups

Dietary

assessment

method

Dietary

outcomes

reported

Results

Reeves et al;

2004 (219) C

ohort U

K

Health care

and security w

orkers

36 M

en (50 S

D7 yrs)

Wom

en (28 S

D6 yrs)

DW

NS

6-day food diary E

nergy, macro

nutrient intake, eating frequency

DW

= fewer m

eals and snacks, shorter period of energy intake

No difference in

energy/nutrient intakes

Morikaw

a et al; 2008 (235)

Cross-

sectional Japan

Manual

factory w

orkers

2,254 M

en

20-59yrs

Fixed DW

SW

(+NS

)

SW

(no NS

)

Self com

pleted diet history questionnaire

Energy derived

from different

food groups, nutrient densities

SW

(+NS

)= higher energy intake (age had confounding im

pact on m

icro nutrient intake)

Esquriol et al;

2009 (222) C

ross-sectional

France P

etrochemi

cal plant 198

Men

Fixed DW

= 100

RS

W = 98

Self com

pleted diet history

Nutrient intakes

and meal pattern

RS

W = higher intake

saturated fat, and higher m

eal frequency afternoon and night.

Barbadoro et al;

2013 (180)

Cross-

sectional Italy

Rail w

orkers 339

Men

Working

age

DW

=229

SW

= 110

Rate your plate

eating pattern questionnaire

Diet quality, food

choices, alcohol N

o significant differences reported

Balieiro et al;

2014 (230) C

ross-sectional

Brazil

Bus drivers

150 M

en D

W = 69

NS

= 81

Sem

i quantitative FFQ

Food intake patterns

NS

= higher meat and

lower vegetable intake

!!

!!

"'!

Reference

Study design

Country

Occupation

/ industry n

Gender /

age range W


parison groups

Dietary

assessment

method

Dietary

outcomes

reported

Results

Tada et al; 2014 (228)

Cross-

sectional Japan

Hospital

nurses 2,758

Female

20-59 yrs

RS

W = 1,579

DW

= 1,179

Sem

i quantitative FFQ

E

nergy and food group intakes

No difference energy

intake. RS

W = H

igher consum

ption of sugar-sw

eetened beverages, alcohol, confectionary and low

er fruit and vegetables

Wirth et al; 2014

(236) C

ross-sectional

US

A

Police

officers (B

CO

PS

S

tudy)

464 M

en (75%)

Mean 42.4

(SD

8.5) yrs

DW

NS

SW

(evening)

FFQ

Dietary

inflamm

atory index

No significant


Hem

io et al; 2015 (227)

Cross-

sectional Finland

Airline

workers

1,478 M

en (55%)

DW

SW

(in flight)

SW

(not in flight)

16 item FFQ

E

nergy, macro

nutrients and food group intakes

Men: S

W (not in flight)

= lower fruit and veg

intake. Wom

en: SW

= higher saturated fat intake

Moreria et al;

2015 (229) C

ross-sectional

Portugal

Airline

ground staff 190

Not stated

DW

= 40

SW

= 150

Sem

i quantitative FFQ

D

ietary intake S

W = higher trans fat,

cholesterol and alcohol intake

Seibt et al; 2015

(225) C

ross-sectional

Germ

any H

otel staff 150

Men (40%

)

20-62yrs

Alternating S

W

=53

Fixed SW

= 97

FFQ

Diet S

election Index D

iet Regularity

Index Food group intake

No significant


!!

!!

""!

Abbreviations: B

CO

PS

: Buffalo C

ardio-metabolic O

ccupational Police S

tress EP

IC: E

uropean prospective Investigation into Cancer and N

utrition. FFQ: food

frequency questionnaire, DW

: day workers, S

W: shift w

orkers, MedD

Mediterranean diet, N

HA

NE

S; N

ational Health and N

utrition Exam

ination Survey, N

S night

work; R

SW

: rotating shift work

Reference

Study design

Country

Occupation

/ industry n

Gender /

age range W


parison groups

Dietary

assessment

method

Dietary

outcomes

reported

Results

Alm

ajwal 2016

(223) C

ross-sectional

Saudi A

rabia

Nurses (non

Saudi)

395 W

omen

Age range

not stated

DW

= 137

SW

=258

Dutch E

ating B

ehaviour Q

uestionnaire

Eating behaviour

NS

associated with

emotional and

restrained eating, higher snacks and low

er fruit and vegetable intake

Hulsegge et al.

2016 (220)

Cross-

sectional N

etherlands

EP

IC cohort

(not occupation specific)

7,856 M

en (30%)

20 -70yrs

DW

= 7,173

SW

=683 (of w

hich NS

= 454)

FFQ

Healthy D

iet Indicator

MedD

score

Energy intake,

macronutrients

and key food groups

NS

= higher energy intake com

pared to D

W. N

o difference in diet quality scores.

NS

compared to S

W

without nights had

higher meat and

energy intake

!

! %&!

1.9 Limitations and gaps in current research

Public health strategies to manage the increasing burden of cardiometabolic disease need

to take account of environmental exposures as well as individual behaviours. There is a

growing need to understand the comparative influence of factors within different

environments on dietary behaviour (237). Due to the amount of time spent in employment,

occupational influences are an important environmental exposure for nutritional research.

In particular the influence of working hours on diet is important to understand as working

hour arrangements have been associated with increased cardiometabolic disease risk.

There are limited dietary studies investigating the association between working hours (time

and length) with dietary quality and eating patterns. Studies that have been conducted

have produced inconsistent results due to methodological limitations and heterogeneity

between the different occupational cohorts studied. A major methodological limitation of

previous research is that prospective dietary data have rarely been used to investigate

dietary behaviour in large-scale occupational cohort studies. The benefit of prospective

measurement methods such as diet diaries, compared to food frequency questionnaires

(FFQs) is that they allow more detailed dietary intake to be captured, as they do not

measure against a predefined food list. Additionally food diaries do not rely on participant

memory / recall ability. Lastly, diet diaries can provide important information about eating

occasions, frequency of eating, regularity, and the combination of foods consumed.

1.10 Chapter summary

The obligation to provide a comprehensive 24-hour service by industry and key public

sector organisations allows limited ability to modify shift work exposure. Research has

associated working ‘atypical hours’ (long hours and/or shift working) with increased

cardiometabolic risk. Dietary behaviour is established as primary prevention and

management of cardiometabolic risk however, how these behaviours vary in relation to

working patterns is currently an under investigated area. The literature review presented

in this chapter is suggestive of a poorer diet in those with atypical working hours compared

to day workers; however methodological limitations have prevented comprehensive

analyses. Gaining knowledge about dietary behaviours and their associations with

specific aspects of work place organisation is of occupational health importance, as it will

potentially guide the development of tailored work place interventions. The overall aims of

!

! %'!

this thesis were to gain an understanding of dietary intakes of British police force

employees and investigate the relationship between working patterns and diet with

markers of cardiometabolic health.

1.11 Hypotheses, research objectives and thesis structure

1.11.1 Study Hypotheses

This thesis tests the following three hypotheses:

H1 British police force employees who work atypical hours (long and/or shift work) report a

poorer diet quality (as evidenced by a lower DASH score) compared to those who work

standard hours (35-40hrs per week / between 7a.m. and 6p.m.).

H2 British police force employees who work atypical hours (long and/or shift work)

compared to those who work standard hours (35-40hrs per week / between 7a.m. and

6p.m.) have a higher cardiometabolic risk profile (as evidenced by anthropometric and

biological risk markers)

H3 Diet quality (measured by DASH score) modifies the association between atypical

working hours and cardiometabolic risk in British police force employees

1.11.2 Research objectives Using data collected from the Airwave Health Monitoring Study the research objectives

were to:

i) Generate nutritional and food intake data from 7-day food records by:

a. Developing a standard operating and quality control protocol for the coding of 7-

day food diaries

b. Recruiting and training a team of diet coders to enter 7-day food diaries into

Dietplan6 nutritional software package

c. Applying composite dish disaggregation to the nutritional data set to enable diet

quality measurement.

ii) Investigate the prevalence of energy intake misreporting and to identify the factors

associated with energy intake misreporting in a police employee cohort

iii) Describe the overall dietary profile of police force employees across different sections

of the force (sex, region, rank and working hours).

!

! %(!

iv) Investigate the association of diet quality (measured by DASH score) and markers of

cardiometabolic risk.

v) Measure the association between number of weekly working hours and markers of

cardiometabolic risk.

vi) Assess whether diet quality modifies markers of cardiometabolic risk in employees with

different working hour schedules.

vii) Develop and pilot a revised food diary and shift work questionnaire to facilitate the

mapping of diet and eating patterns to working hours in future studies.

1.11.3 Thesis structure Baseline data collected from the Airwave Health Monitoring Study - a longitudinal study of

British police force employees, is used for all the analyses presented in this thesis. Details

of the data collection methodology are presented in Chapter 2. In line with Strengthening

the Reporting of Observational Studies in Epidemiology-Nutritional Epidemiology

(STROBE-nut) recommendations (238) Chapter 3 describes in detail the standard dietary

coding protocol that was developed as part of this PhD. Chapter 4 investigates energy

intake misreporting - an acknowledged limitation and potential source of bias in the

measurement of dietary intake. The findings from Chapter 4 are then applied to the

subsequent cross-sectional studies to determine the robustness of findings through

sensitivity analyses by excluding those classified as misreporting energy intake. Chapter

5 describes the dietary intakes across different sections of the police work force and

investigates the association between dietary intakes with working hours. Chapter 6

investigates the utility of using the DASH diet score to measure diet quality in relation to

cardiometabolic disease risk in the Airwave Health Monitoring Study cohort. Chapter 7

draws on the findings of Chapter 5 and 6 to investigate the association between weekly

working hours and cardiometabolic risk, and examines if diet quality modifies this risk

within British police force employees. Chapter 8 describes the development of two data

collection tools that aim to address key methodological limitations in the investigation of

diet and working hours i) an amended food diary that aimed to ‘map’ diet to specific

working hours and ii) a shift work questionnaire. Finally, Chapter 9 synthesizes the main

findings from each chapter and based on interpretation of these findings proposes

recommendations for further research.

!

!

! %)!

CHAPTER 2

2.0 CORE METHODS (NON DIETARY) This chapter provides a summary background to the Airwave Health Monitoring Study

and describes the methods applied to derive the non-dietary explanatory and outcome

variables required for the studies presented in this thesis. All data were managed and

analysed using SAS 9.3 (SAS Institute, Cary, North Carolina USA), unless otherwise

stated. Part of this PhD has been the development of a standard food record coding

protocol therefore dietary variable generation is described in detail in Chapter 3.

2.1 Introduction to the Airwave Health Monitoring Study 2.1.1 Study design The Airwave Health Monitoring Study is a longitudinal study of British police force

employees that commenced in 2004 (239). This study is the largest cohort of police

employees worldwide with 42,112 participants enrolled into the study at the end of

2012. The primary objective of the Airwave Health Monitoring Study is to investigate the

association between Terrestrial Trunked Radio usage and health risks (239).

2.1.2 Participant recruitment The Airwave Health Monitoring Study was open to all police forces in Great Britain.

Forces in the North East of England region declined to take part in the study, Figure

2.1. Recruitment procedures have been described in detail by Elliott et al. (239). The

Airwave Health Monitoring Study is conducted according to the guidelines laid down in

the Declaration of Helsinki. The National Health Service Multi-site Research Ethics

Committee (MREC/13/NW/0588) approved all procedures involving human subjects.

Written informed consent was obtained from all participants.

!

!

! &+!

Figure 2.1 Enrolment in the Airwave Health Monitoring Study per region by the end of 2012

ACPO: Association of Chief Police Officers

Reprinted from Environ Res 134C Elliott et al., The Airwave Health Monitoring Study of police

officers and staff in Great Britain: Rationale, design and methods, page 280 (239).

Copyright 2014 with permission from Elsevier.

2.1.3 Data collection Enrolled participants were invited to attend a regional health-screening clinic. The

duration of the health screen appointment was 40 to 50 minutes, during which time

trained research nurses using a standard protocol conducted all clinical examinations.

Occupational, socio-demographic and lifestyle data were collected from self-

administrated electronic questionnaires completed at the health screen appointment.

Table 2.1 provides a summary of the variables that are used in the studies presented in

this thesis. A complete list of the measurements collected is detailed by Elliott et al.

(239).

8,173

7,453

3,471

7,805

3,232

2,022

7,854

721 980

0

!

!

! &*!

Table 2.1 Summary of the data relevant to this thesis collected by the Airwave Health

Monitoring Study from 2007

2.1.4 Sample selection for nutritional studies From the sample available at the end of 2012 (n = 42,112) 15,404 had completed the

questionnaire, health screen and returned the 7-day food diary. This thesis uses data

collected from the Airwave Health Monitoring Study between 2007 and 2012 to conduct

four cross-sectional studies. The sample size for each study was determined by data

availability at January 2016. Figure 2.2 illustrates the sample selection for each of the

four cross-sectional studies. The sample of participants with diet data available (n =

5,849) was compared across key characteristics to the random (n = 10,380) and total

study sample (n = 42,112) (Appendix A2.1).

Measurement method Data collected

Questionnaire Socio-demographic, work environment, lifestyle, health

behaviours

Nurse interview Medical history, medications

Clinical measurements Cardiovascular system: Blood pressure

Anthropometry: height, body weight, body composition,

waist circumference

Blood measurements (non-

fasted state)

Biochemistry tests (lipid profile, HbA1c,, high sensitivity

CRP)

7-day estimated (unweighed)

food diary

Dietary data

!

!

! &"!

Figure 2.2 Schematic showing the sample selection for inclusion in the four cross-sectional

studies included in this thesis using data from the Airwave Health Monitoring Study

1. Refer to Appendix A2.2 for random selection schematic. 2. Refer to Chapter 3 for details. 3. Chronic

disease diagnosis includes: angina, cancer, chronic obstructive pulmonary disease, liver disease, thyroid

disease and previous stroke or heart disease. 4. Defined as working <35 hours per week.

2.2 Occupational variables All occupational variables were derived from the data collected from the self- completed

questionnaire during the health screen appointment.

!""#$%&'()($%*'+# ),-$".'(*/# 0%+."(*.# 1.%"'1# +)&..*# 233452362# 7!"# 4286629

:%+."(*.# 1.%"'1# +)&..*# ;('1# <,,=# =(%&>#233?52364# 7!"# 6@84349

!"#$%&'()*+'+,-+$),%.%/-)0%/)0%%$)$1"/()"**+**&+#-)2! 34567859

AB)"C=.D# #E,,=# =(%&># *,'#),-$".'.=7!"#"2F8?3G9

AB)"C=.D# H,'# (*#&%*=,-# +.".)'(,*7!"#"@832496

AB)"C=.D#E,,=# =(%&>#*,'# ),=.=# %'#I%*# 236FD# 7!"#"483??9#J&,++#),=(*/# .&&,&+# #7!"#"4@392K@33# L)%"M=%># &.),&=.=# 7! N#49

O%'%# %P%("%0".# <,&#%*%">+(+# 7!"N#@8@2?9!"#$%&'D#O(.'#QC%"('># %*=# )%&=(,-.'%0,"()# &(+L#(*#$,"().# <,&).# .-$",>..+

O%'%# %P%("%0".# <,&#%*%">+(+# 7!"N#@83RF9!"#$%&() S,&L(*/#1,C&+# %*=# )%&=(,-.'%0,"()# &(+L#(*#$,"().# <,&).#.-$",>..+

AB)"C=.D# #T%&'5'(-.# ;,&L.&+##7! N#4U694

AB)"C=.D# #V."<5&.$,&'# )1&,*()# =(+.%+.# =(%/*,+(+#7! N#R229R

AB)"C=.D# #W.$,&'.=# ;.(/1'# ",++#=(.'#7! N#42G9

O%'%# %P%("%0".# <,&#%*%">+(+# 7!"N#@8G4U9!"#$%&*D#O(.'%&># $&,<(".# ,<#$,"().# <,&).#.-$",>..+

O%'%# %P%("%0".# <,&#%*%">+(+# 7!"N#@84269!"#$%&+D#O(.'%&># .*.&/># (*'%L.# -(+&.$,&'(*/

!

!

! &#!

2.2.1 Rank Standard police rank was selected from a predetermined list (constable/sergeant,

inspector/chief inspector, and super intendant or higher). Participants not employed as

police officers were classified as ‘staff’ or ‘other’.

2.2.2 Working environment Job descriptions were collapsed from 28 to two categories of ‘working environment’

based on predominant working environment (mobile, or office based). The classification

was based on information provided by the Police Federation, Table 2.2. However, 20%

of respondents selected ‘other’, and were therefore not designated a working

environment classification.

Table 2.2 Classification of working environment based on job description

Predominant working environment

Job role

Office Detective sergeant/Inspector/ Chief inspector/ Super

intendant

Training officer/ sergeant

Office duties

Custody sergeant/Inspector

Station sergeant

Non-operational support officer/inspector/ chief

Inspector

Policing unit inspector

Control room inspector

Mobile Community support officer

Traffic warden

Beat officer

Mobile patrol officer

Dog handler

Covert officer

Firearms officer

Operations support officer/ Inspector

Traffic officer

Shift sergeant

Patrol inspector

!

!

! &$!

2.2.3 Working hours

2.2.3.1 Number of working hours

Participants were asked ‘How many hours per week do you usually work (excluding

overtime)?’ and to select a value from a drop down list ranging from 10, 11, 12!. //!70

hours per week. They were also asked ‘How many hours per week of overtime do you

usually work?’ and instructed to select a value from a drop down list (0, 1, 2!. //!25+

hours per week). The two values recorded from these questions were summed to

provide total usual weekly working hours. The distribution of the ordinal values of total

weekly working hours was not normally distributed, therefore the values were collapsed

into the following categories based on previous research (169,240): 35-40 hours per

week (standard hours, reference group); 41 – 48 hours per week, 49 – 54 hours per

week, and 55 hours or more per week. Part-time work was defined as reporting

working less than 35 hours per week (241).

As part of this PhD attempts were made by the author to obtain electronic payroll

records from British police forces retrospectively as an objective measure of working

hours. Of the forces with diet diary completion, six forces had data sharing agreements

in place and were contacted by the author. Fragmentation and changes to police force

payroll systems since the Airwave Health Monitoring study commenced resulted in only

one force (Cheshire) being able to provide limited payroll data. Clock on/off times were

provided from December 2006 until May 2008 (health screening appointments were

conducted during 2007/8). Electronic payroll data on 539 employees were provided

and 75 participants from the PhD study sample were matched to self-reported working

hour data. Mean weekly working hours were calculated per participant (excluding

annual leave, sick and maternity leave days). Court days, secondment and training

days were also excluded as no start or end times were reported. A Bland Altman plot

was constructed to determine the agreement between the two measures (242). The

mean difference between the two measurement methods was 5.6 hours, with less

agreement at higher number of weekly working hours. From the sample 96% of

participants were within the limits of agreement (2±SD), Appendix A2.3.

!

!

! &%!

2.2.3.2 Shift work

Classification of being a shift worker was based on a positive response to: Are you a

shift worker? Do you work outside the regular daytime hours of approximately 7 a.m.

and 6 p.m.? on the baseline questionnaire. Classification of a shift worker that worked

nights was based on recording a response of one or more to the question: How many

night shifts in a row do you usually work? All participants who reported working shift

work indicated that they worked nights (positive response to working one or more nights

in a row). Self-reported shift work exposure was collected from participants enrolled

into the study from the start of 2012. The shift work baseline questionnaire data were

available for 732 (7%) participants who completed the food diary up to end 2012.

Due to limited shift work data police radio usage records provided a novel proxy

measure for acute shift work exposure (i.e. within 30 days prior to the health screen

date). Radio records, which detail day, time and duration of each call made on

personal work radios, were available for 3,841 participants with dietary data. From this

sample, radio records for 30 days prior to the date of the health screen were used to

classify acute shift work exposure. To reduce the potential of misclassification

participants with calls recoded on three or more days were included (n = 2,323). This

was an arbitrary cut off to exclude those below the 10% quintile, to reflect irregular

police radio usage (e.g. those that only occasionally use the radio as part of their job).

Each radio call made or received was coded based on the time recorded as day (7 a.m.

to 6.p.m), shift (6 p.m. to 7.a.m), or night (calls recoded between midnight to 5 a.m.).

The classification of ‘shift’ or ‘day’ worker was based on the definition proposed by the

Airwave Health Monitoring Study group in 2007. The definition of night work was based

on UK Government definition (243). Participants were classified as being day workers

(no calls outside 7 a.m. to 6.p.m), shift workers without night work (calls outside of 7

a.m. to 6.p.m, but not during midnight and 5 a.m.) or shift work with night work (calls

recoded between midnight and 5. a.m.). In a sub-group that had both questionnaire

and radio data (n = 188) 96% had a shift work classification based on radio call data

that matched the classification derived from the questionnaire data. Level of agreement

was determined by calculating the kappa statistic. This indicated substantial agreement

between the classifications (k = 0.65, 95%CI 0.40 to 0.89) (244). Participants who

reported not working shifts on the questionnaire were classified as a day worker for

!

!

! &&!

acute shift work exposure in order to increase the number of day workers available for

analysis.

2.2.4 Job control and demand Job strain dimensions were measured using a four point Likert scale in response to six

job content statements derived from the Karosek Job Content Questionnaire (245).

Four statements related to job control and two related to job psychological demand.

The scale was scored 1 to 4 (strongly agree, agree, disagree, strongly disagree).

Cronbach’s alpha (") was used to measure internal consistency: control " = 0.65

(borderline acceptable) and demand " = 0.70 (acceptable). The median scores for total

job control and job demand were calculated, and participants above or below the

median values were defined as having high or low job control/demand. The quadrant

approach was used to define job-strain (246): high (low control, high demand), active

(high control, high demand), passive (low, control, low demand), and low strain (high

control, low demand).

2.2.5 Length of service and time in current job role Length of service was calculated to the nearest year by subtracting the year of the

health screen from the self-reported employment start date as police force employee.

Time in current job role was self-reported to the nearest whole year. This variable was

then grouped based on tertile cut-offs (2 years or less, 3 to 5 years, 6 years or more).

2.2.6 Employment region Employment force was recorded based on Association of Chief Police Officers (ACPO)

region, (Figure 2.1); due to the small frequency of the study sample across ACPO

regions these were collapsed into country: England, Scotland and Wales. Appendix

A.2.4 lists the forces from each country included in the study.

2.3 Measures of cardiometabolic health

2.3.1 Anthropometric measurements Trained nurses took anthropometric measurements following a standard protocol,

during which the participant was dressed in light clothing with no shoes and socks.

!

!

! &'!

Bodyweight was measured to the nearest 0.05kg using digital scales (Marsden digital

weighing scale). Standing height was measured to the nearest 0.1cm (Marsden H226

portable stadiometer). Waist and hip circumference were measured (while lying flat)

using a Wessex-finger/joint measure tape (Seca 201, Seca). Waist measurement was

taken at between the lower rib margin and the iliac crest in the mid-axillary line. For

anthropometric measures two measurements were taken and the mean was reported.

Total percentage body fat was measured via bioelectrical impedance analysis (Tanita

BC-418MA body composition analyser).

2.3.2 Biochemical and clinical measurements

2.3.2.1 Blood pressure

Sitting blood pressure was taken using the Omron HEM 705-CP digital blood pressure

monitor (Omron Health Care). Three measurements were taken 30 seconds apart and

the average recorded.

2.3.2.2 Blood samples

Blood samples (50ml) were drawn from participants in the non-fasted state during the

health screen by a trained phlebotomist using the vacutainer system. Time since eating

was measured by a single question asked at the time of the health screen (‘How many

hours passed since you ate or drank anything other than water?’). The response was

recorded as a numerical variable to the nearest hour. Samples requiring serum were

processed on site (standing for 40 minutes prior to centrifuge at 4,300rpm for 10

minutes). Samples were then aliquoted (2ml) and transported (stored in a thermoporter

at 0-4°C) for processing at a designated study laboratory. The assays reported in this

thesis were conducted using an IL 650 analyser (Instrumentation Laboratory, Bedford,

Massachusetts, USA). Biochemical assays were performed using serum to measure

lipid fractions (reported in mmol/L, with non-HDL calculated as total cholesterol minus

HDL) and CRP (measured by high sensitive -CRP assay). HbA1c was measured using

whole blood collected in ethylenediaminetetraacetic acid (EDTA) anticoagulant. HbA1c

was reported as percentage of glycated haemoglobin (%HbA1c).

!

!

! &(!

2.3.2.3 Repeatability of measurements

A sub-sample of 310 participants was selected to test the repeatability of biological and

clinical measures (mean recall duration 3.4 months, 0.9-6 months) (239); intra-class

correlation coefficients are presented in Table 2.3.

Table 2.3 Intra-class correlation coefficients (ICC) of biological and clinical measurements

taken as part of the Airwave Health Monitoring Study

n First screening Repeat

screening P ICC (95% CI)

Mean SD Mean SD Clinical Variables

Systolic blood pressure

(mmHg)

310 130.3 14.5 126.9 13.8 <0.0001 0.70 (0.64, 0.75)

Diastolic blood pressure

(mmHg)

310 79.4 9.7 76.9 9.6 <0.0001 0.74 (0.69, 0.79)

Weight (kg) 310 82.3 15.6 81.7 15.4 0.008 0.98 (0.97, 0.98)

Height (cm)

310 173.2 9.2 173.2 9.1 0.69 1.00 (0.99, 1.00)

Waist Girth (cm)

306 89.4 11.4 88.7 11.1 0.03 0.89 (0.86, 0.91)

Hip Girth (cm)

306 103.3 7.9 103.2 8.1 0.80 0.82 (0.79, 0.86)

Fat percentage (%)

263 28.4 8.2 27.8 8.2 0.008 0.92 (0.90,0.94)

Blood biochemistry

Total cholesterol (mmol/L)

303 5.4 1.1 5.4 1.0 0.96 0.78 (0.73, 0.82)

High density lipoprotein

(mmol/L)

303 1.52 0.37 1.52 0.37 0.67 0.87 (0.84, 0.90)

Apolipoprotein B (g/L)

303 0.95 0.25 0.93 0.25 0.04 0.83 (0.79, 0.86)

HbA1c (%)

302 5.6 0.5 5.6 0.5 0.66 0.64 (0.57, 0.71)

C reactive protein (mg/L)

269 2.1 3.2 1.9 2.8 0.03 0.65 (0.58, 0.72)

P value from paired t-test.

Reprinted from Environ Res 134C Elliott et al., The Airwave Health Monitoring Study of police

officers and staff in Great Britain: Rationale, design and methods, (239), Supplementary Table 6

Copyright 2014 with permission from Elsevier.

2.3.3 Medical and pharmacological information Research nurses recorded a medical history and current medication list during the

health screen appointment. Diagnosis and date of diagnosis of listed conditions:

angina, cancer, chronic obstructive pulmonary disease, liver disease, thyroid disease,

hypertension, dyslipidaemia, diabetes, arthritis, stroke and heart disease were

recorded. Medication descriptions were matched to British National Formulary category

and clinical indication. For the purpose of the current study medications were

categorised into three groups based on indication of cardiometabolic health

!

!

! &)!

management: lipid management, blood pressure management and blood glucose

management.

2.3.4 Classification of cardiometabolic health Measures of cardiometabolic health are treated in the analyses as both continuous and

categorical variables. Categorical classification of cardiometabolic risk was based on

established parameters for biochemical and anthropometric measurements as detailed

in Table 2.4

Table 2.4 Classification of increased cardiometabolic risk across anthropometric and

biochemical measurement taken as part of the Airwave Health Monitoring Study

References for cut-off values: 1. World Health Organisation (247). 2. International Diabetes Federation

Task Force(28). 3. American Medical Association (50) 4. NICE guideline CG34, Hsia et al. 2007 (57)(59).

5. World Health Organisation, American Diabetes Association (24). 6. US Preventative Services Task

Force (45). *Non-HDL calculated as total cholesterol minus HDL cholesterol.

Measure At risk classification

Body mass index (kg/m

2

)

1

Overweight: 25 – 29.99kg/m

Obese: #30 kg/m

Waist circumference

2

Increased risk: #94cm <102cm (men); #80cm <88cm

(women)

High risk: #102cm (men); #88cm (women)

Dyslipidaemia

3

HDL

Non-HDL*

Total cholesterol: HDL ratio

HDL: 1.0mmol/L (men); 1.3mmol/L (women)

Non HDL #4mmol/L

Ratio #4.5

Blood pressure

4

Systolic pressure mmHg

Diastolic pressure mmHg

Pre-hypertension:

Systolic 120-139mm Hg and/or diastolic 80-89mmHg

Hypertension:

Systolic #140mmHg and/or diastolic #90mmHg

Blood glucose

5

HbA1c

Pre-diabetes: HbA1c 5.7-6.4%

Diabetes (type 2) HbA1c #6.5%

Inflammation

6

CRP mg/L

Moderate increased risk: >1 <3mg/L

High increased risk: #3mg/L

!

!

! '+!

2.4 Additional covariates

2.4.1 Demographic and socio-economic Age (years and months) and ethnicity were self-reported. Ethnicity was collapsed into

two categories based on >95% of the cohort classified as Caucasian. Final attained

level of education and household income was used as proxy measurements of socio-

economic status based on previous studies (248).

2.4.2 Physical activity Physical activity information was collected using The International Physical Activity

Questionnaire Short Form (IPAQ-SF) (249) which calculates metabolic equivalent

(MET) minutes per week across three exercise parameters (walking, moderate and

vigorous). The IPAQ-SF protocol was followed to classify each participant as achieving

a high, moderate or low level of activity (250). To check this classification, MET

classification was compared with self-reported body type ‘athletic’ (#10hours intense

exercise per week) and ‘standard’ (<10hours intense exercise per week) asked by the

nurse as part of the bioelectrical impedance measurement protocol. There was a 100%

agreement between those classified as ‘athletic’ and being in the highest IPAQ-SF MET

category.

2.4.3 Sedentary time Weekly TV viewing time was recorded in multiples of 15 minutes. Hours sitting per

weekday were recorded by 5 hour intervals. These variables were categorised into

three groups (high, moderate and low) based on tertile cut-off values (TV viewing hours

per week: low <6, moderate 6 to 15, high >15; weekly hours sitting on weekdays: low

<20, moderate 20 to 40, high >40).

2.4.3 Additional lifestyle behaviours

2.4.3.1 Smoking

Classification of smoking status (current, previous or never smoking) based on self-

report data; smoking defined as more than five cigarettes smoked per day.

!

!

! '*!

2.4.3.2 Sleep

Participants were asked ‘How many hours sleep do you usually enjoy?’ and to select

one response from a predetermined list (5 hours or less, 6 hours, 7 hours, 8 hours, 9

hours or more).

2.4.3.3 Special diets / change in eating behaviour

Participants were asked if they had experienced a change in appetite in the last 7-days

(either increase or decrease). They were also asked to report if they were currently

following a weight loss or ‘other’ special diet. It was assumed that the response to these

variables related to the dietary recording period.

2.5 Missing data Missing data in studies that rely on self-reported measures is a common problem (251).

It is recommended that both the extent and the pattern of missing data are reported in

observational studies (238). Outcome cardiometabolic variable measurements were

complete for BMI, waist circumference and blood pressure. Biochemical measures

(blood lipid profile and HbA1c) were missing for one participant. Bioelectrical

impedance measures of percentage body fat were missing for 3% of participants (n =

153). CRP was missing for 2% (n = 112). There was a low level of missing data (<5%)

from self-reported covariate variables: marital status (n = 137; 2.3%), smoking status (n

= 20, 0.03%), region of employment (n = 19, 0.03%), income and sleep were missing

data for one participant. Participants with low level missing covariate data were

retained for analyses unless otherwise stated in relevant study chapters.

As a result of a technical issue with data capture occupational exposure measurement

of rank and job description had the most extensive extent of missing data (n = 742,

12.7%). Therefore, the pattern of missing data were explored to determine if it was

random or systematic (251). A dummy code (missing = 0, not missing = 1) was

generated, and then t-tests (continuous data) or chi-squared (categorical data) were

conducted for key participant characteristics to determine significant differences

between responders and non-responders (Appendix Table A2.5). There was no

difference between responders and non-responders for BMI, income, marital status,

physical activity or sex. However, the technical error occurred during data collection in

!

!

! '"!

England therefore there is a significant bias in country, with 95% of missing data from

forces in England (p <0.0001). Non-responders were also older (mean age 42.2 SD

8.7yrs vs. 41.3 SD 9.4yrs; p = 0.022). As this is baseline data and the majority of

missing data relate to self-reported categorical measurements imputation is not a viable

approach (251). Shift work was not collected as part of the baseline health screen until

2012, therefore only 10% of the cohort have shift work data. From using the call record

data available, acute (i.e. during the month prior to the health screen) this increased to

40%. However, this measurement is biased as it could only be estimated for police

radio users. Due to the limitations of shift work data, it is used for exploratory sub-

cohort studies of the main cohort studies presented (Table 2.5).

2.6 Collinearity between explanatory variables Collinearity and multicollinearity exists when two or more explanatory variables used in

a regression model are significantly correlated with one another, thus producing

misleading results (252). To explore collinearity sex stratified correlation coefficients

were used to determine inter-correlations among explanatory variables. Pearson

correlation coefficients were used for normally distributed data, and Spearman Rank

otherwise. Collinearity was defined as strongly correlated explanatory variables (r

>0.50) (253). Cramer’s V was used to measure the association between categorical

variables; collinearity was defined as a strong association based on degrees of freedom

per Cohens’ guidelines (253). Analyses suggested collinearity between the following

variables: police rank and work environment (men V = 0.42, women V = 0.37); police

rank and acute shift work exposure (men V = 0.32, women, V = 0.40), acute shift work

exposure and work environment (men V = 0.27, women V = 0.43), police rank and

household income for men (V = 0.26), marital status and household income in women

(V = 0.33), and age at screening and duration of police force employment (men r =

0.64, women r = 0.53), refer to Appendix A.2.6.!!

2.7 Core statistical methods Statistical analyses were conducted using SAS 9.3 (SAS Institute, Cary NC, USA).

This section gives details of the descriptive statistic methods used across all studies in

this thesis. Additional statistical methods specific to each study are stated in the

!

!

! '#!

relevant chapter. Associations at p <0.05 are considered statistically significant and are

presented to three decimal places. Non-significant p values are presented to two

decimal places.

2.7.1 Descriptive analyses Normality of distribution for continuous variables was assessed graphically (SAS PROC

UNIVARIATE) from frequency and Q-Q plots and statistically by the Anderson-Darling test

for normality (p >0.05 considered normally distributed). Continuous variables that were

not normally distributed were transformed where possible (method of transformation is

stated when applied). To compare characteristics across two-groups independent t-

tests were used for data with normal distribution and Mann-Whitney U-test otherwise.

Mean and standard deviation (SD) were reported for data with normal distribution and

median and inter quartile range (IQR) otherwise. One-way ANOVA (SAS PROC ANOVA)

was used to determine differences in continuous variables between more than two

groups for data with normal distribution. Bonferroni post hoc test was applied to

determine the source of the difference (number of tests corrected for is stated in the

relevant chapter). Otherwise, Wilcoxon Scores with Kruskal-Wallis test for data without

a normal distribution (SAS PROC NPAR1WAY). Categorical data are presented as number

(n) and frequency (%). Associations with categorical variables were analysed using Chi-

Squared test (!$) (SAS PROC FREQUENCY).

2.7.2 Multivariable analyses General linear models (SAS PROC GLM) were conducted to test i) linear trends (p-trend)

across multiple groups (via the contrast statement) and ii) differences between two or

more groups while adjusting for potential cofounding variables. Cross-adjusted sum of

squares (Type III SS) tested the null hypothesis, i.e. that the model did not explain the

variance of the response variable. Type III SS was used to measure the effect of one

variable adjusted for the confounders in the model. Results are presented as adjusted

mean and standard error (±SE).

Logistic regression (SAS PROC LOGISTIC) was used to test associations between

explanatory and binary outcome variables. Odds ratios (OR) with 95%CIs are

presented. Hosmer and Lemeshow test was used to indicate goodness of fit, (p >0.05

indicate a model of good fit). Potential confounders were selected based on their

!

!

! '$!

association with both the exposure and outcome of interest (but not considered to be on

the causal pathway). Confounders were identified based on being reported in previous

studies, or from collinearity associations in the Airwave Health Monitoring Study data.

The covariates used in statistical models were study specific and detailed in the

relevant chapter. Where collinearity was indicated between explanatory variables, the

variable explaining the greatest amount of variability was retained in the model.

2.7.3 Sub group and sensitivity analyses Sensitivity and sub-group analyses were conducted as indicated and are summarised in

Table 2.5. The rationale is discussed in the relevant chapters.

Table 2.5 Sub-group and sensitivity analyses conducted in the cross-sectional studies included

in this thesis

Study Chapter Sensitivity analyses Sub-group

analyses

1: Energy intake

misreporting

4 • Excluding those reporting following

a special diet (not weight loss)

• Excluding those reporting a recent

change in appetite

• Excluding those diagnosed chronic

disease

2: Dietary profile of

British police force

employees

5 • Excluding non-ranked employees

Participants with

shift work data

3: Diet quality and

cardiometabolic risk

in British police

force employees

6 • Excluding those classified as

energy intake misreporters

1

or those

reporting being on a weight-loss

diet

4: Working hours

and cardiometabolic

risk in British police

force employees

7 • Excluding those classified as

energy intake misreporters

1

or those

reporting being on a weight-loss

diet

• Excluding non-ranked employees

(men)

• Including part time employees

(women)

Participants with

shift work data

1.Energy intake misreporters as defined in Study 4 (Chapter 4).

!

!

! '%!

CHAPTER 3

3.0 DIETARY DATA GENERATION: PROTOCOL DEVELOPMENT

3.1 Introduction

3.1.1 Background and rationale Diet diaries provide important information about eating occasions, frequency of eating,

regularity, and the combination of foods consumed. They are also less reliant on

participant memory than retrospective measures such as FFQs and recalls. One of the

limitations of previous research investigating occupational dietary behaviours is that

prospective dietary data have rarely been used. The large-scale collection of 7-day

food records from a single occupational group makes the Airwave Health Monitoring

Study unique, as it allows the comprehensive investigation of diet and various

occupational factors and their association with health outcomes.

Despite the valuable information generated from food diaries it is widely acknowledged

that all current dietary measurement tools present a challenge to nutritional scientists as

they are subject to human error at each stage of the assessment process. Firstly, there

may be either intentional or unintentional misreporting of dietary intake by participants

(254), this limitation is investigated in Study 1 (Chapter 4). The second stage open to

error is the ‘coding’ of the food records (the matching of food and drink items recorded

to a nutritional database code and a portion size) which is prone to subjective decision

making even by experienced coders (255). The translation of food diaries into

estimated food and nutrient intake data is labour intensive and for large-scale dietary

studies numerous coders are required to ensure timely coding completion, which may

result in inter-coder error. Additionally nutritional databases can become unreliable as

formulations of foods change over time and new foods become available.

3.1.2 Standard operating procedure aims

The aim of developing the Airwave Health Monitoring Study protocol was to estimate

nutritional intake in the cohort as accurately as possible by minimising coding errors

!

!

! '&!

and to minimise intra- and inter-coder variability. This chapter describes the design of

the dietary coding protocol, and explains how the dietary variables used in the cross-

sectional studies were generated. The complete protocol is shown in Appendix A.3.1.

3.2 Methods

3.2.1 Dietary data collection

The Airwave Health Monitoring Study used prospective 7-day estimated weighed food

diaries (Appendix A3.2) previously validated against dietary biomarkers in a large UK

epidemiological cohort (256). The 7-day food diary was posted to participants with

detailed written instructions to record all food and drink consumed over seven

consecutive days in predefined eating occasions: ‘pre-breakfast’, ‘breakfast’, ‘mid-

morning’, ‘lunch’, ‘mid-afternoon’, ‘evening meal’, ‘late evening’ and ‘additional

snacks/drinks’. Participants were asked to provide details on cooking methods, brand

names and portion sizes. To aid portion size estimation photographs based on those

developed by Nelson et al. were provided (257). The completed food diaries were

returned at the health screen appointment. Participants were also requested to

complete a selection of questions at the end of the food diary relating to type of milk,

bread and fats usually consumed, dietary supplement use and the addition of salt to

cooking or at the table.

3.2.2 Food record coding protocol The Airwave Health Monitoring Study standard dietary coding protocol was developed

to include elements that address the key limitations of dietary record coding as

highlighted in previous studies (255,258), summarised in Table 3.1.

!

!

! ''!

Table 3.1 Key elements of the Airwave Health Monitoring Study dietary coding protocol

Limitation Airwave Health Monitoring Study protocol

Novel, regional foods with

no nutrient database code

• Standard operating procedure and code

book

Incomplete data recorded

by participant

• Default rules for generic items and missing

quantities

• Code book and standard coding flow-

diagrams

Inter-coder error and

subjective coding

decisions

• Comprehensive coder training

• Standard operating procedure

• Random quality control audit

• Weekly coding team meetings

Human error • Data cleaning protocol to identify gross

coding errors

3.2.2.1 Coding principles

Dietplan6.7 nutritional software programme (Forestfield Software Ltd, Horsham, UK)

was used to code the food diaries. The default nutritional database used is based on

the McCance and Widdowson’s 6th Edition Composition of Foods UK Nutritional

Dataset (UKN). To estimate food and nutrient intakes, each food and beverage is firstly

matched to a UKN database code in Dietplan6.7 and secondly a portion size (grams) is

entered for each item. The UKN database code is linked to the nutrient composition of

that item, therefore facilitating the nutrient intake to be calculated by multiplying the

portion consumed (grams) by the nutrient composition.

The protocol provides a step-by-step guide instructing coders how to enter the raw

dietary data into Dietplan6.7. Each diary is entered by day and meal occasion in

consecutive order (‘pre-breakfast’ = coded as meal 1, ‘breakfast’ = meal 2!etc.).

Diaries were excluded from coding when <1 day was completed or if a meal

replacement diet was recorded. Nutritional supplements with the exception of those

that contained calorific value (e.g. protein supplements) were not entered into the

nutritional software and were recorded separately in Excel (Microsoft Windows). As the

food diaries are unweighed, they rely on participants recording the portion size

consumed. Therefore the protocol provides a series of flow diagrams to guide coders in

!

!

! '(!

the translation of food and drink records to database codes and portion sizes to weights

(grams) using published resources of portion (259,260) and food density information

(260,261). An example algorithm is shown in Figure 3.1 (complete version in Appendix

A3.1).

Figure 3.1 An example of an algorithm from the standard coding protocol to aid portion size

estimation for cold and ambient beverages

!"#$%&'()*+$%*,$%'&$-*.&/0*0.'&)1'&$-*2*3$"4*5*1)6&0-'*5*1"3$($"&3*4%&-7.*40.3%&604*68*1*#"1..9*

*

*

*

*

*

*

*

*

*

*

*

*

*

*

*

*

*

*

!"#$%&#$'("$)"#*+,-",'$)"."))"-$'+$&$

#*"/0.0/$1+234"$.+)$"5&4*2"$6$*0,'7$

%&#$'("$)"#*+,-",'$)"."))"-$'+$#4&22$8$4"-034$+)$2&)9"$92&##7$

:+**

**

**

*

$

$

$

$$

$

$

$$

$$

$

$

$

$

$

**

**

**

*

$

$

$

$$

$

$

$$

$$

$

$

$

$

$

$

C#"$'(0#$1+234"$&,-$&**2!$#*"/0.0/$9)&10'0"#$&#$)"D30)"-B$

!"#$

C#"$-".&32'$1+234"#$#(+E,$@"2+E$&,-$&**2!$#*"/0.0/$9)&10'0"#$&#$)"D30)"-B$

FGG42$H,+,I&2/+(+20/J$

KFL42$HE0,"J$

M"")$8$2&9")$8$/0-")N$OPG42$H&1")&9"$+.$*0,'$Q$6$*0,'JB$

$

$

$

$

?"1..>$$ R.$'(")"$0#$,+'$&$1+234"$8$E"09('$0,$'("$<=>$*+)'0+,$@++A$H*&9"#$PISJ$.+)$'("$#*"/0.0"-$20D30-$3#"$'("$.+22+E0,9$-".&32'$1+234"#$&,-$&**2!$#*"/0.0/$9)&10'0"#$&#$)"D30)"-B$

:+,I&2/+(+20/N$

T&)9"$U$PGG42$H<=>$*+)'0+,$@++AJ$1+234"$.+)$V'34@2")W$

>1")&9"X$4"-034$+)$3,#*"/0.0"-$U$FGG42$1+234"$HY><<$#')&09('$92&##$1+234"J$

=4&22$U$KSG42$HY><<$#4&22$#')&09('$92&##$1+234"J$$

>2/+(+20/N$

Z0,"N$#4&22$8$4"-034$KFL42$H<++-@&#"$FL8LG'($/",'02"JB$T&)9"$FLG42$[$\".N$!"#$%&"'(&!")%&$**+$

M"")$8$2&9")$8$/0-")N$#4&22$F]O42$8$2&)9"$LS]42$

=*0)0'#N$)".")$'+$<=>$.++-$*+)'0+,$@++A$*&9"$P$[$SB$

>$V2&)9"W$#*0)0'$0#$'+$'&A",$&#$&$-+3@2"$4"&#3)"$$

$

$

$

$

:+**

**

**

*

$

$

$

$$

$

$

$$

$$

$

$

$

$

$

**

**

**

*

$

$

$

$$

$

$

$$

$$

$

$

$

$

$

$

Taken from The Airwave Health Monitoring Study. Standard protocol for food diary coding using Dietplan Nutritional

Software. Abbreviations and references: FSA Food Standards portion book (260), MAFF Ministry Agriculture Farms

and Fisheries Food Atlas (262), Carbs and Cals portion book (263). Refer to Appendix 3.1 for complete protocol

!

!

! ')!

3.2.2.2 Standard codebook

In conjunction with the standard protocol a ‘codebook’ was developed to assist

decision-making when no exact UKN code match can be found. The codebook is an

evolving database containing supplementary codes and coding rules following the

principles of the codebook designed by Conway et al. for use in the INTERMAP study

(258). Examples of different scenarios and possible coding solutions are shown in

Table 3.2. The codebook also provides ‘default codes’ for use when only generic

information is recorded by the participant (e.g. ‘chocolate biscuit’). Default codes were

based, where possible, on published UK retail sales survey information that detail the

bestselling UK food items within specific food categories (e.g. Mintel and Keynote

market reports). New codes and rules were added to the codebook following

consensus agreement between four research Dietitians and Nutritionists.

3.2.3 Coder training The translation of food diaries into estimated food and nutrient intake data is labour

intensive. Therefore, numerous coders are required to ensure timely coding

completion. All trainees undergo supervised training by Research Dietitians (RG and

RE) in using the Dietplan6.7 software. New coders are provided with coding scenarios

as part of the trainee induction programme (Appendix A3.1). All trainees were required

to code ten ‘test’ food diaries using the standard protocol and codebook. A Research

Dietitian or Nutritionist checked the completed electronic Dietplan6.7 record against the

written diary for line code errors. Individual feedback was given to the trainee coder

after each test diary is completed. At this time if the total errors are >10% per diary the

coder will be required to complete further test diaries until errors are within tolerance

before progressing to code diaries from the study.

!

!

! (+!

Table 3.2 Example scenarios and possible coding solutions when exact code matches between

recorded foods and UKN* database codes are not available.

Example food item Scenario Codebook solution

White barm cake Regional food name for a bread

roll

UKN code for soft white roll

Hash brown No exact name match. UKN

database contains a different

name

UKN code for potato cake

Reduced salt tomato

sauce

Exact food item not available.

Regular tomato sauce available.

Set up a new database code

by adapting the UKN code for

tomato sauce so that the

sodium level reflects the

manufacturer declaration for

the low salt version.

Choc chip wheat

biscuit breakfast

cereal

Exact food item not available.

Codes for constituent items

available.

Use the UKN code for wheat

biscuit breakfast cereal (92%

of estimated portion weight)

and the UKN code for

chocolate (8% of estimated

portion weight).

Protein shake No exact or similar food items

available in the UKN.

Set up a new food item using

the manufacturer nutrient

declaration

* UK Nutritional Dataset based on McCance and Widdowson’s 6th Edition Composition of

Foods

3.2.4 Quality control An audit cycle was developed to monitor inter-coder reliability with the aim of

continuously improving coding consistency, Figure 3.2. The quality control procedure

used for diary coding in the Airwave Health Monitoring Study was adapted from that

developed by Conway et al. (258). Five per cent of all coded diaries are selected at

random every two to three months (via random number generation). Research

Dietitians and Nutritionists checked the selected electronic Dietplan6.7 record against

the written food diary and classified errors as: ‘code selection error’ (the code selected

in Dietplan6.7 does not match the written record), ‘portion error’ (over +/- 10%

!

!

! (*!

difference of the protocol weight) ‘meal code error’ (item entered into incorrect meal

occasion), ‘missing code error’ (item not coded that is in the written record), and ‘extra

code error’ (item coded that is not in the written record). If the error rate in an audit

check diary is >10% feedback and training are provided to the individual coder.

Following each audit cycle the results are fed back to the team and coding improvement

strategies are implemented as indicated. Example strategies include, staff training, new

codebook entries and the development of additional protocol flow diagrams. Detected

coding errors were not corrected to ensure that the error variance remained constant

across the total sample of diaries coded.!!

Figure 3.2 Schematic of quality control cycle used in the Airwave Health Monitoring Study to

maintain food record coding consistency

!

!

! ("!

3.2.5 Dietary data cleaning To further improve the reliability of the nutritional data, the final data set was screened

for ‘gross coding’ errors. A gross coding error is defined as when the quantity of food

recorded is a clear code error (264), for example, entering 260g of instant coffee

powder rather than 260g instant coffee made with water. To allow for gross error

detection the following procedure was followed: i) maximum portion weights were

manually assigned to each unique database item code (UKN or bespoke Airwave code)

using Excel (Microsoft Windows), ii) the electronic dietary records were linked to the

food item code to the maximum portion information data using SAS 9.3 (via the !"#$%

&'"(' function), iii) if the quantity of any item coded exceeded the set maximum portion

the food diary barcode was identified and, iv) the original diary record checked and the

quantity amended. Maximum portion limits were set by consensus agreement

(between Research Dietitians and Nutritionists) of unrealistic intake values for each

food/drink item. Any item code in the exported dataset not linked to a UKN or bespoke

Airwave code was assumed to be an ex-protocol code (i.e. a code from an alternative

nutritional database available in Dietplan6.7). As shown in Figure 3.3 food diaries that

contained gross coding errors and/or ex-protocol codes were removed from the dataset

pending amendment.

Post screening the data for gross coding errors and ex-protocol codes the mean intakes

per participant were calculated for energy (kcal), fat, protein, carbohydrate and sodium.

Extreme intakes of these nutrients were assessed using SAS (!"#$%)*+,-"+-.'%function).

Potential outliers for each nutrient were considered based on the research teams’

dietary intake knowledge of the extreme values being within a realistic range. From the

data screened four participants were removed from the dataset due to energy intakes

<500kcal/day reported, which is considered physiologically unsustainable.

!

!

! (#!

Figure 3.3 Schematic of the gross quality check and data cleaning procedures applied to the

Airwave Health Monitoring study dietary data !

Export diet data from

Dietplan6.7

Final data set Participant outlier analyses

Import diet data into SAS

software and link using

unique item codes to

maximum portion values

Remove dietary

assessments from the

data set.

Action to take:

•! Check diary /

default code

•! Replacement /

default weight

•! Correct

assessment in

Dietplan6.7

Item codes ! protocol codes

(ex-protocol code error)

Entered portion > maximum portion

identified (gross coding error)

Excluded extreme outliers

3.2.6 Dietary variable generation

3.2.6.1 Composite food disaggregation

To improve the precision in determining potential associations between food and

disease it is important to estimate as accurately as possible the amount of specific

foods consumed. However, Dietplan6.7 does not have the capacity to report dietary

intake at the food level. Therefore manual disaggregation was conducted; >5,000 UKN

and bespoke Airwave food codes were extracted from Dietplan6.7 and imported into

Excel (Windows, Microsoft Office). Previous research has shown that this approach

can improve estimation of meat and fruit and vegetable intake in UK surveys that used

food diaries (265,266). Each food item was then assessed using methodology

previously developed by Bowmen et al (267) and adapted for use in a UK data set.

!

!

! ($!

Figure 3.4 outlines the process of allocating each item into food groups, and Table 3.3

provides an example of food group intake estimation for composite foods. The food

groups selected were based on those required for the DASH diet score (fruit,

vegetables, whole grains, red meat, legumes, fish, nuts, SSBs and low fat dairy). Fruit

and vegetable classifications were based on common UK culinary usage e.g. tomato as

a vegetable, peanut as nut, sweet corn as vegetable. Appendix A3.3 details the foods

classified in each food group. A second researcher checked the composite dish

disaggregation and discrepancies were resolved through consensus agreement

amongst the research team.

Figure 3.4 Schematic of food group allocation procedure applied to the UKN and Airwave

Health Monitoring Study dietary codes.

!"#$ %&$'($"))%*#(+$ &,$"$-,,+$ *.,/0$ 1%2(2$ %&$%)$#,&$ "$3,40,)%&($ -,,+56

789$:$;%.<"=($ -,,+:+.%#>$ %&(4

9,?()

@%)"**.(*"&($ %#&,$ -,,+$*.,/0$ '")(+$ ,#$%#*.(+%(#&):.(3%0(A

;))%*#$"00.,0.%"&($ -,,+$

*.,/0

B/C&%0CD$.(0,.&(+$0,.&%,#$ 'D$-"3&,.$"))%*#(+$ &,$("3E$

-,,+$ *.,/0

A$F(-(.$ &,$G"'C($ H2H

!

!

! (%!

Table 3.3 An example of how composite dishes were disaggregated into each food group of

interest

UKN/

Airwave

item

code Item name

Total

fruit

100%

fruit

juice Legumes

Vegetables

exc.

legumes

Whole

grains

Nuts

and

seeds

Total red

meat Total fish Oily fish Total dairy

Low fat

dairy

19-233

Lamb stir-fried with

vegetables 0.00 0.00 0.10 0.30 0.00 0.00 0.60 0.00 0.00 0.00 0.00

50-175

Samosas,

vegetable 0.00 0.00 0.08 0.32 0.00 0.00 0.00 0.00 0.00 0.00 0.00

15-188

Lasagna,

vegetable, whole

meal 0.00 0.00 0.00 0.50 0.30 0.00 0.00 0.00 0.00 0.10 0.00

The values for each food group represent as a decimal the estimated percentage of that food group each

dish contains. Estimated proportions of each food group contained in a composite dish were based on

recipe information from McCance and Widdowsons and supermarket online ingredient declarations for

prepared meals, for the latter at least three products were checked that matched the description and the

mean estimation applied.

3.2.6.2 Energy and nutrient intake

To account for individual differences in total energy intake, energy adjusted dietary

variables were calculated using the nutrient density method (268). Mean energy intake

(kcal/day) was derived from total energy intake divided by the number of complete food

diary days. Energy adjusted macronutrient intakes (% contribution to total energy

intake) were calculated based on grams of each macronutrient (total carbohydrate, non-

milk extrinsic sugars, total fat, saturated fat, polyunsaturated fat (PUFA),

monounsaturated fat (MUFA), alcohol and protein) recorded multiplied by the

corresponding Atwater factor (kcal/g) of 4, 9, 7 and 4kcal/g for carbohydrate, fat, alcohol

and protein respectively. Dietary fibre is reported based on the Englyst analyses

method for non-starch polysaccharide (NSP) determination. NSP, sodium and food

group intakes are reported as grams per 1000kcal. Dietary energy density was

calculated kcal per gram/food, excluding all beverages (269). Beverages classification

was based on previous research and included all caloric (juices, hot drinks, milk drinks,

alcohol and sweetened beverages) and non-caloric drinks (270).

The sodium intake measured by the food diary corresponds to intrinsic sodium

consumption as discretionary salt intake was rarely recorded. Participants completed

separate questions in the food diary relating to discretionary salt usage (do you usually

!

!

! (&!

add salt to cooking? yes/no/ don’t know; do you usually add salt at the table? yes/no/

don’t know). One salt usage categorical variable was derived from these responses (no

added salt, salt used at the table or in cooking, salt added at the table and during

cooking).

3.2.6.3 Dietary and nutritional supplements

The reported nutritional intakes in the studies conducted for this thesis only include

calorific nutritional supplements (e.g. protein supplements). A database in Excel

(Microsoft, Windows) was created to capture the type, dose and frequency of intake.

Where only brand names were given, manufacturers on-line nutrient declarations were

used to ascertain supplement type. Dichotomous variables (yes/no) for non-calorific

nutritional supplements were created on the data recorded in the database; omega

supplements (fish and plant); mineral supplement (multi mineral and/or single mineral);

vitamin supplement (multi-vitamin and/or single); other supplement (herbal and

complementary health supplements).

3.2.7 Missing dietary data Only participants with dietary data are included in the analyses presented in this thesis.

However, due to participants not completing the questions in the food diary relating to

discretionary salt usage and nutritional supplement usage there was incomplete data

for discretionary salt usage (16% non-responders) and nutritional supplement usage

(10% non-responders).

3.3 Results The food diaries included in the analyses presented in this thesis were coded between

August 2013 and December 2015. During this period Research Dietitians RG and RE

trained 21 coders. RG and RE jointly contributed to and supervised the coding of 6,303

food diaries.

3.3.1 Codebook development Between August 2013 and December 2015 over 600 unique codes and coding rules

were added to the codebook.

!

!

! ('!

3.3.2 Summary coding statistics

The length of time a coder worked on the diaries varied from less than one month to

two years. Table 3.4 summarises the number of food diaries coded per coder and the

number of diaries checked at random. Between 2013 and 2015 there were 258 (4.4%)

food diaries checked. The overall mean code error rate was 3.5% (SD 3.2%) errors per

food diary checked. Table 3.5 shows the mean code error rate per coder, per three-

month period.

Table 3.4 Number of food diaries coded and checked per coder, and overall mean error rate !

Coder ID Months

coding

experience

Number

coded

Number

checked

%

Missing/unknown n/a 11 0 0

01± 24 1872 100 5.3

02 12 433 10 2.3

03 12 162 4 2.4

04± 24 1546 71 4.6

05 1 31 0 0

06 3 142 0 0

07 6 279 7 2.5

08 3 157 9 5.7

09 <1 2 0 0

11 2 107 6 5.6

12 5 448 16 3.6

13 <1 2 0 0

14 2 41 2 4.9

15 5 178 12 6.7

16 5 185 9 4.9

17 <1 14 0 0

18 <1 2 0 0

19 3 72 6 8.3

20 <1 9 0 0

21 3 22 2 9.0

22 3 65 4 6.2

23 <1 6 0 0

24 <1 1 0 0

± Full time members of coding team; coder 10 did not progress from training diaries to

code study food diaries. Random selection made using random number generation, not

all coders were selected.

!

!

! ((!

Table 3.5 Mean food record coding error rates per coder per three-month audit period

Coder

ID* Quarter

±

Year

Mean % code

error per diary

Change on previous

quarter

1 1 2014 2.0

2 0.9 -1.1

3 1.5 0.6

4 1.4 -0.2

1 2015 1.4 0.1

2 0.6 -0.8

3 0.5 -0.1

4 1.5 1.0

2 1 2014 3.5

2 2.8 -0.7

3 9.0 6.3

3 1 2014 6.8

2 3.0 -3.8

4 1 2014 5.6

2 3.2 -2.4

3 2.8 -0.4

4 2.6 -0.2

1 2015 3.5 1.0

3 2.2 -1.4

4 3.0 0.8

7 1 2014 4.5

2 3.5 -1.0

8 1 2014 5.7

2 7.0 1.3

11 1 2014 3.0

2 2.5 0.5

12 3 2014 2.3

4 2.8 0.5

14 3 2014 3.0

15 4 2014 1.0

1 2015 4.2 3.2

3 3.5 -0.7

16 1 2015 8.0

2 5.4 -2.6

19 3 2015 7.5

22 4 2015 3.3

*Random sample generation: not all coders were selected for quality checking. Diaries

inputted by coders 5, 6, 9, 13, 17, 18, 20, 21, 23 and 24 were not selected for checking.

±Quarter refers to three month periods: 1: January to March, 2: April to June, 3: July to

September, 4: October to December.

!

!

! ()!

3.3.3 Quality check of coding Analysis of audit cycle errors found that the most frequent coding errors were portion

weight errors (55% of errors detected), followed by code selection errors (31% of errors

detected). Table 3.6 details examples of the different types of errors identified during

the quality check.

Table 3.6 Examples of different errors found during quality checking procedures !

Category of coding error

Weight / portion error Code / item selection error Missing item

Not following SOP for weights Wrongly matched food code Missing meals

Not applying weight loss/gain

from cooking

Recipe provided in the diary

not used Not coding spread /oil

Not applying specific gravity

to liquids

Non UKN or designated

Airwave code used Sauces and condiments

Spread not added to

appropriate number of slices

of bread

Incorrect type, milk, spread or

bread coded (e.g. not using

the information provided in the

general questions section)

Abbreviations: SOP standard operating procedure for food diary coding; UKN United Kingdom Nutritional

food database.

3.3.4 Data cleaning Gross coding errors and/or ex-protocol codes were detected in 7% of food diaries.

Diaries with gross coding errors and/or protocol codes require checking against the

original dietary record. Due to time constraints these participants were removed from

the final analytical sample used in this thesis.

3.4 Discussion The ‘coding’ of the food records raises numerous and complex issues relating to

incomplete information recorded by participants, the subjectivity and human error of

coders and the limitations of nutritional databases. The development of the Airwave

Health Monitoring Study standard dietary coding protocol aimed to address where

possible these limitations. This chapter has described the standard methodology

!

!

! )+!

developed and employed to increase accuracy in food record coding, this has resulted

in a mean code error rate of less than 10% per food diary. Moreover a codebook of

more than 600 coding rules and unique codes has been developed.

3.4.1 Coding consistency and error rates The rigorous coder training, standard protocol and audit cycle developed and

implemented have maintained a mean error rate below 10%. It is difficult to compare

this value with other nutritional epidemiological studies as in-depth quality control

results are rarely reported. The INTERMAP group reported an average error rate (lines

with errors/total lines) of 3.0% (271); this was based on 24-hr recall data which cannot

be directly compared to 7-day estimated weight food diaries. For example, during 24-hr

recalls participants can be probed to clarify intake information, therefore providing a

more detailed records for coding. Due to the limited tenure of the majority of coders it

was difficult to determine longitudinal improvements in error rates. From the quality

check data only the two permanent coders (coder 1 and coder 4) error rates reduced

following the first quality check and then remained relatively stable. An alternative

method of quality checking would be dual entry of diaries and compare inter-coder for

variation in nutritional values (255). This method was initially explored, however this

increased the time required to check an individual diary (from 20-30 to 45-60 minutes).

Therefore a consensus decision was taken to error check by comparing by line the

original record against the record entered on Dietplan6.7.

3.4.2 Nutritional database limitations

An inherent limitation of nutritional databases is that they can become outdated due to

continual product development and expanding selection of foods available. As part of

the standard operating procedure a codebook was established to overcome this

limitation. Where possible the ingredients and nutritional composition of novel food and

drink items were checked against leading retailer and/or manufacturers on-line

declarations. The ready access and availability to product information is a step-change

from dietary record studies conducted pre the on-line retail era. However, consideration

needs to be given to the accuracy of manufacturer and retailer declarations. An

additional limitation of nutritional databases is that nutritional content of products,

!

!

! )*!

particularly processed foods, can vary over time. For example between 2001 and 2011

salt has been reduced in bread manufacturing in the UK by ~20% (272). The nutrient

values in the UKN 6

th

edition dataset relate to samples tested prior to 2001 and

therefore salt values may be overestimated for the food diaries completed more

recently, particularly as bread is a commonly consumed food item. It has been

suggested that there is an under-estimation of portion sizes in dietary surveys (273).

This is based on research that has found many retailer serving/portion sizes have

increased since the FSA food portion size book was published in 1988 (274), which is

also used to determine default portions in the UKN nutritional database. To try to

account for varying portion sizes the algorithm developed as part of the coding protocol

prioritised participant recorded portion sizes (based on photographic guides provided),

retailer serving sizes and general UK portion size data from 2005 (259) above the 1988

standard FSA food portion sizes (Appendix A3.1).

3.5 Conclusion In summary a reproducible standard dietary coding protocol has been established for

the Airwave Health Monitoring Study that addresses the key limitations of dietary record

coding. However, participant specific factors relating to dietary measurement cannot be

controlled for through standard procedures; these will be investigated in Chapter 4.

!

! ! ! ! )"!

CHAPTER 4

4.0 STUDY 1: DIETARY ENERGY INTAKE MISREPORTING

4.1 Introduction

4.1.1 Background and study rationale The potential misreporting of energy intake by free-living participants is a widely

acknowledged limitation of all current dietary measurement tools (254). This is

problematic as positive energy balance is a key factor in the aetiology of obesity and other

non-communicable diseases. The gold standard method to measure energy expenditure

is doubly labelled water (275); however, this method is expensive and not feasible in large-

scale population studies. Researchers therefore rely on statistical methods to determine

likely accuracy of energy intake reporting.

Previous research has shown that dietary energy intake misreporting is biased towards

specific population groups (276), leading to potential systematic bias and erroneous

associations between diet and health outcomes (277). Exploring dietary energy intake

misreporting is an important part of nutritional epidemiological studies as compliance to

dietary intake recording may vary within and across different study populations (278).

4.1.2 Study aims and objectives The aims of this study were to investigate the prevalence of energy intake misreporting

among the Airwave Health Monitoring Study participants and to identify the participant

characteristics associated with energy intake misreporting. To achieve these aims the

objectives of the study were to:

i) Classify participants in the Airwave Health Monitoring Study as likely/unlikely to be

misreporting energy intake using the Goldberg method (279).

ii) Estimate the prevalence of misreporting energy intake in the Airwave Health

Monitoring Study cohort.

iii) Investigate the participant characteristics associated with energy intake

misreporting in the Airwave Health Monitoring Study cohort.

!

! ! ! ! )#!

4.2 Methods

4.2.1 Participants The study sample selection procedure is detailed in Chapter 2, Figure 2.2. As statistical

dietary energy misreporting methods assume a stable bodyweight, participants were

excluded from the analyses if they were pregnant (n = 0) and/or reported being on a

weight loss diet at the time of the health screen (women n = 286; men n = 142) providing a

final analytical sample of 5,421. Comparison of key participant characteristics of those

excluded and included in the present study is shown Appendix A4.1.

4.2.2 Dietary and non-dietary variables The non-dietary and dietary variables used in this study are detailed in Chapters 2 and 3

respectively.

4.2.3 Classification of energy intake misreporting In the absence of objective measures of energy intake and expenditure in large cohort

studies the plausibility of energy intake reporting is based on the assumption that

participants are weight stable, where estimated energy intake (EI) is equal to estimated

total energy expenditure (TEE).

4.2.3.1 Estimated energy intake

Mean daily EI was calculated for the number of days the food diary was completed

(kcal/day).

4.2.3.2 Estimated total energy expenditure and physical activity

TEE is expressed as estimated basal metabolic rate (BMR) multiplied by an estimated

physical activity level (PAL) value:

TEE = BMR x PAL

BMR (kcal/day) was estimated using Schofield equations based on sex, age, and weight

(280). As the physical activity data generated by IPAQ-SF does not cover a 24-hour

period to permit translation to an overall PAL value, estimated PALs of 1.4, 1.6 and 1.8

were assigned to ‘low’, ‘moderate’ and ‘high’ IPAQ-SF MET classification respectively.

These values are based on published Department of Health guidance representing non

occupational and occupational activity levels, with 1.4 signifying light occupational activity

and sedentary non occupational activity (280).

!

! ! ! ! )$!

4.2.3.3 Estimation of upper and lower confidence limits for energy reporting

The Goldberg equation (279) was used to estimate the likelihood of a participant

misreporting energy intake. This method takes into account the estimated variation in

daily energy intake, BMR and PAL based on previous studies (279) and the number of

days of dietary assessment.

Upper and lower confidence intervals were calculated at the individual level based on PAL

category and days of food diary completion using the following equation:

EIrep:BMR > PAL x exp [ s.d.min x (S/100) ]

" n

EIrep:BMR < PAL x exp [ s.d.max x (S/100) ]

" n

S = " CV2

wEI + CV2

wBMR + CV2

tP

d

CV = coefficient of variation; EI = energy intake, BMR = Basal Metabolic Rate, PAL =

Physical Activity Level, n = number, d = days of record, W = within subject coefficient,

tP = total variation in PAL (279).

Participants with a ratio of EI:BMR below the lower 95% confidence interval cut-off value

were classified as possibly under-reporting and above the higher cut-off as potentially

over-reporting energy intake, Table 4.1.

Table 4.1 Example of upper and lower cut off values* for ratio of energy intake to basal metabolic

rate based on 7-days food diary records across three physical activity levels

*Derived from Goldberg equation (279) ^PAL: physical activity level, value assigned based

on metabolic equivalent classification from self-report measures.

PAL^ Lower cut off Higher cut off

1.4 Under-

reporting

0.95 Plausible

reporting

range

2.05 Over-

reporting

1.6 1.09 2.35

1.8 1.22 2.65

!

! ! ! ! )%!

4.2.4 Statistical methods

4.2.4.1 Descriptive statistics

The sample was stratified by sex to allow estimation of sex-specific associations between

variables and the likelihood of implausible energy reporting. Refer to Chapter 2, section

2.7 for descriptive statistic procedures.

4.2.4.2 Logistic regression

Sex-stratified stepwise logistic regression was performed to assess the impact of

participant characteristics on the likelihood that a participant would be classified as

misreporting energy intake. The model initially contained all variables known to be

previously associated with energy intake misreporting (age, BMI, smoking status,

household income, physical activity, ethnicity, marital status, and education, length of diary

completion) and occupational specific explanatory variables related to the Airwave Health

Monitoring study (police rank, region, work environment, length of working hours, length of

time sitting per weekday). Variables were retained in the final the model if significant (p

<0.05) predictors of energy misreporting classification.

4.2.4.3 Missing data and sensitivity analyses

Acute shift work classification (shift work within 30 days prior to the health screen) was not

significantly associated with misreporting in bivariate logistic regression analyses (women

p = 0.11, men p = 0.11) and was not included in the final model due to the low sample

size. Three sets of sensitivity analyses were conducted. Firstly removing participants who

reported being on a special diet not for weight loss (n = 186), and then removing those that

recorded a change in appetite in previous two weeks (n = 2,050), as these may represent

change in energy intake during the period of the diary. An additional sensitivity analyses

removed those diagnosed with a chronic disease (cancer, diseases of thyroid, chronic liver

disease, angina, other heart, stroke, chronic obstructive pulmonary disease, n = 296), as

these disease may be associated with changes in energy metabolism.

!

! ! ! ! )&!

4.3 Results

4.3.1 Summary characteristics The proportion of food diaries completed for the entire 7-day period was 88.0%; mean

period of diary completion was 6.8 (SD 0.7) days. Men accounted for 62.0% of the

sample and were significantly older than women (42.6 SD 8.9yrs vs. 39.7 SD 9.6yrs,

p<0.0001), Table 4.2. No participants were classified as over-reporting energy intake.

The overall prevalence of likely under-reporting of energy intake was 50%. Sex was

significantly associated with under-reporting energy intake with 56.0% of men compared to

40.7% of women being classified as under-reporters (X2

, p <0.0001). Sex stratified

analyses, in Table 4.3, showed differences in the associations between demographic,

lifestyle and occupational factors when under- and plausible reporters were compared.

Across both men and women potential under-reporters were more likely to be classified as

overweight or obese compared to plausible reporters (p <0.0001), and to be in the highest

category for physical activity (p <0.0001). Male police staff were more likely to be

classified as a plausible reporter, while constable and sergeants were more likely to be

classified as an under-reporter (p = 0.002). Men in the highest category for weekday

sitting (>40 hours per week) were more likely to be classified as a plausible reporter

compared to the lowest group (<20 hours per week) (35% vs. 28%, p <0.0001). For

women, but not men, those classified as under-reporting energy intake were younger than

plausible reporters: 39.1 (SD 9.5) years vs. 40.1 (SD 9.6) years, p = 0.019. Male and

female employees working longer hours were more likely to be classified an under-

reporter, and part-time working women were more likely to be classified as a plausible

reporter (22.5% vs. 16.9%). Women with high job strain compared to passive job strain

were more likely to be classified as under-reporting energy intake (28.0% vs. 22.6%). Men

and women recording a change in appetite over the previous week or being on a non-

weight loss diet were more likely to be classified as under-reporting energy intake.

Ethnicity, marital status, smoking status, shift work, and work environment were not

significantly associated with under-reporting for either men or women.

!

! ! ! ! )'!

Table 4.2 Characteristics of Airwave Health Monitoring Study participants included in the study of

energy intake reporting (n = 5,421)

Men (n =3,355) Women (n =2,066)

Mean (SD) p#

Age at screening, years 42.6 (8.9) 39.7 (9.6) <0.0001

Body mass index kg/m

2

27.7 (3.6) 25.6 (4.5) <0.0001

Basal metabolic rate, kcal/day

-1

1907 (162) 1429 (122) <0.0001

Daily energy, kcal 2088 (472) 1699 (378) <0.0001

Ratio: EI:BMR 1.1 (0.3) 1.2 (0.3) <0.0001

%

Classed as under-reporter

56.0 40.7 <0.0001

Abbreviations: EI energy intake, BMR basal metabolic rate. #Student t-test compared mean

values between male and female participants and Chi squared test compared differences between

men and women for plausible and under reporters of energy intake

!!!

!!

"#!

Table 4.3 Com

parison of demographic, anthropom

etric, lifestyle and occupational characteristics of under- and plausible reporters of energy intake across m

en and wom

en

Men

Wom

en

Under-reporters

Plausible-reporters

p U

nder-reporters P

lausible-reporters p

N (%

)

1880 (56.0)

1475 (44.0)

840

(40.7) 1226

(59.3)

M

ean (SD

) *

Basal m

etabolic rate, kcal day-1

1939 (160)

1859 (150)

<0.0001 1454

(135) 1412

(109) <0.0001

Daily energy intake, kcal

1797 (302)

2460 (379)

<0.0001 1381

(231) 1916

(297) <0.0001

EE

:BM

R ratio

0.93 (0.16)

1.33 (0.19)

<0.0001 0.95

(0.15) 1.36

(0.21) <0.0001

Age, years

42.4 (8.7)

42.7 (9.1)

0.35 39.1

(9.5) 40.1

(9.6) 0.019

N (%

) †

Food diary completed (7-days)

1666 (88.6)

1335 (90.5)

0.78 731

(87.0) 1108

(90.4) 0.017

Ethnicity: W

hite 1812

(96.5) 1434

(97.2) 0.23

822 (97.9)

1204 (98.2)

0.57 B

ody mass index

<0.0001

<0.0001

Healthy (<25kg/m

2) 284

(15.1) 462

(31.3)

379 (45.1)

709 (57.8)

O

ver weight (25 - 30kg/m

2) 1038

(55.2) 816

(55.3)

313 (37.3)

378 (30.80

O

bese (>30kg/m2)

558 (29.7)

197 (13.4)

148

(17.6) 139

(11.3)

Marital status

0.71

0.17

Cohabiting

267 (14.4)

196 (13.5)

197

(24.5) 225

(19.1)

Divorced/separated

125 (6.7)

94 (6.5)

75

(9.3) 132

(11.2)

Married

1331 (71.8)

1072 (73.6)

378

(47.0) 606

(51.4)

Single

130 (7.0)

94 (6.5)

155

(19.2) 217

(18.4)

Missing

27 (1.4)

19 (1.3)

35

(4.2) 46

(3.7)

Annual household incom

e

0.008

0.020 Less than £32,000

169 (9.0)

136 (9.2)

144

(9.2) 114

(9.4)

£32,000 - £47,999 203

(10.8) 192

(13.0)

160 (10.3)

154 (12.7)

£48,000 - £57,999

796 (42.4)

675 (45.8)

666

(42.7) 558

(45.9)

£58,000- £77,999 513

(27.3) 337

(22.8)

419 (26.9)

271 (22.3)

M

ore than £ 78,000 198

(10.5) 135

(9.1)

169 (10.8)

118 (9.7)

!!!

!!

""!

Table 4.3 continued

Men

Wom

en

Under-reporters

Plausible- reporters

p U

nder-reporters P

lausible- reporters p

N

(%) †

E

ducation

0.16

0.08 Left school before taking G

CS

E

94 (5.0)

61 (4.1)

33

(3.9) 38

(3.1)

GC

SE

or equivalent 602

(32.0) 434

(29.4)

244 (29.0)

347 (28.3)

V

ocational qualifications 135

(7.2) 114

(7.7)

68 (8.1)

83 (6.8)

A

levels / Highers or equivalent

606 (32.2)

469 (31.8)

278

(33.1) 391

(31.9)

Bachelor D

egree or equivalent 336

(17.9) 310

(21.0)

175 (20.8)

267 (21.8)

P

ostgraduate qualifications 106

(5.6) 87

(5.9)

42 (5.0)

100 (8.2)

E

mploym

ent (force) country

0.022

0.63 E

ngland 1287

(68.6) 1033

(70.3)

615 (73.5)

900 (73.6)

S

cotland 353

(18.8) 296

(20.1)

120 (14.3)

188 (15.4)

W

ales 236

(13.1) 141

(9.6)

102 (12.2)

135 (11.0)

M

issing 4

(0.1) 5

(0.3)

3 (0.3)

3 (0.2)

E

mploym

ent rank

0.002

0.48 P

olice staff 297

(18.1) 276

(21.5)

385 (52.4)

593 (55.2)

P

olice Constable/S

ergeant 1127

(68.8) 856

(66.7)

306 (41.7)

410 (38.2)

Inspector/C

hief Inspector or above 198

(12.1) 124

(9.7)

18 (2.4)

32 (3.0)

O

ther 16

(1.0) 27

(2.1)

25 (3.4)

39 (3.6)

M

issing 242

(12.9) 192

(13.0)

106 (12.6)

152 (12.4)

S

hift work (last 30 days)

0.99

0.11

Day

115 (13.5)

92 (13.4)

51

(19.5) 92

(23.4)

Shift (no night)

170 (20.0)

138 (20.1)

83

(31.8) 142

(36.1)

Shift w

ith night 565

(66.5) 457

(66.5)

127 (48.7)

159 (55.6)

M

issing 1030

(55.0) 788

(53.0)

579 (69.0)

833 (68.0)

W

ork environment

0.73

0.58

Mainly office duties

582 (35.5)

438 (34.1)

272

(37.1) 415

(38.6)

Mainly m

obile duties 785

(47.9) 629

(49.0)

224 (30.5)

304 (28.3)

U

nclassified 271

(16.5) 216

(16.8)

238 (32.4)

355 (33.0)

M

issing 242

(12.9) 192

(13.0)

106 (12.6)

152 (12.4)

!!!

!!

$%%!

Table 4.3 continued

Men

Wom

en

U

nder-reporters P

lausible-reporters p

Under-reporters

Plausible- reporters

p

N (%

) †

Total hours worked per w

eek

<0.0001

0.001 <35 hours (part tim

e) 18

(1.0) 39

(2.6)

142 (16.9)

276 (22.5)

35 – 40 hours (standard)

643 (34.2)

516 (35.0)

366

(43.6) 536

(43.7)

41 – 48 hours 694

(36.9) 572

(38.8)

213 (25.4)

284 (23.2)

49 – 54 hours

245 (13.0)

194 (13.1)

59

(7.0) 68

(5.5)

55 hours or more

280 (14.9)

154 (10.4)

60

(7.1) 62

(5.1)

Years in police force em

ployment

0.48

0.84

6 years or less 323

(17.2) 251

(17.0)

317 (37.7)

450 (36.7)

6 - 12 years

434 (23.1)

328 (22.2)

208

(24.8) 301

(24.5)

2 – 21 years 478

(25.4) 410

(27.8)

200 (23.8)

290 (23.6)

21 years or m

ore 645

(34.3) 486

(32.9)

115 (13.7)

185 (15.1)

Job strain

0.33

0.002

Low strain (high control, low

demand)

601 (32.0)

494 (33.5)

212

(25.2) 313

(25.5)

Passive (low

control, low dem

and) 322

(17.0) 262

(17.8)

190 (22.6)

344 (28.1)

A

ctive (high demand, high control)

562 (29.9)

398 (27.0)

203

(24.2) 307

(25.0)

High strain (high dem

and, low control)

395 (21.1)

321 (21.8)

235

(28.0) 262

(21.4)

Physical activity classification

§

<0.0001

<0.0001 Low

106

(5.6) 252

(17.0)

52 (6.2)

240 (19.6)

M

oderate 796

(42.3) 688

(46.6)

379 (45.1)

627 (51.1)

H

igh 978

(52.0) 535

(36.3)

409 (37.7)

359 (29.3)

A

lcohol status

0.002

0.16 N

ever drinker 40

(2.1) 32

(2.2)

41 (4.9)

40 (3.3)

P

revious drinker 101

(5.4) 42

(2.8)

58 (6.9)

79 (6.4)

C

urrent drinker 1739

(92.5) 1401

(95.0)

741 (88.2)

1107 (90.3)

!!!

!!

$%$!

Table 4.3 continued

Men

Wom

en

U

nder-reporters P

lausible- reporters p

Under-reporters

Plausible -reporters

p

N (%

) †

Sm

oking status

0.73

0.12 N

ever smoker

1303 (69.6)

1041 (70.8)

568

(67.9) 832

(68.0)

Former sm

oker 437

(23.3) 327

(22.2)

171 (20.4)

278 (22.7)

C

urrent smoker

132 (7.0)

102 (6.9)

98

(11.7) 113

(9.2)

Sitting (total w

eekdays)

<0.0001

0.70 Low

(<20 hours) 602

(32.0) 419

(28.4)

281 (33.4)

390 (31.8)

M

oderate (20 – 40 hours) 744

(39.6) 534

(36.2)

317 (37.7)

481 (39.2)

H

igh (> 40 hours) 534

(28.4) 522

(35.4)

242 (28.8)

355 (39.0)

W

eekly TV view

ing

0.79

0.32 Low

(< 6 hours) 490

(26.1) 369

(25.0)

306 (36.4)

407 (33.2)

M


(44.9) 672

(45.6)

358 (42.6)

550 (44.9)

H

igh (>15 hours) 545

(29.0) 434

(29.4)

176 (20.9)

269 (21.9)

C

hange in appetite last 7 days

0.017

0.032 Y

es 657

(35.0) 458

(31.0)

404 (48.1)

531 (43.3)

D

iagnosed chronic disease‡

0.74

0.65

Yes

86 (4.6)

71 (4.8)

54

(6.4) 85

(6.9)

‘Other’ diet (not w

eight loss)

0.040

0.043 Y

es 74

(4.0) 39

(2.6)

38 (4.7)

35 (2.9)

M

enopause .

. .

. .

0.15 N

o .

. .

. .

700 (83.3)

994 (81.3)

Y

es .

. .

. .

81 (9.6)

151 (12.3)

D

on't know

. .

. .

. 59

(7.0) 78

(6.4)

Abbreviations: G

CS

E: G

eneral Certificate of S

econdary Education *S

tudent t-test compared m

ean values between m

ale and female participants

† Chi squared test com

pared differences between m

en and wom

en for plausible and under-reporters of energy intake. Missing values not included in

chi-squared test. § Based on self reported m

etabolic equivalent IPA

Q classification. ‡ C

hronic disease: cancer, diseases of thyroid, chronic liver disease, angina, other heart, stroke, chronic obstructive pulm

onary disease.

!

! ! ! ! *+"!

4.3.2 Logistic regression Stepwise logistic regression models showed that BMI, physical activity, and age were

significant predictors of energy intake under-reporting classification for both men and

women (Tables 4.4 and 4.5). Across both sexes BMI and physical activity were the

variables that accounted for the highest increase in odds for being classified as an

under-reporter. Those with a BMI over 30kg/m2 had higher odds of being classified as

an under-reporter: OR 2.98 (95%CI 2.22, 4.00) and 5.79 (95%CI 4.57, 7.33) in women

and men respectively compared to those with healthy BMI. Women and men in the

highest physical activity category compared to the lowest were more likely to under-

report: women OR 6.66 (95%CI 4.70, 9.42), men OR 4.85 (95%CI 3.72, 6.32),.

Additional predictors for women were higher education level reducing the odds of

under-reporting classification and high job strain compared to low job strain increasing

them (OR 1.33 95%CI 1.02, 1.72). For men being a current drinker, and higher amount

of weekday sitting were associated with reduced odds of under-reporting classification,

while working long hours (>55 hours per week) increased odds of being classified as

under-reporting (OR 1.39 95%CI 1.09, 1.77). The final models containing all identified

predictors, Tables 4.4 and 4.5 and were statistically significant for women and men (p

<0.0001).

!

! ! ! ! *+#!

Table 4.4 Predictors of under-reporting energy amongst men in the Airwave Health Monitoring Study*

Explanatory variable included in final model Odds Ratio 95% Confidence intervals

Body mass index Ref: Healthy (<25kg/m2) 1.00 Over weight (25 - 30kg/m2) 2.24 (1.87, 2.69) Obese (>30kg/m2) 5.79 (4.57, 7.33)

Physical activity / week Ref: Low 1.00 Moderate 2.90 (2.24, 3.77) High 4.85 (3.72, 6.32) Total hours worked per week Part time 0.35 (0.19, 0.64)

Ref: 35 – 40 hours (standard hours) 1.00 41 – 48 hours 0.95 (0.80, 1.13)

49 – 54 hours 0.96 (0.76, 1.21) >55 hours 1.39 (1.09, 1.77) Job strain Ref: Low strain (high control, low demand) 1.00 Passive (low control, low demand) 1.01 (0.81, 1.25) Active (high demand, high control) 1.06 (0.88, 1.28) High strain (high demand, low control) 0.98 (0.80, 1.21) Alcohol status Ref: never 1.00 Former 1.88 (1.01, 3.51) Current 0.94 (0.57, 1.55) Hours sitting per weekday Ref: Low (<20 hours) 1.00 Moderate (20 – 40 hours) 1.03 (0.86, 1.23) High (> 40 hours) 0.72 (0.60, 0.88) Education Ref: Left school before taking GCSE 1.00 GCSE or equivalent 0.88 (0.61, 1.29) Vocational qualifications 0.72 (0.46, 1.12) A levels / Highers or equivalent 0.82 (0.56, 1.21) Bachelor Degree or equivalent 0.72 (0.48, 1.07) Postgraduate qualifications 0.71 (0.45, 1.14) Age (per 5 year increase) 0.95 (0.91, 0.99)

GCSE: General Certificate of Secondary Education. Variables included in model presented showed significant association (p <0.05) with under-reporting in stepwise logistic regression. Variables in italics were significant predictors for women and are shown to enable comparison across sexes. *1,880 classified as under-reporting energy intake from a sample size of 3,355.

!

! ! ! ! *+$!

Table 4.5 Predictors of under-reporting energy amongst women in the Airwave Health Monitoring Study *

Explanatory variable included in final model Odds Ratio 95% Confidence intervals

Body mass index Ref: Healthy (<25kg/m2) 1.00 Over weight (25 - 30kg/m2) 1.79 (1.45, 2.20) Obese (>30kg/m2) 2.98 (2.22, 4.00)

Physical activity classification

Ref: Low 1.00

Moderate 3.13 (2.24, 4.38) High 6.66 (4.70, 9.42) Total hours worked per week Part time 0.74 (0.57, 1.00) Ref: 35 – 40 hours (standard hours) 1.00 41 – 48 hours 0.99 (0.78, 1.26) 49 – 54 hours 1.16 (0.78, 1.73) >55 hours 1.08 (0.72, 1.61) Job strain Ref: Low strain (high control, low demand) 1.00

Passive (low control, low demand) 0.82 (0.64, 1.07) Active (high demand, high control) 1.00 (0.77, 1.31) High strain (high demand, low control) 1.33 (1.02, 1.72) Alcohol status Ref: never 1.00 Former 0.67 (0.37, 1.20) Current 0.88 (0.43, 1.09) Hours sitting per weekday Ref: Low (<20 hours) 1.00 Moderate (20 – 40 hours) 1.00 (0.80,

1.25)

High (> 40 hours) 1.02 (0.80, 1.30) Education Ref: Left school before taking GCSE 1.00

GCSE or equivalent 0.62 (0.37, 1.06)

Vocational qualifications 0.74 (0.38, 1.30) A levels / Highers or equivalent 0.56 (0.33, 0.96) Bachelor Degree or equivalent 0.51 (0.29, 0.88) Postgraduate qualifications 0.31 (0.16, 0.58)

Age (per 5 year increase) 0.89

(0.84, 0.94)

GCSE: General Certificate of Secondary Education. Variables included in model presented showed significant association (p <0.05) with under-reporting in stepwise logistic regression. Variables in italics were significant predictors for men and are shown to enable comparison across sexes. *840 classified as under-reporting energy intake from a sample size of 2,066.

!

! ! ! ! *+%!

4.3.3 Sensitivity analyses Three sets of sensitivity analyses (Appendix A4.2.1 – A4.2.3) were conducted, i)

excluding participants reporting a special diet not for weight loss ii) excluding

participants reporting a change in appetite over the last two weeks and iii) excluding

participants recording chronic disease diagnosis. For women, excluding participants

with change in appetite and non-weight loss dieters removed job strain as a significant

predictor of under-reporting, while excluding those with chronic disease did not change

the characteristics associated with under-reporting. Models excluding participants that

reported a change in appetite (Appendix A4.2.1) added ethnicity as a significant

predictor of under-reporting for women, with British Caucasians compared to other

Ethnic categories more likely to be classified as under-reporting energy intake (OR

0.18; 95%CI 0.05, 0.72). For men, drinking status no longer remained a significant

predictor of under-reporting in sensitivity models. When participants reporting a change

in appetite were removed age was no longer a significant predictor. BMI and physical

activity remained the strongest predictors of under-reporting in all models.

4.4 Discussion The potential misreporting of energy intake by free-living participants is a widely

acknowledged limitation of all current dietary measurement tools (254). The aims of

this study were to investigate the prevalence of energy intake misreporting among the

Airwave Health Monitoring Study participants and to identify the participant

characteristics associated with energy intake misreporting. By using the Goldberg

equations, one of the most common statistical methods for classifying dietary energy

misreporting, the estimated prevalence of potential energy intake under-reporting is

50% in the Airwave Health Monitoring Study cohort.

4.4.1 Summary of key findings

• The estimated prevalence of potential energy intake under-reporting in the

Airwave Health Monitoring Study cohort was 56% for men and 41% for women.

• In agreement with previous studies under-reporting energy intake is systematic,

with BMI being the strongest predictor of under-reporting energy intake.

!

! ! ! ! *+&!

• Workplace factors such as length working hours and job strain show an

association with energy intake underreporting in the Airwave Health Monitoring

Study cohort.

4.4.2 Discussion of main findings Objectives i) and ii): To classify participants in the Airwave Health Monitoring Study as

likely/unlikely to be misreporting energy intake using the Goldberg method and estimate

the prevalence of misreporting energy intake

As there is no valid or consensus statistical method to measure energy under-reporting

in large-scale surveys published prevalence rates vary greatly, making comparison

between studies problematic. Classification of over-reporting is usually minimal in

dietary assessments (275). No participants were classified as over-reporting energy

intake in the Airwave Health Monitoring Study.

Applying individualised PAL values and calculating 95%CI to estimate acceptability of

energy intake reporting Murakami et al. classified 45% men and 55% women as under-

reporting energy intake from 7-day weighed food diaries using NDNS data (68).

However, these rates are greater than those published in a review of energy intake

misreporting that found the prevalence of under-reporting to be between 12% and 44%

in studies using estimated food records conducted for 3-, 4- or 7-days (281). Studies

included in the review applied estimated PAL values of between 1.05 and 1.35. The

PAL applied can be arbitrary in the absence of objective PAL measurement information;

because of the way that the Goldberg equation derives the under-reporting cut-off

points (assuming energy balance), the PAL used will impact the prevalence of under-

reporting. Lentjes et al. constructed Bland Altman plots to measure agreement

between energy intake and TEE based on varying values of PAL to estimate under-

reporters of energy intake (defined as a difference in TEE-EI >0) in EPIC-Norfolk (282).

Using this method 39-91% of participants were estimated to be under-reporting energy

intake depending on the PAL value applied (282). Due to the likely heterogeneous job

roles within the Airwave Health Monitoring Study cohort (e.g. office based, on the beat

officers, mobile patrol) the metabolic equivalent data were used to estimate PAL levels

rather than apply a universal value. The Whitehall study applied a universal EI:BMR

cut-off of 1.2 to identify low energy reporters in a sample of 7-day food diaries, and

!

! ! ! ! *+'!

estimated 31% prevalence of under-reporting (283). Therefore the positive association

between high physical activity and under-reporting may be due to the self-reported

measure of physical activity used. Moreover, the higher prevalence of under-reporting

observed in comparison to some previous studies is likely due to the considerably

higher PAL (1.4, 1.6 and 1.8) values applied to the Airwave Health Monitoring Study

cohort.

A systematic review found that self-report physical activity measurements have low to

moderate correlations with direct measurements with both under- and over-reporting

observed and no bias towards a specific population characteristics (284). Although the

Airwave Health Monitoring Study data show agreement between high METs

classification (via questionnaire) and self-disclosure of high exercise intensity (nurse

interview) these are both self-report measures. In the current study identical MET

values were applied for each activity level recorded to all classes of BMI based on

IPAQ-SF guidelines (3.3 walking, 4.0 moderate intensity activity and 8.0 vigorous

intensity activity). A recent study has suggested that calculating METs using the

standard resting oxygen uptake of 3.5 ml O2-1 min-1 could overestimate energy

expenditure in overweight and obese people by up to 38.8% (285), consequently

overestimating under-reporting in these individuals. Systematic over estimation of

METs in obese participants may result in misclassification of PAL category,

subsequently over-estimating and biasing classification of under-reporting in obese

participants.

Objective iii) To investigate the participant characteristics associated with energy

intake misreporting in the Airwave Health Monitoring Study cohort.

In agreement with previous studies, classification of under-reporting energy intake in

the Airwave Health Monitoring Study was biased towards specific population groups.

Consistent with EPIC-Norfolk (282) and UK National Diet and Nutrition Survey (254)

under-reporting prevalence in the Airwave Health Monitoring Study is directly

associated with BMI. In agreement with a previous study conducted in a working-age

French cohort, higher education in women, but not men, was associated with plausible

energy intake reporting (286). Previous studies have observed positive associations

!

! ! ! ! *+(!

between smoking and under-reporting (287) however this was not a significant

observation in the Airwave Health Monitoring Study cohort. Advancing age was a weak

but significant predictor of plausible energy intake reporting; this relationship is in

agreement with findings from the EPIC cohort (288), although other studies have shown

advancing age to be associated with under-reporting (287). The Whitehall II study

reported that those in lower employment grades were more likely to be classified as

reporting low energy intakes, following adjustment for BMI (283). Although initial

analyses showed employment as a higher rank officer to be associated with under-

reporting in men, stepwise logistic regression found that this was not a significant

predictor. In the Airwave Health Monitoring Study employment rank is highly associated

with length of work hours (X2, p <0.0001), with 55% of men employed as inspector or

higher working more than 49 or more hours per week compared to 17% of staff, which

may explain why rank was not identified as a significant predictor of under-reporting.

Higher mean weekday sitting hours for men was associated with reduced odds of being

classified as under-reporting. Participants in the highest quartile for weekday sitting

were more likely to be in a job role that is predominantly office based potentially making

it more practical for the participant to record dietary intake compared to being on mobile

duties; however, in the current sample ~12% of participants did not have these data

available.

The reasons for misreporting dietary intake are likely to be multifactorial. Diet diaries

are completed prospectively and this may cause changes in participant dietary habits

as a result of being under observation, for example consuming items that are easier to

record, or selectively consuming items to appear ‘more healthy’. A report investigating

the under-reporting of energy intake in UK nutritional surveys suggested that snacking,

eating out of the home and the under-estimation of portion sizes may all contribute to

lower energy intake reporting (273). The Airwave Health Monitoring Study included

generic photos to aid participants. Although this method has been shown to improve

portion size estimation large portions are more likely to be under-estimated (257).

Without the use of an objective measure of energy intake and expenditure it is difficult

to determine the true extent and therefore the likely mechanisms linking the

characteristics of energy-intake misreporters to lower energy recording.

!

! ! ! ! *+)!

4.4.3 Study strengths and limitations

The main strengths of this study were the large sample of 7-day food dairies and the

objective measurement of bodyweight used to determine BMI. The Airwave Health

Monitoring Study is a cohort of employees working in varied work environments

enabling the investigation of occupational predictors of dietary intake misreporting.

There are a number of limitations to this study. The present investigation only

considers energy intake misreporting; however, under-reporting may not be distributed

equally across all types of foods but may be biased towards ‘unhealthy’ foods (289).

Bias in reporting at the food level in free-living populations cannot be estimated using

statistical methods, or therefore adjusted for in analyses. Previous research has shown

that those classified as restrained eaters are more likely to under-report their energy

intake (290). Although questions about dietary restraint (291) were not included as part

of the health screen, sensitivity analyses were conducted excluding those following a

special diet not for weight loss or a change in appetite during the previous two weeks.

These analyses did not change the overall prevalence of under-reporting of energy

intake but did alter the significance of predictive characteristics for women. Sensitivity

analyses removing those with a recent change in appetite should be interpreted with

caution as a positive response to this questionnaire could infer either an increase or

decrease in appetite. It has been suggested that stress may play a role in restrained

eating behaviours (292), this supports the observation amongst women with higher job

strain being more likely to be classified as under-reporting energy intake. A further

limitation is that the present study is missing data on specific variables, in particular for

shift work. Shift work is highly associated with work environment, rank and work hours

within the cohort. Occupational factors may potentially be important predictors of

energy intake misreporting in this cohort, however the specific aspects (work hours, job

strain, shift work or work environment) that are likely to be important are difficult to

determine due to missing data and collinearity between these variables.

4.5 Study conclusion and relevance to further studies

Despite the acknowledged limitations of self-reported dietary intake it is currently the

only method of dietary measurement that is feasible for deployment in large-scale

nutritional epidemiological studies. The estimated prevalence of energy intake under-

reporting in the Airwave Health monitoring study cohort is comparable to rates reported

!

! ! ! ! **+!

for the general UK population. Although the results of this study reinforce the

suggestion that potential under-reporting of energy intake is not a result of random error

but a systematic bias, it is important to note that these observations may reflect in part,

the result of the analytical procedure used to classify energy intake misreporting. In

particular, the association between PAL and BMI with under-reporting are potentially

subject to statistical artefact. The observed prevalence and potential systematic bias of

under-reporting in the Airwave Health Monitoring Study, as with all large-scale dietary

studies, will be an important consideration when using the data where a health outcome

is associated with total energy intake. The novel observations that workplace factors

such as length of working hours and job strain are potentially associated with energy

intake misreporting will be important to consider when investigating dietary associations

with occupational factors. Therefore, energy adjustment or sensitivity analyses will be

used in the cross-sectional studies in this thesis where indicated.

!

! ***!

CHAPTER 5

5.0 STUDY 2: DIETARY PROFILE OF BRITISH POLICE FORCE EMPLOYEES

5.1 Introduction

5.1.1 Background and study rationale The literature review presented in Chapter 1 (Section 1.8) found suggestive evidence

that the time and number of working hours influenced dietary intake, with those working

longer hours or shift work consuming poorer quality diets and/or irregular eating

patterns. These differences may contribute to the observed variations in

cardiometabolic disease risk across employees working different hours. Although the

workplace is now recognised as an effective channel for the delivery of nutrition and

health promotional initiatives (293) the efficacy of interventions are dependent on

tailoring them to employee characteristics and nutritional needs (294). Gaining

knowledge about the dietary intakes of different occupational groups is necessary to

understand potential disease risk and aid the subsequent development of targeted

nutritional interventions.

The British police force employs over 250,000 men and women (18) across a range of

occupational grade and geographical regions, both of which have previously been

associated with differences in dietary intake. Studies have previously showed that

those living in Scotland have a poorer diet quality compared to those in England (295),

and results from the Whitehall II study found those in higher employment grades

consumed a healthier diet compared to those in lower grades (133). The

heterogeneous nature of British police force employees therefore makes the dietary

assessments across working hours more complex. The strengths of the Airwave Health

Monitoring Study is the large sample size, prospective dietary measurement and the

extensive demographic and lifestyle data that facilitates the investigation of dietary

intakes across different sections of the British police force.

5.1.2 Study aims and objectives

The overall aims of this cross-sectional study were to explore the dietary profile of

British police force employees and describe the differences in dietary intakes across

!

! **"!

region of employment, job role/rank and working hours. To achieve the study aims the

objectives of the study were to:

i) Estimate the nutritional intake of a cross-section of British police force

employees from 7-day food records collected as part of the Airwave Health

Monitoring Study.

ii) Describe the overall dietary profile of the British police force compared to the

general UK population and UK dietary guidelines.

iii) Measure the overall diet quality of the Airwave Health Monitoring cohort using

the Dietary Approaches to Stop Hypertension (DASH) diet quality score.

iv) To measure irregularity of daily energy intake in the Airwave Health Monitoring

cohort using the irregularity score developed by Pot et al. (119).

v) Compare dietary intakes, overall diet quality (DASH score) and irregularity of

daily energy intake across different sections of the police force - geographical

region, job role/rank and working hours.

vi) To determine if dietary intakes, overall diet quality (DASH score) and irregularity

of daily energy intake vary across length of weekly working hours independent

of established predictors of dietary intake (region, rank, age and education).

5.2 Methods

5.2.1 Participants The sampling procedure is detailed in Chapter 2, Figure 2.2. For the purpose of the

present study, all participants were included who had coded dietary data available as at

end December 2015 (n = 5,849), Appendix A2.1.

5.2.2 Dietary measurements The methods used to generate dietary data are detailed in Chapter 3. For the present

study energy adjusted macronutrient and food group intakes were used in the analyses.

5.2.2.2 Measurement of diet quality

The Dietary Approaches to Stop Hypertension (DASH) score was selected to measure

diet quality as it captures the key food groups associated with metabolic disease (whole

grains, low fat dairy, fruit, vegetables, nuts, legumes, SSBs, processed red meat and

sodium) (97,100,101). Participant intake by food group was estimated from the

!

! **#!

disaggregated dietary data (Chapter 3, Figure 3.4). The DASH score was calculated

using the method developed by Fung et al. (296). Mean grams per day intake were

calculated for ‘positive’ and ‘negative’ food groups for each participant (positive food

groups: whole grains, low fat dairy, total fruit, vegetables excluding white potatoes, nuts

legumes and seeds; negative food groups: SSBs and processed red meat). Sodium

intake was calculated as mg/day. The participants were then stratified by sex and the

quintile (Q) of intake for each food group was calculated. The following quintile score

system was then applied: For positive food groups: Q1 = 1 point, Q2= 2 points, Q3= 3

points, Q4= 4 points, Q5= 5 points and for negative food groups reverse scoring was

applied: Q1= 5 points, Q2 = 4 points Q3= 3 points, Q4= 2 points, Q5= 1 point. DASH

quintile cut off values for the total study sample are shown in Appendix A5.1. The sum

of the scores for each food group quintile is then calculated to give an overall score

from eight (least healthy diet) to 40 (most healthy diet). To determine the utility of the

DASH score in characterising foods and nutrients not included in the score calculation

but previously associated with cardiometabolic health (saturated fat, alcohol, energy

density, non-milk extrinsic sugars, and NSP), intake of these dietary variables were

compared across fifths of DASH diet score using general linear models. Dietary energy

density (kcal per g/food), saturated fat, and alcohol showed significant negative linear

associations (ptrend <0.0001), and NSP showed a linear positive relationship (ptrend

<0.0001) across groups of DASH score (Appendix A5.2).

5.2.2.3 Measurement of regularity of energy intake

Regularity of dietary energy intake refers to the day-to-day variability in total energy

intake. The method developed by Pot et al. was applied to measure daily irregularity of

energy intake (119).

For each participant a score was calculated to measure variation in energy intake per

day of recorded intake:

EI irregularity score = [(mean daily EI – EI day)§/ mean daily EI] x 100

EI energy intake, §absolute value

!

! **$!

The mean energy intake irregularity score was then calculated for each participant. The

higher the score, the more irregular the day-to-day energy intake and the lower the

score the lower the variation.

5.2.3 Group level variables Dietary intakes across the different sections of police force employees were explored by

comparing intakes within sex stratified groups i) across geographic region of

employment, ii) job role/rank and iii) working hours. Measurements of these variables

are detailed in Chapter 2, Section 2.2.

To measure dietary profiles across the different job roles the sample was grouped into

non-ranking police employees (‘staff’ and ‘other’), mid rank police officers (constables

and sergeants), and higher ranked police officers (Inspector or higher). As only 57 of

female police officers were employed in higher rank positions these were combined with

mid-rank police officers to create one category for the analyses.

Participants were classified into groups based on previous large scale studies (<35, 35

- 40, 41 – 48, 49 – 54 hours, and !55 hours per week) (15,169). As a low number of

women worked in the highest working hour groups (49 – 54 hours/week = 8.1%; and

>55 hours per week = 7.6%) these groups were collapsed in to one group. Due to the

low number of men reporting part-time work (<35 hours per week) (n = 58), these were

removed from the analyses comparing diet across work hours. Employment region was

based on Police Officers regions (Figure 2.1) collapsed into country: England, Scotland

and Wales (Appendix A.2.4).

5.2.4 Statistical methods

Refer to Chapter 2, Section 2.7 for descriptive statistic procedures. As food group

intake data were not normally distributed partial Spearman correlation coefficients

(adjusted for sex) were used to determine correlations among dietary variables. To

explore differences in dietary intakes across employee groups general linear models

were conducted – firstly, across men and women, and then across sex stratified groups

classified by i) region of employment, ii) job role/rank and iii) weekly working hours.

General linear models were then used to assess dietary intakes across groups of

working hours while controlling for region, rank and age. Post hoc Bonferroni correction

!

! **%!

was applied to correct for multiple testing (to control for Type I errors) when determining

significant differences between more than two groups. The correction (amended alpha

level) was applied based on the number of comparisons made - across three groups

the correction was applied for three tests and for six tests across four groups (i.e. for

adjusted statistical significance the critical p value of 0.05 was divided by the number of

tests corrected for).

Sensitivity and sub group analyses

There were significant differences in socio-demographic and occupational

characteristics across ranked and non-ranked staff (Appendix Table A5.3). Non-ranked

employees were more likely to work shorter working hours. Therefore, sensitivity

analyses investigating working hours and dietary intake were conducted excluding non-

ranked employees. Ranked employees account for 55% of the study sample (n =

3,266).

Sub-group analyses were conducted for participants who had shift work data. Due to

the limited sample size these analyses include all job roles and working hour

categories. Unadjusted dietary intake was assessed across three groups of shift

workers (day workers, shift without night work and night work). General linear models

were then used to measure dietary intake differences across groups of shift workers

while controlling for region, rank, age and weekly working hours.

5.3 Results

5.3.1 Descriptive statistics The current study is based on cross-sectional analyses of 5,849 participants from the

Airwave Health Monitoring Study (2,352 women and 3,497 men). Women were

significantly younger than men (39.8 SD 9.6 vs. 42.6 SD 8.9 years, p <0.0001). With

the exception of ethnicity and number of hours sitting on weekdays there were

significant differences between men and women for all socio-demographic and

occupational characteristics. Compared to women, men were more likely to be married

(72.5% vs. 49.8%), be employed as a ranked police officer (78.8 vs. 42.1%). Women

were more likely to work part-time (20.3% vs. 1.7%) and to be classified as having

passive job strain (25.5% vs. 17.4%), Table 5.1.

!

! **&!

Non-ranked police staff compared to ranked officers were more likely to work standard

(35-40 hours) or part time working hours (74.9% vs. 32.2% mid rank, and 15.9% higher

rank). Participants enrolled from Welsh forces were less likely to be male compared to

English and Scottish forces (59.4% vs. 69.5% and 65.9% respectively). Employees in

Scotland were more likely to work 55 hours or more per week (15.0% vs. 8.8% England

and 11.3% Wales). Comparison of socio-demographic and occupational characteristics

across strata of region, rank/job role and working hours are shown in Appendix A5.3 –

A5.5.

Partial correlations (adjusted for sex) between energy adjusted dietary macronutrients

and food groups are presented Appendix A5.6. DASH score was negatively correlated

with dietary energy density (r = -0.42), energy intake irregularity score (r = -0.17),

saturated fat (r = -0.29) and energy from alcohol (r = -0.08). DASH score showed a

strong positive correlation with dietary fibre (r = 0.70). SSB intake was positively

correlated with energy intake from non-milk extrinsic sugars (NME) (r = 0.50), and

negatively correlated with dietary fibre intake (r = -0.22). Dietary energy density was

negatively correlated with dietary fibre (r = -0.40), fruit (r = -0.42), and vegetable (r = -

0.44) intakes.

!

! **'!

Table 5.1 Comparison of demographic, lifestyle and occupational characteristics across men

and women with dietary data from the Airwave Health Monitoring Study (n = 5,849) Total sample Women Men P* N (%) 5849 (100) 2352 (40.2) 3497 (59.8) Age, years (SD) 41.4 (9.3) 39.8 (9.6) 42.6 (8.9) <0.0001 N (%) White 5691 (97.3) 2307 (98.1) 3384 (96.8) 0.30 Relationship status <0.0001 Cohabiting 953 (16.3) 473 (20.9) 480 (13.9) Divorced/separated 467 (8.0) 236 (10.4) 231 (6.7) Married 3629 (62.0) 1127 (49.8) 2502 (72.5) Single 663 (11.3) 426 (18.8) 237 (6.9) Missing 137 (2.3) 90 (3.8) 47 (1.4) Education 0.012 Left school before taking GCSE 248 (4.2) 83 (3.5) 165 (4.7) GCSE or equivalent 1739 (29.7) 663 (28.2) 1076 (30.8) Vocational qualifications 426 (7.3) 173 (7.4) 253 (7.2) A levels / Highers or equivalent 1892 (32.3) 764 (32.5) 1128 (32.3) Bachelor Degree or equivalent 1173 (20.1) 503 (21.4) 670 (19.2) Postgraduate qualifications 370 (6.3) 166 (7.1) 204 (5.8) Annual household income <0.0001 Less than £32,000 940 (16.1) 624 (26.5) 316 (9.0) £32,000 - £47,999 671 (11.5) 261 (11.1) 410 (11.7) £48,000 - £57,999 2320 (39.7) 784 (31.8) 1536 (43.9) £58,000- £77,999 1340 (22.9) 457 (19.3) 883 (25.3) More than £ 78,000 613 (10.5) 262 (11.1) 351 (10.0) Employment force, country <0.0001 England 4150 (71.2) 1727 (73.7) 2423 (69.5) Scotland 1018 (17.5) 347 (14.8) 671 (19.2) Wales 662 (11.3) 269 (11.5) 393 (11.3) Missing 19 (0.3) 9 (0.4) 10 (0.1) Rank <0.0001 Police staff 1729 (29.6) 1130 (52.7) 599 (19.7) Police Constable/ Sergeant 2876 (49.2) 811 (39.3) 2065 (67.9) Inspector/Chief Inspector or

390 (6.7) 57 (2.8) 333 (10.9)

Other 112 (1.9) 66 (3.2) 46 (1.5) Missing 742 (12.6) 288 (12.2) 454 (13.0) Work environment <0.0001 Mainly office duties 1843 (31.5) 779 (48.3) 1064 (35.0)

Mainly mobile duties 2063 (35.3) 592 (35.0) 1471 (48.2) Unclassified 1201 (20.5) 693 (16.7) 508 (16.8) Missing

702 (12.0) 288 (12.2) 414 (11.8) Total hours worked per week <0.0001 Part time 535 (9.1) 477 (20.3) 58 (1.7) 35-40 hours 2234 (38.2) 1023 (43.5) 1211 (34.6) 41 – 48 hours 1877 (32.1) 561 (23.8) 1316 (37.6) 49 – 54 hours 607 (10.4) 148 (6.3) 459 (13.1) 55 hours or more 596 (10.2) 143 (6.1) 453 (12.9)

!

! **(!

Table 5.1 continued Total sample Women Men P*

Years in police force <0.0001 6 years or less 1454 (24.9) 854 (36.3) 600 (17.2) 6 to 12 years 1384 (23.7) 590 (25.1) 794 (22.7) 12 to 21 years 1485 (25.4) 560 (23.8) 925 (26.4) 21 years or more 1526 (26.1) 348 (14.8) 1178 (33.7) Shift work last 30 days <0.0001 Day only 370 (6.3) 156 (21.5) 214 (13.4) Shift (no nights) 567 (9.7) 244 (33.7) 323 (20.2) Shift (with night work) 1386 (23.7) 324 (44.7) 1062 (66.4) Missing 3526 (60.3) 1628 (69.2) 1898 (54.3) Job Strain

<0.0001 Low (high control, low demand) 1735 (29.7) 603 (25.6) 1132 (32.4) Passive (low control, low

1208 (20.7) 599 (25.5) 609 (17.4)

Active (high demand, high

1586 (27.1) 580 (24.7) 1006 (28.8) High (high demand, low control) 1320 (22.6) 570 (24.2) 750 (21.4) Physical activity† <0.0001 Low 706 (12.1) 334 (14.2) 372 (10.6) Moderate 2686 (45.9) 1146 (48.7) 1540 (44.0) High 2457 (42.0) 872 (37.1) 1585 (45.4) Smoking status <0.001 Never smoker 3829 (65.5) 1589 (67.8) 2240 (70.0) Former smoker 1324 (22.6) 523 (22.3) 801 (23.0) Current smoker 476 (8.1) 233 (9.9) 243 (7.0) Missing 20 (0.4) 7 (0.3) 13 (0.4)

Sleep <0.0001 5 hours or less 318 (5.4) 144 (6.1) 174 (5.0) 6 hours 1578 (27.0) 519 (22.1) 1059 (30.3) 7 hours 2471 (42.2) 956 (40.7) 1515 (43.3) 8 hours 1285 (22.0) 633 (26.7) 652 (18.6) 9 hour or more 196 (3.4) 100 (4.2) 96 (2.7) Sitting (total weekdays) 0.07 Low (<20 hours) 1827 (31.2) 759 (32.3) 1068 (30.5) Moderate (20 – 40 hours) 2240 (38.3) 915 (38.9) 1325 (37.9) High (> 40 hours) 1782 (30.5) 678 (28.8) 1104 (31.6) Weekly TV viewing <0.0001 Low (< 6 hours) 1720 (29.4) 827 (35.2) 893 (22.5) Moderate (6 – 15 hours) 2606 (44.6) 1023 (43.5) 1583 (45.3) High (>15 hours) 1523 (26.0) 502 (21.3) 1021 (29.2)

Abbreviations: SD: standard deviation. GCSE: General certificate of Secondary Education; *Student t-test compared mean age between male and female participants. Chi squared test compared differences between men and women across categorical variables; missing data were not included in the analyses. †METs metabolic equivalents, classification by IPAQ guidelines(250).

!

! **)!

5.3.2 Dietary profile across sex

The mean daily energy intake reported in the Airwave Health Monitoring study was

1674 (SD 386) and 2077 (SD 473) kcal for women and men respectively compared with

1560 (SD 442) and 2032 (SD 617) kcal for UK adults in the National Diet and Nutrition

Survey (NDNS) (89). Women reported a significantly lower mean daily energy intake

compared to men (p <0.0001). There was no significant difference in energy intake

irregularity score between men and women. Men had a higher dietary energy density

1.6 (SD 0.4) kcal/g food vs. 1.5 (SD 0.4) kcal/gram food, p <0.0001 (Table 5.2). There

were significant differences between sources of energy intake between men and

women. Women derived more energy from carbohydrates (48.0 SD 7.0% vs. 46.7 SD

6.9%, p <0.0001) and non-milk extrinsic sugars (12.2 SD 5.2% vs. 11.6 SD 4.8%, p

<0.0001), while men obtained more energy from alcohol (4.4 IQR 7.5% vs. 2.9 IQR

6.5%, p <0.0001). With the exception of whole grains, there were significant

differences between energy-adjusted intakes across all food groups and dietary fibre,

Table 5.2. Women consumed more fish, dairy and dietary fibre per 1000kcal than men

while men consumed more red and processed meat per 1000kcal. Women consumed

diets with a higher concentration of fruit, vegetables, legumes, dairy and SSBs

(g/1000kcal) compared to men. Women were also more likely to report nutritional

supplement usage. Intake of sugar-sweetened beverages was higher than those

recorded in the NDNS, and for women higher energy from total fat and alcohol was

reported in the Airwave Health Monitoring Study compared to the NDNS (89). .

!!"#$!

A

irwa

ve

sa

mp

le

Na

tion

al D

iet a

nd

Nu

trition

Su

rve

y1

UK

Die

tary

gu

ide

line

s2

W

om

en

(n=

2,3

52

)

Me

n (n

=3

,49

7)

p W

om

en

M

en

DA

SH

sco

re, m

ed

ian

(ran

ge

) 2

4

(9 -3

8)

24

(9

-39

)

- -

- -

EI re

gu

larity

sco

re, m

ed

ian

(IQR

) 1

9.0

(1

0.2

) 1

8.5

(1

0.7

) 0

.59

-

- -

-

M

ea

n (S

D)

M

ea

n (S

D)

Me

an

da

ily T

EI, k

ca

l 1

67

4

(38

6)

20

77

(4

73

) <0.0001

15

60

(4

42

) 2

03

2

(61

7)

En

erg

y d

en

sity

foo

d k

ca

l/g

1.5

(0

.4)

1.6

(0

.4)

<0.0001 -

- -

-

% T

EI to

tal fa

t 3

3.8

(5

.6)

33

.5

(5.4

) 0

.09

3

2.9

(6

.8)

33

.0

(7.1

) <

35

% E

I

% T

EI s

atu

rate

d fa

t 1

2.4

(2

.9)

12

.3

(2.8

) 0

.62

1

2.1

(3

.5)

12

.1

(3.6

) <

11

% E

I

% T

EI p

rote

in

17

.0

(3.4

) 1

7.2

(3

.4)

0.027 1

6.5

(2

.5)

16

.5

(4.8

)

% T

EI c

arb

oh

yd

rate

4

8.0

(7

.0)

46

.7

(6.9

) <0.0001

46

.3

(7.9

) 4

4.0

(7

.6)

% T

EI N

ME

s

12

.2

(5.2

) 1

1.6

(4

.8)

<0.0001 1

1.7

(6

.0)

11

.9

(5.6

) <

10

% E

I

NS

P g

/10

00

kca

l 7

.4

(2.0

) 6

.8

(2.3

) <0.0001

7.9

6.8

>2

3g

NS

P/ d

ay (~

>1

2g

/10

00

kca

l.da

y)

M

ed

ian

(IQR

)

Me

dia

n

% T

EI a

lco

ho

l † 2

.9

(6.5

) 4

.4

(7.5

) <0.0001

1.5

3.3

Alc

oh

ol u

nits

/da

y†

0.9

(2

.0)

1.6

(2

.8)

<0.0001 0

.4

1

.2

M

axim

um

2 u

nits

/da

y

SS

Bs g

/10

00

kca

l 3

4.7

(1

09

.3)

29

.4

(10

2.4

) 0.016

70

.5

8

5.1

Lo

w fa

t da

iry g

/10

00

kca

l 9

6.6

(9

2.2

) 9

2.8

(8

3.3

) 0.010

- -

-

Wh

ole

gra

ins g

/10

00

kca

l 1

9.4

(2

8.1

) 1

9.0

(3

2.0

) 0

.75

-

- -

Fru

it g/1

00

0kca

l 7

8.1

(9

5.0

) 6

5.4

(8

4.9

) <0.0001

- -

-

>4

00

g/d

ay (~

20

0g

/10

00

kca

l ba

se

d o

n

2,0

00

kca

l/da

y)

Ve

ge

tab

les g

/10

00

kca

l 7

9.5

(5

8.0

) 5

9.2

(4

3.5

) <0.0001

- -

-

Le

gu

me

s g

/10

00

kca

l 1

1.7

(1

5.4

) 1

0.9

(1

2.9

) 0.013

- -

-

Fis

h g

/10

00

kca

l 9

.6

(20

.4)

8.3

(1

7.3

) <0.0001

Re

d m

ea

t g/1

00

0kca

l 2

9.6

(2

7.4

) 3

6.0

(2

6.4

) <0.0001

35

.9

4

2.4

Pro

ce

sse

d m

ea

t g/1

00

0kca

l 1

4.6

(1

7.3

) 1

9.0

(1

7.2

) <0.0001

M

ea

n (S

D)

So

diu

m m

g/1

00

0kca

l 1

48

3

(32

1)

14

87

(2

94

) 0

.24

<

6g

/da

y s

alt ~

2,4

00

mg

so

diu

m

(1,2

00

mg

/10

0kca

l ba

se

d o

n 2

,00

0

kca

l/da

y)

Dis

cre

tion

ary

sa

lt usa

ge

±

N (%

) 0

.09

No salt added

10

69

(5

2.7

) 1

46

0

(50

.6)

-

- -

Salt added either table or cooking

55

1

(27

.2)

78

2

(27

.1)

-

- -

Salt added both table and cooking

24

0

(11

.8)

41

2

(14

.3)

-

- -

Salt substitute used

16

9

(8.3

) 2

29

(7

.9)

-

- -

Nu

trition

al s

up

ple

me

nt u

se

~

-

- -

Mineral (m

ulti or single) 2

34

(1

0.9

) 2

01

(6

.5)

<0.0001 -

- -

Vitam

in (multi or single)

44

5

(20

.7)

47

6

(15

.3)

<0.0001 -

- -

Om

ega (plant and fish) 2

21

(1

0.3

) 3

94

(1

2.7

) 0.008

- -

-

Other / herbal

28

7

(13

.4)

27

5

(8.9

) <0.0001

- -

-

Ab

bre

via

tion

s:

TE

I to

tal

en

erg

y in

take

, N

ME

: N

on-m

ilk e

xtrin

sic

su

ga

rs.

NS

P:

No

n-s

tarc

h p

oly

sa

cch

arid

es.

SS

Bs su

ga

r sw

ee

ten

ed

b

eve

rag

es.

Stu

de

nt

t-test

co

mp

are

d m

ea

n va

lue

s b

etw

ee

n m

ale

a

nd

fe

ma

le

pa

rticip

an

ts, M

an

n-W

hitn

ey U

-test c

om

pa

red

me

dia

n v

alu

es b

etw

ee

n m

ale

an

d fe

ma

le p

artic

ipa

nts

an

d C

hi s

qu

are

d te

st to

co

mp

are

diffe

ren

ce

s b

etw

ee

n m

en

an

d w

om

en

acro

ss c

ate

go

rica

l va

riab

les. 1

. Na

tion

al D

iet

an

d N

utritio

n S

urv

ey (8

9), e

ne

rgy a

dju

ste

d in

take

s c

alc

ula

ted

ba

se

d o

n 1

9-6

4 y

rs p

ub

lish

ed

inta

ke

s. ‘-‘ d

en

ote

s n

ot c

om

pa

rab

le d

ata

ava

ilab

le. 2

. Die

tary

refe

ren

ce

va

lue

s a

nd

up

da

ted

gu

ide

line

s (7

1,1

06

,28

0).

†In

clu

de

s n

on

-co

nsu

me

rs, u

nit a

lco

ho

l = 8

g e

tha

no

l. ±S

alt q

ue

stio

nn

aire

ava

ilab

le fo

r 4,9

12

. Sa

lt = s

od

ium

x 2

.5. ~

Nu

trition

al s

up

ple

me

nt u

sa

ge

ava

ilab

le fo

r 5,2

52

, mis

sin

g d

ata

no

t inclu

de

d in

an

aly

se

s.

Table 5.2 C

om

pa

riso

n o

f die

tary

inta

ke

s a

cro

ss m

en

an

d w

om

en

in th

e A

irwa

ve

He

alth

Mo

nito

ring

Stu

dy a

nd

co

mp

aris

on

ag

ain

st th

e U

K N

atio

na

l Die

t an

d

Nu

trition

Su

rve

y, a

nd

UK

die

tary

gu

ide

line

s (n = 5

,84

9)

!

! "#"!

5.3.3 Dietary profile across region of employment DASH scores for both men and women were significantly lower for employees in

Scotland compared to England or Wales, Table 5.3. Total mean energy intake, dietary

energy density, meat, fruit and full fat dairy intake did not differ significantly across

countries for men or women.

For men whole grain, vegetable and legume intakes were significantly lower in Scotland

compared to England and Wales. Male employees in Scotland consumed significantly

higher amounts of sugar-sweetened beverages compared to those in England and

Wales. For men fibre intake differed between all countries, with Wales consuming the

highest concentration (7.2 SD 2.2 g/1000kcal) compared to England (6.9 SD 2.0

g/1000kcal) and Scotland (6.3 SD 1.8 g/1000kcal). Men in England were more likely to

report not adding salt to food compared to those in Scotland (53.8% vs. 42.4%), while

those in Scotland were more likely to report adding salt at the table and to cooking

(16.7% vs. England 13.6%). Men in Scotland also had the highest sodium intake

compared to the other countries (1522 SD 297 mg/1000kcal vs. England 1483 SD 295

mg/1000kcal, vs. Wales 1454 SD 267 mg/1000kcal) (Table 5.3).

There were less dietary differences across countries of employment for women

compared to men. Energy from saturated and total fat were higher in Scotland

compared to England and Wales. Fibre intakes were similar across England and Wales

(7.4 SD 2.3 g/1000kcal vs.7.8 SD 2.3 g/1000kcal) but significantly lower in Scotland

(6.9 SD 2.0 g/1000kcal). In women, low fat dairy intake was lower in Scottish (81.0 IQR

87.9 g/1000kcal) compared to English employees (98.7 IQR 93.1 g/1000kcal, p =

0.006). There was no difference in SSB or sodium intake across countries. The pattern

of discretionary salt usage was similar to that observed in men, with female employees

in Scotland most likely to report adding to cooking and at the table (16.0%) and those in

England more likely to report not adding salt (55.3%), Table 5.4.

!

! "##!

Table 5.3 Dietary differences across employment region in the Airwave Health Monitoring Study: men (n = 3,487)*

England Scotland Wales p

N (%) 2423 (69.5) 671 (19.2) 393 (11.3)

Mean (SD)

DASH score 24.4 (4.9) 22.9 (5.0) 24.6 (5.1) <0.0001a Mean daily energy, kcal 2080 (478) 2084 (457) 2042 (459) 0.31 Energy density of food, kcal/g 1.6 (0.4) 1.6 (0.4) 1.5 (0.4) 0.22 Fat, % TEI 33.5 (5.4) 34.6 (5.1) 32.5 (5.5) <0.0001b Saturated fat, % TEI 12.2 (2.8) 12.9 (2.8) 11.8 (2.8) <0.0001b Protein, % TEI 17.3 (3.4) 16.7 (3.5) 17.3 (3.1) 0.001a Carbohydrate, % TEI 46.7 (6.9) 46.5 (6.8) 47.0 (7.1) 0.51 NME, %TEI 11.5 (4.9) 12.0 (4.9) 11.2 (4.6) 0.016c Fibre (NSP) g/1000kcal 6.9 (2.0) 6.3 (1.8) 7.2 (2.2) <0.0001b Sodium mg/1000kcal 1483 (295) 1522 (297) 1454 (267) 0.001a

Median (IQR)

Alcohol % TEI† 4.4 (7.4) 3.9 (7.4) 5.2 (8.7) 0.008b SSBs g/1000kcal† 27.8 (100.7) 42.9 (124.5)

18.7 (79.9) <0.0001a

Full fat dairy g/1000kcal 20.0 (26.1) 18.0 (23.6) 18.4 (24.3) 0.23 Low fat dairy g/1000kcal 93.3 (83.6) 90.8 (82.7) 95.2 (84.5) 0.37 Wholegrain g/1000kcal 20.0 (31.9) 14.3 (27.5) 21.3 (36.8) <0.0001a Total fruit g/1000kcal 65.3 (86.9) 63.3 (78.0) 67.9 (87.4) 0.39 Vegetables g/1000kcal 61.3 (44.8) 52.9 (39.2) 57.5 (41.5) <0.0001a Legume g/1000kcal 11.4 (12.9) 8.7 (12.3) 11.4 (14.2) <0.0001a Total fish per g/1000kcal 8.4 (17.7) 7.1 (14.1) 9.8 (17.0) <0.0001b Total red meat g/1000kcal 36.0 (26.5) 35.6 (26.1) 36.5 (25.3) 0.59 Processed meat g/1000kcal 19.1 (16.8) 19.0 (18.3) 18.8 (17.8) 0.51 EI irregularity score 19.0 (10.6) 18.4 (10.8) 18.6 (11.2) 0.82 N (%) Discretionary salt usage± <0.0001 No salt added 1080 (53.8) 233 (42.4) 142 (44.8) Salt added either table or cooking 521 (26.0) 163 (29.6) 98 (30.9) Salt added table and cooking 273

(13.6) 92 (16.7) 45 (14.2) Salt substitute used 134 (6.7) 62 (11.3) 32 (10.1) Nutritional supplement use~ Mineral (multi or single) 144 (6.6) 38 (6.4) 19 (5.6) 0.73 Vitamin (multi or single) 332 (15.4) 94 (15.9) 49 (14.3) 0.82 Omega (plant and fish) 278 (12.9) 67 (11.3) 48 (14.0) 0.44 Other / herbal 214 (9.9) 35 (5.9) 25 (7.3) 0.006

*10 participants missing region of employment data. Abbreviations: TEI total energy intake, NME: Non-milk extrinsic sugars. NSP: Non-starch polysaccharides. SSBs: sugar sweetened beverages. †Includes non-consumers. ±Salt questionnaire available for 2,875, Nutritional supplement usage ~available for 3,096 (missing data not included in analyses). Chi squared test compared frequencies across categorical variables. To compare means values between groups one-way ANOVA was used for parametric data (values presented as mean and standard deviation). If significance indicated (p<0.05) Bonferroni post hoc test was applied to identify the source of the difference. Wilcoxon rank sum test were conducted for nonparametric data (values presented as median and inter quartile range). If significance indicated (p <0.05) Wilcoxon rank sum tests were then conducted between each group to establish the source of the difference with Bonferroni post hoc test applied to correct for multiple comparisons. a) England vs. Scotland and Scotland vs. Wales b) Between all countries. c) Scotland vs. Wales

!

! "#$!

Table 5.4 Dietary differences across employment region in the Airwave Health Monitoring Study: women (n = 2,343)*

England Scotland Wales p N (%) 1727 (73.7) 347 (14.8) 269 (11.5)

Mean (SD)

DASH score 24.3 (4.9) 23.1 (4.6) 24.4 (5.1) 0.0002a Mean daily energy, kcal 1681 (386) 1678

(394) 1622 (369) 0.06

Energy density of food, kcal/g 1.5 (0.4) 1.5 (0.4) 1.4 (0.4) 0.65 Fat, % TEI 33.7 (5.7) 34.7 (5.6) 33.2 (5.2) 0.003a Saturated fat, % TEI 12.2 (2.9) 13.0 (2.9) 12.3 (2.9) <0.0001a Protein, % TEI 17.0 (3.6) 16.6 (2.9) 16.9 (2.8) 0.09 Carbohydrate, % TEI 48.0 (7.2) 47.8 (6.7) 48.4 (6.7) 0.49 NME, %EI 12.3 (5.3) 12.5 (5.1) 11.5 (5.1) 0.05 Fibre (NSP) g/1000kcal 7.4 (2.3) 6.9 (2.0) 7.8 (2.3) <0.0001b Sodium mg/1000kcal 1475 (322) 1517

(318) 1486 (314) 0.08

Median (IQR)

Alcohol % TEI† 2.9 (6.4) 3.1 (6.0) 2.9 (7.1) 0.72 SSBs g/1000kcal† 34.0 (109.3) 43.0 (128.6) 33.9 (91.1) 0.21 Full fat dairy g/1000kcal 20.7 (27.0) 22.8 (26.5) 23.0 (24.6) 0.65 Low fat dairy g/1000kcal 98.7 (93.1) 81.0 (87.9) 93.9 (92.4) 0.006d

Wholegrain g/1000kcal 19.4 (27.5) 17.0 (26.0) 23.6 (33.8) 0.004b Total fruit g/1000kcal 79.1 (96.6) 74.7 (84.3) 75.7 (95.8) 0.75 Vegetables g/1000kcal 80.4 (58.9) 71.2 (43.6) 85.1 (62.2) 0.0001a Legume g/1000kcal 11.3 (15.2) 11.9 (14.8) 13.5 (16.2) 0.09 Total fish per g/1000kcal 9.8 (20.5) 9.0 (19.0) 10.1 (20.4) 0.80 Total red meat g/1000kcal 29.6 (27.4) 28.8 (27.8) 31.4 (28.6) 0.98 Processed meat g/1000kcal 14.5 (17.3) 15.1 (17.6) 14.4 (18.2) 0.22 EI irregularity score 19.0 (10.1) 19.3 (10.1) 19.0 (10.7) 0.78 N (%) Discretionary salt usage± 0.0001 No salt added 824 (55.3) 147 (47.9) 94 (40.7) Salt added either table or cooking 385 (26.1) 90 (29.3) 73 (31.6) Salt added table and cooking 157 (10.6) 49 (16.0) 34 (14.7) Salt substitute used 118 (8.0) 21 (6.8) 30 (13.0) Nutritional supplement use~ Mineral (multi or single) 177 (11.2) 32 (9.9) 242 (11.9) 0.62 Vitamin (multi or single) 328 (21.0) 71 (22.0) 43 (17.8) 0.44 Omega (plant and fish) 168 (10.7) 26 (8.1) 27 (11.2) 0.34 Other / herbal 218 (13.8) 32 (9.9) 36 (14.9) 0.13

*9 participants missing region of employment data. Abbreviations: TEI total energy intake, NME: Non-milk extrinsic sugars. NSP: Non-starch polysaccharides. SSBs: sugar sweetened beverages. NSP Non-starch polysaccharides. †Includes non-consumers. ±Salt questionnaire available for 2,022, nutritional supplement usage ~available for 2,040 (missing data not included in analyses). Chi squared test compared frequencies across categorical variables. To compare means values between groups one-way ANOVA was used for parametric data (values presented as mean and standard deviation). If significance indicated (p <0.05) Bonferroni post hoc test was applied to identify the source of the difference. Wilcoxon rank sum test were conducted for nonparametric data (values presented as median and inter quartile range). If significance indicated (p <0.05) Wilcoxon rank sum tests were then conducted between each group to establish the source of the difference with Bonferroni post hoc test applied to correct for multiple comparisons. a) England vs. Scotland and Scotland vs. Wales b) Between all countries c) Scotland vs. Wales d) England vs. Scotland

!

! "#%!

5.3.4 Dietary profile across job role / rank Across job role/rank male police officers had a lower DASH score (23.7 SD 4.9)

compared to non-ranked staff (24.6 SD 5.3) and high ranked staff (25.0 SD 4.6).

Energy density and SSB intake was also higher in male police officers compared to

non-ranked (p = 0.001) and higher ranked employees (p <0.0001). Whole grain and

fruit intake were highest among higher-ranking male employees compared to police

officers. There was no difference is salt usage across ranks. Other/herbal and omega

supplement usage was higher in non-ranked staff (28.2% vs. 19.8%, 19.0%), Table 5.5.

Women in non-ranked positions had a lower DASH score compared to ranked officers

(24.4 SD 4.9 vs. 23.8 SD 4.9, p = 0.015). Alcohol intake was higher in ranked

compared to non-ranked employees (3.5 IQR 6.7 %EI, vs. 2.6 IQR 6.0 %TEI, p

<0.0001). Fruit intake was higher is non-ranked compared to ranked employees (81.7

IQR 95.3 g/1000kcal, vs. 74.3 IQR 97.0 g/1000kcal, p <0.019). Ranked female police

officers were more likely to report not adding salt to food (56.0% vs. 49.4%), Table 5.6.

For both men and women ranked officers had a higher daily energy irregularity score

compared to non-ranked employees (men: non-ranked 17.9 IQR10.4 vs. ranked 19.3

IQR 10.9 p = 0.004; women non-ranked 18.6 IQR 9.7 ranked 19.6 IQR 10.9, p = 0.006).

!

! "#&!

Table 5.5 Measurement of dietary differences across job role/rank: men (n = 3,043)* Non-ranking staff

and other roles Police officers

(constables and sergeants)

Higher ranking officers (Inspector

or higher)

p

N (%) 645 (21.2) 2065

(67.8) 333 (11.0)

Mean (SD) DASH score 24.6 (5.3) 23.7 (4.9) 25.0 (4.6) <0.0001a Mean daily energy, kcal 2055 (475) 2076

(476) 2098 (453) 0.38

Energy density of food, kcal/g 1.5 (0.4) 1.6 (0.4) 1.5 (0.4) 0.001b Fat, % TEI 33.5 (5.6) 33.5 (5.4) 33.5 (5.4) 0.99 Saturated fat, % TEI 12.3 (2.9) 12.3 (2.8) 12.2 (2.8) 0.90 Protein, % TEI 17.1 (3.1) 17.2 (3.6) 17.0 (2.8) 0.52 Carbohydrate, % TEI 47.1 (7.1) 46.7 (6.9) 46.1 (6.7) 0.11

NME, %EI 11.5 (5.0) 11.7 (4.9) 10.8 (4.3) 0.007c Fibre (NSP) g/1000kcal 7.1 (2.3) 6.7 (2.0) 7.0 (2.0) <0.0001a Sodium mg/1000kcal 1473 (307) 1500

(293) 1469 (283) 0.05

Median (IQR) Alcohol % TEI† 4.1 (7.6) 4.4 (7.4) 4.9 (7.8) 0.018d SSBs g/1000kcal 19.7 (91.6) 34.2 (109.7) 18.2 (68.6) <0.0001a Full fat dairy g/1000kcal 20.7 (28.8) 19.2 (24.7) 17.5 (22.9) 0.030e Low fat dairy g/1000kcal 94.2 (86.8) 90.9 (81.9) 95.2 (79.8) 0.045c Wholegrain g/1000kcal 18.9 (34.2) 17.9 (30.3) 23.3 (33.3) 0.013c Total fruit g/1000kcal 66.2 (85.4) 62.7 (82.2) 78.1 (88.3) <0.0001f Vegetables g/1000kcal 59.9 (45.4) 58.1 (42.5) 58.9 (41.2) 0.41 Legume g/1000kcal 11.5 (12.8) 10.4 (13.0) 10.9 (11.2) 0.18 Total fish per g/1000kcal 8.7 (17.7) 7.6 (17.5) 10.7 (15.3) 0.35 Total red meat g/1000kcal 37.4 (29.9) 35.9 (25.6) 35.6 (26.5) 0.58 Processed meat g/1000kcal 19.3 (17.2) 19.3 (17.3) 17.6 (16.2) 0.05 EI irregularity score 17.9 (10.4) 19.3 (10.9) 18.1 (10.9) 0.004b N (%) Discretionary salt usage± 0.18 No salt added 241 (45.2) 869 (51.4) 139 (51.7) Salt added either table or cooking 167 (31.3) 439 (26.0) 65 (24.2) Salt added: table and cooking 78 (14.6) 245 (14.5) 42 (15.6) Salt substitute used 47 (8.8) 137 (8.1) 23 (8.6) Nutritional supplement use~ Mineral (multi or single) 44 (7.6) 105 (5.8) 16 (5.4) 0.24 Vitamin (multi or single) 81 (14.1) 278 (15.3) 42 (14.3) 0.71 Omega (plant and fish) 93 (16.2) 212 (11.7) 30 (10.2) 0.009 Other / herbal 69 (12.0) 146 (8.1) 24 (8.2) 0.014

*454 participants missing occupational rank data. Abbreviations: TEI total energy intake, NME: Non-milk extrinsic sugars. NSP: Non-starch polysaccharides. SSBs: sugar sweetened beverages. †Includes non-consumers ±Salt questionnaire data available for 2,492. ~Nutritional supplement usage available for 2,682, missing data not included in analyses. Chi squared test compared frequencies across categorical variables. To compare means values between groups one-way ANOVA was used for parametric data (values presented as mean and standard deviation). If significance indicated (p <0.05) Bonferroni post hoc test was applied to identify the source of the difference. Wilcoxon rank sum test were conducted for nonparametric data (values presented as median and inter quartile range). If significance indicated (p <0.05) Wilcoxon rank sum tests were then conducted between each group to establish the source of the difference with Bonferroni post hoc test applied to correct for multiple comparisons. a) Non-rank vs. officers and officers vs. high rank. b) Non-rank vs. officers c) Officers vs. high rank d) Officers vs. high rank and non-ranked vs. high rank e) Non ranked vs. officer and non-ranked vs. high ranked. f) Between all groups

!

! "#'!

Table 5.6 Measurement of dietary differences across job role/rank in the Airwave Health Monitoring Study: women (n = 2,064)*

Non-ranking staff and other roles

Ranked staff p

N (%) 1196 (58.0) 868 (42.0) Mean (SD) DASH score 24.4 (4.9) 23.8 (4.9) 0.015 Mean daily energy, kcal 1677 (395) 1670 (370) 0.68 Energy density of food, kcal/g 1.5 (0.4) 1.5 (0.4) 0.23 Fat, % TEI 33.7 (5.6) 33.7 (5.7) 0.88 Saturated fat, % TEI 12.5 (3.0) 12.2 (2.9) 0.08 Protein, % TEI 16.9 (3.4) 17.0 (3.2) 0.32

Carbohydrate, % TEI 48.7 (6.8) 47.5 (7.2) 0.0001 NME, %EI 12.3 (5.1) 12.1 (5.4) 0.35

Fibre (NSP) g/1000kcal 7.5 (2.3) 7.3 (2.3) 0.05 Sodium mg/1000kcal 1475 (316) 1501 (324) 0.06 Median (IQR) Alcohol % TEI† 2.6 (6.0) 3.5 (6.7) <0.0001 SSBs g/1000kcal 32.2 (99.2) 38.5 (121.4

0.019

Full fat dairy g/1000kcal 22.4 (28.1) 19.9 (24.5) 0.003 Low fat dairy g/1000kcal 101.8 (93.9) 91.1 (89.0) 0.001 Wholegrain g/1000kcal 19.0 (27.3) 20.7 (28.0) 0.13 Total fruit g/1000kcal 81.7 (95.3) 74.3 (97.0) 0.019 Vegetables g/1000kcal 79.5 (60.8) 78.2 (54.9) 0.64 Legume g/1000kcal 11.9 (15.9) 11.4 (14.4) 0.16 Total fish per g/1000kcal 9.4 (19.5) 10.2 (21.1) 0.11 Total red meat g/1000kcal 29.9 (28.7) 29.6 (26.0) 0.44 Processed meat g/1000kcal 15.5 (17.3) 14.0 (17.4) 0.37 EI irregularity score 18.6 (9.7) 19.6 (10.9) 0.006 N (%) Discretionary salt usage± 0.016 No salt added 513 (49.4) 417 (56.0) Salt added either table or cooking 311 (30.0) 175 (23.5) Salt added: table and cooking 124 (12.0)

91 (12.2) Salt substitute used 89 (8.6) 62 (8.3) Nutritional supplement use~ Mineral (multi or single) 131 (11.9) 78 (10.0) 0.19 Vitamin (multi or single) 231 (21.0) 160 (20.5) 0.78 Omega (plant and fish) 135 (12.3) 63 (8.1) 0.003 Other / herbal 156 (14.2) 99 (12.7) 0.34

*288 participants missing occupational rank data. Abbreviations: TEI total energy intake, NME: Non-milk extrinsic sugars. NSP: Non-starch polysaccharides. SSBs: sugar sweetened beverages. Student t-test compared mean values between ranked and non-ranked employees (values presented as mean and standard deviation). Mann-Whitney U-test compared median values between ranked and non-ranked employees (values presented as median and inter quartile range). Chi squared test compared frequencies across categorical variables. †Includes non-consumers. ±Salt questionnaire available for 1,782. ~Nutritional supplement usage available for 1,048 (missing data not included in analyses).

!

! "#(!

5.3.5 Dietary profile across working hours

5.3.5.1 Dietary differences across working hours: men

Unadjusted analyses showed that men working !55 hours per week compared to

standard working hours (35 - 40 hours per week) had a lower DASH score (23.9 SD 5.0

vs. 24.7 SD 4.9, p = 0.001) and dietary fibre intake (6.7.0 SD 1.9 g/1000kcal vs. 7.0 SD

2.0, p = 0.001). SSBs intake (g/1000kcal) was higher in those working !49 hours per

week compared to standard hours (49 - 54 hours: 35.4 IQR 95.4, !55 hours 34.3 IQR

120.0 vs. 24.1 IQR 95.1, p = 0.004). Daily energy intake was lower in those working

!55 hours compared to all other groups; this was the only difference that remained

significant after adjusting for region and age, Table 5.7. Daily energy intake irregularity

score was higher in those in the higher weekly working hour groups compared to

standard working hours.

Sensitivity analyses in men excluding non-ranked employees found DASH score to be

significantly higher in those working standard hours compared to other groups, and this

remained significant after adjustment for age and region. Whole grain intake was

significantly higher and SSBs lower in those working standard hours compared to all

other groups. After adjustment for age and region SSB intake remained significantly

lower in those working standard hours compared to the highest level of weekly working

hours (p = 0.025); and whole grain intake remained significantly lower in employees

working 41-48, and !55 hours per week compared to those working standard hours,

Appendix A5.7.

5.3.5.2 Dietary differences across working hour: women

DASH score was significantly higher in all groups compared to those working !49 hours

per week (p <0.0001). Fibre intake was lower in those working !49 hours per week

compared to those working <35 and 35-40 hours per week (7.1 SD 2.3 vs. 7.6 SD 2.1

and 7.5 SD 2.4 g/1000kcal, p = 0.009). Energy intake from alcohol was lower in those

working <35 hrs per week compared to women working 41-48 hours per week. Fruit

intake was lower for women working >49 hours per week compared to all other groups

of working hours. Processed meat was significantly lower in women working <35 hours

per week compared to 35-40 hours per week (12.8 IQR 15.9 g/1000kcal vs. 15.5 IQR

17.8 g/1000kcal, p = 0.046). Adjusted analyses (age and region) showed that the

observed associations remained significant for DASH score, alcohol, processed meat,

!

! "#)!

SSB and fat intake. Fibre only remained significant between the lowest and highest

groups of working hours, and the significant difference in fruit intake was attenuated,

Table 5.8.

Sensitivity analyses conducted excluding non-ranked officers showed that women in the

highest category for working hours had a lower DASH score (ptrend = 0.036). After

adjustment for region and age significance was attenuated. Energy from fat and

saturated fat was higher for women working <35hours per week compared to 41-48

hours per week. SSB intake was lowest in those working <35 hours per week (24.4

IQR 107.1 g/1000kcal) and highest in those working 41-48 hours per week (46.1 IQR

141.8 g/1000kcal) Appendix A5.8. No differences were seen across working hours for

salt or nutritional supplement usage in female ranked officers.

!!"#$!

Table 5.7 Dietary profile across w

orking hour groups: men (n = 3,497)*

35 - 40

hrs/week

41 – 48 hrs./w

eek 49 – 54

hrs/week

!55 hrs/w

eek P

1 P

2

N (%

) 1211

(34.6) 1316

(37.6) 459

(13.1) 453

(12.9)

Mean (SD

)

D

AS

H score

24.7 (4.9)

23.9 (5.1)

24.0 (5.0)

23.9 (5.0)

0.001a

0.26 M

ean daily energy, kcal 2070

(466) 2092

(487) 2096

(458) 2020

(464) 0.044

b 0.019

b E

nergy density of food (kcal/g) 1.5

(0.4) 1.6

(0.4) 1.6

(0.4) 1.6

(0.4) 0.030

c 0.23

Fat, % TE

I 33.4

(5.5) 33.7

(5.4) 33.4

(5.1) 33.2

(5.4) 0.22

0.09 S

aturated fat, % TE

I 12.2

(2.9) 12.4

(2.8) 12.2

(2.6) 12.4

(2.8) 0.40

0.40 P

rotein, % TE

I 17.1

(3.3) 17.2

(3.4) 17.2

(3.7) 17.2

(3.3) 0.93

0.86 C

arbohydrate, % TE

I 46.9

(6.7) 46.7

(7.0) 46.1

(7.0) 46.8

(7.0) 0.18

0.09 N

ME

, %E

I 11.4

(4.8) 11.7

(4.9) 11.3

(4.9) 11.8

(5.0) 0.10

0.27 Fibre (N

SP

) g/1000kcal 7.0

(2.0) 6.7

(2.1) 6.8

(2.0) 6.7

(1.9) 0.001

d 0.10

Sodium

mg/1000kcal

1490 (306)

1482 (282)

1501 (306)

1491 (298)

0.77 0.73

M

edian (IQR

)

A

lcohol % TE

I † 4.3

(7.1) 4.4

(7.7) 4.9

(8.3) 4.1

(8.2) 0.22

0.05 S

SB

s g/1000kcal 24.1

(95.1) 33.2

(107.1) 35.4

(95.4) 34.3

(120.0) 0.004

f 0.22

Full fat dairy g/1000kcal 19.3

(25.4) 19.9

(26.5) 18.8

(24.1) 18.9

(23.8) 0.52

0.38 Low

fat dairy g/1000kcal 94.2

(84.4) 92.0

(82.5) 88.3

(84.9) 94.0

(79.2) 0.37

0.21 W

holegrain g/1000kcal 20.5

(32.3) 17.7

(31.2) 21.1

(32.4) 17.5

(31.0) 0.021

d 0.24

Total fruit g/1000kcal 71.6

(88.9) 59.4

(83.1) 69.2

(86.2) 61.2

(80.1) 0.06

0.19 V

egetables g/1000kcal 60.7

(45.3) 57.6

(43.0) 59.2

(43.4) 58.0

(39.5) 0.15

0.55 Legum

e g/1000kcal 11.1

(13.8) 11.0

(11.8) 10.2

(14.5) 10.6

(12.9) 0.81

0.61 Total fish per g/1000kcal

8.6 (17.4)

7.6 (17.2)

8.7 (18.0)

8.4 (17.3)

0.62 0.65

Total red meat g/1000kcal

35.9 (28.0)

35.6 (25.1)

36.3 (25.8)

36.5 (23.8)

0.90 0.69

Processed m

eat g/1000kcal 19.2

(17.6) 18.7

(16.9) 18.9

(17.0) 19.5

(17.3) 0.75

0.89 E

I irregularity score 17.9

(10.1) 19.3

(10.6) 19.2

(11.0) 20.2

(11.1) <0.0001

e 0.005

e

!!"%&!

Table 5.7 continued (m

en) 35 - 40

hrs/week

41 – 48 hrs/w

eek 49 – 54

hrs/week

!55 hrs/w

eek P

N

(%)

Discretionary salt usage±

0.83

No salt added

517 (36.0)

284 (37.3)

139 (34.2)

82 (36.3)


544 (37.9)

283 (37.1)

157 (39.4)

89 (39.4)


206 (14.4)

93 (12.2)

55 (13.5)

26 (11.5)


169 (11.8)

103 (13.4)

56 (13.8)

29 (12.8)

Nutritional supplem

ent use~

Mineral (m

ulti or single) 77

(7.0) 74

(6.5) 22

(5.3) 24

(6.5) 0.71

Vitam


153 (13.8)

198 (12.6)

59 (12.6)

57 (12.2)

0.13

Om


(13.8) 142

(12.4) 47

(11.4) 38

(9.8) 0.18

Other / herbal

102 (9.2)

103 (9.0)

31 (7.5)

28 (7.2)

0.52

*excluding part time w

orkers due to small num

ber. Abbreviations: TE

I total energy intake, NM

E: N

on-milk extrinsic sugars. N

SP

: Non-starch polysaccharides. S

SB

s: sugar sw

eetened beverages. . †Includes non-consumers. ±S

alt n = 2831 ~supplement usage n = 3051. C

hi-squared test to compare differences across categorical variables, m

issing data not included in analyses. To com

pare means values betw

een groups one-way A

NO

VA

was used for param

etric data (values presented as mean and standard deviation). If

significance indicated (p <0.05) Bonferroni post hoc test w

as applied to identify the source of the difference. Wilcoxon rank sum

test were conducted for nonparam

etric data (values presented as m

edian and inter quartile range). If significance indicated (p <0.05) Wilcoxon rank sum

tests were then conducted betw

een each group to establish the source of the difference w

ith Bonferroni post hoc test applied to correct for m

ultiple comparisons. P

1 unadjusted, P2 G

eneral Linear Models used adjusted for age and region of

employm

ent a)

35-40 hrs vs. 41-48 hrs and 35-40 hrs vs. !55hrs b)

49-54hrs vs. !55hrs c)

35-40 hrs vs. 41-48 hrs and 35-40 hrs vs. 49-54hrs d)

35-40 hrs vs. 41-48 hrs e)

35-40 hrs vs. all other working hour categories

f) 35-40hrs vs. !55hrs

!!"%"!

Table 5.8 Dietary profile across w

orking hour groups: wom

en (n = 2,352)

<35 hrs/w

eek 35 - 40

hrs/week

41 – 48 hrs/w

eek !49

hrs/week

P1

P2

N (%

) 477

(20.3) 1023

(43.5) 561

(23.8) 291

(12.4)

Mean (SD

)

DA

SH

score 24.8

(4.8) 24.1

(4.8) 24.2

(5.0) 23.1

(4.9) <0.0001

a 0.026

a M


(369) 1671

(398) 1669

(371) 1646

(398) 0.22

0.28 E

nergy density of food (kcal/g) 1.5

(0.4) 1.5

(0.4) 1.5

(0.4) 1.5

(0.4) 0.60

0.41 Fat, %

TEI

34.5 (5.3)

33.8 (5.7)

33.3 (5.7)

33.8 (6.0)

0.006b

0.020b

Saturated fat, %

TEI

12.8 (2.8)

12.4 (3.0)

12.1 (2.8)

12.1 (3.1)

0.003c

0.001c

Protein, %

TEI

16.7 (2.9)

17.1 (3.6)

17.0 (3.3)

17.0 (3.6)

0.21 0.06

Carbohydrate, %

TEI

48.3 (6.2)

48.0 (7.1)

48.1 (7.3)

47.6 (7.6)

0.57 0.15

NM

E, %

EI

11.7 (4.3)

12.1 (5.3)

12.7 (5.6)

12.5 (5.6)

0.009b

0.31 Fibre (N

SP

) g/1000kcal 7.6

(2.1) 7.5

(2.4) 7.3

(2.2) 7.1

(2.3) 0.009

d 0.030

e S

odium m

g/1000kcal 1480

(308) 1479

(322) 1490

(332) 1484

(317) 0.93

0.97

Median (IQ

R)

Alcohol %

TEI †

2.5 (5.8)

2.8 (6.7)

3.6 (6.3)

3.2 (7.2)

0.014c

0.001c

SS

Bs g/1000kcal

24.1 (85.3)

33.4 (101.3)

44.6 (142.7)

47.6 (127.3)

<0.0001f

0.006f


(25.7) 21.4

(27.6) 19.9

(24.5) 20.5

(26.5) 0.16

0.35 Low


(88.0) 94.7

(91.1) 94.8

(94.8) 86.7

(85.5) 0.17

0.10 W


(26.9) 18.4

(28.8) 22.2

(27.6) 17.9

(28.3) 0.08

0.39 Total fruit g/1000kcal

84.9 (99.4)

79.8 (93.6)

77.1 (98.1)

64.0 (93.9)

0.005a

0.38 V


(52.9) 80.1

(60.2) 79.1

(55.2) 77.8

(52.6) 0.50

0.45 Legum

e g/1000kcal 12.1

(14.8) 12.1

(15.3) 11.0

(15.1) 10.2

(15.0) 0.09


8.7 (17.4)

9.1 (20.2)

9.9 (20.7)

10.1 (20.6)

0.89 0.06


28.3 (27.5)

30.7 (28.0)

29.3 (25.1)

29.5 (29.7)

0.37 0.40

Processed m

eat g/1000kcal

12.8 (15.9)

15.5 (17.8)

14.1 (17.0)

14.8 (19.2)

0.046g

0.016g

EI irregularity score

17.5 (9.8)

19.0 (9.7)

19.5 (10.2)

20.4 (11.7)

<0.0001h

0.001h

!!"%#!

Table 5.8 continued (wom

en) <35

hrs/week

35 - 40 hrs/w

eek 41 – 48

hrs/week

!49 hrs/w

eek p

N

(%)


0.010

No salt added

236 (22.1)

437 (40.9)

268 (25.1)

128 (12.0)


437 (19.4)

251 (45.6)

121 (22.0)

72 (13.1)


36 (15.0)

105 (43.8)

54 (22.5)

45 (18.8)


33 (19.5)

87 (54.5)

33 (19.5)

16 (9.5)

Nutritional supplem

ent use~

Mineral (m

ulti or single) 44

(10.1) 100

(10.8) 64

(12.5) 26

(9.6) 0.53

Vitam


93 (21.4)

178 (19.1)

115 (22.5)

59 (21.7)

0.44

Om


(12.2) 91

(9.8) 50

(9.8) 27

(9.3) 0.54

Other / herbal

63 (14.5)

117 (12.6)

71 (13.9)

36 (13.2)

0.77

Abbreviations: TE


E: N


SP


SB

s: sugar sweetened beverages. . †Includes non-consum

ers. ±Salt

usage n = 2029; ~Nutritional supplem

ent usage n = 2147. Chi-squared test to com

pare differences across categorical variables, missing data not included in analyses. To

compare m

eans values between groups one-w

ay AN

OV

A w

as used for parametric data (values presented as m

ean and standard deviation). If significance indicated (p <0.05) B

onferroni post hoc test was applied to identify the source of the difference. W

ilcoxon rank sum test w

ere conducted for nonparametric data (values presented as m

edian and inter quartile range). If significance indicated (p <0.05) W

ilcoxon rank sum tests w

ere then conducted between each group to establish the source of the difference w

ith Bonferroni

post hoc test applied to correct for multiple com

parisons. P1 unadjusted, P

2 General Linear M

odels used adjusted for age and region of employm

ent a)

<35 hrs vs. >49; 35-40 vs. >49; 41-48 vs. >49 b)

<35 hrs vs. 41-48 c)

<35 hrs vs. >49 and <35 vs. 41-48 d)

<35 hrs vs. >49 and 35-40 vs. >49 e)

<35 hrs vs. >49 f)

<35 hrs vs. >49 and <35 hrs vs. 35-40 and 35-40 vs. 41-48 g)

<35 hrs vs. 35-40 h)

<35 hrs vs. all other groups

!

! *##!

5.3.6 Sub group analysis: Dietary profile across shift work categories

Sub-group analyses were conducted in participants with shift work data (refer to

Appendix A5.9 for sample characteristics). Men working shifts with night work

compared to day work had a lower DASH score, higher SSB intake, lower fruit,

vegetable and alcohol intake, Table 5.9. Differences between shift categories for DASH

score did not remain significant after adjustment for age, region and rank. Women with

night work compared to day work had lower DASH score, higher sodium and SSB

intake. There was no difference in alcohol, fruit and vegetable intake across shift work

groups amongst women, Table 5.10. DASH score was no longer significantly different

after adjusting for age, region and rank while SSB intake remained significantly higher

in night working women compared to day workers.

!

! *#$!

Table 5.9 Sub group analyses: dietary profile across shift work classification: men (n = 1,599) Day work Shift no nights Shift with nights p1 p2 N (%) 214 (13.4) 323 (20.2) 1062 (66.4) Mean (SD) DASH score 24.7 (5.2) 23.8 (4.8) 23.3 (5.0) 0.001a 0.08 Mean daily energy, kcal 2079 (418) 2091 (513) 2086 (483) 0.96 0.89 Energy density of food (kcal/g) 1.5 (0.4) 1.6 (0.4) 1.6 (0.4) 0.001b 0.22 Fat, % TEI 33.9 (4.9) 33.6 (5.1) 33.9 (5.4) 0.80 0.61 Saturated fat, % TEI 12.5 (2.7) 12.4 (2.8) 12.5 (2.8) 0.93 0.34 Protein, % TEI 17.2 (3.2) 17.2 (3.1) 17.0 (3.6) 0.70 0.08 Carbohydrate, % TEI 45.9 (7.2) 46.9 (6.2) 47.3 (6.9) 0.018a 0.016a

NME, %EI 11.0 (4.6) 11.8 (4.8) 12.3 (5.0) 0.001a 0.007d Fibre (NSP) g/1000kcal 6.9 (2.1) 6.9 (2.0) 6.6 (2.0) 0.006c 0.045c Sodium mg/1000kcal 1494 (278) 1524 (309) 1505 (290) 0.45 0.19 Median (IQR) Alcohol % TEI† 4.6 (7.1) 4.6 (6.6) 3.4 (6.8) 0.003a 0.015a SSBs g/1000kcal 22.5 (72.3) 36.2 (115.1) 41.3 (134.6) 0.001b 0.002a Full fat dairy g/1000kcal 21.2 (25.1) 18.7 (25.8) 19.6 (25.2) 0.12 0.34 Low fat dairy g/1000kcal 92.7 (85.4) 90.7 (82.9) 87.8 (80.6) 0.79 0.22 Wholegrain g/1000kcal 18.7 (31.1) 19.0 (32.9) 17.0 (28.7) 0.14 0.34 Total fruit g/1000kcal 79.4 (81.7) 59.3 (77.8) 59.9 (81.1) 0.002b 0.049b Vegetables g/1000kcal 59.4 (45.0) 59.1 (43.2) 55.4 (41.3) 0.013a 0.013a Legume g/1000kcal 10.3 (12.6) 11.2 (13.0) 10.6 (12.8) 0.48 0.89 Total fish per g/1000kcal 10.2 (19.3) 8.3 (17.7) 6.5 (14.8) 0.002d <0.001d Total red meat g/1000kcal 35.0 (30.8) 36.5 (27.0) 35.9 (24.9) 0.60 0.33 Processed meat g/1000kcal 18.2 (15.6) 20.5 (17.1) 19.0 (17.1) 0.11 0.040e EI irregularity score 18.2 (10.7) 19.9 (10.7) 19.2 (10.9) 0.10 0.98 N (%) Discretionary salt usage± 0.49 No salt added 86 (47.3) 127 (47.9) 473 (53.7) Salt added either table or

53 (29.1) 78 (29.4) 212 (24.1)

Salt added both table and

28 (15.4) 39 (14.7) 126 (14.3)

Salt substitute used 15 (8.2) 21 (7.9) 70 (7.9)

Nutritional supplement use~

Mineral (multi or single) 14 (7.3) 21 (7.5) 60 (6.3) 0.73

Vitamin (multi or single) 38 (19.8) 38 (13.7) 135 (14.3) 0.12

Omega (plant and fish) 30 (15.6) 26 (9.4) 100 (10.6) 0.08

Other / herbal 15 (7.8) 23 (8.3) 70 (7.4) 0.89 Abbreviations: TEI total energy intake, NME: Non-milk extrinsic sugars. NSP: Non-starch polysaccharides. SSBs: sugar sweetened beverages. †Includes non-consumers. ~Nutritional supplement data n = 1,416, ±salt usage data n = 1,328. Chi-squared test to compare differences across categorical variables, missing data not included in analyses. To compare means values between groups one-way ANOVA was used for parametric data (values presented as mean and standard deviation). If significance indicated (p <0.05) Bonferroni post hoc test was applied to identify the source of the difference. Wilcoxon rank sum test were conducted for nonparametric data (values presented as median and inter quartile range). If significance indicated (p <0.05) Wilcoxon rank sum tests were then conducted between each group to establish the source of the difference with Bonferroni post hoc test applied to correct for multiple comparisons. p1 unadjusted, p2 General Linear Models used adjusted for age, rank and region of employment (n = 1,398 due to missing rank data).

a) Day workers vs. Shift with nights b) Day vs. shift and day vs. shift with nights c) Shift vs. shift with nights d) Day vs. shift with nights; shift without nights vs. shift with nights e) Day vs. shift no nights

!

! *#%!

Table 5.10 Sub group analyses: dietary profile across shift work classification: women (n = 724) Day work Shift no nights Shift with nights p1 p2 N (%) 156 (21.5) 244 (33.7) 324 (44.7) Mean (SD) DASH score 24.2 (4.7) 23.8 (5.3) 23.0 (4.9) 0.046a 0.39 Mean daily energy, kcal 1729 (421) 1735 (402) 1657 (360) 0.033b 0.15 Energy density of food, kcal/g 1.5 (0.4) 1.5 (0.4) 1.5 (0.4) 0.52 0.55 Fat, % TEI 34.8 (5.7) 33.8 (5.6) 33.7 (5.5) 0.10 0.80 Saturated fat, % TEI 13.0 (3.1) 12.3 (2.9) 12.4 (2.8) 0.05 0.75 Protein, % TEI 16.9 (3.3) 16.4 (3.2) 17.0 (3.5) 0.09 0.37 Carbohydrate, % TEI 47.0 (6.8) 48.5 (7.0) 47.9 (7.1) 0.11 0.45 NME, %EI 12.4 (4.9) 12.8 (5.8) 13.0 (5.6) 0.57 0.56 Fibre (NSP) g/1000kcal 7.1 (2.1) 7.2 (2.2) 7.0 (2.1) 0.58 0.62 Sodium mg/1000kcal 1448 (295) 1441 (300) 1529 (336) 0.002c 0.015c Median (IQR) Alcohol % TEI† 3.2 (5.6) 2.9 (7.0) 3.1 (6.9) 0.93 0.74 SSBs g/1000kcal 20.1 (100.0) 35.3 (112.9) 60.5 (147.4)

0.002a 0.050a

Full fat dairy g/1000kcal 23.6 (32.9) 21.9 (22.9) 22.8 (27.2) 0.51 0.55 Low fat dairy g/1000kcal 97.5 (90.5) 87.0 (90.1) 78.5 (83.0) 0.11 0.79 Wholegrain g/1000kcal 17.5 (26.0) 19.5 (29.2) 17.7 (26.7) 0.43 0.41 Total fruit g/1000kcal 78.5 (93.3) 74.2 (95.0) 66.8 (86.1) 0.31 0.86 Vegetables g/1000kcal 70.5 (50.6) 74.3 (49.5) 74.9 (48.5) 0.83 0.89 Legume g/1000kcal 11.4 (13.2) 9.2 (16.4) 10.7 (14.5) 0.39 0.32 Total fish per g/1000kcal 8.9 (16.7) 8.5 (18.1) 9.5 (19.6) 0.74 0.74 Total red meat g/1000kcal 27.1 (22.0) 28.3 (25.9) 28.7 (24.7) 0.99 0.33 Processed meat g/1000kcal 14.0 (15.1) 14.6 (16.8) 14.6 (18.1) 0.79 0.33 EI irregularity score 18.4 (11.6) 19.2 (9.7) 20.4 (10.5) 0.28 0.08 N (%) Discretionary salt usage± 0.12 No salt added 66 (19.2) 117 (54.9) 160 (56.7) Salt added either table or

39 (23.3) 57 (26.8) 71 (25.2)

Salt added both table and

16 (20.5) 23 (10.8) 39 (13.8) Salt substitute used 16 (36.4) 16 (7.5) 12 (4.3) Nutritional supplement use~ Mineral (multi or single) 15 (10.3) 19 (8.2) 33 (11.1) 0.53 Vitamin (multi or single) 35 (24.1) 36 (15.5) 70 (23.5) 0.045 Omega (plant and fish) 16 (11.0) 15 (6.5) 23 (7.7) 0.27 Other / herbal 18 (12.4) 24 (10.3) 43 (14.4) 0.37

Abbreviations: TEI total energy intake, NME: Non-milk extrinsic sugars. NSP: Non-starch polysaccharides. SSBs: sugar sweetened beverages. †Includes non-consumers. ~Nutritional supplement data n = 675, ±salt usage data n = 632. Chi-squared test to compare differences across categorical variables, missing data not included in analyses. To compare means values between groups one-way ANOVA was used for parametric data (values presented as mean and standard deviation). If significance indicated (p <0.05) Bonferroni post hoc test was applied to identify the source of the difference. Wilcoxon rank sum test were conducted for nonparametric data (values presented as median and inter quartile range). If significance indicated (p <0.05) Wilcoxon rank sum tests were then conducted between each group to establish the source of the difference with Bonferroni post hoc test applied to correct for multiple comparisons. P1 unadjusted, P2 General Linear Models used adjusted for age, rank and region of employment (n = 637 due to missing rank data). a) Day vs. shift with nights b) Shift no night vs. shift with nights c) Day vs. shift no nights, and shift no nights vs. shift with nights

!

! *#&!

5.4 Discussion Large national occupational groups such as the British police force are heterogeneous

in terms of job role, employment grade and geographical location making investigation

of diet and working hours complex. The overall aims of this cross-sectional study were

to explore the dietary profile of British police force employees and describe the

differences in dietary intakes across region of employment, job role/rank and working

hours.


• Macronutrient intakes amongst British police force employees are comparable to

those observed in the British general population.

• Compared to police employees in England those employed in Scotland reported

a poorer quality dietary pattern based on DASH score.

• Ranked officers compared to non-ranked police employees reported a poorer

quality dietary pattern based on DASH score.

• Among ranked male police officers longer weekly working hours were associated

with lower DASH dietary score, and in particular higher SSB intake and lower

whole grain intake (independent of age and region).

• Longer weekly working hours were associated with higher variation in daily

energy intakes.

• Shift work with night work was associated with higher SSB intake for both men

and women.

• For men, but not women night workers had lower fruit and vegetable intake

compared to day workers.

5.4.2 Discussion of main findings

Objectives i) and ii) To estimate the nutritional intake of a cross-section of British

police force employees from 7-day food records collected as part of the Airwave Health

Monitoring Study and describe the overall dietary profile of the British police force

compared to general UK population and UK dietary guidelines.

This study has found that the overall mean energy and macronutrient intakes reported

in the Airwave Health Monitoring Study are similar to those reported in the National Diet

!

! ! *#'!

and Nutrition Survey (NDNS) (89). However, intakes of SSBs were higher in both men

and women, and alcohol was higher in women compared to the NDNS. In line with the

general population energy intake from total fat was less than current public health

recommendations (<35%) (280). Energy intake from non-milk extrinsic sugars and

saturated fat were above recommendations (energy intake from non-milk extrinsic

sugars <10%; saturated fat <11%) for men and women.

There were clear dietary differences across men and women, with female employees

more likely to report a diet with a higher concentration of fruit and vegetables, and lower

concentration of processed and red meat. This observation supports a previous study

that also found that women were more likely report healthier choices (higher fruit and

vegetable intake with lower intake of high fat foods) compared to men (166).

Additionally the Whitehall II Study also reported that men in higher employment grades

and women (regardless of grade) were more likely to report a healthy eating pattern

associated with beneficial HDL and triglyceride levels (297). The dietary differences

between male and female police employees are likely to be multifactorial but may relate

to shorter working hours or different job roles allowing for more control over dietary

choices.

Objectives iii) and iv) To measure the overall diet quality of the Airwave Health

Monitoring cohort using the Dietary Approaches to Stop Hypertension (DASH) diet

quality score. To measure irregularity of daily energy intake in the Airwave Health

Monitoring cohort

The DASH score when applied to the Airwave Health Monitoring Study cohort showed a

dose response association with intakes of key nutrients and foods not included in the

score, but previously associated with cardiometabolic health (i.e. saturated fat, alcohol,

NMEs and fibre). Additionally, a lower (unhealthier) score was associated with self-

reported discretionary salt usage. These observations suggest that DASH score may

be an appropriate measure of overall diet quality in the Airwave Health Monitoring

Study cohort. DASH score had a significant but weak association with irregularity of

daily energy intake; to the author’s best knowledge daily fluctuations in energy intake

have not previously been compared to diet quality measures. However the original

study that developed the energy intake irregularity score showed a significant positive

!

! ! *#(!

association between daily energy intake irregularity with total fat and alcohol intake

(119). Moreover, previous studies have associated not skipping meals (a measure of

irregularity of eating) to be associated with higher fish, fruit and vegetable intakes (118).

Objective v) Compare dietary intakes, overall diet quality (DASH score) and irregularity

of daily energy intake across different sections of the police force across geographical

regions, job role/rank and working hours

The geographical differences in dietary intakes observed across regions of employment

in the Airwave Health Monitoring Study largely reflect those previously reported (295).

Studies in Britain have identified populations in Scotland and Wales as consuming diets

higher in saturated fat and sodium, and lower in fruit and vegetables compared to the

English population (295). This study also attributed the poor diet quality of Scotland

and Wales to the disproportionate incidence of CVD in these regions (295).

Job roles are diverse within the police force. A detailed breakdown of employment

grade and rank was not available for the Airwave Health Monitoring Study cohort at the

time of analysis. Collinearity analyses showed a strong association between police

rank and work environment with those classified as staff/other more likely to be office

based compared to mid-ranked police officers that are more likely to undertake mobile

duties, Appendix A2.6. Moreover, work environment and rank were strongly associated

with shift work, with non-office based employees (ranked officers) more likely to work

shifts. The dietary differences observed between mid-ranked officers and non-ranked

employees may therefore be hypothesised to reflect their working environments and

access to food. Where sample size permitted (male participants), there was an

observed difference between mid- and high- ranked staff, with those employed in higher

ranks reporting more wholegrain and fruit intakes and lower SSB intake. These findings

reflect those observed in the Whitehall II Study, that found higher grade staff consumed

a healthier diet compared to lower grade staff (133).

Following adjustment for region and age the association between poor diet quality and

longer working hours attenuated in men. However, in sensitivity analyses conducted in

mid-rank police officers longer working hours remained significantly associated with a

lower DASH score. In particular SSBs were higher and whole grain intake lower in

those working longer hours. Mid-ranking police officers are more likely to work shifts

!

! ! *#)!

and less likely to be office based (Appendix A2.6). This suggests that job role within the

police force may be an important determinant of how duration of weekly working hours

influence dietary intake in male police force employees. The analyses for female police

employees included part-time workers. Part-time workers tended to have a healthier

diet (higher DASH score and lower alcohol intake) compared to other employees. The

differences across working hours remained significant after adjustment for age and

region.

Objective vi) To determine if dietary intakes, overall diet quality (DASH score) and

irregularity of daily energy intake vary across length of weekly working hours

independent of established confounding factors (region, rank, age and education).

Previous studies investigating dietary intake and number of working hours have mainly

used questionnaires to collect dietary information (Chapter 1, Section 1.8) therefore it is

difficult to directly compare the results with the Airwave Health Monitoring Study.

Previous studies have observed that employees working longer hours had increased

intakes of take-away foods, convenience foods and snacks (210–212) which may be

indicative of a poorer quality diet. The observations reported in this current study

largely support previous studies in that longer working hours are associated with a less

healthy diet compared to those working standard hours. However, the present study

has highlighted that other occupational factors such as job role may determine dietary

intakes across length of weekly working hours. The findings from the present study for

women, but not men, support findings from the Whitehall II study, where higher rates of

overtime were associated with increased alcohol intake (174). This difference in

observations may be due to men police employees who work longer hours being more

likely to work shifts with night work, which was associated with lower median alcohol

consumption.

Regularity of energy intake refers to consistency of consuming a similar amount of

calories at the same eating occasions each day. It is possible that that irregularity of

energy intake may be a proxy measure for skipping meals. In a Swedish cohort regular

eating (defined as not skipping meals) was inversely associated with components of

MetS, with the association remaining significant after adjustment for diet quality (118).

Previous studies using questionnaires have shown long working hours to be associated

!

! ! *$+!

with irregular eating patterns (208,210). In the present study a daily energy irregularity

score was applied based on a previous study (119). The findings presented here

suggest that employees that work longer weekly working hours have greater variations

in day-to-day energy intake. However due to the limitations of the food diary used it

was not possible to measure individual meal irregularity, which has been shown to be a

predictor of MetS (118,119).

The results from the sub group analyses of employees with shift work data suggests

that men working nights have a poor diet quality compared to day workers (driven by

lower fruit and vegetable and higher SSB intakes); this association remained significant

following adjustment for age, region and rank. The lower intake in male shift workers of

fruit and vegetables is in line with existing studies conducted in Brazilian transport

workers (230), Japanese health workers (228) and male airline workers in Finland

(227). Additionally, this observation reflects the reported lower intake of fruit and

vegetables in UK shift workers (185). For men and women SSB intake remained

significantly higher in night workers compared to day workers after adjustment for age

and region. Previous observational studies by Assis et al. (216), and Tada et al. (228)

both reported higher SSB intakes in night and rotating workers respectively compared

to day workers.

There was no association between energy intake and shift work in men. In women

night workers reported less energy intake; however, this association attenuated after

adjustment for established confounders. Results of the EPIC Netherlands cohort did

observe shift workers to have higher energy intakes compared to day workers (56

kcal/day); however, they did not observe any difference in diet quality when shift

workers were compared to non-shift workers (220). The EPIC study sample included a

relatively small sample of shift workers (10%) from various occupational backgrounds.

Moreover the study used the Healthy Diet Indicator score (mainly nutrient based) and

MedD as a determinate of diet quality, while these scores include similar food groups to

DASH they are not directly comparable (see Chapter 1, Section 1.3.3).

5.4.3 Study strengths and limitations

The strength of the present study is the large-scale collection of 7-day food records

from a single occupational group and the standard coding protocol used for dietary

coding (detailed in Chapter 3). To the author’s best knowledge 7-day food diaries have

!

! ! *$*!

not been previously used to assess the impact of duration of weekly working hours on

dietary intake or to investigate the diet of British police force employees. This study has

a number of limitations. The cross-sectional design of this study cannot ascertain

causality as employees eating habits may change over time. This study also assumes

that participants were working their usual working hours during the recording of dietary

intake. The findings of this study may also be subject to residual confounding, as

information on staff canteen availability or existing work place health initiatives was not

collected; this could potentially bias the association observed between geographic force

region and diet intakes. As with all current dietary recording tools, bias in reporting is an acknowledged

limitation in nutritional research. Excluding those classified as under-reporting energy

intake did not alter the significance of the observed associations between diet quality

and food group intakes across weekly working hours. However, bias in reporting at the

food level in free-living populations cannot be estimated using statistical methods.

Therefore, misclassification of diet quality is a potential source of bias, e.g. over-

reporting ‘healthier’ foods and under-reporting of ‘unhealthy’ foods. Additionally, the

observations reported may be subject to selection bias, as although the characteristics

of the cohort used in this study are similar to the overall study cohort, they may not be

representative of the total British police force.

5.5 Study conclusions and relevance to further studies

In conclusion this study has profiled the dietary intakes of British police force

employees. In general, the observed differences in dietary intakes across different

population groups (sex, region and occupational rank/grade) in the police force reflect

previous research in the general UK population. The novel aspect to this study has

been the application of 7-day food diaries in the investigation of duration of weekly

working hours and diet. Controlling for the potential confounders the findings presented

suggest that longer working hours in ranked male police officers is associated with

dietary intakes indicative of a poorer diet quality. Additionally, longer weekly working

hours was associated with higher daily irregularity of energy intake. The observations

that employees who work night shifts have a poorer diet quality compared to day and

other shift workers is also in agreement with previous observations.

!

! *$"!

CHAPTER 6

6.0 STUDY 3: DIET QUALITY AND CARDIOMETABOLIC RISK IN

BRITISH POLICE FORCE EMPLOYEES

6.1 Introduction

6.1.1 Background and study rationale

A previous study in America found front line police officer duties to be associated with

increased risk of sudden cardiac death (19). It has also been suggested that police

officers have a higher prevalence of traditional cardiometabolic risk factors (20,21). The

reasons for this increase in risk are likely to be multifactorial, however it is important to

investigate established modifiable risk factors such as diet within this occupational

group.

As discussed in Chapter 1 (Section 1.3.3) it has been suggested that the DASH diet is

one of the most robust dietary patterns (298) based on the combination of RCT and

cohort study data showing cardiometabolic health benefits (97,100). In Study 2

(Chapter 5) the DASH score when applied to the Airwave Health Monitoring Study

cohort showed a dose response association with intakes of key nutrients and foods not

included in the score, but previously associated with cardiometabolic health (Appendix

A5.2). This observation suggests that DASH may be an informative measure of diet

quality in the Airwave Health Monitoring study cohort, and potentially in the general UK

population. Despite DASH being commonly applied to US cohorts and its

recommendation by the American Heart Association, to date the DASH score has not

been widely applied to UK cohorts to determine cardiometabolic risk (100).

6.1.2 Study aims and objectives

The overall aim of this cross-sectional study was to gain an understanding of how diet

quality measured by the DASH score impacts on markers of cardiometabolic risk in

British police force employees. To achieve this aim the objectives of the study were to:

i) Measure the association between diet quality (determined by the DASH score)

and cardiometabolic risk in British police force employees.

ii) Identify employee characteristics associated with reporting a dietary pattern

associated with elevated cardiometabolic disease risk.

!

! ! *$#!

6.2 Methods

6.2.1 Participants

The sampling procedure is detailed in Chapter 2, Figure 2.2. For the purpose of this

present study participants diagnosed with a chronic disease were excluded from the

analyses if they reported a chronic disease diagnosis (angina, heart disease, angina,

chronic obstructive pulmonary disease, cancer, chronic liver disease, thyroid disease

and/or previous stroke) as these diseases may affect cardiometabolic markers of

interest. The analytical sample included for analyses was 5,527.

6.2.2 Dietary variables

Calculations of dietary variables are detailed in Chapter 5 (Section 5.2.2). Participants

were classified into five ascending diet quality groups based on quintile cut offs of

DASH score distribution.

6.2.3 Outcome measurements: markers of cardiometabolic risk

The data collection methodology for the biomarkers and anthropometric measurements

included in the present study are explained in detailed in Chapter 2. Biomarkers (HDL,

non-HDL cholesterol, blood pressures, CRP and HbA1c) and anthropometric markers

(BMI, waist circumference, percentage body fat) of cardiometabolic health were treated

as continuous and as categorical variables (where applicable) based on established

cut-off values associated with increased cardiometabolic risk (Chapter 2 Table 2.4).

The data collected as part of the Airwave Health Monitoring Study did not permit the

calculation of MetS based on the consensus definition (28) due to fasting glucose and

blood triglycerides not being collected. Therefore an amended MetS score was

calculated for each participant based on established cardiometabolic risk markers

shown in Table 6.1. The amended score included non-HDL and CRP to address the

limitations of the traditional MetS previously cited by The CardioMetabolic Health

Alliance (41). Participants with missing CRP (n = 112) were excluded from these

analyses.

6.2.4 Covariate measures

The socio-demographic, lifestyle and occupational variables used in this study were all

self-reported. The measurement of these variables is detailed in Chapter 2.

!

! ! *$$!

Table 6.1 Cut off values for cardiometabolic risk factor calculation

References for cut-off values: 1. International Diabetes Federation Task Force (28). 2. Non-HDL calculated as total cholesterol minus HDL cholesterol, NICE Guideline CG181 (299). 3. American Diabetes Association (54). 4. US Preventative Services Task Force (45) 6.2.5 Statistical methods Refer to Chapter 2, Section 2.7 for descriptive statistic procedures. Partial Spearman

rank correlation coefficients (sex and age adjusted) were calculated to determine

correlations among markers of cardiometabolic health, dietary variables and continuous

covariate measures. To assess the association between DASH score and biomarkers

of cardiometabolic risk general linear models were used to test linear trends (ptrend)

across quintiles of DASH score (via the contrast statement). Multivariate models made

sequential adjustment for established cofounding variables as listed in Table 6.2.

Confounding variables were selected a priori based on existing research (refer to

Chapter 1) that has shown these variables to have an association with diet quality and

cardiometabolic risk but are not suspected to be on the causal pathway (age, physical

activity, education, job strain, smoking, alcohol intake, energy intake). Additional

variables were selected if they made a significant contribution to the model and could

be theoretically defined as a confounder (TV viewing). Energy intake irregularity score

Risk marker At risk classification

Waist circumference1 Increased risk: !94cm (men); !80cm (women)

Blood lipids2

HDL: <1.0mmol/L (men); <1.3mmol/L (women) AND/OR Non HDL !4mmol/L AND/OR Reported diagnosed dyslipidaemia AND/OR On lipid lowering medication

Blood pressure1

Systolic !130mmHg and/or diastolic !85mmHg AND/OR Reported diagnosed hypertension AND/OR On hypotensive medication

Blood glucose3

HbA1c !5.7% AND/OR Reported diagnosed diabetes AND/OR On medication for glucose control Inflammation4

CRP !3mg/L <10mg/L

!

! ! *$%!

was not included, as it did not make a significant contribution to the model. Model 3

additionally adjusted for self-reported diagnosis or medication for blood pressure,

glucose and/or lipid management as these could potentially confound the relationship

with diet (potential dietary modifications made as a result of diagnoses) and with

cardiometabolic risk (e.g. better control of the cardiometabolic risk markers). BMI was

included in a separate model (Model 4) as it potentially lies on the causal pathway

between diet and markers of cardiometabolic health. Bonferroni adjustment was

applied to correct for multiple testing across quintile groups (corrected for 10 tests,

significance indicated at p <0.005)

Due to skewed distribution CRP was transformed using the logarithmic function to

obtain a normal distribution; the results were back transformed for results presentation.

Results are presented as adjusted mean and standard error (±SE).

Table 6.2 Model variables for general linear models investigating the relationship between diet

quality (DASH score) and markers of cardiometabolic risk Model Covariates included

Model 1 Age (continuous)

Model 2 As model 1 + categorical variables: physical activity, smoking status,

education, TV viewing, job strain, menopause status (women only); continuous variables: mean energy intake and mean alcohol g/day

Model 3 As Model 2 + diagnosed with hypertension, high cholesterol, T2DM and/or taking medication for blood pressure control, lipid control and/or for glucose control (categorical variables: yes/no)

Model 4 As Model 3 + Body mass index (continuous)

Sex stratified logistic regression was first conducted to estimate the odds of having

three or more markers of cardiometabolic risk per quintile of DASH score. Secondly, it

was used to investigate participant characteristics associated with consuming a

relatively poor quality dietary pattern (lowest DASH score group). Initially, bivariate

logistic regression was conducted to determine the relationship between poor diet

quality (yes, no) with each covariate. Then established confounders based on previous

studies (age, smoking status, household income, physical activity, marital status,

education, TV viewing, and region of employment) were included in multivariable

models. To allow adequate cases per strata work hours and education were collapsed

into three and four categories respectively and participants classified as ‘other’ police

!

! ! *$&!

rank (n =112) were excluded due to low frequency. Length of service in the police force

was not included as an explanatory variable due to the strong correlation with age (men

r = 0.63, women r = 0.48).

Sensitivity analyses were conducted by excluding those classed as under-reporting

energy intake (men = 1,880, women = 840; refer to Chapter 4 for classification

methodology). Due to the reduced sample size following removal of under-reporters of

energy intake, linear regression was conducted with DASH as a continuous variable

when measuring the association between diet quality and cardiometabolic risk.

6.3 Results

6.3.1 Descriptive statistics

For men and women there was a positive linear trend across DASH score groups and

age (ptrend <0.0001). Attained final education level was not significantly associated with

DASH score in women. Annual household income was only significantly associated

with DASH score amongst men, with those in the two lowest DASH score groups more

likely to have an income of <£32,000. Men and women employed in Scotland were

more likely to be in the lowest DASH score group compared to those employed in

England or Wales (men p <0.0001, women p =0.002). Standard weekly working hours

(35-40 hours) were associated with being in the highest DASH score group, while those

working more than 48 hours per week were less likely to be in the highest DASH score

group (p <0.0001). Men working shifts with nights were more likely to be in the lowest

group of lower DASH score compared to the highest group (70.5% vs. 62.0%). High

job strain was associated with being in the unhealthiest DASH score group for men and

women. Current smokers amongst men and women were more likely to be in the

lowest compared to the highest DASH score category (men 11.4% vs. 3.7% women; p

<0.0001; 21.4% vs. 4.6%; p >0.0001). Higher physical activity for both men and women

was associated with being in the healthiest DASH group. Duration of weekday sitting

and sleep were not associated with DASH score. Refer to Appendix A6.1 for tables

showing cohort characteristics across quintile groups of DASH score.

Partial Pearson correlations (controlled for sex and age) showed that DASH score had

the highest correlation with waist circumference and percentage body fat (r = -0.14) out

!

! ! *$'!

of all the cardiometabolic measures. However no moderate or strong associations (r

>0.30) between food, macronutrient intakes or DASH score with cardiometabolic

markers or physical activity were observed (Appendix A6.2).

6.3.3 Association between DASH score and markers of cardiometabolic health

In men DASH score was negatively associated with BMI, waist circumference, body fat,

HbA1c, non-HDL cholesterol, CRP and diastolic blood pressure and these associations

remained significant after adjustment for established confounders and total energy

intake, Table 6.3. Although the strength of the associations attenuated in Model 4

(adjusted for BMI) they still remained significant. The association was linear across

quintile groups. There was no association between systolic blood pressure and DASH

score. HDL cholesterol showed a positive relationship with DASH when the score was

kept continuous, but there was no association across quintile groups.

For women DASH was negatively associated with BMI, waist circumference, body fat,

non-HDL, diastolic blood pressure and C-reactive protein (Table 6.4). This association

did not remain significant for diastolic blood pressures or body fat after adjustment for

BMI. The relationship between DASH and systolic blood pressure did not remain

significant after adjustment for other lifestyle variables. There was no significant

association between DASH score with HbA1c or HDL cholesterol.

Further analyses excluding participants classified as under-reporting energy intake

showed an attenuation in significance of the relationship between DASH score and ratio

of total cholesterol to HDL (TC:HDL) for women. For men the negative association

between DASH score and BMI, waist circumference, body fat percentage and CRP

remained after removal of under-reporters. However, the associations between DASH

and HbA1c and HDL were no longer significant after adjusting for lifestyle factors

(Model 2), (Appendix A6.3.1 and A6.3.2).!!

Logistic regression showed a dose response relationship across quintiles of DASH

score and the odds of having three or more markers of cardiometabolic risk for men and

women (ptrend <0.0001), Tables 6.5 and 6.6. The association remained for men and

women in the lowest DASH score group with an OR of 1.84 (95%CI 1.43, 2.38) and

1.92 (95%CI 1.35, 2.84) respectively for recording three or more cardiometabolic risk

factors. There was an attenuation of the OR following further adjustment for BMI,

!

! ! *$(!

however the significant dose-response relationship across quintile groups of DASH

score and three or more metabolic risk factors was maintained. Effect size attenuated

in sensitivity analyses excluding under-reporters and after adjustment for BMI the

significant association between diet quality and high cardiometabolic risk was lost for

men (sensitivity analyses results presented in Appendix A6.3).

!!"#$!

Table 6.3 Association of D

AS

H score w

ith markers of cardiom

etabolic health in the Airw

ave Health M

onitoring Study: M

en (n =3,332)*

least healthy Q

uintile of DA

SH

score most healthy

Q1

Q2

Q3

Q4

Q5

P-trend

N (%

) 590

(17.7) 650

(19.5) 771

(23.0) 671

(20.1) 650

(19.5)

Adjusted m

ean (SE

) B

ody mass index, kg/m

2

Model 1

28.3 (0.1)

28.0 (0.1)

27.9 (0.1)

27.5 (0.1)

27.2 (0.1)

<0.0001 M

odel 2 28.3

(0.2) 28.1

(0.2) 28.0

(0.2) 27.6

(0.2) 27.3

(0.2) <0.0001

Model 3

30.1 (0.4)

30.0 (0.4)

29.9 (0.3)

29.5 (0.4)

29.2 (0.4)

<0.0001 W

aist circumference, cm

M

odel 1 95.7

(0.4) 95.0

(0.4) 94.1

(0.3) 92.8

(0.4) 91.8

(0.4) <0.0001

Model 2

96.5 (0.5)

96.1 (0.5)

95.4 (0.4)

94.2 (0.5)

93.5 (0.5)

<0.0001 M

odel 3 101.5

(0.9) 101.0

(0.9) 100.4

(0.9) 99.2

(0.9) 98.4

(0.9) <0.0001

Model 4

96.3 (0.5)

96.3 (0.5)

95.8 (0.5)

95.5 (0.5)

95.3 (0.5)

<0.0001 B

ody fat, %

M

odel 1 23.6

(0.2) 23.2

(0.2) 22.5

(0.2) 21.7

(0.2) 20.9

(0.2) <0.0001

Model 2

23.9 (0.3)

23.6 (0.2)

23.2 (0.2)

22.4 (0.3)

21.9 (0.3)

<0.0001 M

odel 3 26.1

(0.6) 25.8

(0.6) 25.5

(0.6) 24.6

(0.6) 24.1

(0.6) <0.0001

Model 4

23.1 (0.4)

23.2 (0.4)

22.8 (0.4)

22.5 (0.4)

22.3 (0.4)

<0.0001 H

bA1c, %

Model 1

5.65 (0.02)

5.65 (0.02)

5.63 (0.02)

5.59 (0.02)

5.55 (0.02)

<0.0001 M

odel 2 5.67

(0.03) 5.67

(0.02) 5.65

(0.02) 5.60

(0.02) 5.56

(0.03) <0.0001

Model 3

6.71 (0.05)

6.71 (0.05)

6.68 (0.05)

6.65 (0.05)

6.60 (0.05)

<0.0001 M

odel 4 6.66

(0.05) 6.65

(0.05) 6.63

(0.05) 6.61

(0.05) 6.56

(0.05) 0.0003

HD

L mm

ol/L

Model 1

1.35 (0.01)

1.36 (0.01)

1.39 (0.01)

1.37 (0.01)

1.38 (0.01)

0.06 M

odel 2 1.31

(0.01) 1.33

(0.01) 1.36

(0.01) 1.33

(0.01) 1.35

(0.02) 0.08

Model 3

1.23 (0.03)

1.25 (0.03)

1.27 (0.03)

1.25 (0.03)

1.27 (0.03)

0.08 M

odel 4 1.29

(0.03) 1.31

(0.03) 1.33

(0.03) 1.29

(0.03) 1.31

(0.03) 0.80

!!"%&!

Table 6.3

least healthy Quintile of D

AS

H score m

ost healthy

Continued (m

en)

Q1

Q2

Q3

Q4

Q5

P-trend

Non H

DL m

mol/L

M

odel 1 4.15

(0.04) 4.10

(0.04) 3.93

(0.03) 3.94

(0.04) 3.79

(0.04) <0.0001

Model 2

4.17 (0.04)

4.13 (0.05)

4.01 (0.05)

4.02 (0.05)

3.89 (0.05)

<0.0001 M

odel 3 3.55

(0.10) 3.52

(0.01) 3.39

(0.10) 3.39

(0.10) 3.27

(0.10) <0.0001

Model 4

3.44 (0.10)

3.42 (0.10)

3.29 (0.10)

3.31 (0.10)

3.20 (0.10)

<0.0001 TC

:HD

L ratio

Model 1

4.26 (0.04)

4.19 (0.04)

3.99 (0.04)

4.05 (0.04)

3.87 (0.04)

<0.0001 M

odel 2 4.38

(0.05) 4.30

(0.05) 4.14

(0.05) 4.21

(0.05) 4.05

(0.05) <0.0001

Model 3

4.05 (0.11)

4.00 (0.11)

3.81 (0.11)

3.88 (0.11)

3.79 (0.11)

<0.0001 M

odel 4 3.84

(0.11) 3.78

(0.11) 3.63

(0.01) 3.73

(0.11) 3.59

(0.11) <0.0001

Diastolic blood pressure m

mH

g

M

odel 1 83.1

(0.4) 82.3

(0.4) 81.8

(0.3) 80.9

(0.4) 80.3

(0.4) <0.0001

Model 2

83.0 (0.4)

82.4 (0.4)

82.3 (0.4)

81.5 (0.4)

81.2 (0.5)

0.0002 M

odel 3 83.7

(0.9) 83.1

(0.9) 83.1

(0.9) 82.2

(0.9) 81.9

(0.9) 0.0002

Model 4

82.0 (0.9)

81.5 (0.9)

81.6 (0.9)

80.9 (0.9)

80.9 (0.9)

0.021 S

ystolic blood pressure mm

Hg

Model 1

135.6 (0.6)

135.7 (0.5)

136.2 (0.5)

135.7 (0.5)

135.5 (0.5)

0.24 M

odel 2 136.4

(0.7) 135.6

(0.6) 136.2

(0.6) 135.9

(0.6) 135.9

(0.7) 0.74

Model 3

140.1 (1.4)

139.5 (1.3)

140.3 (1.3)

139.8 (1.4)

139.9 (1.4)

0.86 M

odel 4 138.2

(1.3) 137.8

(1.3) 138.6

(1.3) 138.4

(1.3) 138.7

(1.3) 0.37

CR

Pm

g/L±

M

odel 1 1.14

(0.93) 1.06

(0.93) 0.98

(0.93) 0.84

(0.93) 0.78

(0.93) <0.0001

Model 2

1.37 (0.95)

1.27 (0.95)

1.19 (0.95)

1.02 (0.95)

0.96

(0.95) <0.0001

Model 3

1.27 (1.04)

1.20 (1.04)

1.14 (1.04)

1.00 (1.04)

0.94 (1.04)

<0.0001 M

odel 4 1.04

(1.08) 0.99

(1.08) 0.96

(1.08) 0.86

(1.08) 0.83

(1.08) <0.0001

*Excluding participants w

ith self-reported chronic disease diagnosis: cancer, diseases of thyroid, chronic liver disease, angina, other heart, stroke and chronic obstructive pulm

onary disease. Abbreviations: H

DL H

igh density lipoprotein, TC total cholesterol, C

RP

high sensitivity C-reactive protein. ±C

RP

log transformed to

allow param

etric testing, untransformed values presented. M

odel 1 adjusted for age, Model 2 + physical activity, sm

oking, education, TV view

ing, job strain, mean

energy intake, alcohol; Model 3 + diagnosed ± treatm

ent for diabetes, lipids or blood pressure. Model 4 + body m

ass index (continuous).

!!"%"!

Table 6.4 Association of D

AS

H score w

ith markers of cardiom

etabolic health in the Airw

ave Health M

onitoring Study: W

omen

(n = 2,195)*

least healthy Q

uintile of DA

SH

score most healthy

Q1

Q2

Q3

Q4

Q5

P-trend

N (%

) 397

(18.0) 448

(20.0) 480

(22.0) 457

(21.0) 413

(19.0)

Adjusted m

ean (SE

) B

ody mass index kg/m

2

Model 1

26.4 (0.2)

26.2 (0.2)

25.9 (0.2)

25.6 (0.2)

25.1 (0.2)

<0.0001 M

odel 2 26.6

(0.3) 26.5

(0.3) 26.3

(0.3) 26.0

(0.3) 25.5

(0.3) 0.001

Model 3

28.6 (0.7)

28.5 (0.7)

28.3 (0.7)

28.0 (0.6)

27.6 (0.7)

0.001 W


M

odel 1 83.4

(0.5) 82.8

(0.5) 82.0

(0.5) 81.1

(0.5) 79.6

(0.5) <0.0001

Model 2

83.7 (0.7)

83.5 (0.7)

82.8 (0.7)

82.1 (0.7)

80.8 (0.7)

<0.0001 M

odel 3 88.9

(1.6) 88.6

(1.6) 88.0

(1.6) 87.3

(1.6) 85.9

(1.6) <0.0001

Model 4

83.2 (0.8)

83.1 (0.8)

82.9 (0.8)

82.8 (0.8)

82.4 (0.8)

0.019 B

ody fat percentage

M

odel 1 33.4

(0.4) 33.1

(0.3) 32.9

(0.3) 32.1

(0.3) 31.1

(0.4) <0.0001

Model 2

33.6 (0.4)

33.5 (0.4)

33.4 (0.5)

32.8 (0.5)

32.0 (0.5)

0.001 M

odel 3 35.8

(1.1) 35.6

(1.1) 35.5

(1.1) 34.9

(1.1) 34.1

(1.1) 0.001

Model 4

32.2 (0.6)

32.2 (0.6)

32.3 (0.6)

31.9 (0.6)

31.8 (0.6)

0.11 H

bA1c, %

Model 1

5.68 (0.03)

5.66 (0.02)

5.64 (0.02)

5.65 (0.02)

5.70 (0.02)

0.94 M

odel 2 5.71

(0.03) 5.69

(0.03) 5.66

(0.03) 5.67

(0.03) 5.70

(0.03) 0.75

Model 3

6.75 (0.07)

6.74 (0.07)

6.71 (0.07)

6.72 (0.07)

6.73 (0.07)

0.49 M

odel 4 6.72

(0.07) 6.71

(0.07) 6.68

(0.07) 6.70

(0.07) 6.72

(0.07) 0.75

HD

L, mm

ol/L

M

odel 1 1.70

(0.02) 1.70

(0.02) 1.74

(0.02) 1.73

(0.02) 1.71

(0.02) 0.42

Model 2

1.69 (0.02)

1.70 (0.02)

1.73 (0.02)

1.72 (0.02)

1.71 (0.03)

0.33 M

odel 3 1.59

(0.06) 1.59

(0.06) 1.63

(0.06) 1.62

(0.06) 1.61

(0.06) 0.30

Model 4

1.66 (0.60)

1.67 (0.05)

1.69 (0.05)

1.68 (0.06)

1.65 (0.05)

0.98

!!"%'!

Table6.4 continued least healthy Q

uintile of DA

SH

score most healthy

(w

omen)

Q1

Q2

Q3

Q4

Q5

P-trend

Non H

DL, m

mol/L

M

odel 1 3.46

(0.04) 3.40

(0.04) 3.40

(0.04) 3.33

(0.04) 3.21

(0.04) <0.0001

Model 2

3.59 (0.06)

3.54 (0.05)

3.56 (0.06)

3.47 (0.06)

3.36 (0.06)

0.001 M

odel 3 3.67

(0.13) 3.62

(0.13) 3.64

(0.13) 3.56

(0.13) 3.44

(0.13) 0.003

M

odel 4 3.55

(0.13) 3.51

(0.13) 3.54

(0.13) 3.47

(0.13) 3.37

(0.13) 0.004

TC:H

DL ratio

M

odel 1 3.18

(0.05) 3.20

(0.04) 3.11

(0.04) 3.03

(0.04) 3.00

(0.05) 0.001

Model 2

3.30 (0.06)

3.29 (0.06)

3.21 (0.06)

3.15 (0.06)

3.10 (0.06)

0.002 M

odel 3 3.43

(0.14) 3.44

(0.14) 3.36

(0.14) 3.31

(0.14) 3.25

(0.14) 0.002

Model 4

3.27 (0.14)

3.28 (0.14)

3.21 (0.14)

3.17 (0.14)

3.14 (0.14)

0.022 D

iastolic blood pressure mm

Hg

M

odel 1 77.5

(0.5) 77.4

(0.4) 76.8

(0.4) 76.7

(0.4) 74.6

(0.5) <0.0001

Model 2

77.9 (0.6)

78.3 (0.6)

77.7 (0.6)

77.8 (0.6)

75.9 (0.6)

0.005 M

odel 3 78.5

(1.4) 78.8

(1.4) 78.3

(1.4) 78.3

(1.4) 76.6

(1.4) 0.006

Model 4

76.7 (1.3)

77.1 (1.3)

76.7 (1.3)

76.9 (1.3)

75.4 (1.3)

0.08 S

ystolic blood pressure mm

Hg

M

odel 1 124.2

(0.7) 123.3

(0.6) 123.8

(0.6) 123.8

(0.6) 121.7

(0.7) 0.03

Model 2

125.9 (0.9)

125.5 (0.8)

126.1 (0.9)

126.3 (0.9)

124.4 (0.9)

0.35 M

odel 3 129.2

(2.0) 128.7

(2.0) 129.4

(2.0) 129.6

(2.0) 127.8

(2.0) 0.39

Model 4

126.9 (1.9)

126.5 (1.9)

127.4 (1.9)

127.7 (1.9)

126.4 (1.9)

0.91 C

RP

mg/L

±

Model 1

1.28 (0.95)

1.07 (0.94)

1.16 (0.94)

1.00 (0.94)

0.85 (0.95)

<0.0001 M

odel 2 1.53

(1.06) 1.32

(1.06) 1.43

(1.06) 1.25

(1.06) 1.10

(1.06) <0.0001

Model 3

1.82 (1.15)

1.57 (1.15)

1.71 (1.15)

1.49 (1.15)

1.31 (1.15)

<0.0001 M

odel 4 1.48

(1.14) 1.26

(1.14) 1.39

(1.14) 1.24

(1.14) 1.13

(1.14) <0.0001


ith self-reported chronic disease diagnosis: cancer, diseases of thyroid, chronic liver disease, angina, other heart, stroke and chronic obstructive pulm

onary disease. Abbreviations: H

DL H

igh density lipoprotein, TC total cholesterol, C

RP

high sensitivity C-reactive protein. ±C

RP

log transformed to

allow param

etric testing, untransformed values presented. M

odel 1: age, Model 2 + physical activity, sm


ing, job strain, menopause status,

mean energy intake, alcohol; M

odel 3 + diagnosed ± treatment for diabetes, lipids or blood pressure. M

odel 4 + body mass index (continuous)

!!"%(!

Table 6.5 Odds ratio of having three or m

ore markers of m

etabolic risk per quartile of DA

SH

score: men

M

inimal adjusted†

Fully adjusted§ Fully adjusted§ +B

MI

C

ases/N

OR

(95%

CI)

O

R

(95%C

I) O

R

(95%C

I)

Ref: Q

5 (healthiest) 235/636

1.00

1.00

1.00

Q4

265/658 1.24

(0.98, 1.57)

1.19

(0.93, 1.50)

1.10 (0.84,

1.44)

Q3

330/763 1.62

(1.29, 2.03)

1.54

(1.22, 1.92)

1.32 (1.02,

1.71)

Q2

271/640 1.72

(1.36, 2.19)

1.52

(1.19, 1.94)

1.22 (0.92,

1.60)

Q1 (unhealthiest)

256/581 2.16

(1.67, 2.77)

1.84

(1.43, 2.38)

1.50 (1.12,

2.00) p-trend

<0.0001

<0.0001

0.0006

A

bbreviations: BM

I body mass index; C

I confidence intervals; OR

Odds R

atio †M

inimal adjusted = age (n = 3,278; 54 participants did not have C

RP

available for metabolic risk calculation)

§Fully adjusted: age +physical activity, smoking status, education and TV

viewing, job strain, continuous variables: m

ean energy intake and mean alcohol g/day

Table 6.6 Odds ratio of having three or m

ore markers of m


SH

score: wom

en

M

inimal adjusted†

Fully adjusted§ Fully adjusted§ +B

MI

C

ases/N

OR

(95%

CI)

O

R

(95%C

I) O

R

(95%C

I)

Ref: Q


1.00

1.00

1.00

Q4

107/444 1.39

(0.98, 1.95)

1.35

(0.92, 1.92)

1.35 (0.91,

2.02)

Q3

126/473 1.85

(1.32, 2.60)

1.82

(1.29, 2.58)

1.74 (1.17,

2.57)

Q2

108/438 1.84

(1.30, 2.61)

1.72

(1.21, 2.48)

1.57 (1.04

2.35)

Q1 (unhealthiest)

103/378 2.21

(1.53, 3.19)

1.92

(1.35, 2.84)

1.84 (1.19

2.97)

p-trend

<0.0001

0.0004

0.005

Abbreviations: B

MI body m

ass index; CI confidence intervals; O

R O

dds Ratio

†Minim

al adjusted = age (n = 2,139; 56 participants did not have CR

P available for m

etabolic risk calculation) §Fully adjusted: age + physical activity, sm

oking status, education and TV view

ing, job strain, menopause; continuous variables: m

ean energy intake and mean

alcohol g/day

!

! "#$!

6.3.4 Participant characteristics associated with a poor diet quality Bivariate logistic regression for each potential explanatory characteristic did not find

household income to be associated with poor diet quality in women, and in men the

negative association observed in the highest income brackets did not remain significant

after adjustment. In men the highest category of education (bachelor degree or above)

was negatively associated with poor diet quality, and remained significant after

adjustment. For men and women region of employment and smoking status were

positively associated with poor diet quality (Scotland vs. England: men OR 1.88 95%CI

1.53, 2.32; women OR 1.49 95%CI 1.11, 2.00; current vs. never smoker men OR 1.90

95%CI 1.41, 2.58, women OR 3.35 95%CI 2.47, 4.55), Table 6.7. These associations

remained significant after adjustment for established confounders. Advancing age was

associated with reduced odds of having a poor diet (men OR 0.77, 95%CI 0.73, 0.82;

women OR 0.79 95% CI 0.74; 0.82 per 5 years) as was being in the highest category for

physical activity (men OR 0.58 95%CI 0.43, 0.78; women OR 0.51 95%CI 0.37, 0.72).

High and passive job strain (vs. low) was associated with increased odds of having a

poor diet for men, after adjustment only high job strain reminded significant (OR 1.38;

95%CI 1.06, 1.58). Working 49 hours or more per week was associated with poor diet

quality in men and women, and in men increased odds of a poor diet quality were also

observed with working between 41 and 49 hours per week. Following adjustment for

potential confounders, significance attenuated in women but not in men; compared to

working 40 hours or less per week those working !49 hours per week had OR 1.36

(95%CI 1.06, 1.75) of being classified as having a poor quality dietary pattern. For men,

but not women, being in the highest categories for weekly TV viewing hours was

associated with increased odds of a poor diet (moderate viewing OR 1.65 95%CI 1.29,

2.12; high viewing OR 1.74 95%CI 1.33, 2.28).

!!"##!

Ta

ble

6.7

Od

ds ra

tio (O

R) o

f be

ing

in th

e lo

we

st D

AS

H d

iet q

ua

lity g

rou

p fo

r me

n a

nd

wo

me

n

W

om

en

M

en

U

na

dju

ste

d†

A

dju

ste

d§

U

na

dju

ste

d^

Ad

juste

d¶

O

R

(95

%C

I) O

R

(95

%C

I) O

R

(95

%C

I) O

R

(95

%C

I)

Ag

e (p

er 5

ye

ar in

cre

as

e)

0.7

9

(0.7

4,

0.8

2)**

0.7

4

(0.6

8

0.8

0)**

0.7

7

(0.7

3,

0.8

2)**

0.7

6

(0.7

1,

0.8

1)**

Re

latio

ns

hip

sta

tus

R

ef: C

oh

ab

iting

1

.00

1

.00

1

.00

1

.00

D

ivo

rce

d/S

ep

ara

ted

/Oth

er

0.5

6

(0.3

6,

0.8

8)*

0.8

3

(0.4

9,

1.4

1)

0.7

1

(0.4

8,

1.0

6)*

1.0

0

(0.6

4,

1.5

6)

Ma

rried

0

.62

(0

.47

, 0

.82

)* 0

.91

(0

.66

, 1

.25

) 0

.58

(0

.46

, 0

.74

)** 0

.77

(0

.59

, 1

.03

)

Sin

gle

1

.09

(0

.79

, 1

.50

) 1

.09

(0

.75

, 1

.60

) 1

.17

(0

.82

, 1

.68

) 1

.17

(0

.78

, 1

.73

)

Ed

uc

atio

n

Re

f: GC

SE

or b

elo

w

1.0

0

1.0

0

1.0

0

1.0

0

Vo

ca

tion

al q

ua

lifica

tion

s

0.9

5

(0.6

7,

1.4

7)

0.6

7

(0.4

1,

1.0

8)

1.3

2

(0.9

4,

1.8

4)

1.1

0

(0.7

7,

1.5

6)

A le

ve

ls / H

igh

ers

or e

qu

iva

len

t 0

.92

(0

.71

, 1

.20

) 0

.68

(0

.50

, 0

.91

)* 1

.07

(0

.87

, 1

.32

) 0

.83

(0

.65

, 1

.04

)

Ba

ch

elo

r De

gre

e o

r hig

he

r 0

.68

(0

.51

, 0

.91

)* 0

.43

(0

.31

, 0

.60

)**±

0.7

4

(0.5

8,

0.9

4)*

0.5

3

(0.4

1,

0.7

0)**

An

nu

al h

ou

se

ho

ld in

co

me

R

ef: le

ss th

an

£3

2,0

00

1

.00

1

.00

1

.00

1

.00

£

32

,00

0 - £

47

,99

9

0.8

8

(0.6

0,

1.2

9)

1.0

0

(0.6

5,

1.5

3)

0.9

6

(0.6

6,

1.4

0)

1.0

7

(0.7

1,

1.6

1)

£4

8,0

00

- £5

7,9

99

0

.91

(0

.69

, 1

.20

) 1

.03

(0

.72

, 1

.44

) 0

.95

(0

.70

, 1

.29

) 1

.07

(0

.75

, 1

.53

)

£5

8,0

00

- £7

9,9

99

0

.89

(0

.64

, 1

.22

) 0

.94

(0

.63

, 1

.43

) 0

.69

(0

.49

, 0

.97

)* 0

.84

(0

.58

, 1

.23

)

Mo

re th

an

£7

8,0

00

0

.64

(0

.42

, 0

.97

)* 0

.90

(0

.55

, 1

.46

) 0

.64

(0

.42

, 0

.98

)* 0

.93

(0

.59

, 1

.48

)

To

tal h

ou

rs w

ork

ed

pe

r we

ek

Re

f: <4

0 h

ou

rs

1.0

0

1.0

0

1.0

0

1.0

0

>4

0 <

48

ho

urs

1

.04

(0

.80

, 1

.35

) 0

.97

(0

.72

, 1

.30

) 1

.37

(1

.11

, 1

.70

)* 1

.23

(0

.97

, 1

.53

)

> 4

9 h

ou

rs

1.5

3

(1.1

2,

2.0

9)*

1.3

3

(0.9

3,

1.9

1)

1.5

3

(1.2

1,

1.9

2)*

1.3

6

(1.0

6,

1.7

5)*

±

Em

plo

ym

en

t (forc

e) c

ou

ntry

Re

f: En

gla

nd

1

.00

1

.00

1

.00

1

.00

Sco

tlan

d

1.4

9

(1.1

1,

2.0

0)*

1.9

0

(1.3

8,

2.6

3)**

1.8

8

(1.5

3,

2.3

2)**

2.2

5

(1.8

0,

2.8

1)**

Wa

les

1.0

4

(0.7

3,

1.4

8)

1.0

0

(0.6

8,

1.4

6)

1.1

1

(0.8

3,

1.4

8)

1.2

4

(0.9

2,

1.6

9)

!!"#$!

Ta

ble

6.7

co

ntin

ue

d

Wo

me

n

Me

n

Un

ad

juste

d†

A

dju

ste

d§

U

na

dju

ste

d^

Ad

juste

d¶

O

R

(95

%C

I) O

R

(95

%C

I) O

R

(95

%C

I) O

R

(95

%C

I)

Ra

nk

R

ef: P

olic

e s

taff

1.0

0

1.0

0

1.0

0

1.0

0

Po

lice

Co

nsta

ble

/Se

rge

an

t 1

.10

(0

.87

, 1

.40

) 1

.04

(0

.79

, 1

.38

) 1

.22

(0

.95

, 1

.56

) 0

.95

(0

.71

, 1

.26

)

Insp

ecto

r/Ch

ief In

sp

ecto

r or a

bo

ve

0

.82

(0

.38

, 1

.78

) 1

.82

(0

.79

, 4

.26

) 0

.77

(0

.52

, 1

.14

) 0

.98

(0

.63

, 1

.53

)

Jo

b s

train

R

ef: L

ow

(hig

h c

on

trol, lo

w d

em

an

d)

1.0

0

1.0

0

1.0

0

1.0

0

Pa

ssiv

e (lo

w c

on

trol, lo

w d

em

an

d)

1.0

9

(0.8

1,

1.4

5)

1.0

1

(0.7

2,

1.4

2)

1.4

1

(1.0

8,

1.8

4)*

1.1

6

(0.8

8,

1.5

9)

Activ

e (h

igh

de

ma

nd

, hig

h c

on

trol)

0.8

4

(0.6

1,

1.1

6)

0.8

5

(0.6

0,

1.2

2)

1.2

7

(0.9

9,

1.6

1)

1.2

4

(0.9

7,

1.7

9)

Hig

h (h

igh

de

ma

nd

, low

co

ntro

l) 1

.34

(1

.00

, 1

.81

)* 1

.13

(0

.82

, 1

.57

) 1

.66

(1

.30

, 2

.12

)** 1

.38

(1

.06

, 1

.58

)*

Wo

rk e

nv

iron

me

nt

Re

f: Ma

inly

offic

e d

utie

s

1.0

0

1.0

0

1.0

0

1.0

0

Ma

inly

mo

bile

du

ties

1.2

6

(0.9

6,

1.6

7)

0.8

3

(0.6

0,

1.1

5)

1.7

1

(1.3

8,

2.1

2)**

1.1

6

(0.9

1,

1.4

7)

Ph

ys

ica

l ac

tivity

/ we

ek

R

ef: L

ow

1

.00

1

.00

1

.00

1

.00

M

od

era

te

0.8

6

(0.6

3,

1.1

6)

0.8

0

(0.5

8,

1.1

0)

1.0

2

(0.7

6,

1.3

5)

1.0

4

(0.7

7,

1.4

0)

Hig

h

0.5

1

(0.3

7,

0.7

2)**

0.4

8

(0.3

4,

0.6

9) **

0.5

8

(0.4

3,

0.7

8)*

0.6

0

(0.4

4,

0.8

2)*

Sm

ok

ing

sta

tus

R

ef: N

eve

r sm

oke

r 1

.00

1

.00

1

.00

1

.00

F

orm

er s

mo

ke

r 1

.29

(0

.98

, 1

.69

) 1

.37

(1

.02

, 1

.83

)* 0

.85

(0

.67

, 1

.06

) 1

.00

(0

.79

, 1

.27

)

Cu

rren

t sm

oke

r 3

.35

(2

.47

, 4

.55

)** 3

.07

(2

.19

, 4

.29

)** ±

1.9

0

(1.4

1,

2.5

8)**

1.7

1

(1.2

3,

2.3

7)*

Sle

ep

5

ho

urs

or le

ss

1.0

0

1.0

0

1.0

0

1.0

0

6 h

ou

rs

0.9

8

(0.6

0,

1.6

1)

0.9

2

(0.5

4,

1.5

8)

0.8

0

(0.5

4,

1.2

0)

0.8

1

(0.5

2,

1.2

4)

7 h

ou

rs

0.9

8

(0.6

1,

1.5

5)

0.9

9

(0.6

0,

1.6

4)

0.7

4

(0.5

0,

1.0

9)

0.6

9

(0.4

5,

1.0

6)

8 h

ou

rs

0.9

8

(0.6

1,

1.5

9)

0.8

1

(0.4

8,

1.3

8)

0.7

3

(0.4

8,

1.1

2)

0.6

9

(0.4

4,

1.0

9)

9 h

ou

r or m

ore

1

.92

(1

.04

, 3

.56

)* 1

.60

(0

.82

, 3

.13

) 0

.85

(0

.44

, 1

.61

) 0

.61

(0

.30

, 1

.23

)

We

ek

ly h

ou

rs s

itting

(we

ek

da

ys

)

R

ef: L

ow

(<2

0 h

ou

rs)

1.0

0

1.0

0

1.0

0

1.0

0

Mo

de

rate

(20

– 4

0 h

ou

rs)

0.9

5

(0.7

3

1.2

3)

1.0

5

(0.7

8,

1.3

9)

1.1

1

(0.8

9,

1.3

7)*

1.0

8

(0.8

6,

1.3

5)

Hig

h (>

40

ho

urs

) 1

.03

(0

.78

1

.36

) 1

.18

(0

.86

, 1

.62

) 0

.90

(0

.71

, 1

.13

)* 0

.87

(0

.68

, 1

.12

)

!!"#%!

Ta

ble

6.7

co

ntin

ue

d

Wo

me

n

Me

n

U

na

dju

ste

d†

A

dju

ste

d§

U

na

dju

ste

d^

Ad

juste

d¶

O

R

(95

%C

I) O

R

(95

%C

I) O

R

(95

%C

I) O

R

(95

%C

I)

TV

vie

win

g, h

ou

r pe

r we

ek

Re

f: Lo

w (<

6 h

ou

rs)

1.0

0

1.0

0

1.0

0

1.0

0

Mo

de

rate

(6 –

15

ho

urs

) 1

.00

(0

.78

, 1

.28

) 1

.11

(0

.85

, 1

.46

) 1

.51

(1

.20

, 1

.92

)* 1

.65

(1

.29

, 2

.12

)*

Hig

h (>

15

ho

urs

) 1

.17

(0

.87

, 1

.57

) 1

.27

(0

.92

, 1

.76

) 1

.57

(1

.22

, 2

.02

)* 1

.74

(1

.33

, 2

.28

)** ±

Cla

ss

ified

as

un

de

r-rep

orte

r EI

Re

f: Ye

s

1.0

0

1.0

0

1.0

0

1.0

0

No

1

.09

(0

.87

, 1

.38

) 1

.14

(0

.88

, 1

.48

) 1

.20

(1

.00

, 1

.44

)* 1

.22

(0

.99

, 1

.62

)

Me

no

pa

us

e

Re

f: No

1

.00

1

.00

-

-

Ye

s

0.4

6

(0.2

9,

0.7

2)**

1.1

0

(0.6

4,

1.9

0)

- -

- -

- -

Do

n't k

no

w

0.7

9

(0.5

0,

1.2

5)

0.9

9

(0.5

7,

1.7

4)

- -

- -

- -

Ab

bre

via

tion

s: G

CS

E G

en

era

l Ce

rtifica

te o

f Se

co

nd

ary

Ed

uca

tion

; EI: e

ne

rgy in

take

Wo

me

n: †3

97

/2,1

95

cla

ssifie

d a

s re

co

din

g a

‘po

or d

iet q

ua

lity’, w

ith e

xce

ptio

n o

f ran

k/w

ork

en

viro

nm

en

t 34

2/1

,86

1 d

ue

to m

issin

g d

ata

. §A

dju

ste

d fo

r ag

e, e

du

ca

tion

, ho

use

ho

ld in

co

me

, ma

rital s

tatu

s, p

hysic

al a

ctiv

ity, re

gio

n, s

mo

kin

g s

tatu

s, a

nd

da

ily T

V v

iew

ing

; 36

9/2

,09

9 c

lassifie

d

as re

co

rdin

g a

‘po

or d

iet q

ua

lity’ d

ue

to m

issin

g d

ata

, with

exce

ptio

n o

f ran

k/w

ork

en

viro

nm

en

t 32

1/1

,78

8 d

ue

to m

issin

g d

ata

.

Me

n: ^5

90

/3,3

32

cla

ssifie

d a

s re

co

rdin

g a

‘po

or d

iet q

ua

lity’, w

ith e

xce

ptio

n o

f ran

k/w

ork

en

viro

nm

en

t 52

7/2

,86

5 d

ue

to m

issin

g d

ata

. ¶A

dju

ste

d fo

r

ag

e, e

du

ca

tion

, ho

use

ho

ld in

co

me

, ma

rital s

tatu

s, p

hysic

al a

ctiv

ity, re

gio

n, s

mo

kin

g s

tatu

s a

nd

da

ily T

V v

iew

ing

, 57

7/3

,26

7 c

lassifie

d a

s re

co

rdin

g a

‘po

or d

iet q

ua

lity’ d

ue

to m

issin

g d

ata

, with

exce

ptio

n o

f ran

k/w

ork

en

viro

nm

en

t 51

8/2

,81

4 d

ue

to m

issin

g d

ata

.

*p<

0.0

5, ** p

<0

.00

01

±Ptrend a

cro

ss s

trata

<0

.05

!

! *%(!

6.4 Discussion This study has described the characteristics of British police force employees against

ascending groups of DASH diet quality scores, a score negatively associated with

increased cardiometabolic risk in mainly US cohorts. This study has tested the

association between DASH score and markers of cardiometabolic risk within British

police force employees. Lastly this study has identified police employee characteristics

associated with reporting a dietary pattern associated with elevated cardiometabolic

disease risk.


• Diet quality classified by the DASH diet score was negativity associated with

measures of adiposity, inflammation, diastolic blood pressure and non-HDL

cholesterol British police force employees.

• Men and women in the lowest fifth of DASH score distribution had almost twice

the odds of having three or more markers of cardiometabolic risk independent of

other lifestyle factors and BMI.

• A poorer diet quality was associated with other negative lifestyle behaviours

(inactivity, smoking and TV viewing).

• For men long working hours and high job strain were associated with increased

odds of having a diet quality associated with increased cardiometabolic risk

(independent of established predictors of diet quality).


Objective i) To measure the association between diet quality (determined by the DASH

score) and cardiometabolic risk in British police force employees.

The present study observed that the DASH score when applied to the Airwave Health

Monitoring Study cohort to have a negative linear association with measures of

adiposity (BMI, waist circumference and body fat), non-HDL cholesterol and CRP

(marker of inflammation) for men and women, independent of mean energy intake, age

and activity levels. Additionally with the exception of body fat and diastolic blood

pressure in women these significant associations remained after adjustment for BMI.

These findings are in agreement with previous observational studies showing

!

! *%)!

adherence to the principles of the DASH dietary pattern to be associated with a

reduction in long-term weight gain (107) and reduced incidence of CVD (300).

Observational evidence demonstrating a beneficial effect of DASH on inflammation is

lacking however, an eight-week RCT reported reduction in CRP (-26.9 ±3.5%) following

a DASH dietary intervention in diabetic patients (301).

Although there was a significant negative association between DASH and non-LDL

cholesterol this study did not observe any significant association between HDL and

DASH score. This is in agreement with a meta-analysis of 15 RCTs investigating

DASH and markers of cardiometabolic risk. Pooled analyses showed an overall mean

difference 0.1mmol/L (95%CI -2.0, 2.1) in HDL and -4.0mmol/L (95%CI -7.7, -0.3) in

LDL (100). The lack of relationship between DASH and HDL could be due to PUFA

intake not being positively associated with DASH score in the Airwave Health

Monitoring Study (Appendix A5.2), as studies have suggested PUFA intakes have a

greater effect on HDL compared to MUFA (302).

Despite DASH being developed to manage hypertension there was no significant

relationship between DASH score and systolic blood pressure. This finding contradicts

previous observational (303) and RCTs (100). One possible explanation is the

accuracy of sodium intake measurements. Although in the Airwave Health Monitoring

Study cohort table and cooking salt usage showed a significant negative association

with DASH score (Appendix A5.2), the sodium content of processed/manufactured

foods can be difficult to estimate due to the inherent limitations of nutritional databases

(304) potentially leading to sodium score misclassification in DASH. Additionally, blood

pressure measurements can vary diurnally and in the repeated measurement of blood

pressure in a cohort sub-sample there were significant differences between initial and

repeat measurements (Table 2.3).

Although previous studies have showed a negative association between DASH score

and T2DM incidence (300) the current study did not find a consistent relationship with

HbA1c. In women there was no relationship between DASH and HbA1c in any

analyses, and in men significance attenuated after removing under-reporters. This

latter observation may be the result in bias towards a healthy BMI as under-reporting

was strongly associated with being overweight and obese. A potential reason for the

lack of association is the cross-sectional nature of the current study, as the Airwave

!

! *&+!

Health Monitoring Study is a relatively young cohort so cumulative effects of dietary

exposure may not yet manifest in glucose metabolism abnormalities. Moreover,

although HbA1c is established for diagnosis for T2DM, it’s utility over an oral glucose

tolerance test (OGTT) in determining pre-diabetes has been questioned (305). It has

been suggested that HbA1c can be recorded in an acceptable ‘healthy’ range (<5.6%)

while OGTT results show abnormal glucose metabolism (305). A further explanation

may relate to alcohol consumption, with large prospective studies showing a u-shaped

curve between alcohol intake and T2DM (306,307). In the current study alcohol had a

negative linear relationship with DASH score (Appendix A5.2).

In the present study, classification of having a poorer quality dietary pattern (being in

the lowest fifth for DASH score) was characterised by a total fat, saturated fat and NME

intake above UK dietary guidelines, high intakes of alcohol, and less than 2 portions of

fruit or vegetables per day (refer to Appendix A5.2 for detailed nutritional profile by

quintile of DASH score). The present study found that those in the lowest quintile of

DASH score had over double the odds of having three or more metabolic risk markers

compared to those in the highest DASH score category (age adjusted). The

relationship across DASH score groups showed a dose-response with only slight

attenuation in the strength of association after adjusting for other lifestyle factors

(smoking, physical activity, TV viewing, energy intake) and BMI. Although the

cardiometabolic scoring system used in the present study is not previously validated in

predicting future cardiometabolic disease, it contains individual markers of risk

previously independently associated with cardiometabolic risk. The observation that

DASH score was negatively associated with having three or more cardiometabolic risk

markers is in agreement with a previous study in Iranian nurses that found participants

in the highest DASH score tertile had 81% lower odds of MetS (308).

Objective ii) To identify employee characteristics associated with reporting a dietary

pattern associated with elevated cardiometabolic disease risk.

Previous studies have suggested low socio-economic status and education level to be

a predictor of poor diet quality (248,309). Multivariate analyses in the present study did

not demonstrate an association between education and household income (proxy

determinants of socio-economic status) with poor diet quality in police force employees,

!

! *&*!

with the exception of the highest income bracket for men. A possible explanation may

be because studying a single occupational cohort limits the influence of socioeconomic

status on diet differences that may be observed in studies across a wider socio-

economic spread. Employment in the Scottish police force was associated with

increased odds compared to employment in England of being classified as having a

poor dietary pattern. This supports the findings reported in Chapter 5 and previous

observations in the general UK population which found Scottish residents to have diets

higher in saturated fat, sodium and lower in vegetables (295). The association between

DASH score and low physical activity and current smoking status are suggestive of

overall clustering of healthy lifestyle behaviours, which has been reported previously

(310). Similarly a small cross-sectional study in police officers in Pennsylvania, USA (n

= 247) observed that fruit and vegetable intake was positively associated with physical

activity (311). In the present study TV-viewing, but not weekday sitting, was associated

with a poor quality dietary pattern; however, after adjustment this observation only

remained significant in men. The Nurses’ Health Study (female only cohort) found

higher rates of weekly TV viewing to be associated with a diet high in energy, saturated

fats, snacks and sweets (140). Advancing age has also been previously shown to be

related to healthier diet quality scores (310). In The Food Choice at Work Study in

Ireland found employees with higher nutritional knowledge had a higher DASH score

(312). In the current study participants reporting a weight loss diet were less likely to

record a poor diet quality, which may suggest increased knowledge or awareness of

nutrition.

Compared to working 40 hours or less per week, working longer hours in men was

associated with poor quality dietary pattern, again supporting observations reported in

Chapter 5. For women part-time work was negatively associated with recoding a poor

diet. High compared to low job strain remained a significant predictor of poor diet

quality in men after adjustment for known confounders. Previous research has shown

psychological job demands to be positively associated with high fat food intakes in men,

but not women (136). The observations from the present study relating to these

occupational factors should be interpreted with caution as sensitivity analyses,

excluding those classified as under reporting dietary energy, showed an attenuation of

significance for high job strain (OR 1.24, 95%CI 0.85, 1.81) and work hours (OR 1.32,

95%CI 0.92, 1.90) in adjusted logistic regression analyses. The loss of significance may

!

! *&"!

in part, be due to a reduction in statistical power due to loss of sample size (~50%).

Additionally, the odds of being classified as under-reporting energy intake were

significantly associated with long working hours (Chapter 4). Nevertheless the findings

of the current study are of potential occupational health importance as research has

shown both job strain and long working hours to be associated with higher risk of

adverse cardiometabolic health (15, 205).

6.4.3 Study strengths and limitations The strength of the present study is the large-scale collection of 7-day food records

from a single occupational group and the standard coding protocol used for dietary

coding (detailed in Chapter 3). The DASH score is based mainly on food group intakes,

so therefore reduces the reliance on nutritional databases to estimate nutrient intakes

(313).

This study has a number of limitations. The DASH score is calculated based on median

cohort specific intake values and therefore it is not directly comparable to other cohorts.

However, this study has demonstrated that within this cohort it reflects overall diet

quality with those in the lowest fifth consuming a diet that is of a higher energy density,

exceeds fat intake recommendations as well as being low in fruits, vegetables and

whole grains. A further limitation of the current study was missing data on specific

variables in particular shift work information was currently only available for 30% of the

cohort. Although chi-squared analyses in men showed an association between the

lowest DASH score group and shift work with nights (and conversely, working days and

being in the healthiest DASH group). Shift work exposure could not be analysed in the

logistic regression models due to the low sample size. In the Airwave Health Monitoring

Study shift work prevalence is strongly associated with job role and rank: employees

with predominately mobile duties are more likely to be constables or sergeants and

undertake shift work. Therefore, in the current study it was difficult to determine the

individual potential occupational factors associated with diet quality due to

multicollinearity and missing data. There was a borderline association between

plausible energy reporting in men and classification as having a poor diet quality.

However, removing under-reporters did not change the significance of associations

between measures of adiposity and inflammation.

!

! *&#!

A final consideration is that the blood lipid measurements were taken using non-fasting

serum samples. Exploratory analyses including ‘hours since consuming anything

except water’ (continuous variable) did not make a significant contribution to the model

for HDL, non-LDL or TC:HDL. It is important to note that from 2014 the National

Institute for Health and Care Excellence recommended non-fasting serum to be the

standard method of lipid measurement to screen for cardiovascular risk (299).

6.5 Study conclusions and relevance to further studies

In conclusion this study has measured the dietary pattern quality of British police force

employees. The DASH score when applied to the Airwave Health Monitoring study was

negatively associated with markers of increased cardiometabolic risk (adiposity,

inflammation and non-HDL cholesterol) independent of other lifestyle factors. Being in

the lowest fifth of DASH score was associated with almost double the odds of having

three or more markers of metabolic risk. For male police force employees the

observations reported here suggest that working long hours and higher job strain are

associated with reporting a diet quality indicative of increased cardiometabolic risk.

This study also suggests that the DASH diet may potentially provide a beneficial dietary

intervention in British police employees to improve cardiometabolic health.

!

! *&$!

CHAPTER 7

7.0 STUDY 4: WORKING HOURS AND CARDIOMETABOLIC RISK IN BRITISH POLICE FORCE EMPLOYEES 7.1 Introduction

7.1.1 Background and study rationale As discussed in Chapter 1 (Section 1.7) previous research has shown that employees

working longer hours compared to standard hours (35 – 40 hours per week) or those

working shifts are at increased cardiometabolic risk. The European Working Time

Directive (2003/88/EC) aimed to limit working hours to an average of 48 hours per week

(calculated across a 17-week period). In the UK emergency service employees,

including many police force employees, are exempt from the directive (171). In 2014

the proportion of UK employees working over 40 hours and 48 hours per week was

44% and 13% respectively (172).

The observations from the Airwave Health Monitoring Study Cohort presented in this

thesis have shown poorer diet quality to be associated with longer working hours (Study

2, Chapter 5), and working long hours was predictive of reporting a diet associated with

increased chance of having three or more cardiometabolic risk markers for men (Study

3, Chapter 6). To the author’s best knowledge the contribution of diet to

cardiometabolic risk factors across length of weekly working hours has not been

investigated.

7.1.1 Study aims and objectives The overall aims of this cross-sectional study were to gain an understanding of the

influence that the number of weekly working hours has on cardiometabolic risk in British

police force employees and to investigate if diet modifies the risk.

To achieve the study aims the objectives of the study were to:

i) Describe the cardiometabolic health profile of British police force employees.

ii) Investigate the association between long working hours and markers of

cardiometabolic disease risk in British police force employees (and in a sub-

group with shift work data).

!

! *&%!

iii) Establish if diet quality (measured by DASH score) modifies the relationship

between long working hours and markers of cardiometabolic disease risk in

British police force employees.

7.2 Methods

7.2.1 Participants The sample selection is detailed in Chapter 2, Figure 2.2. Exclusion criteria for the

present study were i) part-time work (<35hrs per week) as police work may not be only

source of employment and/or ii) self-reported chronic disease diagnosis (angina, heart

disease, angina, chronic obstructive pulmonary disease (COPD), cancer, chronic liver

disease, thyroid disease and/or history of stroke) as these diseases may effect

metabolic markers of interest. Appendix A7.1 compares those reporting part-time work

and/or chronic disease diagnosis with the rest of the cohort. The final analytical sample

used in the present study was 5,036.

7.2.2 Working hour exposure measures The primary independent variable for this study is self-reported usual weekly working

hours (regular hours and regular overtime hours), (refer to Chapter 2). Participants

were classified into groups based on previous large scale studies (35-40, 41–48, 49–54

hours, and !55 hours per week) (15,169). As a low number of women worked in the

highest working hour groups (49–54 hours/week = 8.1%; and >55 hours per week =

7.6%) these groups were collapsed in to one group. Due to the frequency of reported

part-time work in female employees (20%) additional analyses were conducted for

female employees that included part-time workers.

7.2.3 Outcome measurements of cardiometabolic risk

The data collection methodology for the biomarkers and anthropometric measurements

used in this study are explained in Chapter 2. Biomarkers (HDL, non-HDL cholesterol,

blood pressures, CRP and HbA1c) and anthropometric markers (BMI, waist

circumference, percentage body fat) of cardiometabolic health were treated as

continuous and as categorical variables based on established cut-off values associated

with increased cardiometabolic risk (Table 2.4). To determine the combinations of

cardiometabolic risk factors that participants had dichotomous variables were created

(yes/no) for the cardiometabolic risk factors shown in Chapter 6, Table 6.1.

!

! *&&!

7.2.4 Dietary variables Chapter 6 demonstrated that dietary pattern quality determined by the DASH score was

associated with key cardiometabolic risk markers. Calculations of dietary variables are

detailed in Chapter 5 (Section 5.2.2).

7.2.5 Covariate measurements The collection of these variables is detailed in Chapter 2.

7.2.6 Statistical methods Refer to Chapter 2, Section 2.7 for descriptive statistic procedures. General linear

models tested the linearity of the relationship across the groups of working hours (ptrend)

and markers of cardiometabolic risk. Bonferroni post hoc test was applied to correct for

multiple comparisons and determine the source of differences between groups (alpha

level corrected for three tests across three groups and six tests across four groups).

Sequential analyses were conducted to assess the effect of covariate adjustment on the

association between working hours and markers of cardiometabolic health (Table 7.1).

Covariates were selected for inclusion into the models by either i) an observed

significant statistical association with both the independent variable and dependent

variable under investigation (and plausibly classified as a confounder) or ii) a priori

based on an association determined in previous cohort studies (refer to Chapter 1).

BMI was included in a separate model (Model 5) as it potentially lies on the causal

pathway between work hours and markers of cardiometabolic health. Type III sum of

squares was used to measure the effect of one variable adjusted for the confounders in

the regression model. Results are presented as adjusted mean ±SE.

!

! *&'!

Table 7.1 Model variables in general linear models investigating the association between

weekly number of working hours and markers of cardiometabolic risk Model Covariates included

Model 1 Age (continuous)

Model 2 As model 1 + categorical variables: physical activity, smoking status, education, household income, TV viewing, job strain (+ menopause for women)

Model 3 As Model 2 + DASH score, mean energy intake

Model 4 As Model 3 + diagnosed with hypertension, high cholesterol, T2DM and/or taking medication for blood pressure control, lipid control and/or for glucose control (categorical variables: yes/no) Model 5 As Model 4 + Body mass index (continuous)

Diet quality was significantly associated with length of weekly working hours (Chapter 5

and Chapter 6) and was therefore hypothesised to be a potential moderating variable.

Baron and Kenny define a moderating variable as a variable that changes the

magnitude and/or direction of the association between the independent and dependent

variable (314). To test the hypothesis that diet quality (categorical: DASH score group)

modified the association between length of working hours (categorical: working hours

group) and cardiometabolic disease risk Model 3 was repeated with a cross-over

interaction term (314), Figure 7.1.

Figure 7.1 Model of effect modification

Based on the model of effect modification in Figure 7.1: path a is the association

observed between working hours (predictor variable) and the cardiometabolic risk

marker of interest, and path b is the observed association between diet quality

(potential moderating variable) and the cardiometabolic risk marker of interest. Path c

!"#$%&' (")#*

+,-. */"#0

!"#$%&' (")# '#")12

+,-. */"#0 '#")1

34#5%"60748"9%/ #%*$ 604*)#0

!

"

#

!

! *&(!

signifies the association between the interaction of diet quality and working hours with

the cardiometabolic risk marker of interest. If path c is shows a statistically significant

(p <0.05) association then there is an interaction between the predictor and moderating

variables (314).

Sensitivity analyses were conducted by excluding those classed as under-reporting

energy intake (men = 1,880, women = 840; refer to Chapter 4 for classification

methodology) and repeating general linear models testing the relationship between

number of weekly working hours and cardiometabolic risk. Based on the findings

reported in Study 2 (Chapter 5) which showed significant diet quality differences across

working hours in mid ranked male police officers additional sensitivity analyses were

conducted removing non- and higher ranked officers in men. Sub group analyses were

conducted in participants with shift work data available (men and women were

combined to increase study power - therefore all models were adjusted for sex).

7.3 Results

7.3.1 Descriptive statistics The mean age of the sample included in the cross-sectional study was 41.1 years (SD

9.2), 65.0 % (n = 3,280) of the sample were male. Men worked significantly longer

hours per week compared to women (p <0.0001). For men the mean number of weekly

hours worked were 42.2 SD 5.4 hrs/excluding overtime and 3.1 SD 4.2 hrs/overtime, for

women the values were 40.5 SD 5.2 hrs and 2.0 SD 3.2 hrs respectively. From the

sample 26.5% of men and 15.7% women worked more than 48 hours per week. Refer

to Appendix 7.2 for summary characteristics.

Men compared to women were more likely to have a BMI classified as being obese

(23.0% vs. 15.5%), while women were more likely to have a waist circumference in the

highest risk category (26.0% vs. 18.2%) (Appendix A7.3). From the sample 3.3% of

participants had an HbA1c value to meet T2DM criteria and 0.8% reported being

diagnosed and/or on medication for T2DM. Women were less likely to have a blood

pressure reading in the hypertensive range (14.4% vs. 38.9%). Men were more likely to

have three or more markers of cardiometabolic risk compared to women (41.3 vs.

25.1%). In participants that had three or more markers of cardiometabolic risk the most

!

! *&)!

common combination was dyslipidaemia, elevated blood pressure and elevated waist

circumference (26%), Figure 7.1. For women the most frequently observed

combination was elevated waist circumference, HbA1c and blood pressure (16.8%).

!!"#$!

Figure 7.2 Prevalence of cardiom

etabolic phenotypes for men and w

omen w

ith three or more cardiom

etabolic risk factors1

0.0

5.0

10.0

15.0

20.0

25.0

30.0

All

Blood pressure + CRP + lipid

Blood pressure + CRP + lipid +HbA1c

Blood pressure + CRP+ HbA1c

Blood pressure + lipid + HbA1c CRP + lipid +HbA1c

Waist circ. + blood pressure + CRP

Waist circ. + blood pressure + CRP + HbA1c

Waist circ. + blood pressure + lipid

Waist circ. + blood pressure + lipid + HbA1c

Waist circ. + blood pressure +HbA1c Waist circ. + CRP + HbA1c

Waist circ. + CRP + lipid

Waist circ. + CRP + lipid + HbA1c Waist circ. + lipid + HbA1c

Proportion (%)

men

wom

en

A

bbreviations: CR

P high sensitivity C

-Reactive protein. H

bA1c glycated haem

oglobin, 1E

levated risk classification for waist circum

ference, HbA

1c (or T2DM

diagnosis/treatm

ent), blood pressure (or hypertension diagnoses/treatment), Lipid: H

DL and/or non-H

DL risk (or dyslipidaem

ia diagnosis/treatment), elevated C

RP

. 1M

en (n = 1,332), wom

en (n = 420)

!

! "#"!

7.3.2 Characteristics across working hour groups Across men and women a higher number of weekly working hours were associated with

being younger, with men working 55 hours or more a week being on average 3.2 years

younger than those working standard hours, this figure was 3.7 years for women. Men in

the highest category of weekly working hours were more likely to have achieved a higher

level of education (p <0.0001), for women there was no association between number of

working hours and education (p = 0.32). Men and women working 55 hours or more a

week were more likely to be in the highest five categories for household income

compared to those working standard hours (p <0.0001). For both genders those working

standard hours (35-40 hours per week) were more likely to be employed as police staff,

whilst those working the highest number of weekly hours category were most likely to be

police sergeants or constables (p <0.001). For women working standard hours was

significantly associated with reporting an office based job role compared to the highest

category of working hours (40.1% vs. 34.0%, p <0.0001). Men working 55 hours or more

per week were more likely to have a mobile job role compared to those working standard

working hours (49.0 vs. 41.4%, p <0.0001). There was no association between number

of working hours and sleep, physical activity, smoking status, or drinking status across

men or women. Men and women working 55 hours or more per week were less likely to

be in the highest classification for weekly TV viewing (men 32.9% vs. 35.1% p = 0.013;

women 26.6% vs. 36.3%, p <0.0001). No association was observed between weekday

sitting for men and number of working hours, however women working 55 or more hours

per week compared to those working standard hours were more likely to be in the lowest

category for hours sitting per week (44.4% vs. 33.0%, p = 0.004). For men employees

working standard hours were more likely to report being diagnosed or on treatment for

hypertension compared to those in the highest category of working hours (8.9% vs. 6.7%,

p = 0.001). Refer to Appendix A7.4 for comparison of participant characteristics across

working hour groups.

7.3.3 Working hours and markers of cardiometabolic health

The linear associations between usual weekly working hours and cardiometabolic risk

markers are presented in Tables 7.2 and 7.3 for men and women respectively. For men

there was a significant positive association between BMI and number of working hours

(ptrend <0.0001). The association did not change after adjustment for demographic, socio-

economic and lifestyle confounders. There was also a positive association between

number of weekly working hours with waist circumference, CRP and percentage body fat

!

!!!

"#$!

for men, however significance attenuated in Model 4 which was adjusted for BMI.

Sensitivity analyses (excluding non ranked and higher rank employees) only attenuated

the association between CRP and working hours in Model 3 onwards (dietary

adjustment), with no significant interaction between DASH score and long hours

observed (Appendix A7.5). A second set of sensitivity analyses removing participants

classified as under-reporting energy intake did not change the significance of any of the

relationships observed with the exception of waist circumference, that no longer showed

a significant linear relationship across working hours in Model 5 (adjusted for BMI),

Appendix A7.6.

In women there was a positive association across working hours and HDL, and negative

association across ratio TC:HDL. Sensitivity analyses (including part-time workers)

showed comparable results, however these analyses also found women working part-

time to have a significantly lower BMI compared to those working over 49 hours per week

(Appendix A7.7). These associations did not remain significant in further sensitivity

analyses (excluding energy intake under-reporters), Appendix A7.6.

There were no significant interactions between DASH score and working hours on

markers of cardiometabolic risk for men or women. Sub-group analyses were conducted

using participants with shift work data available from police radio records. There was no

observed association between shift work and markers of cardiometabolic risk, Appendix

A7.8.

!!"#$!

Table 7.2 Association betw

een number of w

eekly working hours and m

arkers of cardiometabolic health: m

en (n = 3,280)

35 - 40

hrs/week

41 – 48 hrs/w

eek 49 – 54

hrs/week

>55 hrs/w

eek P

trend P

interaction

N (%

) 1145

(34.9) 1263

(38.5) 441

(13.5) 431

(13.4)

B


2 A

djusted mean (SE)

M

odel 1 27.6

(0.1) 27.7

(0.1) 27.9

(0.2) 28.5

(0.2) <0.0001

M

odel 2 27.6

(0.2) 27.7

(0.2) 27.9

(0.2) 28.5

(0.2) <0.0001

M

odel 3 27.6

(0.2) 27.6

(0.2) 27.9

(0.2) 28.4

(0.2) <0.0001

0.45 M

odel 4 29.4

(0.4) 29.4

(0.4) 29.7

(0.4) 30.2

(0.4) <0.0001

W


M

odel 1 93.5

(0.3) 93.6

(0.3) 94.1

(0.4) 95.3

(0.5) 0.001

M

odel 2 94.0

(0.4) 94.0

(0.4) 94.4

(0.6) 95.6

(0.5) 0.004

M

odel 3 93.8

(0.4) 93.7

(0.4) 94.1

(0.5) 95.2

(0.5) 0.008

0.28 M

odel 4 99.4

(0.9) 99.3

(0.9) 99.7

(1.0) 100.8

(1.0) 0.007

M

odel 5 96.0

(0.5) 95.8

(0.5) 95.6

(0.6) 95.6

(0.6) 0.17

Percentage body fat

M

odel 1 22.1

(0.2) 22.1

(0.2) 22.6

(0.3) 23.4

(0.3) <0.0001

M

odel 2 22.4

(0.3) 22.3

(0.3) 22.8

(0.3) 23.6

(0.3) <0.0001

M

odel 3 22.2

(0.2) 22.1

(0.2) 22.6

(0.3) 23.3

(0.3) 0.001

0.20 M

odel 4 24.4

(0.6) 24.4

(0.6) 24.8

(0.6) 25.5

(0.6) <0.0001

M

odel 5 22.6

(0.4) 22.5

(0.4) 22.7

(0.4) 22.7

(0.4) 0.39

TC

:HD

L

M

odel 1 4.02

(0.03) 4.09

(0.03) 4.10

(0.05) 4.12

(0.05) 0.12

M

odel 2 4.19

(0.05) 4.23

(0.05) 4.23

(0.06) 4.25

(0.06) 0.39

M

odel 3 4.18

(0.05) 4.21

(0.05) 4.21

(0.06) 4.22

(0.06) 0.52

0.30 M

odel 4 3.83

(0.11) 3.87

(0.11) 3.87

(0.12) 3.88

(0.12) 0.48

M

odel 5 3.69

(0.11) 3.72

(0.11) 3.70

(0.11) 3.67

(0.11) 0.62

!!!!

"#%!


en)

35 - 40

hrs/week

41 – 48

hrs/week

49 – 54

hrs/week

>55

hrs/week

HD

L mm

ol/L

M

odel 1 1.38

(0.01) 1.37

(0.01) 1.37

(0.01) 1.36

(0.01) 0.38

M

odel 2 1.34

(0.01) 1.34

(0.01) 1.34

(0.02) 1.33

(0.02) 0.55

M

odel 3 1.34

(0.01) 0.34

(0.01) 1.34

(0.02) 1.33

(0.02) 0.61

0.63 M

odel 4 1.27

(0.03) 1.27

(0.03) 1.27

(0.03) 1.26

(0.03) 0.61

M

odel 5 1.31

(0.03) 1.31

(0.03) 1.32

(0.03) 1.32

(0.03) 0.49

N

on HD

L mm

ol/L

M

odel 1 3.94

(0.03) 3.99

(0.02) 4.02

(0.05) 4.02

(0.05) 0.13

M

odel 2 4.03

(0.05) 4.07

(0.05) 4.08

(0.06) 4.07

(0.06) 0.45

M

odel 3 4.01

(0.05) 4.05

(0.05) 4.06

(0.06) 4.04

(0.06) 0.58

0.09 M

odel 4 3.38

(0.10) 3.42

(0.10) 3.44

(0.11) 3.42

(0.11) 0.51

M

odel 5 3.31

(0.10) 3.34

(0.10) 3.35

(0.11) 3.30

(0.11) 0.95

H

bA1c, %

M

odel 1 5.61

(0.02) 5.61

(0.01) 5.64

(0.02) 5.60

(0.03) 0.94

M

odel 2 5.63

(0.02) 5.62

(0.02) 5.65

(0.03) 5.62

(0.03) 0.97

M

odel 3 5.62

(0.02) 5.61

(0.02) 5.64

(0.03) 5.61

(0.03) 0.86

0.50 M

odel 4 6.62

(0.05) 6.60

(0.05) 6.64

(0.05) 6.60

(0.05) 0.62

M

odel 5 6.59

(0.05) 6.57

(0.05) 6.59

(0.05) 6.54

(0.05) 0.22

!!!!

"#&!


en) 35 - 40

hrs/week

41 – 48 hrs/w

eek 49 – 54

hrs/week

>55 hrs/w

eek P

trend P

interaction

Diastolic blood pressure, m

mH

g

Model 1

81.2 (0.3)

81.8 (0.3)

81.5 (0.4)

82.3 (0.5)

0.07

Model 2

81.7 (0.4)

82.1 (0.4)

81.8 (0.6)

82.7 (0.6)

0.15

Model 3

81.6 (0.4)

81.9 (0.4)

81.6 (0.6)

82.4 (0.6)

0.22 0.09

Model 4

82.5 (0.9)

83.0 (0.9)

82.6 (1.0)

83.4 (1.0)

0.18

Model 5

81.3 (0.9)

81.8 (0.9)

81.2 (1.0)

81.6 (1.0)

0.86

Systolic blood pressure, mm

Hg

M

odel 1 136.0

(0.4) 135.8

(0.4) 135.4

(0.6) 136.1

(0.6) 0.91

M

odel 2 135.8

(0.6) 135.7

(0.6) 135.2

(0.8) 136.0

(0.8) 0.97

M

odel 3 135.7

(0.6) 135.6

(0.6) 135.1

(0.8) 135.8

(0.8) 0.97

0.52 M

odel 4 139.8

(1.4) 139.9

(1.4) 139.4

(1.4) 140.0

(1.4) 0.97

M

odel 5 138.5

(1.3) 138.6

(1.3) 137.8

(1.4) 138.0

(1.4) 0.37

C

RP, m

g/L±

M

odel 1 0.91

(0.92) 0.94

(0.92) 0.97

(0.94) 1.03

(0.94) 0.016

M

odel 2 0.97

(0.94) 1.00

(0.94) 1.03

(0.95) 1.10

(0.95) 0.012

M

odel 3 0.96

(0.94) 0.97

(0.94) 1.00

(0.95) 1.05

(0.95) 0.05

0.23 M

odel 4 0.94

(0.99) 0.95

(0.99) 0.98

(0.99) 1.03

(0.99) 0.047

M

odel 5 0.82

(0.98) 0.83

(0.98) 0.84

(0.99) 0.84

(0.99) 0.51

A

bbreviations HD

L High density lipoprotein, TC

total cholesterol, CR

P high sensitivity c reactive protein.

±CR

P log transform

ed to allow param

etric testing, untransformed values presented.

Model 1 adjusted for age, M

odel 2 + physical activity, smoking, education, TV

viewing, household incom

e, job strain. Model 3 + D

AS

H score m

ean energy intake, alcohol (continuous variables). M


odel 5 + body mass index (continuous).

!

! *'&!

Table 7.3 Association between number of weekly working hours and markers of cardiometabolic health: women (n = 1,756) 35 - 40

hrs/week 41 – 48

hrs/week >49

hrs/week Ptrend P interaction

N (%) 948 (54.0) 533 (30.0) 275 (13.0) Body mass index, kg/m2 Adjusted mean (SE) Model 1 26.1 (0.2) 25.6 (0.2) 26.0 (0.3) 0.77 Model 2 26.3 (0.3) 25.8 (0.3) 26.3 (0.4) 0.97 Model 3 26.2 (0.3) 25.8 (0.3) 26.2 (0.4) 0.92 0.60 Model 4 28.3 (0.7) 27.8 (0.8) 28.3 (0.8) 0.96 Waist circumference, cm Model 1 82.5 (0.4) 80.8 (0.5) 81.5 (0.7) 0.22 Model 2 82.9 (0.7) 81.3 (0.7) 82.3 (0.9) 0.42 Model 3 82.7 (0.7) 81.2 (0.7) 81.9 (0.9) 0.32 Model 4 87.4 (1.8) 85.8 (1.8) 86.7 (1.9) 0.34 0.80 Model 5 82.5 (0.9) 81.9 (0.9) 81.8 (0.9) 0.06 Percentage body fat Model 1 33.0 (0.2) 32.1 (0.3) 32.6 (0.4) 0.46 Model 2 33.3 (0.4) 32.6 (0.5) 33.1 (0.6) 0.75 Model 3 33.2 (0.4) 32.5 (0.5) 32.9 (0.6) 0.60 Model 4 34.8 (1.2) 34.1 (1.2) 34.6 (1.2) 0.60 0.29 Model 5 31.9 (0.6) 31.7 (0.6) 31.5 (0.7) 0.24 HDL, mmol/L Model 1 1.70 (0.01) 1.75 (0.02) 1.77 (0.02) 0.021 Model 2 1.71 (0.02) 1.77 (0.03) 1.78 (0.03) 0.023 Model 3 1.72 (0.02) 1.77 (0.03) 1.78 (0.03) 0.017 0.95 Model 4 1.57 (0.06) 1.62 (0.07) 1.63 (0.07) 0.020 Model 5 1.63 (0.06) 1.67 (0.06) 1.69 (0.07) 0.016 TC:HDL Model 1 3.2 (0.0) 3.0 (0.0) 3.0 (0.1) 0.010 Model 2 3.2 (0.1) 3.1 (0.1) 3.0 (0.1) 0.010 Model 3 3.2 (0.1) 3.1 (0.1) 3.0 (0.1) 0.006 0.60 Model 4 3.4 (0.2) 3.3 (0.2) 3.2 (0.2) 0.008 Model 5 3.3 (0.2) 3.2 (0.2) 3.1 (0.2) 0.006 Non HDL, mmol/l Model 1 3.40 (0.03) 3.31 (0.04) 3.32 (0.05) 0.18 Model 2 3.50 (0.05) 3.42 (0.06) 3.43 (0.07) 0.22 Model 3 3.49 (0.05) 3.41 (0.06) 3.40 (0.07) 0.15 0.65 Model 4 3.54 (0.14) 3.46 (0.14) 3.46 (0.15) 0.19 Model 5 3.46 (0.14) 3.39 (0.14) 3.38 (0.15) 0.19

HbA1c, % Model 1 5.69 (0.02) 5.62 (0.02) 5.62 (0.03) 0.046 Model 2 5.67 (0.03) 5.62 (0.04) 5.61 (0.04) 0.14 Model 3 5.67 (0.03) 5.62 (0.04) 5.61 (0.04) 0.15 0.67 Model 4 6.80 (0.08) 6.76 (0.08) 6.76 (0.09) 0.28 Model 5 6.77 (0.08) 6.74 (0.08) 6.73 (0.09)

0.28

!

!!!

*''!

Table 7.3 continued (women)

35 – 40 hrs/week

41 – 48 hrs/week

>49 hrs/week

Ptrend P interaction

Diastolic blood pressure, mmHg Model 1 76.8 (0.3) 76.7 (0.4) 76.9 (0.6) 0.88 Model 2 77.1 (0.6) 77.1 (0.6) 77.5 (0.7) 0.55 Model 3 77.0 (0.6) 77.1 (0.6) 77.3 (0.7) 0.65 0.74 Model 4 77.3 (1.5) 77.3 (1.5) 77.5 (1.6) 0.70 Model 5 75.8 (1.4) 76.1 (1.5) 76.1 (1.5) 0.67 Systolic blood pressure, mmHg Model 1 123.9 (0.4) 123.0 (0.6) 122.9 (0.8) 0.29 Model 2 124.8 (0.8) 124.3 (0.9) 124.5 (1.1) 0.76 Model 3 124.7 (0.8) 124.2 (0.9) 124.4 (1.1) 0.70 0.47 Model 4 127.6 (2.2) 127.0 (2.2) 127.2 (2.3) 0.70 Model 5 125.6 (2.1) 125.4 (2.1) 125.3 (2.2) 0.66 CRP, mg/L Model 1 1.26 (1.03) 1.10 (1.04) 1.15 (1.06) 0.17 Model 2 1.38 (1.06) 1.19 (1.07) 1.29 (1.08) 0.33 Model 3 1.35 (1.06) 1.18 (1.07) 1.25 (1.08) 0.22 0.72 Model 4 1.72 (1.17) 1.49 (1.17) 1.59 (1.18) 0.24 Model 5 1.43 (1.16) 1.28 (1.16) 1.31 (1.16)

0.18

Abbreviations HDL High density lipoprotein, TC total cholesterol, CRP high sensitivity c reactive protein. P1, P-for-trend across groups of working hours. ±CRP log transformed to allow parametric testing, untransformed values presented. Model 1 adjusted for age, Model 2 + physical activity, smoking, education, TV viewing, household income, job strain, menopause status. Model 3 + DASH score mean energy intake, alcohol (continuous variables). Model 4 + diagnosed ± treatment for diabetes, lipids or blood pressure. Model 5 + body mass index (continuous).

!

!!!

*'(!

7.4 Discussion This cross-sectional study has described the cardiometabolic profile of British police

force employees and examined the association between long working hours and

markers of cardiometabolic disease risk. Finally it investigated if diet quality modified

the relationship between long working hours and markers cardiometabolic disease risk.


• Around a quarter of male employees reported working more than 48 hours per

week.

• Male police officers were more likely to have three or more markers of

cardiometabolic risk compared to women.

• Length of weekly working hours was positively associated with measures of

adiposity (BMI, waist circumference and body fat) and inflammation in male

police employees.

• Longer weekly working hours in female police force employees was associated

with a preferable lipid profile (higher HDL and lower total cholesterol to HDL

ratio).

• Diet quality (determined by DASH score) did not modify the relationship between

working hours and markers of cardiometabolic risk.


Objective i) To describe the cardiometabolic health profile of British police force

employees

The prevalence of overweight and obesity in the Airwave Health Monitoring study was

55.5% and 23.0% respectively for men and 34.0% and 15.5% for women. These rates

of obesity are lower compared to the general population, though overweight is higher

(men: 41.5%/24.9% overweight/obese, women: 32.3%/25.2% overweight/obese) (185).

Based on blood pressure readings obtained at the Airwave Health Monitoring Study

health screen 38.9% men and 14.4% of women had a reading within hypertension

criteria, this compares to 31.5% men and 29.0% women in the general population

(185). The differences in prevalence between the Airwave Health Monitoring Study

cohort compared to the general population is possibly due to the lower age profile of the

!

!!!

*')!

study cohort and potentially occupational differences. Two fifths of male police force

employees had three or more cardiometabolic risk factors. This is higher than the

estimated prevalence of MetS for the general European population of ~20-30% (31).

The difference in the prevalence of increased metabolic risk may be due to the criteria

applied in the present study (excluding triglycerides as unavailable and inclusion of

CRP). However, studies in German and US police officers have observed prevalence

to be higher in this occupational group compared to the general population (315,316).

Less than 10% of male participants were classified as having no markers of

cardiometabolic risk. Those who were classified with three or more markers were most

likely to have a combination of markers that included an elevated waist circumference

and blood pressure. Women who had three or more markers were likely to have

elevated waist circumference plus an additional risk marker (HbA1c, CRP, blood

pressure or dyslipidaemia). Although the data analysed in the present study are cross-

sectional the common feature of elevated waist circumference across the main

cardiometabolic risk phenotypes supports the hypothesis that excess visceral adipose

tissue is central to metabolic abnormalities associated with cardiometabolic disease

development (41). The difference in the cardiometabolic risk profile between men and

women is likely to be multifactorial. Some of the difference may be explained by age,

as female participants were on average almost three years younger than male

participants. However, lifestyle factors are likely to be important. Female police

employees were more likely to be in the lowest category for weekly TV viewing and as

observed in Study 2 (Chapter 5) women had a dietary profile higher in fruit and

vegetable intake. Female police employees also had significantly different working

patterns than male employees, typically working less hours per week and less likely to

be classified as a shift worker.

Objective ii) To investigate the association between long working hours and markers of

cardiometabolic disease risk in British police force employees

In the Airwave Health Monitoring Study working hours showed a dose-response

relationship with markers of adiposity (BMI, waist circumference, body fat) and CRP for

men only. Potentially suggesting that male police force employees working longer

hours may be at increased risk of future cardiometabolic diseases. The sex specific

!

!!!

*(+!

observation supports previous findings from an Australian working cohort that found

long working hours in men but not women to be associated with increased BMI

independent of established confounders (317). Findings from large meta-analyses

found a positive association between length of working hours and risk of stroke and

coronary heart disease; however, in comparison to the present study the meta-analysis

did not observe any difference in risk between men and women (15). One explanation

for the lack of positive association between measures of adiposity and working hours in

female employees could be that they work shorter hours therefore the effects of

extended working hours on women cannot be compared directly to. It could also be

hypothesised that the cumulative exposure to working hours may be reduced in women

due to child rearing activities, particularly as female employees reported shorter

durations of police force employment compared to their male counterparts. However,

when sensitivity analyses were conducted including women who worked part-time,

women working more than 49 hours per week had a higher BMI compared to part-time

workers. This observation supports research in Australian nurses that reported higher

weight gain in those working >49 hours per week compared to part-time workers (173).

Female employees in the Airwave Health Monitoring Study that worked longer hours

had a potentially better lipid profile than those working longer hours as evidenced by

higher HDL (improved total to HDL ratio). The reasons for this observation are likely to

be multifactorial and subject to residual confounding, for example, this study did not

control for oral contraceptive use which has been previously shown to modify lipid

profiles (318).

It has been suggested that the stress may be a physiological factor in conjunction with

lifestyle (high level of weekly sitting and alcohol intake) that links long working hours to

incidence of stroke and heart disease (15). Men and women in the Airwave Health

Monitoring Study working longer weekly working hours are more likely to report high job

strain compared to those working standard hours. This observation in male employees

may contribute to the relationship between working hours and increased adiposity,

however statistical models controlled for stress. Moreover, there was low-level

collinearity in men between job strain and diet quality. It also should be noted that

employees working longer weekly hours reported higher levels of physical activity and

!

!!!

*(*!

less sitting time (potentially linked to differences in job role) which may increased

energy expenditure levels.

Objectives iii) To establish if diet quality modifies the relationship between long

working hours and markers cardiometabolic disease risk.

Long working hours could be hypothesized to reduce employee recovery periods and

adversely impact health behaviours. In the Airwave Health Monitoring Study sleep

duration was not associated with working hours. In Studies 2 and 3 presented in this

thesis it was observed that long weekly working hours in men was associated with a

lower DASH score (poorer diet quality); and a lower DASH score was associated with

increased measures of adiposity and inflammation. However, in the current study diet

quality (DASH score) did not modify the relationship between working hours and

markers of cardiometabolic disease risk. There are numerous possible explanations for

this observation. First the association between working hours and diet quality, although

significant in men, was also strongly associated with high TV viewing and low physical

activity, which were not associated with longer weekly working hours. It is possible that

less inactivity and more activity may offset potentially negative effects of a poor diet. As

with all occupational cohort studies consideration must be given to the ‘healthy worker

effect’ (319) where ‘unhealthy’ workers may reduce their hours due to illness, therefore

reducing the likelihood of observing an association where one exists. There is potential

evidence of this in the Airwave Health Monitoring Study; with a significant association

between part-time work and chronic disease diagnosis observed. This could bias

results towards the null, with previous long hour workers now categorised in a shorter

working hour group.

The findings may be subject to residual confounding (where there is a variable of

potential explanation that has not been measured, or has been poorly measured). In

the Airwave Health Monitoring Study cohort long working hours is significantly

associated with working shifts. Sub-group analyses exploring short-term shift exposure

(last 30 days) and markers of cardiometabolic risk did not observe any significant

findings. This lack of association is contrary to previous studies (181,191,195) however

these studies were not conducted in comparable occupational groups (e.g.

manufacturing and healthcare). An additional reason may be due to bias in the method

!

!!!

*("!

of exposure measurement, namely that it only captured exposure in employees using

the radio as part of their job role. A small study of 98 US police officers found that

midnight work to be associated with increased MetS; however, that study used payroll

records to determine shift work exposure. Due to the limited sample size it was not

possible to establish if shift work is a modifying risk factor in the relationship between

working hours and adiposity. The Buffalo Cardio-Metabolic Occupational Police Stress

(BCOPS) study did show an interaction between working hours and night work with

waist circumference and BMI (190). However, the working hour categories used in the

BCOPS study were based on part-time, standard hours and 40 hours or more per

week, which is a much lower and narrower range than the categories used in the

present study. Interestingly the BCOPS study also did not observe any association

between working hours and measures of adiposity in women (190). In common with

the Airwave Health Monitoring Study the BCOPS study also found that women worked

significantly shorter hours compared to their male counterparts. A limitation of the

BCOPS study is that they only measured total energy intake and alcohol based on FFQ

data, neither of which they observed to be significantly correlated with working hours; it

also had a relatively small sample size (n = 710) (190). Lastly, previous observational

studies have shown time of eating (123) and frequency of energy intake (112) to be

associated with BMI and/or central obesity. Although energy intake irregularity was

measured it was not found to be an independent predictor of cardiometabolic risk

markers (exploratory data not presented). However, this was the only aspect of eating

patterns that was measured due to the design of the initial food diary used in the

Airwave Health Monitoring Study.

7.4.3 Study strengths and limitations A key strength of this study is the large sample size with comprehensive 7-day dietary

data. Additionally the outcome measures of the study were collected from a standard

health screening protocol. However, the present study has a number of methodological

limitations. Firstly, the study does not take into account cumulative exposure to working

long hours. A further methodological limitation of the Airwave Health Monitoring Study

is that the number of weekly working hours was self-reported rather than being obtained

from payroll data as in the BCOPS study (190). However the BCOPS study has a

considerably smaller sample size focused on one force. In a small sub-sample of one

!

!!!

*(#!

police force enrolled in the Airwave Health Monitoring Study there was a strong

agreement between payroll and self-reported working hours (Chapter 2, 2.2.3). The

present study also assumed that weekly working hours remain stable over time;

however, changes in job role and external demands on emergency service employees

needs to be considered when interpreting the findings reported here. In common with

other studies total weekly working hours was used as the exposure of interest (15,169);

however, other factors including length of individual shifts or number of weekly working

days and rest days should be an important consideration in future studies. As

discussed in Chapter 6 serum lipid measurements are based on non-fasting samples.

However in line with current guidelines (299) and the high intra-class correlation

coefficients (Chapter 2, Table 2.3) suggesting that these are robust measures of blood

cholesterol concentrations. Additionally the cross-sectional nature of the study data

cannot be used to determine mediation and therefore inform causality between long

weekly working hours and increased cardiometabolic disease risk. Longitudinal studies

are needed to determine if changes in cardiometabolic markers are observed when

employees move from working standard weekly hours to longer hours.

7. 5 Study conclusion and relevance to further studies

This study has observed that longer weekly working hours are positively associated

with measures of adiposity and inflammation in British male police force employees

independent of established risk factors. However, this association was not observed in

female police force employees. The results of the present study suggest that diet

quality (measured by the DASH score) does not modify the relationship between long

working hours and cardiometabolic markers in British police force employees.

!

!!!

*($!

CHAPTER 8

8.0 METHODOLOGICAL DEVELOPMENT - MAPPING DIET TO WORKING HOURS

8.1 Introduction

The time and frequency of dietary intake have been previously associated with

cardiometabolic disease risk independently of energy intake (114,126). Shift workers

compared to day workers have been shown to have a temporal redistribution of nutrient

intake and differences in eating frequency compared to day workers (213,214,216,221–

223). However there are limited published studies that have investigated how these

dietary behaviours differ between the various types of shifts that employees work.

It has been suggested that food diaries currently provide the most reliable method of

measuring dietary behaviour in shift workers (320). The food diary adopted by the

Airwave Health Monitoring Study was based on the version used in the EPIC study that

commenced data collection circa.1990. A limitation of the diary design (Appendix A3.2)

is that it does not facilitate the recording of dietary intake by time of day. Additionally,

the diary is segmented into predefined occasions (e.g. breakfast, lunch and dinner).

These subjective eating occasions may not fully characterise modern ‘grazing’ eating

patterns (273) or the altered eating structures previously observed in shift workers

(213,214,216,221–223). Therefore this may bias participant dietary reporting towards

traditional eating patterns. A further methodological limitation is that the baseline

Airwave Health Monitoring Study questionnaire did not capture the type, duration and

previous shift work exposure. Although radio usage data provided a proxy measure for

the studies included in this thesis it only captured shift work data for employees that

used radios as part of their job role. At the time of undertaking this PhD there was no

standardised or validated shift work questionnaire. Researchers have previously

recommended that future shift work exposure questionnaires should clarify: i) the type

of shift work schedule, ii) the frequency of shift work and iii) previous work history (321).

Therefore as part of this PhD the following measurement tools were developed to

address these methodological limitations: i) an amended food diary and coding protocol

!

!!!

*(%!

to allow the mapping of temporal dietary intake to working hours and ii) a shift work

exposure questionnaire.

8.2 Time of day food diary

8.2.1 Design and pilot The original Airwave Health Monitoring Study diary (Appendix A3.2) was adapted to

allow participants to record their dietary intake by time. The redesigned diet diary

included a space to record daily working hours (start/end time of shift, or day off), and

also included a revised example and instruction page (Appendix 8.1). The new food

diary was piloted in April 2014 to check clarity of instruction (determined by level of

information completion) and comparison of energy intake reporting compared to the

original food diary. The pilot of the diary and results are detailed in Appendix A8.2.

Following the pilot the new food diary replaced the original version from January 2015.

8.2.2 Exploratory cross-sectional study: Mapping eating behaviours to working hours

8.2.2.1 Aims

The aims of the exploratory analyses were to:

i) Estimate the prevalence of energy intake misreporting using the new food diary

ii) Characterise dietary behaviours (nutrients, foods and eating occasions) by the

different types of shifts worked by police force employees

iii) Characterise the temporal distribution of energy intake across the different types

of shifts worked by police force employees

8.2.2.1 Methods

Participants

Participants included in this exploratory study were enrolled into the Airwave Health

Monitoring Study between September and November 2014 from the London

Metropolitan Police Force. A sample cohort of 300 participant food diaries were

selected for assessment based on diary availability. A final sample of 240 participants

was included for exploratory cross-sectional analyses, Figure 8.1.

!

!!!

*(&!

Dietary intake and eating pattern variables

The dietary intake variables were then generated as detailed in Chapter 3. An eating

occasion was defined as the reported consumption of any food/beverage >50kcals with

a minimum of 15 minutes between eating occasions (114). To determine dietary intake

by time each 24-hour period (midnight – midnight) was split into six 4-hourly periods

and the energy and nutrient intakes were calculated for each of these periods.

Figure 8.1 Schematic showing sample selection for inclusion in exploratory cross-sectional

analyses investigating the mapping of dietary intake to working hours

Sample of food diaries completed by London Metropolitan enrolled Sept – Nov 2014 (n = 300)

Food diaries coded (n = 256)

Exclude: Food diary completed during annual leave (n = 20) Incomplete information (n = 24)

Exclude: Not entered into Dietplan by time (n = 2)

Gross coding / ex-protocol code error (n = 14)

Sample available for analyses (n = 240) Working hour data

Daily working hours as recorded in the food diary were logged into an Excel database

and shifts classified as shown in Table 8.1. The length of each shift was calculated

based on the start and end times recorded. This database was then merged with the

diet and available meta-data in SAS 9.3 (SAS PROC MERGE). Limited meta-data were

available at the time of analysis. Age, sex and body weight were available and used to

estimate prevalence of plausible energy intake reporting.

Measuring energy intake misreporting

The Goldberg equation (279) (refer to Chapter 4 for details) was used to determine

energy intake mis-reporting. As physical activity data were not available at the time of

analysis a PAL of 1.5 was universally applied based on published Department of Health

guidance representing moderate occupational activity (280). This provided upper and

lower confidence intervals of 0.98 and 2.30. Appetite change and weight loss diet

!

!!!

*('!

information were unavailable so all participants with the exception of three with missing

date of birth information were included for energy intake misreporting classification.

Table 8.1 Criteria for the classification of each shift recoded by participants in the Airwave

Health Monitoring Study

Shift classification Criteria

Late Shift ends after 7p.m., but before midnight

Early Shift starts before 7a.m.

Day Shift starts and ends between 7a.m. and 7p.m.

Rest day Day off / rest day recorded by participant

Night -Night Night-Night includes the end of one night shift and the start of the next within a 24-hr period

Rest - Night Includes the start of the night shift (hours worked before midnight), i.e. this is the last day recorded or the shift type is not recorded for the day following the start of the night shift.

Night – Rest Night shift ending in a rest day (i.e. the start of the rest day includes completion of the night shift)

Exploratory statistical analyses

The mean nutrient and food intakes per participant per shift type recorded were treated

as separate data points (providing 679 mean daily intakes). General linear models

were conducted to determine differences in nutrient and food intakes across the

different shift types. The models were conducted unadjusted and then adjusted for age

and sex. To investigate the difference in the number of eating episodes and energy

intake across time of day and shift worked factorial ANOVA were conducted

(independent variables: shift worked and 4-hourly time period). !Post hoc Bonferroni

correction was applied to correct for multi comparisons when determining significant

difference between groups (corrected for 21 multiple tests across seven groups with

adjusted statistical significance between groups taken as p <0.003).

!

!!!

*((!

8.2.2.2 Results

Sample characteristics and energy intake reporting

The sample included 240 participants; 66% were male. The estimated prevalence of

dietary intake under-reporting was 35.6%, Table 8.2. No participants were classified as

over-reporting energy intake.

Working patterns

From the sample of 240 participants 76 (32%) recorded working day shifts only,

Appendix A8.3. The most common shift combination was late and day shifts with rest

days recorded by 25% of participants. Those recording two consecutive night shifts

work accounted for 14% of participants. The mean length of a day shift was 8.2 hours

(SD1.0) while night shifts had the longest duration lasting mean length of 13.3 hours

(SD2.1), Appendix A8.4.

Table 8.2 Summary characteristics of participants enrolled into the Airwave Health Monitoring

study from the London Metropolitan police force in 2014 that completed the baseline food diary

(n = 233)±

Men (n =153) Women (n = 80) Mean (SD) Age at screening, years 40.1 (8.5) 38.5 (8.6) Body mass index kg/m2 27.5 (3.3) 25.6 (4.7) Est. Basal metabolic rate, kcal/day-1 1906 (143) 1438 (135)

Daily energy, kcal 2072 (505) 1661 (381) Ratio: EI:BMR 1.1 (0.3) 1.2 (0.3) % Classed as under-reporter±

37.2 32.5 Current smoker§ 11.1 7.5 ±Excluding participants with age or weight missing (n = 7)

Dietary intakes across different shifts

Dietary intakes are compared across different shifts and rest days in Table 8.3. Mean

energy intake was lowest when rest-night days were recorded (1702 SD 641kcal/day)

this was significantly lower than energy intake recorded on rest days (2035 SD

682kcal/day). The mean number of eating occasions was lowest for those starting a

night shift following a rest day (3.7, SD 1.6) while the highest number of eating

occasions was recorded on day shifts (4.9 SD 1.6). With the exception of

!

!!!

*()!

carbohydrates, macronutrient intakes were not significantly different across different

shifts. Energy intake from alcohol varied across shift type with those working a night

shift followed by a day shift consuming a higher percentage of energy from alcohol

compared to day workers and late workers. Employees were more likely to record not

drinking alcohol during the day preceding a night shift (rest-night) compared to working

a day shift (95.8% vs. 43.0%) – data not tabulated. Intake from SSBs showed a

consistent pattern of intake when consumers were included and excluded. Lowest

intakes of SSBs were on rest days, day and early shifts. Intakes of SSBs were

significantly higher for those on a night shift followed by a rest day compared to day

shifts, rest days and late shifts. Fruit intake was higher on day shifts compared to rest

days. There was no difference in meat, vegetable or fish intake across different shifts.

Energy intake and eating occasions by time across different shifts

Table 8.4 shows the percentage of participants working a specific shift recording intake

at each 4-hourly time point. Of those working a night shift followed by another night

shift or a rest day 79% and 62% respectively recorded energy intake between midnight

and 4.a.m while 7% of those on a day shift reported energy intake during this time

period. Of the participants working a day shift 98% recorded energy intake between

midday and 4p.m. compared to 67% of participants working a night shift followed by a

further night shift.

Between midnight and 4a.m. participants working a late shift consumed 22% of their

daily energy intake. Regardless of shift worked the highest amount of energy intake (on

average 30% or more) was consumed between 4p.m. and 8p.m. Participants working a

night shift followed by a night shift recorded the highest percentage of their daily energy

intake during the period 8.p.m to midnight, Figure 8.2.

!!!!

"#$!

Table 8.3 Differences in average nutrient, food group and eating occasions across different shifts for m

en and wom

en (n = 240)

D

ay Late

Early

Night - N

ight R

est-Night

Night-R

est R

est P

1 P

2 N

193

106

24

33

48

42

233

Mean (S

D)

Energy, kcal

1916 (597)

2007 (751)

1908 (624)

1745 (662)

1702 (641)

2117 (817)

2035 (682)

0.010a

0.005a

EI episodes

4.9 (1.6)

4.5 (1.4)

4.4 (1.6)

4.3 (2.0)

3.7 (1.6)

4.8 (2.1)

4.6 (1.6)

0.001a,b,c

0.004a,b,c

Kcal/gram

food 1.9

(0.5) 2.1

(0.7) 1.9

(0.7) 2.2

(1.2) 2.1

(0.9) 2.2

(0.8) 2.1

(1.1) 0.06

0.06 %

EI carbohydrate

46.8 (8.2)

48.0 (9.2)

47.4 (11.4)

48.6 (11.7)

46.9 (11.5)

45.4 (13.6)

44.8 (8.7)

0.05 0.040

%E

I total fat 34.1

(7.1) 35.2

(8.2) 32.0

(10.1) 35.2

(9.5) 36.4

(8.0) 35.5

(9.6) 34.9

(6.8) 0.27

0.23 %

EI saturated fat

12.0 (3.6)

12.7 (4.4)

10.9 (4.0)

13.2 (4.8)

12.5 (3.7)

13.4 (5.4)

12.5 (3.6)

0.13 0.08

%E

I protein 18.0

(4.9) 18.2

(5.9) 16.5

(4.0) 18.2

(7.0) 19.0

(6.9) 17.4

(6.8) 17.0

(4.3) 0.11

0.12 D

ietary fibre, g/1000kcal 7.3

(2.6) 6.9

(2.8) 7.7

(4.1) 6.5

(3.0) 6.8

(3.6) 6.3

(3.0) 6.3

(2.4) 0.009

d 0.008

d S

odium, m

g/1000kcal 1393

(412) 1517

(587) 148

(623) 1376

(636) 1325

(613) 1273

(465) 1351

(421) 0.046

e 0.040

e

Median (IQ

R)

% E

I Alcohol

1.3 (6.6)

0.0 (0.0)

0.0 (7.5)

0.0 (0.0)

0.0 (0.0)

0.0 (4.2)

3.3 (10.1)

<0.0001a, d-i

<0.0001a, d-i

% E

I Alcohol (consum

ers) 5.9

(6.5) 4.0

(7.4) 7.5

(19.8) 1.8

(19.9) 15.5

(19.6)* 12.9

(15.3) 8.3

(10.8) 0.001

j 0.002

j,k S

SB

s, g /1000kcal 1 0.0

(119.3) 0.0

(124.1) 0.0

(125. 0.0

(176.1) 0.0

(124.1) 31.7

(201.5) 0.0

(111.6) 0.004

j,k,l 0.010

j,k,l S

SB

s, g /1000kcal 2

159.4 (211.5)

143.2 (193.8)

129.4 (172.

186.4 (473.6)

247.9 (235.0)

201.5 (430.1)

131.7 (200.2)

0.001 j,k,l

0.005 j,k,l

Total red meat, g/1000kcal

26.6 (42.1)

31.7 (51.3)

28.4 (35.6)

24.5 (49.4)

31.7 (51.3)

14.6 (73.4)

32.9 (42.7)

0.36 0.44

Processed red m

eat

13.4 (27.9)

14.2 (32.8)

16.3 (36.0)

9.7 (28.6)

14.2 (32.8)

0.0 (13.3)

17.3 (31.0)

0.26 0.20

Dairy, g/1000kcal

102.9 (95.6)

110.2 (117.9)

99.6 (82.7)

126.7 (143.5)

110.2 (117.9)

108.6 (143.2)

94.6 (97.2)

0.027a

0.003a

Total fish, g/1000kcal 5.1

(21.7) 0.0

(13.8) 0.0

(18.6) 0.0

(0.0) 0.0

(13.8) 0.0

(0.0) 0.0

(16.0) 0.20

0.22 W

hole grains, g/1000kcal 22.5

(38.3) 25.3

(49.9) 38.9

(78.8) 0.0

(20.5) 25.3

(49.9) 0.0

(23.8) 11.9

(32.4) <0.0001

h,k,n-q <0.0001

h,k,n-q Legum

es, g/1000kcal 9.7

(22.5) 0.0

(12.3) 10.0

(30.8) 0.0

(1.1) 0.0

(12.3) 0.0

(15.8) 7.3

(19.6) 0.012

g,f,n 0.021

g,f Total Fruit, g/1000kcal

57.8 (103.5)

49.7 (119.6)

29.9 (87.0)

17.3 (110.9)

49.7 (119.6)

5.5 (81.1)

25.6 (74.7)

0.007d

0.006d

Vegetables, g/1000kcal

41.4 (66.6)

34.0 (59.7)

35.4 (65.7)

19.7 (81.0)

34.0 (59.7)

42.3 (79.4)

38.0 (62.5)

0.54 0.57

Abbreviations E

I energy intake, SS

Bs sugar sw

eetened beverages, NM

Es non m

ilk extrinsic sugars. 1.Includes non-consum

ers, 2. Excludes non-consum

ers. To compare m

eans values betw

een groups one-way A

NO

VA

was used for param

etric data (values presented as mean

and standard deviation). If significance indicated (p <0.05) Bonferroni post hoc test w

as applied to identify the source of the difference. W


ere conducted for nonparam

etric data (values presented as median and inter quartile range). If significance

indicated (p <0.05) Wilcoxon rank sum

tests were then conducted betw

een each group to establish the source of the difference w

ith Bonferroni post hoc test applied to correct for

multiple com

parisons. p1 unadjusted, p2 General Linear M

odels used adjusted for age and sex. S

uperscript letters indicate source of differences as indicated in the grid opposite. *2 observations !

Day

Late

Early

Night-Night

Rest-Night

Night-Rest

Late m

-

Early -

f -

N

ight-Night

- q

g -

Rest-N

ight b

p h

- -

N

ight-Rest

j k

o c

- -

Rest

d e

n i

a l

!

!!!!

"#"!

Table 8.4 Percentage of m

ale and female participants recording energy intake by four-hourly periods per shift w

orked (n = 240)

D

ay Late

Early

Night - N

ight R

est-Night

Night-R

est R

est

n 193

106 24

33 48

42 233

%

Participants

Midnight (00:00) – 03:59

7 11

17 79

25 62

12

04:00 – 07:59 87

38 92

52 25

43 39

08:00 – 11:59 94

93 88

48 69

64 91

Midday (12:00) – 15:59

98 95

75 67

81 86

94

16:00 – 19:59 97

90 79

91 81

90 94

20:00 – 23:59 85

89 83

88 81

79 78

The darker the green shade of cells indicates the higher percentage of participants recording energy intake during this time period. C

hi squared p <0.0001

!!!!

"#%!

Figure 8.2 Percentage of participants’ energy intake by four-hourly periods per shift type (n = 240)

The mean percentage of total energy intake (error bars indicate ±S

E) is show

n per 4-hour period. Factorial AN

OV

A w

as conduced to assess if type of shift w

orked was associated w

ith the percentage of total daily energy intake consumed per 4-hour period (p = 0.0004)

! " #! #" $! $" %! %" &! &" "!

!!'!!()!%'"*

!&'!!()!+'"*

!,'!!()##'"*

#$'!!()#"'"*

#-'!!()#*'"*

$!'!!()$%'"*

!"#$"%&'()" *+*,-)"'+.&/*

0'1*"'+"2"3#4,()"5*,'#%6

./012

1/3.

0.43567893

67893567893

6789350.43

0.43

:/2

!

!

!!!

"#$!

8.2.2.2 Discussion

The exploratory analyses presented in this Chapter demonstrate the additional

eating characteristics captured by using a food diary that allows participants to

record dietary intake by time. The energy intakes reported in the current sample of

London Metropolitan police force employees were comparable to those reported in

the larger sample (Chapter 5, Table .2 - men: 2,072 SD 505 kcal day vs. 2,077 SD

473; women: 1,661 SD 381 vs. 1,674 SD 386kcal/day). The results also support

existing research that suggests a temporal redistribution of energy intake in those

working shifts (219, 222). In particular those working late shifts (shift that finishes

before midnight) consumed on average 22% of energy intake between midnight and

4a.m. This observation could have potentially adverse metabolic consequences as a

RCT found that post-prandial blood triglyceride and glucose concentrations, were

significantly higher following an identical snack consumed at 4a.m. compared to

4p.m. (129). In general, these exploratory analyses showed a trend in temporal

distribution of energy intake towards the end of the day (between 4p.m and

midnight). Consuming a higher percentage of daily energy intake during the evening

has been associated with MetS and obesity (123). In line with the results reported in

Chapter 6 SSB were consistently reported at higher intake levels by those working a

night shift. As research increasingly suggests an association between SSB intake

and adverse glycaemic control (322) reducing SSB intake may be an important

occupational health message for those working night shifts. Study 2 (Chapter 6)

found lower alcohol intake reported by shift workers working nights. Exploring intake

across different shifts found energy intake from alcohol varied across shift type with

those working a night shift followed by a day shift consuming a higher percentage of

energy from alcohol compared to day workers and late workers. Notable limitations

of this exploratory study are firstly, the small sample size and the low number of

early shift workers. Secondly the sample is drawn from a specific police employee

population (London Metropolitan) that may not be representative of the British police

force in terms of demographic and also the food environment. Lastly, questionnaire

data were not available at the time of conducting the analyses to investigate job

strain, sleep and working environment (mobile or office based) on dietary

behaviours. Nevertheless, this initial study provides an insight into how the eating

patterns of police force employees vary by shift worked.

!

!!!

"#%!

8.3 Shift work questionnaire

8.3.1 Design phase

A standardised structured self-administered questionnaire to capture shift work was

developed as part of the cohort follow-up by the author. The key aspects of shift to

measure were determined from reviewing the literature (Chapter 1, Table 1.3) and

gaining the opinions of experts in the field of shift work and health (Dr Lesley

Rushton from Imperial College, Dr Ruth Travis from Oxford University and Professor

Simon Folkard from Swansea University). The key variables of shift work identified

were:

1) Arrangement of shift work (e.g. rotating, irregular, on call)

2) Type of shifts based on standard definitions (e.g. nights, late, early, day)

3) Total shift work exposure (e.g. including previous employment)

4) Rest days between shifts

Additionally the questionnaire needed to meet the criteria listed in Table 8.5.

Table 8.5 Criteria for the Airwave Health Monitoring Health Monitoring Study shift work

questionnaire

Criteria Questionnaire development Structured questions that can be completed by participants electronically

• Closed/multiple choice questions • Clear layout

Ability to link to the original baseline questionnaire

• Keep key screening questions as per baseline questionnaire (e.g. Are you a shift worker?)

Capture historical shift work data

• Questions prior to baseline screening • Recall question covering all years of employment

Capture details on intensity and type of shift patterns

• Multiple choice questions relating to type and arrangement of shifts worked

• Providing standard definitions of shifts (i.e. morning/early shift starts before 6a.m.)

Minimise reporting error • Parameters were provide to highlight invalid responses

• Filter/contingency questions included • Lists for numeric responses • Pilot to check understanding of terminology

!

!!!

"#&!

8.3.2 Testing phase

The questionnaire was tested in colleagues who worked shifts (no identifiable or

personal details were collected). They were asked to complete the questionnaire

and time how long it took to finish and also provide any feedback regarding the

wording of questions (e.g. too technical or vague). The mean time for completion

was 9 minutes (range 4 – 15 minutes). Following amendments to wording and

questionnaire flow the online questionnaire was developed (Dennis McRobie,

Imperial College London) and then tested in-house using various scenarios

developed by the research team. The final questionnaire (Appendix 8.5) was

implemented in March 2016 into the follow-up questionnaire.

8.3.3 Preliminary data

Preliminary data from 2,161 completed shift work sections of the follow-up

questionnaire were available at the end of July 2016. The data were not linked to

baseline data and no demographic data were available to the author at the time of

analysis. Participants were classified as per the classification used in the cross-

sectional studies presented in this thesis (day worker, shift worker without nights,

shift worker with nights), and the additional classification of ‘occasional shift worker’

was added based on the screening question. Chi-squared analyses were conducted

to compare occupational characteristics across these classifications of shift work

exposure.

8.3.4 Preliminary results

The characteristics by shift work classification are presented in Table 8.6. From the

sample 32% reported not currently working shifts. Those who reported working

shifts were more likely to be employed as a constable/sergeant and have a ‘mobile’

working environment. Non-shift workers were more likely to be employed as ‘staff’

and work standard or part-time hours. Those reporting being a shift worker at follow-

up were more likely to report being a shift worker at baseline. Of participants

reporting being a current non-shift worker 69.6% reported previous exposure to shift

work, with 95.2% having previously worked nights. The most frequently reported

shift pattern was rotating (70% of shift workers without nights and 80% of shift

workers with night work).

!

!!!

"#'!

Table 8.6 Selected occupational characteristics obtained using the new shift work questionnaire during follow-up of the Airwave Health Monitoring Study (n = 2,161)

Not a shift worker

Shift worker without night work

Shift worker with night work

Occasional 2-3x year

p

N (%) 682 (31.6) 547 (25.3) 727 (33.6) 205 (9.5) N (%) Rank <0.0001 Staff 348 (51.0) 137 (25.1) 77 (10.6) 16 (2.8) Constable/Sergeant 250 (36.7) 376 (68.7) 584 (80.3) 126

(9.4)

Inspector or above 75 (11.0) 26 (4.8) 61 (8.4) 61 (27.4) Other 9 (1.3) 8 (1.5) 5 (0.7) 2 (8.3) Work environment <0.0001 Office 575 (84.3) 278 (50.8) 233 (32.1) 153 (74.6) Mobile 107 (15.7) 269 (49.2) 494 (67.9) 52 (25.4) Current shifts worked Early

- - 270 (49.4) 609 (83.8) - - Late - - 450 (82.3) 669 (92.0) - - Night - - - - 727 (100) - - On-call - - 61 (11.1) 64 (5.0) - - Other - - 48 (8.8) 0 (0.0) - - Shift pattern <0.0001 Fixed

-! -! 95 (17.4) 58 (8.0) - - Irregular

-! -! 47 (8.6) 73 (10.0) - - Rotating -! -! 386 (70.6) 582 (80.1) - - Other -! -! 19 (3.5) 14 (1.9) - - If night worker: Number of night shifts per month: 1-2 -! -! -! -! 126 (17.3) -! -! - 3-5 -! -! -! -! 91 (12.5) -! -! 6-10 -! -! -! -! 444 (61.1) -! -! 11-15 -! -! -! -! 22 (3.0) -! -! 16-20 -! -! -! -! 27 (3.7) -! -! More than 20 -! -! -! -! 17 (2.3) -! -! Weekly working hours <0.0001 Part time (<35 hrs.) 114 (16.7) 65 (11.9) 46 (6.3) 9 (3.9) 35-40 hrs. 302 (44.3) 216 (39.5) 180 (24.8) 47 (6.3) 41-48 hrs. 165 (24.2) 151 (27.6) 289 (39.8) 85 (12.3) 49 -54 hrs. 63 (9.2) 60 (11.0) 103 (14.2) 36 (13.7) ! 55hrs 38 (5.6) 55 (10.0) 109 (15.0) 28 (12.2) Shift worker at baseline* <0.0001 No 442 (64.8) 72 (13.2) 45 (6.2) 20 (9.8) Yes 213 (31.2) 464 (84.8) 666 (91.6) 110 (53.7) Occasional 27 (4.0) 11 (2.0) 16 (2.2) 75 (36.6) Total years shift work N (%) <0.0001 Nil

207 (30.4) 4 (0.2) 0 (0.0) 2 (0.9) 6 months – 2 years

24 (3.5) 4 (0.2) 4 (0.6) 2 (0.9) 3 – 9 years

133 (19.5) 97 (4.5) 88 (12.1) 58 (28.3) 10 – 19 years

202 (29.6) 244 (11.3) 377 (51.9) 90 (44.0) 20 years or more 116 (17.0) 198 (9.2) 258 (35.5) 53 (25.8) Total years night work <0.0001 Nil

23 (4.8) 74 (13.6) 2 (0.3) 3 (1.4) 6 months – 2 years

34 (7.2) 38 (7.0) 26 (3.6) 12 (5.8) 3 – 9 years

179 (37.7) 179 (33.0) 148 (20.4) 75 (36.5) 10 – 19 years

158 (33.3) 174 (32.0) 347 (47.7) 80 (39.0) 20 years or more 81 (17.1) 78 (14.4) 204 (28.1) 33 (16.1)

Chi-squared compared differences across different shift classifications for categorical variables. ‘-‘ signifies not applicable.*Retrospective data collection- participants asked to recall shift status at their screening date.

!

!!!

"#(!

8.3.5 Discussion and future work

While the results from the new shift work questionnaire support observations

reported in this thesis – e.g. associations between shift work and job role and weekly

working hours. It also provides previous shift work exposure information. This is

valuable for future research when considering the potential legacy effect of shift

work. This is important as retired shift workers have been shown to be at increased

risk of diabetes and hypertension (196).

Future research is needed to access the validity of the questionnaire (i.e. does it

actually measure shift work exposure). In particular questions relating to historical

shift work may be subject to recall bias. However, further collection and

interpretation of the radio usage data offers a potential objective measure of working

patterns over the follow-up study period. Lastly the reliability of the questionnaire

should be determined by repeating the administration of the questionnaire (test-

retest) in a cross-section of participants.

8.4 Chapter summary This chapter has detailed the development and preliminary data exploration of two

measurement tools that will facilitate the investigation of how shift work influences

eating behaviour and ultimately health outcomes. The measurement tools

developed are of value to future studies using data collected from the Airwave Health

Monitoring Study as well as other nutritional occupational cohorts.

!

!!!

"#)!

CHAPTER 9

9.0 SYNTHESIS OF RESULTS AND OVERALL CONCLUSION

9.1 Synthesis of results This thesis has characterised the diets of British police force employees and

examined the association between working hours and diet quality. It has also

investigated the relationship between working hours and cardiometabolic risk and

tested if diet quality modifies this association. This Chapter summarises the main

observations from each study presented in this thesis and provides an overall

interpretation of the results. It also discusses the key limitations of the studies

undertaken and recommends future research directions.

9.1.1 Summary of main findings

• Diet quality classified by the DASH diet score was negativity associated with

adiposity, inflammation, diastolic blood pressure and non-HDL cholesterol in

British police force employees independently of established risk factors.

• For male employees long working hours were associated with increased odds

of having a diet quality associated with elevated cardiometabolic risk.

• Shift work with night work was associated with higher SSB intake for both

men and women.

• Duration of weekly working hours was positively associated with measures of

adiposity (BMI, waist circumference and body fat) and inflammation in male

police employees.

• Diet quality determined by DASH score did not modify the relationship

between working hours and markers of cardiometabolic risk.

An overview of each research objective and respective hypothesis outcome is given

in Table 9.1.

!!!!

"##!

Table 9.1 Sum

mary of the research objectives achieved in this thesis and the outcom

e of each hypothesis

Research

objective C

hapter Sum

mary

Related hypothesis outcom

e

i) Generate

nutritional and food intake data from

7-day food records.

Chapter 3

A staff training and audit procedure w

as developed. Implem

enting the protocol resulted in a m

ean error rate of 3.5 % (S

D 3.2) per food diary

checked. The systematic application of com

posite item disaggregation

allowed for the calculation of food group intakes.

ii) Investigate the prevalence of energy intake m

isreporting and to identify the factors associated w

ith energy intake m

isreporting.

Chapter

4 (S

tudy 1) The estim

ated prevalence of potential energy intake under-reporting was

56% for m

en and 41% for w

omen, w

hich is comparable to rates reported

for the general UK

population.

In agreement w

ith previous studies under-reporting energy intake was

found to be systematic. B

MI w

as the strongest predictor of under-reporting energy intake. W

orkplace factors such as length working hours

and job strain also showed an association w

ith energy intake underreporting.

iii) Describe the

overall dietary profile of police force em

ployees across different sections of the force (sex, region, rank and w

orking hours).

Chapter

5 (S

tudy 2) C

ompared to police em

ployees in England those em

ployed in Scotland

record a poorer quality dietary pattern based on DA

SH

score. Ranked

officers compared to non-ranked police em

ployees reported a poorer quality dietary pattern based on D

AS

H score. These results reflect

observations reported in the general UK

population.

Am

ong ranked male police officers longer w

eekly working hours w

ere associated w

ith lower D

AS

H dietary score, and in particular higher sugar

sweetened beverage intake and low

er whole grain intake (independent of

age and region). Shift w

ork with night w

ork was associated w

ith higher sugar sw

eetened beverage intake for both men and w

omen.

H1 B

ritish police force employees w

ho work

atypical hours (long and/or shift work) report

a poorer diet quality (as evidenced by a low

er DA

SH

score) compared to those w

ho w

ork standard hours (35-40hrs per week /

between 7a.m

. and 6p.m.).

Hypothesis partially supported: The

positive association between diet and

working hours w

as stronger for male

employees, particularly am

ongst mid-ranked

officers.

!!!!

%$$!

Research

objective C

hapter Sum

mary


e

iv) Investigate the association of diet quality (m

easured by D

AS

H score)

and markers of

cardiometabolic

risk.

Chapter

6 (S

tudy 3) D

AS

H score w

as negatively associated with m

arkers of increased cardiom

etabolic risk (adiposity, inflamm

ation and non-HD

L cholesterol) independent of other lifestyle factors and B

MI. B

eing in the lowest fifth of

DA

SH

score was associated w

ith almost double the odds of having three

or more m

arkers of cardiometabolic risk.

Duration of w


as positively associated with a poorer

quality dietary intake (independent of established predictors of diet quality).

Supports the partial acceptance of H

1 for poor diet quality and longer w

orking hours in m

en.

v) Measure the

association betw

een number of

working hours and

markers of

cardiometabolic

risk.

Chapter

7 (S

tudy 4) D

uration of weekly w

orking hours was positively associated w

ith m

easures of adiposity (BM

I, waist circum

ference and body fat) and CR

P

in men. These associations rem

ained significant after controlling for established risk factors. In w

omen longer w


ere associated w

ith an improved lipid profile; although, com

pared to part-time

workers, w

orking >49 hours per week w

as associated with a higher B

MI.

No association betw

een acute (previous 30 days) shift work exposure

and cardiometabolic risk m

arkers were observed.

H2 B

ritish police force employees w

ho work

atypical hours (long and/or shift work)

compared to those w

ho work standard hours

(35-40hrs per week / or betw

een 7a.m. and

6p.m.) have a w

orse cardiometabolic risk

profile (as evidenced by anthropometric and

biological risk markers)

Hypothesis partially supported: The

influence of working hours on health m

ay differ betw

een sexes. The association in m

en for markers of adiposity w

as dose-response and rem

ained after controlling for established confounders. The healthy w

orker effect could potentially have biased results tow

ards the null. Further longitudinal research is needed.

!!!!

%$"!

Research

objective C

hapter Sum

mary


e

vi) Assess w

hether diet quality m

odifies markers

of cardiometabolic

risk in employees

with different

working hour

schedules.

Chapter

7 (S

tudy 4) In S

tudy 4 interaction analyses (DA

SH

score x duration of weekly

working hours) w

ere not found to be significant.

H3 D

iet quality (measured by D

AS

H score)

modifies the association betw

een atypical w

orking hours and cardiometabolic risk in

British police force em

ployees

Hypothesis not supported

vii) Develop and

pilot a revised food diary and shift w

ork questionnaire to facilitate the m

apping of diet and eating patterns to w

orking hours in future studies.

Chapter 8

A revised food diary design w

as shown to capture dietary intakes, eating

occasions and temporal eating patterns across different types of shift

work. For exam

ple late shift workers are m

ore likely to report energy intakes betw

een midnight and 4a.m

. Exploratory analyses supported the

observation from S

tudy 5 that night workers intake a higher quantity of

SS

Bs com

pared to other shifts.

The preliminary analyses of a new

shift work questionnaire highlighted

the more detailed exposure data that w

ill be captured for future studies.

Supports the partial acceptance of H

1 for a poorer diet quality in shift w

orkers.

!

!!!

"+"!

9.2 Overall interpretation and implications of findings

The findings presented in this thesis are of value to occupational and public health

practitioners. Firstly, this thesis has highlighted specific employee groups as having a

poorer diet quality, in particular male employees, employees in Scotland and those

working long weekly working hours and shift work. Secondly, this is one of a few

studies that have applied the DASH score to a large UK cohort and demonstrated an

association with cardiometabolic risk independent of other lifestyle factors and BMI.

This observation in conjunction with a recent trial in the UK conferring the DASH diet

adaptability and acceptability to a UK population (323) suggests that the DASH diet

could provide an effective intervention to improve cardiometabolic risk factors as part of

a workplace intervention. The results reported in this thesis indicate that the level of

SSBs is higher in police employees than general UK adults (based on NDNS data). In

particular higher consumption was observed in those working long hours and in those

working night shifts. The reasons for these observations cannot be determined from

the cross-sectional analyses presented in this thesis. Focus group research conducted

in the US reported that fire fighters used high calorie energy drinks to help them stay

alert during long shifts (324). The higher consumption of SSBs in UK police employees

is of potential concern given the evidence associating SSBs intake with T2DM and BMI

(71,322). Additionally the effect of SSBs on metabolic risk may be potentiated by the

time of consumption. A RCT showed that a high GI meal eaten in the evening resulted

in a greater increase in post prandial blood glucose compared to a low GI meal

consumed at the same time (325).

The findings presented in this thesis suggest that for men, working more than standard

working hours (35-40 hours per week) independently of established risk factors is

associated with a dose-response increase in measures of adiposity (BMI, waist

circumference, percentage body fat and CRP). As these cardiometabolic markers are

established risk factors for CVD and T2DM further research is needed to understand

this association. This is of particular importance as around a quarter of male

employees reported working more than 48 hours per week (compared to ~13% of the

general UK population). The results support a large meta-analysis that found weekly

working hours to be associated with adverse cardiometabolic end points (stroke and

coronary heart disease) (15). This thesis did not find diet quality to modify the

!

!!!

"+#!

association between working hours and cardiometabolic risk. However the timing and

frequency of energy intake collected in future from the Airwave Health Monitoring Study

will be an important factor to consider. Research suggests that nutrient intake can

entrain peripheral ‘clocks’ including gene transcription in hepatic cells (326). Therefore,

eating out of ‘sync’ with the biological day results (which may occur in those working

long hours and /or shift work) in a ‘clash’ between the central and peripheral clocks

resulting in metabolic derangements. In a landmark study Arble et al. demonstrated

that mice (nocturnal species) when fed in the light phase gained more weight

independent of energy intake compared to being fed in the dark phase (327). More

recent RCTs in humans have shown nocturnal eating to be associated with decreased

energy expenditure (130).

It is possible that shift work may be a potential confounding factor in the association

between long weekly working hours as shift work has a strong association with length of

hours in the Airwave Health Monitoring Study. However, insufficient shift work data

were available to test this hypothesis. The BCOPS study observed the combination of

night work and long work hours to increase obesity risk (190). Conversely a previous

meta-analysis found the association between long weekly working hours to be

independent of shift work exposure (328).

9.3 Overall strengths and limitations

One novel aspect of this thesis has been the application of 7-day food diaries in the

investigation of duration of weekly working hours and diet in a UK occupational cohort.

A strength of the studies presented in this thesis is that the dietary data were generated

from the food diaries by a team of coders adhering to a standard operating procedure

developed to improve reliability and consistency of 7-day diet diary coding. Additionally

the Airwave Health Monitoring Study includes a high proportion of male participants in

early adulthood who are unrepresented in existing UK longitudinal studies (329).

Chapter 4 discussed the acknowledged limitations in all current dietary measurement

tools - namely the accuracy of reporting by participants. Random spot urine samples

were collected from a sub-group of participants included in the studies presented.

However at the time of conducting the studies included in this thesis no reliable dietary

biomarkers were known to validate dietary intake from a random spot urine sample. A

!

!!!

"+$!

further limitation of the dietary data is responder bias. The return rate of the food

diaries was 50% and baseline diary data were not collected from forces enrolled into

the study before 2007, thus potentially limiting the internal generalizability of these

findings to all British police force employees. Although the findings presented in this

thesis are from the British police force, based on the heterogeneity of this occupational

cohort they may be generalizable to comparable occupational groups for example

paramedics and fire service.

A further methodological limitation of this thesis is that the working hour data were only

collected for shift work on a small section of the cohort. Moreover the definition used in

the baseline questionnaire (‘working outside of regular working hours’) failed to

measure the type of shift work (e.g. night or morning shift work). To address this

limitation the author used data collected from police radio records; however, these data

are biased towards radio users within the police force. Previous atypical working hour

exposure is important to measure due its potential to exert long term health

consequences (330) and also the ‘healthy worker effect’ - where employees with poor

health may be transferred to day work, therefore diminishing the strength of observed

associations (331). During the course of this PhD the author developed a shift work

questionnaire (Chapter 8) to address these limitations. An additional limitation of the

baseline data is that details were not collected from participants regarding secondary

employment, which may be an important factor in considering the link between working

hours and health outcomes. Additionally, the baseline data collected did not include

time spent commuting to/from work, which can lengthen the working day and negatively

impact health behaviours (332).

9.4 Future work In response to the results presented in this thesis further investigations specific to the

Airwave Health Monitoring Study are proposed as follows:

i) An investigation into dietary record responder bias (identification of participant

characteristics associated with returners and non-returners of the food diary)

ii) Longitudinal analyses (when follow-up data are available), which will permit

temporal investigations (e.g. is a change in working hours associated with a change

in diet quality?)

!

!!!

"+%!

iii) To investigate the association between temporal eating patterns and markers of

cardiometabolic risk (using data collected from the revised food diary)

iv) Validation and reliability testing of the shift work questionnaire

v) The potential for the future application of dietary biomarkers to validate reported

dietary intakes using the spot urine samples collected

In addition future studies exploring shift work should include qualitative as well as

quantitative data to contribute to the current understanding of how working hours

influences dietary behaviours (e.g. accessibility, psychological and physiological

factors). This would facilitate the development of occupation specific nutritional

interventions where required.

9.5 Overall conclusion This thesis has found that longer working hours in men are positively associated with

measures of adiposity, which may increase the risk of cardiometabolic diseases. This

thesis also provides evidence to suggest that police employees working longer weekly

hours or shift work have a poorer quality diet - a finding that was particularly evident in

male employees. Although this thesis has demonstrated that diet quality (measured by

the DASH score) had a significant negative association with markers of cardiometabolic

risk, diet quality did not modify the relationship between long working hours and

markers of cardiometabolic risk. These findings contribute to the knowledge of how

occupational factors influence diet, and highlights the need for further research to

understand the causal factors linking long working hours to increased health risks in

male employees. Finally, this thesis has developed a standard dietary coding protocol,

food diary and shift work questionnaire that will assist future researchers in the

investigation of diet and working hours with health outcomes.

!

! "+&!

REFERENCES

1. Office for National Statistics. Deaths Registered in England and Wales, 2013.

Statistical Bulletin. 2014.

2. Sarwar N, Gao P, Seshasai SRK, Gobin R, Kaptoge S, Di Angelantonio E, et al. Diabetes mellitus, fasting blood glucose concentration, and risk of vascular disease: a collaborative meta-analysis of 102 prospective studies. Lancet. Elsevier; 2010 Jun 26;375(9733):2215–22.

3. Diabetes UK. Diabetes in the UK 2012: Key statistics on diabetes. 2012.

4. British Heart Foundation. Coronary Heart Disease Statistics 2012.

5. Public Health England. Adult obesity and type 2 diabetes [Internet]. 2014 [cited 2016 Mar 30]. Available from: https://www.gov.uk/government/uploads/system/uploads/attachment_data/file/338934/Adult_obesity_and_type_2_diabetes_.pdf

6. Ezzati M, Riboli E. Behavioral and Dietary Risk Factors for Noncommunicable Diseases. N Engl J Med. 2013 Sep 4;369(10):954–64.

7. Prevention of cardiovascular disease at population level [Internet]. [cited 2014 Apr 6]. Available from: http://www.nice.org.uk/nicemedia/live/13024/49273/49273.pdf

8. Maintaining a healthy weight and preventing excess weight gain among children and adults | Guidance and guidelines | NICE [Internet]. NICE; 2014 [cited 2014 Dec 29]. Available from: https://www.nice.org.uk/guidance/indevelopment/gid-phg78

9. Diabetes UK. Evidence-based nutrition guidelines for the prevention and management of diabetes. 2011.

10. Yudkin J. Man’s choice of food. Lancet (London, England). 1956 May 12;270(6924):645–9.

11. Booth SL, Sallis JF, Ritenbaugh C, Hill JO, Birch LL, Frank LD, et al. Environmental and Societal Factors Affect Food Choice and Physical Activity: Rationale, Influences, and Leverage Points. Nutr Rev. 2001 Apr 27;59(3):S21–36.

12. World Health Organization. Identification and control of work-related diseases. Geneva; 1985.

13. Gan Y, Yang C, Tong X, Sun H, Cong Y, Yin X, et al. Shift work and diabetes mellitus: a meta-analysis of observational studies. Occup Environ Med. 2014 Jul 16;oemed-2014-102150-.

14. Vyas M V, Garg AX, Iansavichus A V, Costella J, Donner A, Laugsand LE, et al. Shift work and vascular events: systematic review and meta-analysis. BMJ. 2012;345:e4800.

15. Kivimäki M, Jokela M, Nyberg ST, Singh-Manoux A, Fransson EI, Alfredsson L, et al. Long working hours and risk of coronary heart disease and stroke: a systematic review and meta-analysis of published and unpublished data for 603!838 individuals. Lancet. Elsevier; 2015 Aug;386(10005):1739–46.

!

! "+'!

16. NHS. Boost your health at work [Internet]. Choices. Department of Health; 2014 [cited 2015 Jan 5]. Available from: http://www.nhs.uk/Livewell/workplacehealth/Pages/workplaceoverview.aspx

17. Workplace policy and management practices to improve the health and wellbeing of employees | Guidance and guidelines | NICE [Internet]. NICE; [cited 2015 Jul 23]. Available from: https://www.nice.org.uk/guidance/ng13

18. Association of Chief Police Officers. UK Police Directory 2012. In Chichester UK: Keyways Publishing Ltd; 2012.

19. Varvarigou V, Farioli A, Korre M, Sato S, Dahabreh IJ, Kales SN. Law enforcement duties and sudden cardiac death among police officers in United States: case distribution study. BMJ. 2014 Nov 18;349(nov18 2):g6534–g6534.

20. Zimmerman FH. Cardiovascular disease and risk factors in law enforcement personnel: a comprehensive review. Cardiol Rev. Jan;20(4):159–66.

21. Thayyil J, Jayakrishnan TT, Raja M, Cherumanalil JM. Metabolic syndrome and other cardiovascular risk factors among police officers. N Am J Med Sci. 2012 Dec;4(12):630–5.

22. Health and Safety Executive. Managing sickness absence in the police service A review of current practices. 2007.

23. Stary HC, Chandler AB, Dinsmore RE, Fuster V, Glagov S, Insull W, et al. A Definition of Advanced Types of Atherosclerotic Lesions and a Histological Classification of Atherosclerosis": A Report From the Committee on Vascular Lesions of the Council on Arteriosclerosis, American Heart Association. Circulation. 1995 Sep 1;92(5):1355–74.

24. American Diabetes Association. 2. Classification and Diagnosis of Diabetes. Diabetes Care. 2014 Dec 23;38(Supplement_1):S8–16.

25. Wilson PWF, D’Agostino RB, Parise H, Sullivan L, Meigs JB. Metabolic syndrome as a precursor of cardiovascular disease and type 2 diabetes mellitus. Circulation. 2005 Nov 15;112(20):3066–72.

26. Srivastava AK. Challenges in the treatment of cardiometabolic syndrome. Indian J Pharmacol. Medknow Publications; 2012 Mar;44(2):155–6.

27. Kaur J. A Comprehensive Review on Metabolic Syndrome. Cardiol Res Pract. 2014 Jan;2014:1–21.

28. Alberti KGMM, Eckel RH, Grundy SM, Zimmet PZ, Cleeman JI, Donato K a, et al. Harmonizing the metabolic syndrome: a joint interim statement of the International Diabetes Federation Task Force on Epidemiology and Prevention; National Heart, Lung, and Blood Institute; American Heart Association; World Heart Federation; International . Circulation. 2009 Oct 20;120(16):1640–5.

29. Ford ES, Schulze MB, Pischon T, Bergmann MM, Joost H-G, Boeing H. Metabolic syndrome and risk of incident diabetes: findings from the European Prospective Investigation into Cancer and Nutrition-Potsdam Study. Cardiovasc Diabetol. 2008;7:35.

!

! "+(!

30. Dekker JM, Girman C, Rhodes T, Nijpels G, Stehouwer CDA, Bouter LM, et al. Metabolic syndrome and 10-year cardiovascular disease risk in the Hoorn Study. Circulation. 2005 Aug 2;112(5):666–73.

31. Grundy SM. Metabolic syndrome pandemic. Arterioscler Thromb Vasc Biol. 2008 Apr 1;28(4):629–36.

32. Singh GM, Danaei G, Farzadfar F, Stevens GA, Woodward M, Wormser D, et al. The age-specific quantitative effects of metabolic risk factors on cardiovascular diseases and diabetes: a pooled analysis. PLoS One. 2013 Jan;8(7):e65174.

33. World Health Organisation. Physical Status: The Use and Interpretation of Anthropometry. Geneva; 1995.

34. Public Health England. Adult Obesity [Internet]. 2016 [cited 2016 Apr 24]. Available from: http://www.noo.org.uk/NOO_about_obesity/adult_obesity

35. Sahakyan KR, Somers VK, Rodriguez-Escudero JP, Hodge DO, Carter RE, Sochor O, et al. Normal-Weight Central Obesity: Implications for Total and Cardiovascular Mortality. Ann Intern Med. 2015 Dec 1;163(11):827–35.

36. Tanamas SK, Permatahati V, Ng WL, Backholer K, Wolfe R, Shaw JE, et al. Estimating the proportion of metabolic health outcomes attributable to obesity: a cross-sectional exploration of body mass index and waist circumference combinations. BMC Obes. BioMed Central; 2015 Jan 29;3(1):4.

37. Romero-Corral A, Lopez-Jimenez F, Sierra-Johnson J, Somers VK. Differentiating between body fat and lean mass-how should we measure obesity? Nat Clin Pract Endocrinol Metab. 2008 Jun;4(6):322–3.

38. Prado CM, Siervo M, Mire E, Heymsfield SB, Stephan BC, Broyles S, et al. A population-based approach to define body-composition phenotypes. Am J Clin Nutr. 2014 Apr 23;ajcn.113.078576-.

39. Cornier M-A, Dabelea D, Hernandez TL, Lindstrom RC, Steig AJ, Stob NR, et al. The metabolic syndrome. Endocr Rev. 2008 Dec;29(7):777–822.

40. Després J-P, Lemieux I. Abdominal obesity and metabolic syndrome. Nature. 2006 Dec 14;444(7121):881–7.

41. Sperling LS, Mechanick JI, Neeland IJ, Herrick CJ, Després J-P, Ndumele CE, et al. The CardioMetabolic Health Alliance: Working Toward a New Care Model for the Metabolic Syndrome. J Am Coll Cardiol. 2015 Sep 1;66(9):1050–67.

42. Eckel RH, Cornier M-A. Update on the NCEP ATP-III emerging cardiometabolic risk factors. BMC Med. BioMed Central; 2014 Jan 26;12(1):115.

43. Libby P, Ridker PM, Hansson GK. Inflammation in atherosclerosis: from pathophysiology to practice. J Am Coll Cardiol. 2009 Dec 1;54(23):2129–38.

44. Ross R. Atherosclerosis--an inflammatory disease. N Engl J Med. Massachusetts Medical Society; 1999 Jan 14;340(2):115–26.

45. Buckley DI, Fu R, Freeman M, Rogers K, Helfand M. C-Reactive Protein as a Risk Factor for Coronary Heart Disease: A Systematic Review and Meta-analyses for the U.S. Preventive Services Task Force. Ann Intern Med. American College of Physicians; 2009 Oct 6;151(7):483.

!

! "+)!

46. Pradhan AD, Manson JE, Rifai N, Buring JE, Ridker PM. C-reactive protein, interleukin 6, and risk of developing type 2 diabetes mellitus. JAMA. 2001 Jul 18;286(3):327–34.

47. Gordon DJ, Probstfield JL, Garrison RJ, Neaton JD, Castelli WP, Knoke JD, et al. High-density lipoprotein cholesterol and cardiovascular disease. Four prospective American studies. Circulation. 1989 Jan;79(1):8–15.

48. Navab M, Hama SY, Cooke CJ, Anantharamaiah GM, Chaddha M, Jin L, et al. Normal high density lipoprotein inhibits three steps in the formation of mildly oxidized low density lipoprotein: step 1. J Lipid Res. 2000 Sep;41(9):1481–94.

49. Third Report of the National Cholesterol Education Program (NCEP) Expert Panel on Detection, Evaluation, and Treatment of High Blood Cholesterol in Adults (Adult Treatment Panel III) final report. Circulation. 2002 Dec 17;106(25):3143–421.

50. The Emerging Risk Factors Collaboration. Major Lipids, Apolipoproteins, and Risk of Vascular Disease. Vol. 302, JAMA. American Medical Association; 2009 Nov.

51. Tabák AG, Herder C, Rathmann W, Brunner EJ, Kivimäki M. Prediabetes: a high-risk state for diabetes development. Lancet (London, England). 2012 Jun 16;379(9833):2279–90.

52. World Health Organization. Global Report on Diabetes. 2016.

53. The Expert Committee on the Diagnosis and Classification of Diabetes Mellitus*. Follow-up Report on the Diagnosis of Diabetes Mellitus. Diabetes Care. 2003 Oct 24;26(11):3160–7.

54. American Diabetes Association. Diagnosis and classification of diabetes mellitus. Diabetes Care. 2010 Jan 1;33 Suppl 1(Supplement_1):S62-9.

55. de Vegt F, Dekker JM, Jager A, Hienkens E, Kostense PJ, Stehouwer CD, et al. Relation of impaired fasting and postload glucose with incident type 2 diabetes in a Dutch population: The Hoorn Study. JAMA. 2001 Apr 25;285(16):2109–13.

56. Sattar N. Revisiting the links between glycaemia, diabetes and cardiovascular disease. Diabetologia. 2013;56(4):686–95.

57. NICE. Hypertension in adults: diagnosis and management | 1-Guidance | Guidance and guidelines | NICE. NICE; 2011;

58. Stamler J. Blood Pressure, Systolic and Diastolic, and Cardiovascular Risks. Arch Intern Med. American Medical Association; 1993 Mar 8;153(5):598.

59. Hsia J, Margolis KL, Eaton CB, Wenger NK, Allison M, Wu L, et al. Prehypertension and cardiovascular disease risk in the Women’s Health Initiative. Circulation. 2007 Feb 20;115(7):855–60.

60. World Health Organisation. WHO | Preventing chronic diseases: a vital investment. Geneva: World Health Organization; 2005.

61. Bauer UE, Briss PA, Goodman RA, Bowman BA. Prevention of chronic disease in the 21st century: elimination of the leading preventable causes of premature death and disability in the USA. Lancet. 2014;384(9937):45–52.

!

! "*+!

62. World Health Organization. Diet, Nutrition and the prevention of chronic diseases. WHO Tech Rep Ser. 2003;(916).

63. Eckel RH, Grundy SM, Zimmet PZ, Cameron A, Shaw J, Zimmet P, et al. The metabolic syndrome. Lancet (London, England). Elsevier; 2004;365(9468):1415–28.

64. Maintaining a healthy weight and preventing excess weight gain among adults and children | Guidance and guidelines | NICE. NICE;

65. Magkos F, Fraterrigo G, Yoshino J, Luecking C, Kirbach K, Kelly SC, et al. Effects of Moderate and Subsequent Progressive Weight Loss on Metabolic Function and Adipose Tissue Biology in Humans with Obesity. Cell Metab. Elsevier; 2016 Feb 12;23(4):591–601.

66. Hooper L, Martin N, Abdelhamid A, Davey Smith G. Reduction in saturated fat intake for cardiovascular disease. In: Hooper L, editor. Cochrane Database of Systematic Reviews. Chichester, UK: John Wiley & Sons, Ltd; 2015.

67. Mozaffarian D, Micha R, Wallace S. Effects on coronary heart disease of increasing polyunsaturated fat in place of saturated fat: a systematic review and meta-analysis of randomized controlled trials. PLoS Med. 2010 Mar;7(3):e1000252.

68. Murakami K, McCaffrey TA, Livingstone MBE. Associations of dietary glycaemic index and glycaemic load with food and nutrient intake and general and central obesity in British adults. Br J Nutr. Cambridge University Press; 2013 Dec 14;110(11):2047–57.

69. Livesey G, Taylor R, Livesey H, Liu S. Is there a dose-response relation of dietary glycemic load to risk of type 2 diabetes? Meta-analysis of prospective cohort studies. Am J Clin Nutr. 2013;97(3):584–96.

70. Pi-Sunyer FX. Glycemic index and disease. Am J Clin Nutr. American Society for Nutrition; 2002 Jul;76(1):290S–8S.

71. Draft Carbohydrates and Health report. Scientific consultation: 26 June to 1 September 2014. 2014.

72. Dekker MJ, Su Q, Baker C, Rutledge AC, Adeli K. Fructose: a highly lipogenic nutrient implicated in insulin resistance, hepatic steatosis, and the metabolic syndrome. Am J Physiol Endocrinol Metab. 2010 Nov 1;299(5):E685-94.

73. Tappy L, Lê K-A, Abdel-Sayed A, Binnert C, Le K, Bortolotti M, et al. Metabolic effects of fructose and the worldwide increase in obesity. Physiol Rev. American Physiological Society; 2010 Jan;90(1):23–46.

74. Bray GA. How bad is fructose? Am J Clin Nutr. American Society for Nutrition; 2007 Oct;86(4):895–6.

75. Drewnowski A. The Nutrient Rich Foods Index helps to identify healthy, affordable foods. Am J Clin Nutr. American Society for Nutrition; 2010 Apr 1;91(4):1095S–1101S.

76. Streppel MT, Sluik D, van Yperen JF, Geelen A, Hofman A, Franco OH, et al. Nutrient-rich foods, cardiovascular diseases and all-cause mortality: the Rotterdam study. Eur J Clin Nutr. Nature Publishing Group; 2014 Jun 19;68(6):741–7.

!

! "**!

77. Elliott P, Stamler J, Dyer AR, Appel L, Dennis B, Kesteloot H, et al. Association Between Protein Intake and Blood Pressure The INTERMAP Study.

78. Alkerwi A, Sauvageot N, Buckley JD, Donneau A-F, Albert A, Guillaume M, et al. The potential impact of animal protein intake on global and abdominal obesity: evidence from the Observation of Cardiovascular Risk Factors in Luxembourg (ORISCAV-LUX) study. Public Health Nutr. 2015 Jul;18(10):1831–8.

79. Jonvik KL, Nyakayiru J, Pinckaers PJ, Senden JM, van Loon LJ, Verdijk LB. Nitrate-Rich Vegetables Increase Plasma Nitrate and Nitrite Concentrations and Lower Blood Pressure in Healthy Adults. J Nutr. 2016 May 1;146(5):986–93.

80. Trichopoulou A, Costacou T, Bamia C, Trichopoulos D. Adherence to a Mediterranean diet and survival in a Greek population. N Engl J Med. 2003 Jun 26;348(26):2599–608.

81. Matsumoto C, Miedema MD, Ofman P, Gaziano JM, Sesso HD. An expanding knowledge of the mechanisms and effects of alcohol consumption on cardiovascular disease. J Cardiopulm Rehabil Prev. 34(3):159–71.

82. Chambers ES, Viardot A, Psichas A, Morrison DJ, Murphy KG, Zac-Varghese SEK, et al. Effects of targeted delivery of propionate to the human colon on appetite regulation, body weight maintenance and adiposity in overweight adults. Gut. 2015 Nov;64(11):1744–54.

83. Frost G, Sleeth ML, Sahuri-Arisoylu M, Lizarbe B, Cerdan S, Brody L, et al. The short-chain fatty acid acetate reduces appetite via a central homeostatic mechanism. Nat Commun. Nature Publishing Group; 2014 Apr 29;5.

84. Othman RA, Moghadasian MH, Jones PJ. Cholesterol-lowering effects of oat #-glucan. Nutr Rev. 2011 Jun;69(6):299–309.

85. He FJ, Li J, Macgregor GA. Effect of longer term modest salt reduction on blood pressure: Cochrane systematic review and meta-analysis of randomised trials. BMJ. British Medical Journal Publishing Group; 2013;346(5):f1325.

86. He FJ, Burnier M, Macgregor GA. Nutrition in cardiovascular disease: salt in hypertension and heart failure. Eur Heart J. 2011 Dec;32(24):3073–80.

87. Aburto NJ, Ziolkovska A, Hooper L, Elliott P, Cappuccio FP, Meerpohl JJ. Effect of lower sodium intake on health: systematic review and meta-analyses. BMJ. British Medical Journal Publishing Group; 2013;346(7041):f1326.

88. Blaustein MP, Zhang J, Chen L, Hamilton BP. How does salt retention raise blood pressure? Am J Physiol Regul Integr Comp Physiol. 2006 Mar;290(3):R514-23.

89. Bates B, Lennox A, Prentice A, Bates C, Page P, Nicholson S, et al. National Diet and Nutrition Survey: Results from Years 1-4 (combined) of the Rolling Programme (2008/2009 – 2011/12). London; 2014.

90. Mozaffarian D, Ludwig DS. Dietary guidelines in the 21st century--a time for food. JAMA. American Medical Association; 2010 Aug 11;304(6):681–2.

91. Fardet A, Boirie Y. Associations between food and beverage groups and major diet-related chronic diseases: an exhaustive review of pooled/meta-analyses and systematic reviews. Nutr Rev. 2014 Nov 18;72(12):741–62.

!

! "*"!

92. Brunkwall L, Hellstand S, Orho-Melander M, Ericson U. Overall dietary characteristics of individuals with high consumption of beverages previously associated with risk of type 2 diabetes. Diabetologia. 2015;Abstracts:S96.

93. Tucker KL, Hallfrisch J, Qiao N, Muller D, Andres R, Fleg JL. The combination of high fruit and vegetable and low saturated fat intakes is more protective against mortality in aging men than is either alone: the Baltimore Longitudinal Study of Aging. J Nutr. 2005 Mar;135(3):556–61.

94. Jacobs DR, Gross MD, Tapsell LC. Food synergy: an operational concept for understanding nutrition. Am J Clin Nutr. 2009 May 1;89(5):1543S–1548S.

95. Charles RL, Rudyk O, Prysyazhna O, Kamynina A, Yang J, Morisseau C, et al. Protection from hypertension in mice by the Mediterranean diet is mediated by nitro fatty acid inhibition of soluble epoxide hydrolase. Proc Natl Acad Sci U S A. 2014 Jun 3;111(22):8167–72.

96. Jacobs DR, Steffen LM. Nutrients, foods, and dietary patterns as exposures in research: a framework for food synergy. Am J Clin Nutr. American Society for Nutrition; 2003 Sep;78(3 Suppl):508S–513S.

97. Calton EK, James AP, Pannu PK, Soares MJ. Certain dietary patterns are beneficial for the metabolic syndrome: reviewing the evidence. Nutr Res. 2014 Jul;34(7):559–68.

98. Esposito K, Kastorini C-M, Panagiotakos DB, Giugliano D. Mediterranean diet and metabolic syndrome: an updated systematic review. Rev Endocr Metab Disord. 2013 Sep;14(3):255–63.

99. Appel LJ, Moore TJ, Obarzanek E, Vollmer WM, Svetkey LP, Sacks FM, et al. A Clinical Trial of the Effects of Dietary Patterns on Blood Pressure. N Engl J Med. 1997 Apr 17;336(16):1117–24.

100. Siervo M, Lara J, Chowdhury S, Ashor A, Oggioni C, Mathers JC. Effects of the Dietary Approach to Stop Hypertension (DASH) diet on cardiovascular risk factors: a systematic review and meta-analysis. Br J Nutr. 2015 Mar 26;113(1):1–15.

101. Liese AD, Nichols M, Sun X, D’Agostino RB, Haffner SM. Adherence to the DASH Diet is inversely associated with incidence of type 2 diabetes: the insulin resistance atherosclerosis study. Diabetes Care. 2009 Aug 1;32(8):1434–6.

102. de Koning L, Chiuve SE, Fung TT, Willett WC, Rimm EB, Hu FB. Diet-quality scores and the risk of type 2 diabetes in men. Diabetes Care. 2011 May;34(5):1150–6.

103. Malik VS, Popkin BM, Bray GA, Després J-P, Willett WC, Hu FB. Sugar-sweetened beverages and risk of metabolic syndrome and type 2 diabetes: a meta-analysis. Diabetes Care. 2010 Nov 1;33(11):2477–83.

104. Soedamah-Muthu SS, Ding EL, Al-Delaimy WK, Hu FB, Engberink MF, Willett WC, et al. Milk and dairy consumption and incidence of cardiovascular diseases and all-cause mortality: dose-response meta-analysis of prospective cohort studies. Am J Clin Nutr. 2011 Jan 1;93(1):158–71.

!

! "*#!

105. O’Connor LM, Lentjes MAH, Luben RN, Khaw K-T, Wareham NJ, Forouhi NG. Dietary dairy product intake and incident type 2 diabetes: a prospective study using dietary data from a 7-day food diary. Diabetologia. 2014 May;57(5):909–17.

106. Department of Health. UK Chief Medical Officers Alcohol Guidelines Review. 2016.

107. Fung TT, Pan A, Hou T, Chiuve SE, Tobias DK, Mozaffarian D, et al. Long-Term Change in Diet Quality Is Associated with Body Weight Change in Men and Women. J Nutr. 2015 Aug;145(8):1850–6.

108. Satija A, Bhupathiraju SN, Rimm EB, Spiegelman D, Chiuve SE, Borgi L, et al. Plant-Based Dietary Patterns and Incidence of Type 2 Diabetes in US Men and Women: Results from Three Prospective Cohort Studies. Moore SC, editor. PLOS Med. Public Library of Science; 2016 Jun 14;13(6):e1002039.

109. Bertoia ML, Rimm EB, Mukamal KJ, Hu FB, Willett WC, Cassidy A. Dietary flavonoid intake and weight maintenance: three prospective cohorts of 124"086 US men and women followed for up to 24 years. BMJ. 2016 Jan 28;352(jan28_1):i17.

110. Titan SM, Bingham S, Welch A, Luben R, Oakes S, Day N, et al. Frequency of eating and concentrations of serum cholesterol in the Norfolk population of the European prospective investigation into cancer (EPIC-Norfolk): cross sectional study. BMJ. 2001;323(7324):1286–8.

111. Kim S, Park G-H, Yang JH, Chun SH, Yoon H-J, Park M-S. Eating frequency is inversely associated with blood pressure and hypertension in Korean adults: analysis of the Third Korean National Health and Nutrition Examination Survey. Eur J Clin Nutr. Nature Publishing Group; 2014 Apr;68(4):481–9.

112. Murakami K, Livingstone MBE. Eating Frequency Is Positively Associated with Overweight and Central Obesity in U.S. Adults. J Nutr. 2015 Dec;145(12):2715–24.

113. Murakami K, Livingstone MBE. Associations between meal and snack frequency and diet quality and adiposity measures in British adults: findings from the National Diet and Nutrition Survey. Public Health Nutr. Cambridge University Press; 2015 Oct 16;1–11.

114. Aljuraiban GS, Chan Q, Oude Griep LM, Brown IJ, Daviglus ML, Stamler J, et al. The Impact of Eating Frequency and Time of Intake on Nutrient Quality and Body Mass Index: The INTERMAP Study, a Population-Based Study. J Acad Nutr Diet. 2015 Jan 22;

115. Raynor HA, Goff MR, Poole SA, Chen G. Eating Frequency, Food Intake, and Weight: A Systematic Review of Human and Animal Experimental Studies. Front Nutr. 2015 Jan;2:38.

116. Hutchison AT, Heilbronn LK. Metabolic impacts of altering meal frequency and timing - Does when we eat matter? Biochimie. 2015 Jul 29;124:187–97.

117. Leech RM, Worsley A, Timperio A, McNaughton SA. Characterizing eating patterns: a comparison of eating occasion definitions. Am J Clin Nutr. 2015 Nov 7;102(5):1229–37.

118. Sierra-Johnson J, Unden AL, Linestrand M, Rosell M, Sjogren P, Kolak M, et al. Eating meals irregularly: a novel environmental risk factor for the metabolic syndrome. Obes (Silver Spring). 2008;16(6):1302–7.

!

! "*$!

119. Pot GK, Hardy R, Stephen a M. Irregular consumption of energy intake in meals is associated with a higher cardio-metabolic risk in adults of a British birth cohort. Int J Obes (Lond). Nature Publishing Group; 2014 Mar 28;(September 2013):1–23.

120. Mekary RA, Giovannucci E, Cahill L, Willett WC, van Dam RM, Hu FB. Eating patterns and type 2 diabetes risk in older women: breakfast consumption and eating frequency. Am J Clin Nutr. 2013;98(2):436–43.

121. Farshchi HR, Taylor MA, Macdonald IA. Beneficial metabolic effects of regular meal frequency on dietary thermogenesis, insulin sensitivity, and fasting lipid profiles in healthy obese women. Am J Clin Nutr. 2005 Jan;81(1):16–24.

122. Garaulet M, Gómez-Abellán P. Timing of food intake and obesity: a novel association. Physiol Behav. 2014 Jul;134:44–50.

123. Bo S, Musso G, Beccuti G, Fadda M, Fedele D, Gambino R, et al. Consuming More of Daily Caloric Intake at Dinner Predisposes to Obesity. A 6-Year Population-Based Prospective Cohort Study. PLoS One. 2014 Jan;9(9):e108467.

124. Almoosawi S, Prynne CJ, Hardy R, Stephen AM. Time-of-day of energy intake: association with hypertension and blood pressure 10 years later in the 1946 British Birth Cohort. J Hypertens. 2013 May;31(5):882–92.

125. Odegaard AO, Jacobs DR, Steffen LM, Van Horn L, Ludwig DS, Pereira MA. Breakfast frequency and development of metabolic risk. Diabetes Care. 2013 Oct;36(10):3100–6.

126. Almoosawi S, Prynne CJ, Hardy R, Stephen AM. Time-of-day and nutrient composition of eating occasions: prospective association with the metabolic syndrome in the 1946 British birth cohort. Int J Obes (Lond). Macmillan Publishers Limited; 2013 May;37(5):725–31.

127. Almoosawi S, Prynne CJ, Hardy R, Stephen AM. Diurnal eating rhythms: association with long-term development of diabetes in the 1946 British birth cohort. Nutr Metab Cardiovasc Dis. 2013 Oct;23(10):1025–30.

128. Betts JA, Richardson JD, Chowdhury EA, Holman GD, Tsintzas K, Thompson D. The causal role of breakfast in energy balance and health: a randomized controlled trial in lean adults. Am J Clin Nutr. 2014 Jun 4;100(2):539–47.

129. Al-Naimi S, Hampton SM, Richard P, Tzung C, Morgan LM. Postprandial metabolic profiles following meals and snacks eaten during simulated night and day shift work. Chronobiol Int. 2004 Jan;21(6):937–47.

130. Bandín C, Scheer FAJL, Luque AJ, Ávila-Gandía V, Zamora S, Madrid JA, et al. Meal timing affects glucose tolerance, substrate oxidation and circadian-related variables: A randomized, crossover trial. Int J Obes (Lond). 2015 May;39(5):828–33.

131. Garaulet M, Madrid JA. Chronobiological aspects of nutrition, metabolic syndrome and obesity. Adv Drug Deliv Rev. Elsevier B.V.; 2010 Jul 31;62(9–10):967–78.

132. Bolton-Smith C, Smith WC, Woodward M, Tunstall-Pedoe H. Nutrient intakes of different social-class groups: results from the Scottish Heart Health Study (SHHS). Br J Nutr. 1991 May;65(3):321–35.

!

! "*%!

133. Marmot MG, Stansfeld S, Patel C, North F, Head J, White I, et al. Health inequalities among British civil servants: the Whitehall II study. Lancet. Elsevier; 1991 Jun;337(8754):1387–93.

134. Buden JC, Dugan AG, Namazi S, Huedo-Medina TB, Cherniack MG, Faghri PD. Work Characteristics as Predictors of Correctional Supervisors’ Health Outcomes. J Occup Environ Med. 2016 Aug;1.

135. Clark MM, Jenkins SM, Hagen PT, Riley BA, Eriksen CA, Heath AL, et al. High Stress and Negative Health Behaviors. J Occup Environ Med. 2016 Jul;1.

136. Hellerstedt WL, Jeffery RW. The association of job strain and health behaviours in men and women. Int J Epidemiol. 1997 Jun;26(3):575–83.

137. Anstey S, Tweedie J, Lord B. Qualitative study of Queensland paramedics’ perceived influences on their food and meal choices during shift work. Nutr Diet. 2016 Feb;73(1):43–9.

138. Catenacci VA, Wyatt HR. The role of physical activity in producing and maintaining weight loss. Nat Clin Pract Endocrinol Metab. Nature Publishing Group; 2007 Jul;3(7):518–29.

139. Kokkinos PF, Fernhall B. Physical activity and high density lipoprotein cholesterol levels: what is the relationship? Sports Med. 1999 Nov;28(5):307–14.

140. Hu FB. Television Watching and Other Sedentary Behaviors in Relation to Risk of Obesity and Type 2 Diabetes Mellitus in Women. JAMA. American Medical Association; 2003 Apr 9;289(14):1785.

141. US Department of Health and Human Services. The health consequences of smoking—50 years of progress [Internet]. Surgeon General. 2014 [cited 2016 May 5]. Available from: https://scholar.google.com/scholar?q=The+Health+Consequences+of+Smoking?50+Years+of+Progress:+A+Report+of+the+Surgeon+General,+2014&btnG=&hl=en&as_sdt=0,3#1

142. Craig WY, Palomaki GE, Haddow JE. Cigarette smoking and serum lipid and lipoprotein concentrations: an analysis of published data. BMJ. 1989 Mar 25;298(6676):784–8.

143. Chiolero A, Faeh D, Paccaud F, Cornuz J. Consequences of smoking for body weight, body fat distribution, and insulin resistance. Am J Clin Nutr. American Society for Nutrition; 2008 Apr;87(4):801–9.

144. Spiegel K, Tasali E, Leproult R, Van Cauter E. Effects of poor and short sleep on glucose metabolism and obesity risk. Nat Rev Endocrinol. Nature Publishing Group; 2009 May;5(5):253–61.

145. Gangwisch JE, Heymsfield SB, Boden-Albala B, Buijs RM, Kreier F, Pickering TG, et al. Short sleep duration as a risk factor for hypertension: analyses of the first National Health and Nutrition Examination Survey. Hypertension. 2006 May 1;47(5):833–9.

146. Ayas NT, White DP, Al-Delaimy WK, Manson JE, Stampfer MJ, Speizer FE, et al. A Prospective Study of Self-Reported Sleep Duration and Incident Diabetes in Women. Diabetes Care. 2003 Feb 1;26(2):380–4.

!

! "*&!

147. McEwen BS. Stress and the Individual. Arch Intern Med. American Medical Association; 1993 Sep 27;153(18):2093.

148. Brunner EJ, Hemingway H, Walker BR, Page M, Clarke P, Juneja M, et al. Adrenocortical, autonomic, and inflammatory causes of the metabolic syndrome: nested case-control study. Circulation. 2002 Nov 19;106(21):2659–65.

149. Heraclides A, Chandola T, Witte DR, Brunner EJ. Psychosocial stress at work doubles the risk of type 2 diabetes in middle-aged women: evidence from the Whitehall II study. Diabetes Care. 2009 Dec;32(12):2230–5.

150. Conry MC, Morgan K, Curry P, McGee H, Harrington J, Ward M, et al. The clustering of health behaviours in Ireland and their relationship with mental health, self-rated health and quality of life. BMC Public Health. BioMed Central; 2011 Dec 6;11(1):692.

151. Ding D, Rogers K, van der Ploeg H, Stamatakis E, Bauman AE, Danaei G, et al. Traditional and Emerging Lifestyle Risk Behaviors and All-Cause Mortality in Middle-Aged and Older Adults: Evidence from a Large Population-Based Australian Cohort. Tumwine JK, editor. PLOS Med. Public Library of Science; 2015 Dec 8;12(12):e1001917.

152. Yu E, Rimm E, Qi L, Rexrode K, Albert CM, Sun Q, et al. Diet, Lifestyle, Biomarkers, Genetic Factors, and Risk of Cardiovascular Disease in the Nurses’ Health Studies. Am J Public Health. 2016 Sep;106(9):1616–23.

153. North BJ, Sinclair DA. The intersection between aging and cardiovascular disease. Circ Res. 2012 Apr 13;110(8):1097–108.

154. Castelli WP. Epidemiology of coronary heart disease: the Framingham study. Am J Med. 1984 Feb 27;76(2A):4–12.

155. Gershon H, Gershon D. Paradigms in aging research: a critical review and assessment. Mech Ageing Dev. 2000 Aug;117(1–3):21–8.

156. Jousilahti P, Vartiainen E, Tuomilehto J, Puska P. Sex, Age, Cardiovascular Risk Factors, and Coronary Heart Disease": A Prospective Follow-Up Study of 14 786 Middle-Aged Men and Women in Finland. Circulation. 1999 Mar 9;99(9):1165–72.

157. Bonithon-Kopp C, Scarabin P-Y, Darne B, Malmejac A, Guize L. Menopause-Related Changes in Lipoproteins and Some Other Cardiovascular Risk Factors. Int J Epidemiol. 1990 Mar 1;19(1):42–8.

158. Senior PA, Bhopal R. Ethnicity as a variable in epidemiological research. BMJ. 1994 Jul 30;309(6950):327–30.

159. Gatineau M, Mathrani S. Obesity and Ethnicity [Internet]. Oxford: National Obesity Observatory; 2011 [cited 2016 Jun 18]. Available from: http://www.noo.org.uk/NOO-pub/breifing_papers

160. Kaprio J, Tuomilehto J, Koskenvuo M, Romanov K, Reunanen A, Eriksson J, et al. Concordance for type 1 (insulin-dependent) and type 2 (non-insulin-dependent) diabetes mellitus in a population-based cohort of twins in Finland. Diabetologia. 1992 Nov;35(11):1060–7.

161. Morris AP, Voight BF, Teslovich TM, Ferreira T, Segrè A V, Steinthorsdottir V, et al. Large-scale association analysis provides insights into the genetic architecture and

!

! "*'!

pathophysiology of type 2 diabetes. Nat Genet. 2012 Sep;44(9):981–90.

162. Bao W, Hu FB, Rong S, Rong Y, Bowers K, Schisterman EF, et al. Predicting risk of type 2 diabetes mellitus with genetic risk models on the basis of established genome-wide association markers: a systematic review. Am J Epidemiol. 2013 Oct 15;178(8):1197–207.

163. Frayling TM, Timpson NJ, Weedon MN, Zeggini E, Freathy RM, Lindgren CM, et al. A common variant in the FTO gene is associated with body mass index and predisposes to childhood and adult obesity. Science. 2007 May 11;316(5826):889–94.

164. Corella D, Ortega-Azorín C, Sorlí J V., Covas MI, Carrasco P, Salas-Salvadó J, et al. Statistical and Biological Gene-Lifestyle Interactions of MC4R and FTO with Diet and Physical Activity on Obesity: New Effects on Alcohol Consumption. Dasgupta K, editor. PLoS One. Public Library of Science; 2012 Dec 21;7(12):e52344.

165. Marmot MG. Fair Society , Healthy Lives [Internet]. 2010 [cited 2014 Dec 10]. Available from: http://www.instituteofhealthequity.org/projects/fair-society-healthy-lives-the-marmot-review

166. Wardle J, Haase AM, Steptoe A, Nillapun M, Jonwutiwes K, Bellisie F. Gender differences in food choice: The contribution of health beliefs and dieting. Ann Behav Med. Springer-Verlag; 2004 Apr;27(2):107–16.

167. Maguire ER, Monsivais P. Socio-economic dietary inequalities in UK adults: an updated picture of key food groups and nutrients from national surveillance data.

168. Grimm ER, Steinle NI. Genetics of eating behavior: established and emerging concepts. Nutr Rev. NIH Public Access; 2011 Jan;69(1):52–60.

169. Kivimäki M, Virtanen M, Kawachi I, Nyberg ST, Alfredsson L, Batty GD, et al. Long working hours, socioeconomic status, and the risk of incident type 2 diabetes: a meta-analysis of published and unpublished data from 222!120 individuals. lancet Diabetes Endocrinol. 2014 Sep 24;3(1):27–34.

170. Okechukwu CA. Long working hours are linked to risky alcohol consumption. BMJ. 2015 Jan 13;350(jan12_13):g7800.

171. UK Government. Maximum weekly working hours [Internet]. 2015 [cited 2016 Apr 2]. Available from: https://www.gov.uk/maximum-weekly-working-hours/overview

172. Department for Business Innovation and Skills. The Impact of the Working Time Regulations on the UK labour market: A review of evidence. 2014.

173. Au N, Hauck K, Hollingsworth B. Employment, work hours and weight gain among middle-aged women. Int J Obes (Lond). 2013 May;37(5):718–24.

174. Virtanen M, Ferrie JE, Singh-Manoux A, Shipley MJ, Vahtera J, Marmot MG, et al. Overtime work and incident coronary heart disease: the Whitehall II prospective cohort study. Eur Heart J. 2010 Jul;31(14):1737–44.

175. Gu F, Han J, Laden F, Pan A, Caporaso NE, Stampfer MJ, et al. Total and Cause-Specific Mortality of U.S. Nurses Working Rotating Night Shifts. Am J Prev Med. Elsevier; 2015 Jan 6;

176. van Drongelen A, Boot CRL, Merkus SL, Smid T, van der Beek AJ. The effects of

!

! "*(!

shift work on body weight change - a systematic review of longitudinal studies. Scand J Work Environ Health. 2011 Jul;37(4):263–75.

177. Zhao I, Bogossian F, Song S, Turner C. The association between shift work and unhealthy weight: a cross-sectional analysis from the Nurses and Midwives’ e-cohort Study. J Occup Env Med. 2011;53(2):153–8.

178. Di Lorenzo L, De Pergola G, Zocchetti C, L’Abbate N, Basso A, Pannacciulli N, et al. Effect of shift work on body mass index: results of a study performed in 319 glucose-tolerant men working in a Southern Italian industry. Int J Obes Relat Metab Disord. 2003 Nov;27(11):1353–8.

179. van Amelsvoort LG, Schouten EG, Kok FJ. Duration of shiftwork related to body mass index and waist to hip ratio. Int J Obes Relat Metab Disord. 1999 Sep;23(9):973–8.

180. Barbadoro P, Santarelli L, Croce N, Bracci M, Vincitorio D, Prospero E, et al. Rotating shift-work as an independent risk factor for overweight Italian workers: a cross-sectional study. Gong Y, editor. PLoS One. Public Library of Science; 2013 Jan;8(5):e63289.

181. De Bacquer D, Van Risseghem M, Clays E, Kittel F, De Backer G, Braeckman L. Rotating shift work and the metabolic syndrome: a prospective study. Int J Epidemiol. 2009;38(3):848–54.

182. Sookoian S, Gemma C, Fernández Gianotti T, Burgueño A, Alvarez A, González CD, et al. Effects of rotating shift work on biomarkers of metabolic syndrome and inflammation. J Intern Med. 2007 Mar;261(3):285–92.

183. Karlsson B, Knutsson A, Lindahl B. Is there an association between shift work and having a metabolic syndrome? Results from a population based study of 27,485 people. Occup Environ Med. 2001 Nov;58(11):747–52.

184. Pietroiusti a, Neri a, Somma G, Coppeta L, Iavicoli I, Bergamaschi a, et al. Incidence of metabolic syndrome among night-shift healthcare workers. Occup Environ Med. 2010 Jan;67(1):54–7.

185. Health and Social Care Information Centre. Health Survey for England - 2013. Health and Social Care Information Centre, 1 Trevelyan Square, Boar Lane, Leeds, LS1 6AE, United Kingdom; 2014.

186. Office for National Statistics. Changes in shift work patterns over the last ten years ( 1999 to 2009 ) RR887 [Internet]. 2011 [cited 2014 Dec 30]. Available from: http://www.hse.gov.uk/research/rrpdf/rr887.pdf

187. Health and Safety Executive. New guidance on managing shift work [Internet]. Vol. 2013. [cited 2013 Oct 30]. Available from: http://www.hse.gov.uk/humanfactors/resources/articles/managing-shift-work.htm

188. Hansen AB, Stayner L, Hansen J, Andersen ZJ. Night shift work and incidence of diabetes in the Danish Nurse Cohort. Occup Environ Med. BMJ Publishing Group Ltd; 2016 Apr;73(4):262–8.

189. Härmä M, Ropponen A, Hakola T, Koskinen A, Vanttola P, Puttonen S, et al. Developing register-based measures for assessment of working time patterns for epidemiologic studies. Scand J Work Environ Health. 2015;41(3):268–79.

!

! "*)!

190. Gu JK, Charles LE, Burchfiel CM, Fekedulegn D, Sarkisian K, Andrew ME, et al. Long work hours and adiposity among police officers in a US northeast city. J Occup Environ Med. 2012 Nov;54(11):1374–81.

191. Wang F, Zhang L, Zhang Y, Zhang B, He Y, Xie S, et al. Meta-analysis on night shift work and risk of metabolic syndrome. Obes Rev. 2014 Sep;15(9):709–20.

192. Brown DL, Feskanich D, Sánchez BN, Rexrode KM, Schernhammer ES, Lisabeth LD. Rotating night shift work and the risk of ischemic stroke. Am J Epidemiol. 2009 Jun 1;169(11):1370–7.

193. Pan A, Schernhammer ES, Sun Q, Hu FB. Rotating night shift work and risk of type 2 diabetes: two prospective cohort studies in women. PLoS Med. 2011;8(12):e1001141.

194. Lin Y-C, Hsiao T-J, Chen P-C. Persistent rotating shift-work exposure accelerates development of metabolic syndrome among middle-aged female employees: a five-year follow-up. Chronobiol Int. 2009 May;26(4):740–55.

195. Esquirol Y, Bongard V, Ferrieres J, Verdier H, Perret B. Shiftwork and higher pancreatic secretion: early detection of an intermediate state of insulin resistance? Chronobiol Int. Informa Healthcare New York, USA; 2012 Nov 8;29(9):1258–66.

196. Guo Y, Liu Y, Huang X, Rong Y, He M, Wang Y, et al. The effects of shift work on sleeping quality, hypertension and diabetes in retired workers. Behrens T, editor. PLoS One. Public Library of Science; 2013 Jan;8(8):e71107.

197. Scheer FAJL, Hilton MF, Mantzoros CS, Shea SA. Adverse metabolic and cardiovascular consequences of circadian misalignment. Proc Natl Acad Sci U S A. 2009 Mar 17;106(11):4453–8.

198. Kivimäki M, Batty GD, Hublin C. Shift work as a risk factor for future type 2 diabetes: evidence, mechanisms, implications, and future research directions. PLoS Med. Public Library of Science; 2011 Dec;8(12):e1001138.

199. Roenneberg T, Allebrandt K V, Merrow M, Vetter C. Social jetlag and obesity. Curr Biol. Elsevier; 2012 May 22;22(10):939–43.

200. Pimenta AM, Bes-rastrollo M, Sayon-orea C, Gea A, Aguinaga-ontoso E, Lopez-iracheta R, et al. Working hours and incidence of metabolic syndrome and its components in a Mediterranean cohort": the SUN project. Eur J Public Health. 2015;(February):1–6.

201. Collins J, Hinshaw JL, Fine E, Albanese MA. Diagnostic radiology resident compliance with recommended health guidelines: effect of resident work environment. Acad Radiol. 2008 Aug;15(8):1046–57.

202. van Amelsvoort LGPM, Jansen NWH, Kant I. Smoking among shift workers: More than a confounding factor. Chronobiol Int. 2006;23(6):1105–13.

203. Lajoie P, Aronson KJ, Day A, Tranmer J. A cross-sectional study of shift work, sleep quality and cardiometabolic risk in female hospital employees. BMJ Open. 2015 Jan 10;5(3):e007327.

204. Virtanen M, Ferrie JE, Gimeno D, Vahtera J, Elovainio M, Singh-Manoux A, et al. Long working hours and sleep disturbances: the Whitehall II prospective cohort study. Sleep. 2009 Jun;32(6):737–45.

!

! ""+!

205. Chandola T, Brunner E, Marmot M. Chronic stress at work and the metabolic syndrome: prospective study. BMJ. 2006 Mar 4;332(7540):521–5.

206. Fujishiro K, Lawson CC, Hibert EL, Chavarro JE, Rich-Edwards JW. Job strain and changes in the body mass index among working women: a prospective study. Int J Obes (Lond). 2015 Sep;39(9):1395–400.

207. Marmot MG, Bosma H, Hemingway H, Brunner E, Stansfeld S. Contribution of job control and other risk factors to social variations in coronary heart disease incidence. Lancet. 1997 Jul 26;350(9073):235–9.

208. Maruyama S, Morimoto K. Effects of long workhours on life-style, stress and quality of life among intermediate Japanese managers. Scand J Work Environ Health. 1996 Oct;22(5):353–9.

209. Nakamura K, Shimai S, Kikuchi S, Takahashi H, Tanaka M, Nakano S, et al. Increases in body mass index and waist circumference as outcomes of working overtime. Occup Med (Lond). 1998 Apr;48(3):169–73.

210. Devine CM, Farrell TJ, Blake CE, Jastran M, Wethington E, Bisogni CA. Work conditions and the food choice coping strategies of employed parents. J Nutr Educ Behav. Jan;41(5):365–70.

211. Escoto KH, French SA, Harnack LJ, Toomey TL, Hannan PJ, Mitchell NR. Work hours, weight status, and weight-related behaviors: a study of metro transit workers. Int J Behav Nutr Phys Act. 2010 Jan;7:91.

212. Fan W, Lam J, Moen P, Kelly E, King R, McHale S. Constrained choices? Linking employees’ and spouses’ work time to health behaviors. Soc Sci Med. 2015 Mar;126:99–109.

213. Reinberg A, Migraine C, Apfelbaum M, Brigant L, Ghata J, Vieux N, et al. Circadian and ultradian rhythms in the feeding behaviour and nutrient intakes of oil refinery operators with shift-work every 3--4 days. Diabète & métabolisme. 1979 Mar;5(1):33–41.

214. Lennernas L; Akerstedt, T; MH. Nutrient intake in day workers and shift workers. Work Stress An Int J Work Heal Organ. 1994;8(4):332–42.

215. Sudo N, Ohtsuka R. Nutrient intake among female shift workers in a computer factory in Japan. Int J Food Sci Nutr. 2001 Jul;52(4):367–78.

216. de Assis MAA, Nahas M V, Bellisle F, Kupek E. Meals, snacks and food choices in Brazilian shift workers with high energy expenditure. J Hum Nutr Diet. 2003 Aug;16(4):283–9.

217. de Assis MAA, Kupek E, Nahas MV, Bellisle F. Food intake and circadian rhythms in shift workers with a high workload. Appetite. 2003;40(2):175–83.

218. Pasqua IC, Moreno CR. The nutritional status and eating habits of shift workers: a chronobiological approach. Chronobiol Int. 2004;21(6):949–60.

219. Reeves SL, Newling!Ward E, Gissane C. The effect of shift!work on food intake and eating habits. Nutr Food Sci. Emerald Group Publishing Limited; 2004 Oct 13;34(5):216–21.

220. Hulsegge G, Boer JM, van der Beek AJ, Verschuren WM, Sluijs I, Vermeulen R, et

!

! ""*!

al. Shift workers have a similar diet quality but higher energy intake than day workers. Scand J Work Environ Health. 2016 Sep 15;

221. Waterhouse J, Buckley P, Edwards B, Reilly T. Measurement of, and some reasons for, differences in eating habits between night and day workers. Chronobiol Int. 2003;20(6):1075–92.

222. Esquirol Y, Bongard V, Mabile L, Jonnier B, Soulat J-M, Perret B. Shift work and metabolic syndrome: respective impacts of job strain, physical activity, and dietary rhythms. Chronobiol Int. 2009 Apr;26(3):544–59.

223. Almajwal AM. Stress, shift duty, and eating behavior among nurses in Central Saudi Arabia. Saudi Med J. 2016 Feb;37(2):191–8.

224. Wirth MD, Burch J, Shivappa N, Steck SE, Hurley TG, Vena JE, et al. Dietary Inflammatory Index Scores Differ by Shift Work Status: NHANES 2005 to 2010. J Occup Environ Med. 2014 Jan 21;

225. Seibt R, Süße T, Spitzer S, Hunger B, Rudolf M. Nutrition and health in hotel staff on different shift patterns. Occup Med (Lond). 2015 Aug;65(6):477–84.

226. Esquirol Y, Perret B, Ruidavets JB, Marquie JC, Dienne E, Niezborala M, et al. Shift work and cardiovascular risk factors: new knowledge from the past decade. Arch Cardiovasc Dis. 2011 Dec;104(12):636–68.

227. Hemiö K, Puttonen S, Viitasalo K, Härmä M, Peltonen M, Lindström J. Food and nutrient intake among workers with different shift systems. Occup Environ Med. 2015 Jul;72(7):513–20.

228. Tada Y, Kawano Y, Maeda I, Yoshizaki T, Sunami A, Yokoyama Y, et al. Association of body mass index with lifestyle and rotating shift work in Japanese female nurses. Obesity (Silver Spring). 2014 Dec;22(12):2489–93.

229. Moreira CA, Guerreiro CS, Simões E, Carolino E, David E, Mendes L, et al. SUN-PP230: The Influence of Shift Work on the Diet of Ground Workers of tap Portugal Portuguese air Transport. Clin Nutr. Elsevier; 2015 Sep 1;34:S109.

230. Balieiro LCT, Rossato LT, Waterhouse J, Paim SL, Mota MC, Crispim CA. Nutritional status and eating habits of bus drivers during the day and night. Chronobiol Int. 2014 Dec;31(10):1123–9.

231. Phiri LP, Draper CE, Lambert E V, Kolbe-Alexander TL. Nurses’ lifestyle behaviours, health priorities and barriers to living a healthy lifestyle: a qualitative descriptive study. BMC Nurs. 2014 Nov 28;13(1):38.

232. Crispim CA, Waterhouse J, Dâmaso AR, Zimberg IZ, Padilha HG, Oyama LM, et al. Hormonal appetite control is altered by shift work: a preliminary study. Metabolism. 2011 Dec;60(12):1726–35.

233. Geliebter A, Gluck ME, Tanowitz M, Aronoff NJ, Zammit GK. Work-shift period and weight change. Nutrition. 2000;16(1):27–9.

234. Escoto KH, Laska MN, Larson N, Neumark-Sztainer D, Hannan PJ. Work hours and perceived time barriers to healthful eating among young adults. Am J Health Behav. 2012 Nov;36(6):786–96.

!

! """!

235. Morikawa Y, Miura K, Sasaki S, Yoshita K, Yoneyama S, Sakurai M, et al. Evaluation of the effects of shift work on nutrient intake: a cross-sectional study. J Occup Heal. 2008;50(3):270–8.

236. Wirth MD, Burch J, Shivappa N, Violanti JM, Burchfiel CM, Fekedulegn D, et al. Association of a dietary inflammatory index with inflammatory indices and metabolic syndrome among police officers. J Occup Environ Med. 2014 Sep;56(9):986–9.

237. Ball K, Timperio AF, Crawford DA. Understanding environmental influences on nutrition and physical activity behaviors: where should we look and what should we count? Int J Behav Nutr Phys Act. 2006;3:33.

238. Lachat C, Hawwash D, Ocké MC, Berg C, Forsum E, Hörnell A, et al. Strengthening the Reporting of Observational Studies in Epidemiology-Nutritional Epidemiology (STROBE-nut): An Extension of the STROBE Statement. PLoS Med. 2016 Jun;13(6):e1002036.

239. Elliott P, Vergnaud A-C, Singh D, Neasham D, Spear J, Heard A. The Airwave Health Monitoring Study of police officers and staff in Great Britain: Rationale, design and methods. Environ Res. 2014 Sep 3;134C:280–5.

240. Kang M-Y, Park H, Seo J-C, Kim D, Lim Y-H, Lim S, et al. Long working hours and cardiovascular disease: a meta-analysis of epidemiologic studies. J Occup Environ Med. 2012 May;54(5):532–7.

241. UK Government. Part-time workers’ rights [Internet]. 2014 [cited 2016 Apr 2]. Available from: https://www.gov.uk/part-time-worker-rights

242. Altman DG, Bland JM. Measurement in Medicine: The Analysis of Method Comparison Studies. Stat. 1983 Sep;32(3):307.

243. UK Government. Night working Hours [Internet]. 2016 [cited 2016 Mar 15]. Available from: https://www.gov.uk/night-working-hours/hours-and-limits

244. Viera AJ, Garrett JM. Understanding interobserver agreement: the kappa statistic. Fam Med. 2005 May;37(5):360–3.

245. Landsbergis PA, Schnall PL, Pickering TG, Schwartz JE. Validity and reliability of a work history questionnaire derived from the Job Content Questionnaire. J Occup Environ Med. 2002 Nov;44(11):1037–47.

246. Ingre M. Excuse me, but did the IPD-work consortium just “falsify” the job-strain model? Scand J Work Environ Health. 2015 Sep 1;41(5):504–5.

247. World Health Organization. Obesity: Preventing and Managing the Global Epidemic. Report of a WHO consultation on Obesity, Geneva, 3-5 June 1997. Geneva; 1998.

248. Galobardes B. Diet and socioeconomic position: does the use of different indicators matter? Int J Epidemiol. 2001 Apr 1;30(2):334–40.

249. Craig CL, Marshall AL, Sjöström M, Bauman AE, Booth ML, Ainsworth BE, et al. International physical activity questionnaire: 12-country reliability and validity. Med Sci Sports Exerc. 2003 Aug;35(8):1381–95.

250. IPAQ scoring protocol [Internet]. 2005 [cited 2014 Oct 20]. Available from: http://www.ipaq.ki.se/scoring.htm

!

! ""#!

251. Fox-Wasylyshyn SM, El-Masri MM. Handling missing data in self-report measures. Res Nurs Health. 2005 Dec;28(6):488–95.

252. Elmståhl S, Gullberg B. Bias in diet assessment methods--consequences of collinearity and measurement errors on power and observed relative risks. Int J Epidemiol. 1997 Oct;26(5):1071–9.

253. Cohen J. Statistical Power Analysis for the Behavioral Sciences. 2nd ed. New Jersey: L. Erlbaum Associates; 1988.

254. Rennie KL, Coward A, Jebb SA. Estimating under-reporting of energy intake in dietary surveys using an individualised method. Br J Nutr. 2007 Jun;97(6):1169–76.

255. Braakhuis AJ, Stewart KL, Cox GR, Hopkins WG, Burke LM. Variability in estimation of self-reported dietary intake data from elite athletes resulting from coding by different sports dietitians. International Journal of Sport Nutrition and Exercise Metabolism. Human Kinetics Publishers, Inc.; 2003.

256. Bingham S a, Gill C, Welch a, Cassidy a, Runswick S a, Oakes S, et al. Validation of dietary assessment methods in the UK arm of EPIC using weighed records, and 24-hour urinary nitrogen and potassium and serum vitamin C and carotenoids as biomarkers. Int J Epidemiol. 1997 Jan;26 Suppl 1(1):S137-51.

257. Nelson M, Atkinson M, Darbyshire S. Food photography II: use of food photographs for estimating portion size and the nutrient content of meals. Br J Nutr. 1996 Jul;76(1):31–49.

258. Conway R, Robertson C, Dennis B, Stamler J, Elliott P. Standardised coding of diet records: experiences from INTERMAP UK. Br J Nutr. Cambridge University Press; 2004 May 1;91(5):765–71.

259. Food Standards Agency. Foodbase [Internet]. 2005 [cited 2013 Nov 10]. Available from: http://www.foodbase.org.uk/results.php?f_report_id=82

260. Food Standards Agency. Food Portion Sizes. 3rd ed. London: TSO; 2010.

261. Charrondiere R, Haytowitz D, Stadlmayr B. FAO/INFOODS Density Database Version 2.0. 2012.

262. Nelson M, Atkinson M, Meyer J. A Photographic Atlas of Food Portion Sizes. In London: MAFF Publications; 1997.

263. Cheyette C, Balolia Y. Carbs and cals and protein and fats. London: Chello Publishing; 2010.

264. Australian Bureau of Statistics. Australian Health Survey: Users’ Guide, 2011-13 [Internet]. Australian Health Survey: Users’ Guide, 2011-13. [cited 2015 Aug 6]. Available from: http://www.abs.gov.au/ausstats/[email protected]/Lookup/4363.0.55.001Chapter65142011-13

265. Prynne CJ, Wagemakers JJMF, Stephen AM, Wadsworth MEJ. Meat consumption after disaggregation of meat dishes in a cohort of British adults in 1989 and 1999 in relation to diet quality. Eur J Clin Nutr. 2009 May;63(5):660–6.

266. O’Brien MM, Kiely M, Galvin M, Flynn A. The importance of composite foods for estimates of vegetable and fruit intakes. Public Health Nutr. 2003 Oct;6(7):711–26.

!

! ""$!

267. Bowman SA, Clemens JC, Thoerig RC, Friday JE, Shimizu M, Moshfegh AJ. Food Patterns Equivalents Database 2009-10": Methodology and User Guide. 2013.

268. Willett W, Howe G, Kushi L. Adjustment for total energy intake in epidemiologic studies. Am J Clin Nutr. 1997 Apr 1;65(4):1220S–1228.

269. Mendoza JA, Drewnowski A, Christakis DA. Dietary energy density is associated with obesity and the metabolic syndrome in U.S. adults. Diabetes Care. 2007 Apr 1;30(4):974–9.

270. Ledikwe JH, Blanck HM, Khan LK, Serdula MK, Seymour JD, Tohill BC, et al. Dietary energy density determined by eight calculation methods in a nationally representative United States population. J Nutr. 2005 Feb;135(2):273–8.

271. Dennis B, Stamler J, Buzzard M, Conway R, Elliott P, Moag-Stahlberg A, et al. INTERMAP: the dietary data--process and quality control. J Hum Hypertens. 2003 Sep;17(9):609–22.

272. Brinsden HC, He FJ, Jenner KH, MacGregor GA. Surveys of the salt content in UK bread: progress made and further reductions possible. BMJ Open. British Medical Journal Publishing Group; 2013 Jun;3(6):e002936.

273. Harper H, Hallsworth M. Counting Calories: How under-reporting can explain the apparent fall in calorie intake. 2016.

274. British Heart Foundation. Portion distortion. 2013.

275. Livingstone MBE, Black AE. Markers of the Validity of Reported Energy Intake. J Nutr. 2003;133(3):895s–920S.

276. Pryer JA, Vrijheid M, Nichols R, Kiggins M, Elliott P. Who are the “low energy reporters” in the dietary and nutritional survey of British adults? Int J Epidemiol. 1997 Feb;26(1):146–54.

277. Rosell MS, Hellénius M-LB, de Faire UH, Johansson GK. Associations between diet and the metabolic syndrome vary with the validity of dietary intake data. Am J Clin Nutr. 2003 Jul;78(1):84–90.

278. Murakami K, Livingstone MBE. Prevalence and characteristics of misreporting of energy intake in US adults: NHANES 2003-2012. Br J Nutr. 2015 Oct 28;114(8):1294–303.

279. Black AE. Critical evaluation of energy intake using the Goldberg cut-off for energy intake:basal metabolic rate. A practical guide to its calculation, use and limitations. Int J Obes Relat Metab Disord. 2000 Sep;24(9):1119–30.

280. Committee on Medical Aspects of Food Policy. Dietary Reference Values for Food Energy and Nutrients for the United Kingdom: Report of the Panel on Dietary Reference Values of the Committee on Medical Aspects of Food Policy. H.M. Stationery Office; 1991.

281. Poslusna K, Ruprich J, de Vries JHM, Jakubikova M, van’t Veer P. Misreporting of energy and micronutrient intake estimated by food records and 24 hour recalls, control and adjustment methods in practice. Br J Nutr. 2009 Jul;101 Suppl:S73-85.

!

! ""%!

282. Lentjes MAH, McTaggart A, Mulligan AA, Powell NA, Parry-Smith D, Luben RN, et al. Dietary intake measurement using 7 d diet diaries in British men and women in the European Prospective Investigation into Cancer-Norfolk study: a focus on methodological issues. Br J Nutr. 2014 Feb;111(3):516–26.

283. Stallone DD, Brunner EJ, Bingham SA, Marmot MG. Dietary assessment in Whitehall II: the influence of reporting bias on apparent socioeconomic variation in nutrient intakes. Eur J Clin Nutr. 1997 Dec;51(12):815–25.

284. Prince SA, Adamo KB, Hamel ME, Hardt J, Connor Gorber S, Tremblay M. A comparison of direct versus self-report measures for assessing physical activity in adults: a systematic review. Int J Behav Nutr Phys Act. 2008 Jan;5(1):56.

285. Wilms B, Ernst B, Thurnheer M, Weisser B, Schultes B. Correction factors for the calculation of metabolic equivalents (MET) in overweight to extremely obese subjects. Int J Obes (Lond). 2014 Nov;38(11):1383–7.

286. Berta Vanrullen I, Volatier J-L, Bertaut A, Dufour A, Dallongeville J. Characteristics of energy intake under-reporting in French adults. Br J Nutr. Cambridge University Press; 2014 Apr 14;111(7):1292–302.

287. Briefel RR, Sempos CT, McDowell MA, Chien S, Alaimo K. Dietary methods research in the third National Health and Nutrition Examination Survey: underreporting of energy intake. Am J Clin Nutr. 1997 Apr;65(4 Suppl):1203S–1209S.

288. Ferrari P, Slimani N, Ciampi A, Trichopoulou A, Naska A, Lauria C, et al. Evaluation of under- and overreporting of energy intake in the 24-hour diet recalls in the European Prospective Investigation into Cancer and Nutrition (EPIC). Public Health Nutr. 2002 Dec;5(6B):1329–45.

289. Lafay L, Mennen L, Basdevant A, Charles MA, Borys JM, Eschwège E, et al. Does energy intake underreporting involve all kinds of food or only specific food items? Results from the Fleurbaix Laventie Ville Santé (FLVS) study. Int J Obes Relat Metab Disord. 2000 Nov;24(11):1500–6.

290. McCrory M a, Hajduk CL, Roberts SB. Procedures for screening out inaccurate reports of dietary energy intake. Public Health Nutr. 2002 Dec;5(6A):873–82.

291. Stunkard AJ, Messick S. The three-factor eating questionnaire to measure dietary restraint, disinhibition and hunger. J Psychosom Res. 1985 Jan;29(1):71–83.

292. Yau YHC, Potenza MN. Stress and eating behaviors. Minerva Endocrinol. NIH Public Access; 2013 Sep;38(3):255–67.

293. Mechelen KP& W van. Effectiveness and economic impact of worksite interventions to promote physical activity and healthy diet. Background paper prepared for the WHO/WEF Joint Event on Preventing Noncommunicable Diseases in the Workplace (Dalian/ China, September 2007). 2007.

294. Geaney F, Kelly C, Greiner BA, Harrington JM, Perry IJ, Beirne P. The effectiveness of workplace dietary modification interventions: a systematic review. Prev Med (Baltim). 2013 Nov;57(5):438–47.

!

! ""&!

295. Scarborough P, Morgan RD, Webster P, Rayner M. Differences in coronary heart disease, stroke and cancer mortality rates between England, Wales, Scotland and Northern Ireland: the role of diet and nutrition. BMJ Open. 2011 Jan 1;1(1):e000263.

296. Fung TT, Chiuve SE, McCullough ML, Rexrode KM, Logroscino G, Hu FB. Adherence to a DASH-style diet and risk of coronary heart disease and stroke in women. Arch Intern Med. 2008 Apr 14;168(7):713–20.

297. Martikainen P, Brunner E, Marmot M. Socioeconomic differences in dietary patterns among middle-aged men and women. Soc Sci Med. 2003;56(7):1397–410.

298. Monsivais P, Scarborough P, Lloyd T, Mizdrak A, Luben R, Mulligan AA, et al. Greater accordance with the Dietary Approaches to Stop Hypertension dietary pattern is associated with lower diet-related greenhouse gas production but higher dietary costs in the United Kingdom. Am J Clin Nutr. American Society for Nutrition; 2015 Jul;102(1):138–45.

299. National Institute for Health and Care Excellence. Cardiovascular disease: risk assessment and reduction, including lipid modification | 2-implementation-getting-started | Guidance and guidelines | NICE [Internet]. NICE; 2014 [cited 2017 Mar 4]. Available from: https://www.nice.org.uk/guidance/cg181/chapter/2-Implementation-getting-started

300. Schwingshackl L, Hoffmann G. Diet Quality as Assessed by the Healthy Eating Index, the Alternate Healthy Eating Index, the Dietary Approaches to Stop Hypertension Score, and Health Outcomes: A Systematic Review and Meta-Analysis of Cohort Studies. J Acad Nutr Diet. 2015;115(5):780–800.e5.

301. Azadbakht L, Surkan PJ, Esmaillzadeh A, Willett WC. The Dietary Approaches to Stop Hypertension eating plan affects C-reactive protein, coagulation abnormalities, and hepatic function tests among type 2 diabetic patients. J Nutr. 2011 Jun;141(6):1083–8.

302. Yanai H, Katsuyama H, Hamasaki H, Abe S, Tada N, Sako A. Effects of Dietary Fat Intake on HDL Metabolism. J Clin Med Res. Elmer Press; 2015 Mar;7(3):145–9.

303. Toledo E, de A Carmona-Torre F, Alonso A, Puchau B, Zulet MA, Martinez JA, et al. Hypothesis-oriented food patterns and incidence of hypertension: 6-year follow-up of the SUN (Seguimiento Universidad de Navarra) prospective cohort. Public Health Nutr. 2010 Mar;13(3):338–49.

304. Champagne CM, Cash KC. Assessment of salt intake: how accurate is it? Proc Nutr Soc. Cambridge University Press; 2013 Aug 1;72(3):342–7.

305. Stefan N, Fritsche A, Schick F, Häring H-U, Collaborators G 2013 RF, Menke A, et al. Phenotypes of prediabetes and stratification of cardiometabolic risk. Lancet Diabetes Endocrinol. Elsevier; 2016 Sep;4(9):789–98.

306. Conigrave KM, Hu BF, Camargo CA, Stampfer MJ, Willett WC, Rimm EB. A prospective study of drinking patterns in relation to risk of type 2 diabetes among men. Diabetes. 2001 Oct;50(10):2390–5.

!

! ""'!

307. Ardisson Korat A V, Willett WC, Hu FB. Diet, lifestyle, and genetic risk factors for type 2 diabetes: a review from the Nurses’ Health Study, Nurses’ Health Study 2, and Health Professionals’ Follow-up Study. Curr Nutr Rep. NIH Public Access; 2014 Dec 1;3(4):345–54.

308. Saneei P, Fallahi E, Barak F, Ghasemifard N, Keshteli AH, Yazdannik AR, et al. Adherence to the DASH diet and prevalence of the metabolic syndrome among Iranian women. Eur J Nutr. 2015 Apr 7;54(3):421–8.

309. Groth M V, Fagt S, Brøndsted L. Social determinants of dietary habits in Denmark. Eur J Clin Nutr. 2001 Nov;55(11):959–66.

310. Alkerwi A, Vernier C, Crichton GE, Sauvageot N, Shivappa N, Hébert JR. Cross-comparison of diet quality indices for predicting chronic disease risk: findings from the Observation of Cardiovascular Risk Factors in Luxembourg (ORISCAV-LUX) study. Br J Nutr. 2014 Dec 5;113(2):1–11.

311. Can SH, Hendy HM. Behavioral Variables Associated with Obesity in Police Officers. Ind Health. 2014 Jul 5;52(3):240–7.

312. Geaney F, Fitzgerald S, Harrington JM, Kelly C, Greiner BA, Perry IJ. Nutrition knowledge, diet quality and hypertension in a working population. Prev Med reports. 2015 Jan;2:105–13.

313. Fung TT, Pan A, Hou T, Mozaffarian D, Rexrode KM, Willett WC, et al. Food quality score and the risk of coronary artery disease: a prospective analysis in 3 cohorts. Am J Clin Nutr. 2016 Jul;104(1):65–72.

314. Baron RM, Kenny DA. The moderator-mediator variable distinction in social psychological research: conceptual, strategic, and statistical considerations. J Pers Soc Psychol. 1986 Dec;51(6):1173–82.

315. Anderson AA, Yoo H, Franke WD. Associations of Physical Activity and Obesity With the Risk of Developing the Metabolic Syndrome in Law Enforcement Officers. J Occup Environ Med. 2016 Sep;58(9):946–51.

316. Humbarger CD, Crouse SF, Womack JW, Green JS. Frequency of Metabolic Syndrome in Police Officers compared to Ncep Iii Prevalence Values. Med Sci Sport Exerc. 2004;36(5):S161.

317. Ostry AS, Radi S, Louie AM, LaMontagne AD, Torrance G, Hooper M, et al. Psychosocial and other working conditions in relation to body mass index in a representative sample of Australian workers. BMC Public Health. BioMed Central; 2006 Dec 2;6(1):53.

318. Naz F, Jyoti S, Akhtar N, Afzal M, Siddique YH. Lipid profile of women using oral contraceptive pills. Pakistan J Biol Sci PJBS. 2012 Oct 1;15(19):947–50.

319. Baillargeon J. Characteristics of the healthy worker effect. Occup Med. Jan;16(2):359–66.

320. Lindstrom J. Does higher energy intake explain weight gain and increased metabolic risks among shift workers? Scand J Work Environ Health. 2016 Jun;42(6):455–7.

321. Bøggild H. Settling the question - the next review on shift work and heart disease in 2019. Scand J Work Environ Health. 2009 May;35(3):157–61.

!

! ""(!

322. Ma J, Jacques PF, Meigs JB, Fox CS, Rogers GT, Smith CE, et al. Sugar-Sweetened Beverage but Not Diet Soda Consumption Is Positively Associated with Progression of Insulin Resistance and Prediabetes. J Nutr. American Society for Nutrition; 2016 Nov 9;jn234047.

323. Harnden KE, Frayn KN, Hodson L. Dietary Approaches to Stop Hypertension (DASH) diet: applicability and acceptability to a UK population. J Hum Nutr Diet. 2010 Feb;23(1):3–10.

324. Dobson M, Choi B, Schnall PL, Wigger E, Garcia-Rivas J, Israel L, et al. Exploring occupational and health behavioral causes of firefighter obesity: a qualitative study. Am J Ind Med. 2013 Jul;56(7):776–90.

325. Morgan LM, Shi J-W, Hampton SM, Frost G. Effect of meal timing and glycaemic index on glucose control and insulin secretion in healthy volunteers. Br J Nutr. 2012 Oct;108(7):1286–91.

326. Vollmers C, Gill S, DiTacchio L, Pulivarthy SR, Le HD, Panda S. Time of feeding and the intrinsic circadian clock drive rhythms in hepatic gene expression. Proc Natl Acad Sci U S A. 2009 Dec 15;106(50):21453–8.

327. Arble DM, Bass J, Laposky AD, Vitaterna MH, Turek FW. Circadian timing of food intake contributes to weight gain. Obes (Silver Spring). 2009;17(11):2100–2.

328. Kivimäki M, Virtanen M, Kawachi I, Nyberg ST, Alfredsson L, Batty GD, et al. Long working hours, socioeconomic status, and the risk of incident type 2 diabetes: a meta-analysis of published and unpublished data from 222!120 individuals. lancet Diabetes Endocrinol. 2015 Jan;3(1):27–34.

329. Cohort Strategic Review Subgroup. Medical Research Council. Maximising the Value of UK Population Cohorts. MRC Strategic Review of the Largest UK Population Cohort Studies. 2014.

330. Puttonen S, Viitasalo K, Harma M. The relationship between current and former shift work and the metabolic syndrome. Scand J Work Env Heal. 2012;38(4):343–8.

331. Li C-Y, Sung F-C. A review of the healthy worker effect in occupational epidemiology. Occup Med (Chic Ill). 1999 May 1;49(4):225–9.

332. Royal Society for Public Health. Health in a hurry. London; 2016.

!

! "")!

APPENDICES

A1.1 Working hours and diet literature search

Search methods

An extensive literature search was conducted using PubMed and EMBASE. The initial

search conducted 26/07/15 and repeated 30/05/16 using the following search criteria:

Search terms: Length of working hours and diet

Exposure search terms (restricted to title and/or abstract)

LONG AND WORK OR WORK AND HOURS OR OVERTIME AND WORK

Outcome search terms (restricted to title and abstract)

DIET OR DIETARY OR NUTRITION OR NUTRITIONAL OR NUTRIENTS OR NUTRITION SURVEYS

OR EATING OR FOOD OR APPETITE OR MEALS

!Search terms: Shift work and diet

Exposure search terms (restricted to title and/or abstract):

NIGHT SHIFT OR NIGHT WORK OR SHIFT WORK OR WORK SCHEDULE OR WORK PATTERN

Outcome search terms (restricted to title and abstract)

DIET OR DIETARY OR NUTRITION OR NUTRITIONAL OR NUTRIENTS OR NUTRITION SURVEYS

OR EATING OR FOOD OR APPETITE OR MEALS

Inclusion criteria:

• English language • Human • Original article • Observational study in free living workers • Diet or nutrition outcome reported • Dietary data method collection stated • Comparison group clearly stated

Additional e-mail notifications (PubMed) June – Oct 2016 included in final review.

230

Results: Length of w

orking hours and diet

Results: S

hift w

ork and diet

Articles retrieved for title and abstract review

PubMed = 41

EMBASE = 242

Articles excluded:

Inclusion criteria not met = 248

Duplicates = 10

25 for review of full text

Articles excluded:


15 articles selected

2 article identified from

reference lists

2 article from e-m

ail alerts

19 articles included in final review

Articles retrieved for title and abstract review

PubMed = 130

EMBASE = 12

Articles excluded:


Duplicates = 1

10 for review of full text

Articles excluded:


5 articles selected

2 articles identified from

reference lists

7articles included in final

review

!!"#% !

A2.1 K

ey participant characteristics compared across study sam

ples from the A

irwave H

ealth Monitoring S

tudy

A2.1.1 C

omparison betw

een cohort included (5,849 coded diaries) and excluded from the random

sample (n =10,308, refer to A

2.2)

E

xcluded from thesis sam

ple (diet data not available) n = 4,459

Included in the thesis study sample (diet data

available) n = 5,849 p

Ag

e, years (SD

) 41.3

9.1 41.4

9.3 0.46

N, %

Men

2795

62.7 3497

59.8 0.0030

Wh

ite 4322

97.1 5691

97.3 0.40

Marital statu

s

0.06

Cohabiting

720 16.6

953 16.7

D

ivorced/separated 345

7.9 467

8.2

Married

2811 64.6

3629 63.5

S

ingle 473

10.9 663

11.6

Missing*

110 2.5

137 2.3

E

du

cation

0.049

Left school before taking GC

SE

160

3.6 248

4.2

GC

SE

or equivalent 1333

29.9 1739

29.7

Vocational qualifications

287 6.4

426 7.3

A


1406 31.5

1892 32.3

B

achelor Degree or equivalent

987 22.1

1173 20.1

P


6.4 370

6.3

Ran

k

0.010

Police staff

975 30.5

1729 33.9

P

olice Constable/S

ergeant 1878

58.3 2876

56.3

Inspector/Chief Inspector

238 7.5

324 6.3

S

uper Intendant or above 41

1.3 66

1.3

Other

60 1.9

112 2.2

M

issing 1267

28.4 742

12.7

Reg

ion

0.001

England

3257 73.2

4150 71.2

S

cotland 639

14.3 1018

17.5

Wales

555 12.4

662 11.4

M

issing 4

0.1 19

0.1

!!"#" !

E

xcluded from thesis sam

ple (diet data not available) n = 4,459

Included in the thesis study sample (diet data

available) n = 5,849

p

N, %

To

tal ho

urs w

orked

per w

eek

0.71

P

art time

391 8.8

535 9.1

40 hours or less (including part tim

e) 1672

37.5 2234

38.2

41 – 48 hours 1436

32.2 1877

32.1

49 – 54 hours 495

11.1 607

10.4

55 hours or more

465 10.4

596 10.2

P

hysical activity (M

ET

s)

0.001

Low

465 10.4

706 12.1

M

oderate 1974

44.3 2686

45.9

High

2020 45.3

2457 42

B

od

y mass in

dex

0.038 H

ealthy (<25kg/m2)

1366 30.6

1927 32.9

O


2) 2142

48.0 2741

46.9

Obese (>30kg/m

2) 951

21.3 1181

20.2

Abbreviations: S

D: standard deviation. G

CS

E: G

eneral certificate of Secondary E

ducation. ME

Ts metabolic equivalents, classification by IP

AQ

guidelines. S

tudent t-test compared m

ean age between participant groups. C

hi squared test compared differences betw

een participant groups across categorical variables; m

issing data was not included in the analyses.

!!"## !

A2.1.2 C

omparison of sub-cohort w

ith the total Airw

ave Health M

onitoring Study cohort

S

ub

-coh

ort A

irwave H

ealth M

on

itorin

g S

tud

y T

otal A

irwave H

ealth M

on

itorin

g S

tud

y1

W

om

en

Men

W

om

en

Men

N (%

) 2352

3497

15,640

37.1 26,472

62.9 A

ge, years (S

D)

39.8 (9.6)

42.6 (8.9)

38.5 (9.4)

40.9 (8.9)

Marital statu

s (%)

Married

49.8

72.5

47.7

69.6

Cohabiting

20.9

13.9

20.8

14.7

Single

18.8

6.9

18.2

7.3

Divorced/separated

10.4

6.7

10.5

6.8

Wh

ite (%)

98.9

96.8

97.5

96.5

Ed

ucatio

n (%

)

Left school before taking G

CS

E

3.5

4.7

3.4

4.3

GC

SE

or equivalent 28.2

30.8

27.9

32.0

V

ocational qualifications 7.4

7.2

7.3

6.8

A


32.5

32.3

31.6

32.4

Bachelor D

egree or equivalent 21.4

19.2

21.3

19.0

P

ostgraduate qualifications 7.1

5.8

7.4

5.4

R

ank (%

)

P

olice staff 54.7

19.7

50.1

16.6

P

olice Constable/S

ergeant 39.3

67.9

43.8

71.3

Inspector/C

hief Inspector 2.6

8.9

2.7

8.7

O

ther 3.2

3.5

3.4

3.4

S

mo

king

status (%

)

N

ever smoker

67.8

41.9

65.2

68.7

Former sm

oker 22.3

13.7

21.9

21.6

C

urrent smoker

9.9

4.2

12.9

9.7

Bo

dy m

ass ind

ex (%)

Healthy (<25kg/m

2) 49.8

21.6

48.7

19.1

O


2) 34.5

55.2

33.9

54.6

O

bese (>30kg/m2)

15.7

23.2

17.3

26.3

GC

SE

: General certificate of S

econdary Education. 1. E

lliott P et al. 2014 E

nviron Res. 2014 S

ep 3;134C:280–5. R

aw data for the full enrolled cohort

not available at time of analyses for statistical com

parison.

!

! "#$!

A2.2 Random selection criteria applied for sample selection of food diaries to code from the Airwave Health Monitoring Study

!

!

%&'()*!+,-*!

./'0*1!23!(&14,5,(&64!*75)/8*8!

9!23!(&14,5,(&64!*75)/8*8!

9!23!424&)!

+&'()*!:24&)!(&14,5,(&64+!41&5*8!;*68!"<=">! $"=="!

! !=<<!

!?! #@AA! B!

!C12'!3215*+!D,4E!5),6,5+!12))*8!2/4!F!D,4E!GHI!),6J&K*!! #L#A@!! !

B=!

!?! AB$@! =M!

!N,4E!E*&)4E!+51**6,6K! #"$=<!! !

@@!

!?! #L@<! ="!

!N,4E!&/42!O/*+4,266&,1*! "LA$<!! !

ML!

!?! B=L! #!

!N,4E!52'()*4*!&/42!O/*+4,266&,1*! "@M""!! !

MM!

!?! "@"A! =<!

!N,4E!P52'()*4*P!5),6,5&)!&68!0,2)2K,5&)!8&4&!! "$LB@!! !

AB!

!?! AA$! "!

!N,4E!QH:R!DE,4*!5*))!+&'()*!! "$#$#!! !

AL!

!?! B$! <!

!N,4E!STQU!()&+'&!+&'()*!! "$"$B!! !

AL!

!?! #A! <!

!N,4E!/1,6*!+&'()*!! "$"=$!! !

A@!

!?! ==M=@! $L!

!N,4E!3228!8,&1V!! ="AB@!! !

#<!

!?! "AB@! "=!

!I&682'!+*)*54,26!23!(&14,5,(&64!321!3228!8,&1V!528,6K! =<<<<W!! !

"$!

!Randomisation procedure conducted by Dr Anne-Claire Vergnaud, Imperial College London, March 2014 W=X#"B!(&14,5,(&64!3228!8,&1,*+!D*1*!528*8!R/K!"<="!YZ&15E!"<=$!;(1,21!42!(1,21!42!1&682',+&4,26>X!23!4E*+*!#L<!D*1*!624!,65)/8*8!,6!4E*!1&682'!+*)*54,26[!!:E*!8*5,+,26!D&+!'&8*!42!,65)/8*!4E*+*!,6!4E*!3,6&)!+&'()*!42!,651*&+*!4E*!+&'()*!+,-*!&\&,)&0)*!321!4E*!+4/8,*+!5268/54*8!,6!4E,+!4E*+,+[

! ! ! ! !

!

! "#A!

A2.3 Bland-Altman Plot: Weekly working hours measured by payroll data and self reported values in a selection of police force employees enrolled in the Airwave Health Monitoring Study 2007-2008.

!"#$#

!%&$#

!%#$#

!&$#

#$#

&$#

%#$#

%&$#

"#$#

"&$#

'#$#

'&$#

(#$#

!%#$# #$# %#$# "#$# '#$# (#$# &#$# )#$#

!"##$

%$&'$()*$+,$

$&)-

$+./0()1.

%(2,

$$34

5$6&()/#)*/+.)-$+./0() 1.%(2,$$34

*%$+),-./.%+$&

0123.&$)

!%$+),-./.!4$+

!

! "#M!

A2.4 Police force regions by country

Country Forces included

England Cheshire Constabulary

Devon & Cornwall Police

Essex Police

Hertfordshire Constabulary

Merseyside Police

Norfolk Constabulary

Staffordshire Police

Suffolk Constabulary

Tayside Police

Warwickshire Police

West Midlands Police

Dorset Police

Lincolnshire Police

Metropolitan Police Service

Northern Constabulary

Scotland Lothian & Borders Police

Dumfries & Galloway Police

Fife Constabulary

Grampian Police

Central Scotland Police

Strathclyde Police

Wales Dyfed Powys Police

Gwent Police

North Wales Police

South Wales Police

!

! "#@!

A2.5 Investigation of missing data for job role/ rank by participant characteristics

Non-responder to job role/rank questionnaire

Responder to job role/rank questionnaire p

N , % 742 12.7 5107 Mean age, years (SD) 42.2 8.7 41.3 9.4 0.022 N (%) Male 454 61.2 3043.0 59.6 0.41 White 724 97.6 4967 97.3 0.66 Marital status 0.30 Cohabiting 108 15.1 845 16.9 Divorced/separated 68 9.5 399 8 Married 463 64.7 3166 63.4 Single 77 10.7 586 11.7 Missing* 26 3.5 111 2.2 Education 0.043 Left school before taking GCSE 24 3.2 224 4.4 GCSE or equivalent 249 33.6 1490 29.2 Vocational qualifications 44 5.9 382 7.5 A levels / Highers or equivalent 247 33.3 1645 32.2 Bachelor Degree or equivalent 141 19.0 1032 20.2 Postgraduate qualifications 37 5.0 333 6.5 Income (household per anum) 0.18 Less than £32,000 105 14.2 835 16.3 £32,000 - £47,999 87 11.7 584 11.4 £48,000 - £57,999 286 38.5 1998 39.1 £58,000- £77,999 196 26.1 1147 22.5 More than £ 78,000 71 9.6 542 10.3 Employment (force) country <0.0001 England 738 95.0 3337 65.3 Wales 21 2.1 997 21.0 Scotland 16 1.9 646 13.7 Physical activity 0.60 Low 89 12.0 617 12.1 Moderate 329 44.3 2357 46.2 High 617 43.7 2133 41.8 Body mass index 0.66 Healthy (<25kg/m2) 245 33.0 1682 32.9 Over weight (25 - 30kg/m2) 356 48.0 2385 46.7 Obese (>30kg/m2) 141 19.0 1040 20.4

A dummy code (missing = 0, not missing = 1) was generated, and then t-tests (continuous data) or chi-

squared (categorical data) were conducted for key participant characteristics to determine significant

differences between responders and non-responders to rank and job role

!

!!"#$ !

A2.6 Tests of collinearity betw

een categorical variables

A2.6.1 A

ssociations between categorical lifestyle and occupational variables: m

en

!"#$%"&'(%"%)('

*+),"%$-.'

/..)"&'0-)(10-&+'$.,-21

31"#'$.'4-#,1

*25&-621.%'74-#,18',-).%#6

9".:'

;-<'(%#"$.

=-#:'1.>$#-.21.%

?06($,"&'",%$>$%6'@'A11:

B2-:$.C'(%"%)('

B&115

=11:&6'0-)#('($%%$.C'7A11:+"6(8

DE'>$1A$.CF'0-)#'51#'A11:

/,)%1'(04%'1G5-()#1

'

!"#$%"&'(%"%)('

*+),"%$-.'HIHJ

/..)"&'0-)(10-&+'$.,-21

HIKLHILL

31"#('$.'4-#,1HIKH

HILHHIKM

*25&-62

1.%'74-#,18',-).%#6HIHN

HILHHIHJ

9".:'HIHO

HILPHIKP

HIKMHIHJ

;-<'(%#"$.HIHP

HIHQHIHO

HILHHIHK

HILQ=-#:'1.>$#-.2

1.%HIHJ

HILLHILP

HIKLHIHR

!"#$HILN

?06($,"&'",%$>$%6'@'A11:

HIHNHIHQ

HIHNHIHK

HIHNHIHQ

HIHMHIHM

B2-:$.C'(%"%)('

HILHHILH

HIHQHIHN

HIHQHIHO

HIHNHIHP

HIHNB&115

HIHNHIHQ

HIHQHIHM

HIHNHIHM

HIHPHIHN

HIHRHIHN

=11:&6'0-)#('($%%$.C'7A

11:+"6(8HIHQ

HIHRHIHP

HIHRHIHM

HIHRHIHP

HILNHIHJ

HIHNHIHN

DE'>$1A$.CF'0-)#'51#'A

11:HIHP

HIHPHIHN

HIHPHIHK

HIHNHIHQ

HIHKHIHO

HIHLHIHQ

HILO/,)%1'(04%'1G5-()#1

HIHOHILH

HILNHILR

HILP!"%$

HILP!"#&

HIHQHIHM

HIHNHIHJ

HIHQ=-#:'0-)#(

HIHQHIHP

HILMHIHP

HIHO!"$'

HILKHILP

HIHMHIHN

HIHQHIHP

HIHRHILP

C

ramer’s V

was used to m

easure association between categorical variables, collinearity w

as defined as a strong association based on degrees of freedom

per Cohens’ guidelines (S

tatistical Pow

er Analysis for the B

ehavioural Sciences, 1988). B

old values indicate strongly associated variables. !

!!"#% !

A2.6.2 A

ssociations between categorical lifestyle and occupational variables: w

omen

!"#$%"&'(%"%)('

*+),"%$-.'

/..)"&'0-)(10-&+'$.,-21

31"#'$.'4-#,1

*25&-621.%'74-#,18',-).%#6

9".:'

;-<'(%#"$.

=-#:'1.>$#-.21.%

?06($,"&'",%$>$%6'@'A11:

B2-:$.C'(%"%)('

B&115

=11:&6'0-)#('($%%$.C'7A11:+"6(8

DE'>$1A$.CF'0-)#'51#'A11:

/,)%1'(04%'1G5-()#1

'

*+),"%$-.'HILK

/..)"&'0-)(10-&+'$.,-21

!"##HIHO

31"#('$.'4-#,1HILN

HILLHILN

*25&-62

1.%'74-#,18',-).%#6HIHN

HILLHILH

9".:'HIHJ

HILKHIKH

HILRHIHJ

;-<'(%#"$.HIHP

HIHJHIHJ

HIHRHIHN

HIHR=-#:'1.>$#-.2

1.%HILK

HILHHILM

HILOHIHQ

!"#$HIHJ

?06($,"&'",%$>$%6'@'A11:

HIHQHIHP

HIHRHIHQ

HIHKHIHO

HIHMHIHR

B2-:$.C'(%"%)('

HILHHIHR

HIHRHIHN

HIHQHIHP

HIHQHIHQ

HIHPB&115

HIHPHIHR

HIHPHIHQ

HIHPHIHR

HIHPHIHP

HIHPHIHP

=11:&6'0-)#('($%%$.C'7A

11:+"6(8HIHM

HIHPHIHN

HIHPHIHM

HILMHIHM

HILOHIHP

HIHQHIHP

DE'>$1A$.CF'0-)#'51#'A

11:HIHP

HIHNHIHN

HIHQHIHQ

HIHRHIHN

HIHNHIHR

HIHMHIHM

HILO/,)%1'(04%'1G5-()#1

HILKHILK

HILQHIKH

HIKP!"%!

HILM!"%#

HIHQHIHR

HIHJHILO

HIHQ=-#:'0-)#(

HILQHIHO

HIHJHIHP

HIHN!"&'

HILQHILO

HIHPHIHK

HIHNHIHQ

HIHNHIKQ

Cram

er’s V w

as used to measure association betw

een categorical variables, collinearity was defined as a strong association based on degrees of

freedom per C

ohens’ guidelines (Statistical P

ower A

nalysis for the Behavioural S

ciences, 1988). Bold values indicate strongly associated variables.

!

! "#$!

A3.1 Standard operating procedure for coding of food diaries in the Airwave Health Monitoring Study

!"#$%&'()'%*+,(-./"+.#"/0(1+234(5(1+%/3%#3(6#.+.7.*(8.#(8..3(3"%#4(7.3"/0(29"/0(:"'+6*%/( ( (

!"#$%!$%!&"

(

(

!"#$%&'()'%*+,(-./"+.#"/0(1+234(

(

1+%/3%#3(6#.+.7.*(8.#(8..3(3"%#4(7.3"/0(29"/0(:"'+6*%/(

/2+#"+"./%*(9.8+$%#'(

("

'()(*+,(-"./0"1234(*"56.7+89"1(.(32":;6<7(89"=2>4;/8"?2@."28-"A)+88("B3B((*"

CD>;6>6+8"28-"'6(>(>637"1(7(2;34"5;+D,"

E@,(;62*"F+**(G("

?+8-+8"

!

! "#%!

!"#$%&'()'%*+,(-./"+.#"/0(1+234(5(1+%/3%#3(6#.+.7.*(8.#(8..3(3"%#4(7.3"/0(29"/0(:"'+6*%/( ( (

!"!#$%#$%&"

(

;./+'/+9( ( ( ( ( ( ( ( ( ( <%0'"

(=> ?/+#.327+"./( ( ( ( ( ( ( ( ( @( (

A> 1'++"/0(26(+,'(%99'99B'/+(./(:"'+6*%/( ( ( ( ( C(

@> 1'*'7+"/0(8..3(7.3'9( ( ( ( ( ( ( ( D(

C> <.#+"./(9"E'(%*0.#"+,B9( ( ( ( ( ( ( =D(

D> F##.#(7,'7G"/0( ( ( ( ( ( ( ( ( A=(

H'8'#'/7'9( ( ( ( ( ( ( ( ( ( AA(

!66'/3"7'9( ( ( ( ( ( ( ( ( ( A@( (

(

(

(

(

(

(

(

(

(

(

(

(

(

(

(

(

!

! "#"!

!"#$%&'()'%*+,(-./"+.#"/0(1+234(5(1+%/3%#3(6#.+.7.*(8.#(8..3(3"%#4(7.3"/0(29"/0(:"'+6*%/( ( (

!"#$%&$%&'"

;<=(>/+#.327+"./((

()*"+,--,./01"231*4"25,6/7*"30",89-/0*",+"),."9,"/0289"307":,7*"9)*"+,,7"7/35/*4"+5,;"9)*"</5.36*"=*3-9)">,0/9,5/01"?987@"A*B3;2-*"322*07/BCD""()*"25/;35@"3/;",+"9)/4"25,9,:,-"/4"9,"5*78:*"*55,5"307"9,"49307357/4*":,7*"307"2,59/,0"4*-*:9/,0D""E-*34*"5*37"9)5,81)"F*+,5*"@,8"49359"9,":,7*"307"/0289"9)*"+,,7"7/35/*4D"""G9)*5"5*4,85:*4"@,8"./--"0**7"9,"5*+*5"9,"+,5":,7/01H""</5.36*"?987@"80/I8*"+,,7":,7*"F,,J"307"9)*"5*4,85:*4"-/49*7",0"231*"##",+"9)/4"7,:8;*09D""<--":,7*54"4),8-7"J**2"3".**J-@"-,1",+"30@"/448*4"5*1357/01":,7*"%"2,59/,0"4*-*:9/,0"9,"7/4:844"39"9)*".**J-@";**9/01D""""KLMNO?PGQ"MRP(KRP<"+,5"+,,7"7/35@"*095@"/09,"S/*92-30H"

• N*44"9)30",0*"73@":,;2-*9*7"• >*3-"5*2-3:*;*09"7/*9"5*:,57*7"F@"9)*"2359/:/2309"• P0:,;2-*9*"7/35/*4"35*":,7*7"F89"/0:,;2-*9*"73@4"35*"0,9"*D1D"/+"73@4"&T'"35*":,;2-*9*"

F89"73@"U"7,*4"0,9"+,--,."9)*"43;*"2399*50"34"9)*"25*6/,84"73@4V""73@4"&T'"35*":,7*7"F89"U"/4"0,9"

P+"3"7/35@"/4"0,9"*09*5*7",09,"S/*92-30"9)*0"2-*34*":,;2-*9*"9)*"W,,7"S/35@"N,1"./9)"9)*"3225,25/39*"5*34,0"AKB:*-"+/-*C""

"""""""""""""

((

(

(

(

!

! "$#!

!"#$%&'()'%*+,(-./"+.#"/0(1+234(5(1+%/3%#3(6#.+.7.*(8.#(8..3(3"%#4(7.3"/0(29"/0(:"'+6*%/( ( (

!"#$%&$%&'"

;<=(>/+'#"/0(8..39("/+.(:"'+6*%/((

1'++"/0(26(3%4(%/3(?'%*(9+#27+2#'(

(

()*+,"+-./"0+.*12)"23"*/+"4-156"41-,6"7)4+,"*/+"8+-5")789+,0"-0"32552:0;"

:"'+(6*%/( @..3(3"%#4(9'7+"./(<+-5"&" =+32,+"9,+->3-0*"<+-5"#" =,+->3-0*"<+-5"?" <14@82,)1)A"<+-5"!" B7)./"<+-5"C" D+-"<+-5"'" (E+)1)A"8+-5"<+-5"F" B-*+,"+E+)1)A"<+-5"G" HI)6*/1)A"+50+JK"""

"

1+%#+(8#.?(3%4(A(?'%*(A(%/3('/+'#(+,'("+'?9(%9($#"++'/("/(+,'(3"%#4(

! L+8+89+,"*2"0-E+"627,":2,>"-0"627"A2"*2"-E214"-)6"4-*-"5200K"

(

(

!

! "$$!

!"#$%&'()'%*+,(-./"+.#"/0(1+234(5(1+%/3%#3(6#.+.7.*(8.#(8..3(3"%#4(7.3"/0(29"/0(:"'+6*%/( ( (

!"#$%&$%&'"

;<=(1'*'7+"/0(8..3(7.3'9(

;<>(?'/'#%*(#2*'9(

! ()"*+,-.,./012"31,145,"6()*7".4",85"95:1+2,"91,1;145<""=02>"+45"6;54?/@5"-5A.?57"BCD"109"6+45-"19959"://94E".<5<"F10+12"91,1;145"50,->"6(GH7"91,1;1454".:".,I4"1"://9":-/F",85"J.-K1L5"M512,8"N/0.,/-.0O""G,+9>"6JP7""A/95;//@"109"?-5:.Q59"K.,8"RJPI"

! *5L5-"4525A,"1"A/95":/-"10".,5F",81,"4,1,54"S6:1,4"/02>7T"1,",85"509"/:",85"954A-.?,./0<"! U,5F4"4+A8"14"9-.0@4"109"4109K.A854"48/+29";5";-/@50".0,/",85.-"A/04,.,+50,"?1-,4""! U,5F4"48/+29";5"A/959".0",85":/-F".0"K8.A8",85>"K5-5"51,50"5<O<"-1K",/F1,/"%";/.259"?/,1,/"

K85-5"?/44.;25"+02544",85"/?,./0".4"0/,"1L1.21;25<""

!*0.#"+,@(8.#(7.3'(9'*'7+"./A((

!

"

"

"

"

"

"

"

"

"

"

"

"

"

(

(

(

(

(

U4",85-5"1"A/95".0",85"()*"91,1;145",81,"F1,A854",85"://9"-5A/-959V"

>54"

G525A,",8.4"A/95"109"-5:5-",/",85"?/-,./0"12O/-.,8F4"

*/"

G51-A8",85"JP"A/95";//@<""U4",85-5"1"A/95",81,"F1,A854",85"://9V"

>54"

G525A,",8.4"A/95"109"-5:5-",/",85"?/-,./0"12O/-.,8F4"

*/"

U4".,"1"A/F?/4.,5"9.48"/-"-5A.?5V" >54"

B5:5-",/",85"O+.952.054":/-"A/F?/4.,5"9.4854"/0",85":/22/K.0O"?1O54"

*/"

U4",85-5"10"5W+.L1250,"()*"A/95"/-"A/F;.01,./0"/:"A/954",81,"A10";5"+459"65<O<"5W+.L1250,"://9"A1,5O/->"109"0+,-.,./012"?-/:.257V"

3.4A+44"K.,8"-5451-A8",51F"

>54"

(45",8.4%,8545"A/95"109"F1@5"1"0/,5".0">/+-"2/O";//@",/"1O-55"1,",85"K55@2>"F55,.0O<""=0"1O-55F50,",8.4"K.22";5"19959",/",85"JP"A/95;//@"

*/"

!

! "$%!

!"#$%&'()'%*+,(-./"+.#"/0(1+234(5(1+%/3%#3(6#.+.7.*(8.#(8..3(3"%#4(7.3"/0(29"/0(:"'+6*%/( ( (

!"#$%&$%&!"

;<=(>.?6.9"+'(:"9,'9((

'()"*+,,+-./0".1)23"3(+4,5"6)"67+8)/"./1+"1().7"9+/31.14)/1":;713"-()7)":+33.6,)<"

= >+1"6)?)7;0)3"@)A9):1"1(+3)"2;5)"4:"-.1("-;1)7B"= C;/5-.9()3""= >+2)"9++8)5"2);,"-()/"*4,,"7)9.:)":7+?.5)5""= >+2)2;5)"6470)73"

"o @A%?6*'(BC(

Instant coffee with semi-skimmed milk and white sugar"D/1)7";3<"EFG"&H=&IJ"./31;/1K"2;5)"4:"-.1("-;1)7"EFG"&#=L&L"C)2.=38.22)5K":;31)47.3)5"2.,8K";?)7;0)"EFG"&H=$!L"340;7K"-(.1)"

D2*'9(8.#('/+'#"/0(7.?6.9"+'(3"9,'9C(

&MN;//)5"7);5O"2);,3<".M)M"3:;0()11.K"3+4:K"6);/3")19M"(

C),)91"EFG"9+5)"-(.9("31;1)3"P9;//)5"Q7)();1)5RM""

#MS.3("31;1)5";3"P(+2)2;5)P";/5"-()7)";"7)9.:)".3":7+?.5)5<""

D/1)7";,,"./07)5.)/13";3"1()";2+4/1"9+/342)5";/5"./"1().7"9+77)91"31;1)Q".M)M"9++8)5"%"7;-"5.7)91,O"./1+"1()";33)332)/1"@.M)M"5+"/+1"3)1"4:";"3):;7;1)"7)9.:)"9+5)BM""

LMS.3("31;1)5";3"R(+2)2;5)R";/5"9+2:,)AT"7)9.:)":7+?.5)5<".M)M"./07)5.)/13";7)"/+1";?;.,;6,)"./"1().7"9+77)91"31;1)"./"1()"5;1;6;3)""UGS"1()"7)9.:)":7+?.5)5".3"/+1"9+2:;7;6,)"1+"1(+3)"./"1()"EFG"5;1;6;3)M"N()98"1()"31;/5;75"7)9.:)3"43)5".*",.31)5"./"V9N;/9)";/5"W.55+-3+/3"@#BM"

T)M0M""7)9.:)"-.1("*4,,"X4;/1.1.)3":7+?.5)5K"641"5.**.94,1"1+")31.2;1)";914;,":7+:+71.+/");1)/M"

Y/"1(.3"./31;/9)";"7)9.:)"3(+4,5"6)"3)1"4:"43./0"1()"PZ)9.:)";/;,O3.3R"+:1.+/<"

'()"9+5)"*+7"1()"7)9.:)"3(+4,5"6)"1()":;71.9.:;/1"9+5)":7)*.A)5"6O"UWM"""

Y*"24,1.:,)"7)9.:)3";7)"7)X4.7)5"*+7"+/)":;71.9.:;/1"1()/";55"P;RK"P6RK"P9R[")19MM"";1"1()")/5"+*"1()"9+5)M"

!"#"$%&'()*+$

'()"7)9.:)"/;2)"3(+4,5"31;1)"1()"5)397.:1.+/"+*"1()"5.3(";/5"1()":;71.9.:;/1"9+5)"

!"#"$$$,-./0$1+2!$3$-!1/4!$56-$%&'()*$

\,);3)")/347)"1(;1"1()"9+77)91"P07+4:"9+5)R".3";33.0/)5"1+"1()"7)9.:)M"

\,);3)"/+1)"1()3)"7)9.:)3"3(+4,5"/+1"6)"./9,45)5"./"1()"U.7-;?)"9+5)6++8";3"1()O";7)"4/.X4)"1+");9(":;71.9.:;/1M"

!

! "$&!

!"#$%&'()'%*+,(-./"+.#"/0(1+234(5(1+%/3%#3(6#.+.7.*(8.#(8..3(3"%#4(7.3"/0(29"/0(:"'+6*%/( ( (

!"#$%&$%&'"

()*+,-",./.01"/,"2-3404/10"2"/51"53""607+80"8639+101:""

;0<07."=>?"7310"@-+7-",./.0,"/"2-3404/10A"906,+35"3B".-+,"1+,-C"+B"/9/+</D<0)"

E)F0/1G%H+763@/90"40/<"I+57<J1+5K"60/1G".3"0/."8+LL/C"8+0,M"@+.-"D6/51"5/40"8639+101:"""

N-07O".-0"60./+<06"%"4/5JB/7.J606"@0D,+.0"/51",0<07.".-0"7<3,0,."0PJ+9/<05."7310"B634"1+0."8</5".-/."60B<07.,".-0"B331"0/.05"/51".-0"4/7635J.6+05."863B+<0)"

?Q:".-+,"4/G"+593<90",0<07.+5K"#"36"4360"7310,".3"734D+50".3"8639+10"/",+4+</6"B331%5J.6+05."863B+<0)"""

;00"D0<3@"B36"-3@".3"4/.7-"B331,"53."35".-0"1/./D/,0)(

')R/,."B331"DJ6K06,:""

S5.06"/,".-0"7348<0.0"B331"I0)K)"Q+K"H/7"0.7)M"

TB".-060"+,"53."/"7310"+5"=>?"1/./D/,0"8<0/,0"7-07O".-0"0U70<"V+6@/90"7310"D33O)""TB".-060"+,"53"7310"+5".-0"7310"D33O".-05"/"607+80"@+<<"D0"60PJ+601"I,00"607+80",0."J8"863.373<M)"

!)"F0./+<(,/51@+7-0,:"

*3"53."J,0",/51@+7-0,"+5"*+0.8</5C"4/.7-".-0",/51@+7-".3".-3,0"35<+50)";/51@+7-0,"7/5"D0"B3J51"35"W0,73"/51"X/+.63,0"@0D,+.0,)"=,0"X/+.63,0"B36"HY;"/51"X/+.63,0",/51@+7-0,"35<G"I+)0)"Z604+J4%<JUJ6G".G80",/51@+7-0,M"/51"W0,73A,"B36"/<<"3.-06"D6/51,"+B"D6/51"53.",./.01"J,0"W0,73"/,"10B/J<.)""

;..39(%/3(<'%*9(/.+("/(+,'(3%+%=%9'(

• ;/51@+7-0,""• X0+K-."@/.7-06,%"601J701"B/.C"7/<36+0C",JK/6"0.7)"40/<,"• R331,"B634"7/B[,"/51"B/,."B331,"0)K)"73,./C"DJ6K06"O+5KC">RN"0.7)"

"\3@".3"7310"

• R+51".-0"5J.6+.+35/<"+5B364/.+35:"&) N-07O"V+6@/90"7310D33O"B36"+.04""#) N-07O"Z36.+35"T5B364/.+35"B3<106"]) N-07O"60./+<06,"@0D,+.0""() ;0/67-"35<+50""

"• N/<36+0"4/.7-"I,00",76005",-3."D0<3@M"

&) N<+7O"2J,0"5J.6+05."0U7-/5K0,A"DJ..35"#) N-33,0"O7/<""]) WG80"+5".-0"/43J5."3B"7/<36+0,"G3J"@/5.".3"4/.7-"+5"20U7-/5K0"9/<J0A"() N<+7O"29+0@"73483,+.+35A"E) TB".-0"863.0+5C"7/6D3-G16/.0,C",JK/6,C"B/."/60"@+.-"&$^".-+,"/43J5."7/5"D0"J,01""

• TB".-0"9/<J0,"B36"863.0+5C"7/6D3-G16/.0,C",JK/6,C"B/."/60"4360".-/5"&$^"1+BB0605."I-+K-06"36"<3@06M".-/5".-/.",./.01"35".-0"60./+<06%4/5JB/7.J606"107</6/.+35:"""

o N-07O"@-/."5J.6+05."+,"3J.,+10"3B".-0"&$^"6/5K0C"0)K)"863.0+5"+,".33"<3@"

!

! "$'!

!"#$%&'()'%*+,(-./"+.#"/0(1+234(5(1+%/3%#3(6#.+.7.*(8.#(8..3(3"%#4(7.3"/0(29"/0(:"'+6*%/( ( (

!"#$%&$%&'"

o ())"*++),"-,./0"123""/0#'3"'/+9(*"9+"4+/56"./75-)3"*++),".*"1236"893"+/"123"./093).3/1"5.,1:"&; <*"=+93">9+13./".,"93?-.93)"8))"=+93"+*"123">9+13./",+-973"

8; <*"789@+26)9813,"893"93?-.93)"8))"=+93"+*"123"789@+26)9813",+-973""! A3"7893*-5"8,",+=3"3;0;">8,18B"@938)"85,+"7+/18./">9+13./""! <*",-089".,"/33)3)"8))",-089;""

• C2393"893").**393/1"16>3,"+*",-089;"D+=3">9+)-71,"28E3"./093).3/1,"F2.72"893"123",8=3"3;0;"./E3913)",-089B"F2.13",-089"8/)"2+/36B"05-7+,3"*9-71+,3",69->",+"+/56"7+)3"F2.13",-089;";'(<"/382*(.8(%33"/0(920%#(%9("+($"**("/7#'%9'(7%#=.,43#%+'9>(

• <*"*81".,"/33)3)"8))"+.5"+9"*81"4*9+="123",+-973"+*"*81"./"123">9+)-71:B"7237G",81-9813)"*81;""! H/3"16>3"+*"*81"=86"/+1"@3",-.18@53B"@+12">85="+.5"8/)"98>3,33)"+.5"=86"@3"

93?-.93);""

""D+=3"*++),"3;0;"A-9039"I./0"@-9039J,B">8,18",858),B"D-@F86B",8/)F.723,"=-,1"@3"7+)3)",3>8981356"• K+)3"123"=8./"./093).3/1,"-,./0"123"./093).3/1"5.,1"

&; K85+9.3"=8172"123"./093).3/1,"-,./0"123.9">9.=896"/-19.3/1"" " L;0;"8"28=",858)",8/)F.72""

! K85+9.3"=8172"@938)"1+"789@+26)9813,"! K85+9.3"=8172"28="1+">9+13./""! ())"0933/",858)">+91.+/"8,"*+9"2+=3=8)3",8/)F.72"

#; K+=>893"*81,B"789@+26)9813,B",-089B">9+13./"*+9"123"=385"1+"/-19.1.+/85"./*+9=81.+/""M; ()N-,1"F3.021,"+*"./093).3/1,",+"/-19.3/1,"893"F.12"&$O""P; D+=3"3Q198"./093).3/1,"*9+="123"./093).3/1,"5.,1"=86"/33)"1+"@3"1+"@3"8))3)"3;0;"+.5"+9"

,-089""

"

"

<1".,">+,,.@53"1+"7857-5813"123">+91.+/",.R3"@6"-,./0"123"S/-19.3/1"3Q728/03J"*-/71.+/"./"T.31>58/'"

!

! "$(!

!"#$%&'()'%*+,(-./"+.#"/0(1+234(5(1+%/3%#3(6#.+.7.*(8.#(8..3(3"%#4(7.3"/0(29"/0(:"'+6*%/( ( (

!"#$%&$%&'"

()"*+,-,./-"0.1/2345-06"

718-00"+95-:230-"09.9-/6"

#;",-/3<,"0834-0"+=">:-./"?.0"@-:"95-"A-1-:.8"B<-093+10":-0@+10-"31"95-">.4C"+="95-"=++/"/3.:D"+:"/-=.<89"4+/-E"F.9"0@:-./"?3="09.9-/"+:"09.9-/"31"95-"A-1-:.8"B<-093+10"31"95-">.4C"+="95-"=++/"/3.:DE6""1+9-"95-"B<.1939D"30"<0-/"=+:"-.45"0834-"F38831A6"<18-00"+95-:230-"09.9-/"<0-"G.H-:.A-I".,+<190"=+:"45--0-"%",-.90"%"=305"%"-AA"718-00"+95-:230-"09.9-/"#$A"8-99<4-"%"JKA"9+,.9+"%"#JA"4<4<,>-:"GL.8./I"M"/-=.<89"@+:93+1"#$A"+="GA:--1"0.8./I"?7NO""&PQJ($E+95-:".//393+10"9+"0.1/2345-0"<0-"R&PA)"

(

;<(-"='3(1%*%39(

S="95-:-"30"1+9"-1+<A5"31=+:,.93+1"@:+H3/-/"9+">-".>8-"9+"-19-:"95-"2-3A59"+="-.45"H-A-9.>8-"3148</-/"<0-"GA:--1"0.8./I"7NO"&PQJ($T"0--"4+/-">++C"=+:"/-=.<89"@+:93+1"03U-0)(

(

(

(

(

(

(

(

(

(

(

(

(

(

(

!

! "$)!

!"#$%&'()'%*+,(-./"+.#"/0(1+234(5(1+%/3%#3(6#.+.7.*(8.#(8..3(3"%#4(7.3"/0(29"/0(:"'+6*%/( ( (

!"#$"%!"%!&#

;<;(=..>"/0(?'+,.39@(

'()#*++#),-#./00-1-2)#3((4/25#(6)/(27#0(1#-*3,#0((.#)86-#*1-#*9*/+*:+-#/2#),-#;<'#.*)*:*7-=#),-1-0(1-#*#3(>6*1*:+-#3((4/25#>-),(.#>*8#2--.#)(#:-#7-+-3)-.#/0#),-#(2-#1-3(1.-.#/2#),-#0((.#./*18#/7#2()#*9*/+*:+-?#

A'0'+%B*'9(

@7)-*>-.A##(1#@>/31(B*9-.A# C#-2)-1#*7#:(/+-.#@D27*+)-.#B*)-1A#/0#@7)-*>-.A#/7#2()#*9*/+*:+-##

@7)/1#01/-.A## C#/0#01/-.#(6)/(2#/7#2()#*9*/+*:+-#-2)-1#*7#@1*BA#E7-+-3)#)86-#(0#(/+#*7#76-3/0/-.#/2#),-#5-2-1*+#FD-7)/(27#*7#),-#@*9-1*5-#*>(D2)AG#1-0-1#)(#),-#3(.-:((4#0(1#),-#.-0*D+)#3(.-H#

@7)/1#01/-.A#/2#3((4#/2#7*D3-# C#-2)-1#*7#:(/+-.#@D27*+)-.#B*)-1A#/0#@7)-*>-.A#/7#2()#*9*/+*:+-#

@1(*7)-.A# C#7-+-3)#@:*4-.A#(1#51/++-.#/0#*9*/+*:+-#I#(/+#E7-+-3)#)86-#(0#(/+#*7#76-3/0/-.#/2#),-#5-2-1*+#FD-7)/(27#*7#),-#@*9-1*5-#*>(D2)AG#1-0-1#)(#),-#3(.-:((4#0(1#),-#.-0*D+)#3(.-H#

@3*77-1(+-.A#(1#@7)-B-.A## C#7-+-3)#@:(/+-.A#

@:(/+-.A#J#.-0*D+)#3((4/25#>-),(.#D2+-77#9-5-)*:+-7#*1-#-*)-2#*7#6*1)#(0#*#7*+*.#

-'%+(C(8"9,(

@7)-*>-.A### # # C#-2)-1#*7#@3*77-1(+-.A#(1#@7)-B-.A#/0#@7)-*>-.A#2()#*9*/+*:+-#

@01/-.A## C#7-+-3)#@51/++-.A#I#(/+#/0#01/-.#(6)/(2#/7#2()#*9*/+*:+-#E7-+-3)#)86-#(0#(/+#*7#76-3/0/-.#/2#),-#5-2-1*+#FD-7)/(27#*7#),-#@*9-1*5-#*>(D2)AG#1-0-1#)(#),-#3(.-:((4#0(1#),-#.-0*D+)#3(.-H#

@7)/1#01/-.#/2#3((4#/2#7*D3-A# C#-2)-1#*7#@3*77-1(+-.A##(1#@7)-B-.A#

@1(*7)A# # # # C#-2)-1#*7#@51/++-.A#/0#@1(*7)A#(6)/(2#2()#*9*/+*:+-# # # #

@51/++-.A## # # C#-2)-1#*7#@1(*7)A#/0#@51/++-.A#(6)/(2#2()#*9*/+*:+-# # #

#

K-0*D+)#3((4/25#>-),(.7#/0#2()#76-3/0/-.L#

M()*)(G#1(()#9-5-)*:+-7G#6-*7###C#@:(/+-.A#D27*+)-.#B*)-1#

N-*)#*2.#0/7,#O#51/++-.#

(

(

(

!

! "%+!

!"#$%&'()'%*+,(-./"+.#"/0(1+234(5(1+%/3%#3(6#.+.7.*(8.#(8..3(3"%#4(7.3"/0(29"/0(:"'+6*%/( ( (

!!"#$%!$%!&"

;<=(>'"0,+(7,%/0'9(./(7..?"/0(

• '(")*+",-.".+/0*)"1("-"(112"/3"0/4+5")*+"6117+2".+/0*)"5++23")1"8+"6-96:9-)+2";)1"-661:5)"(1,".-)+,"9133"2:,/50"6117/50<"-52"+5)+,+2"/5)1"=/+)>9-5?""

• '(")*+".+/0*)"/3".,/))+5"/5")*+"@"-A1:5)B"619:A5".+"-33:A+")*/3"/3"-"6117+2".+/0*)":59+33"/)"/3">-,)"1(",+6/>+""

C+/0*)"D-/5""

• E1123".*/6*"0-/5".+/0*)":>15"6117/50";)*+F"+G>-52".*+5"6117+2<"! H?0?">-3)-I",/6+I"61:361:3I"9+5)/93""

"• E/52")*+".-)+,"0-/5"/5"J6K-56+"-52"C/221.315B3"L*+"K1A>13/)/15"1("E1123"• L*+"5:A8+,"0/4+5"/3")*+">+,6+5)-0+"1(".-)+,"0-/5+2I")*/3"A+-53")*+"-A1:5)"1(".-)+,"-22+2?"

L*+,+(1,+"/("/)"0/4+3"M!NN".-)+,"0-/5")*+")1)-9"/3"G#?NN""

! H?0?"=,/+2">-3)-"81/9+2"M!#OP""#$$0"2,F">-3)-I".-)+,"0-/5"!#OP"

#$$"G"#?#O"Q"NN&0"81/9+2">-3)-"""R,";-5"-9)+,5-)/4+"A+)*12<"#$$0"M"!#OP""

! H?0?"=,/+2",/6+"81/9+2"M!S#P""#N!0"2,F">-3)-I".-)+,"0-/5"!S#P"

#N!"G"#?S#"Q"&TT?T#0"81/9+2",/6+""R,""#N!"M"!S#P""

'(".-)+,"-52"-".-)+,"0-/5"/50,+2/+5)3"-,+"0/4+5"/5")*+",+6/>+"+?0?",/6+"/5"-",/31))1I">-3)-"/5"-"31:>""

• K-96:9-)+")*+".-)+,"0-/5""• L-7+")*+".-)+,"0-/5"-.-F"(,1A")*+"-A1:5)"1(".-)+,"/5")*+",+6/>+""• U22")*+".-)+,"0-/5")1")*+".+/0*)"1(")*+"/50,+2/+5)".*/6*"0-/53".-)+,""

H?0?"V/6+"/5"-",/31))1"""" V+3>152+5)"*-3"+-)+5"!%N"1(")*+",+6/>+I",-.",/6+"T$$0I"4+0+)-89+3"T$$0"-52".-)+,"!$$$0"" " C-)+,"0-/5",/6+"M!S#P"" " " T$$0"G"!?S#"Q"W&$0""" " UA1:5)"1(".-)+,"/5",+6/>+""" " " !$$$0"X"W&$0"0-/5+2"8F")*+",/6+"Q"!N$0"" " UA1:5)"1(",/6+"/5",+6/>+""" " " V/6+"T$$0"M".-)+,"0-/5+2"W&$0"" " " " Q"!O&$0""

V+3>152+5)"*-3"+-)+5"-"Y:-,)+,"1(",/6+"!O&$0I"4+0+)-89+3"T$$0"-52".-)+,"!N$0"

"

!

! "%*!

!"#$%&'()'%*+,(-./"+.#"/0(1+234(5(1+%/3%#3(6#.+.7.*(8.#(8..3(3"%#4(7.3"/0(29"/0(:"'+6*%/( ( (

!"#"$%!$%!&#

'()*+#,-..#

• /--0.#12342#,-.*#1*352)#67-8#4--9385##! :;5;#<*()=#>3.2##

#• /380#)2*#1()*+#,-..#38#?4@(84*#(80#'300-1.-8A.#B2*#@-<7-.3)3-8#->#/--0.#• B2*#86<C*+#53D*8#3.#)2*#E#->#1()*+#,-.)#06+385#4--9385#

! :;5;#F(1#.)*1385#C**>#GH&E#IJ$5#.)*1385#C**>=#1()*+#,-..#GH&E#

# # IJ$5#K#$;H&#L#!&"5#,-.)### # # L#IJ$5#+(1#.)*1385#C**>#G#!&"5#,-..## # # # L#"MM5#*()*8###N+#3>#6.385#(#4(,46,()-+#IJ$#O#H&E#L#"MM5##

! :;5;#F(1#42349*8#4(..*+-,*0##!PJ5#.)*1385#C**>=#1()*+#,-..#G"JE#

# # !PJ5#K#$;"J#L#IH;PJ5#,-.)### # # L#!PJ5#+(1#42349*8#O#IH;PJ5#,-..## # # # L#!H!;"J5#*()*8###N+#3>#6.385#(#4(,46,()-+#!PJ5#O#"JE#L#!H!;"J5#

#

!**(9%29%0'9#)-#C*#*8)*+*0#(.#4--9*0#1*352).#68,*..#.7*43>3*0#-)2*+13.*#*;5;#C+(80#(80#1*352);##Q.*#)2*#>-,,-1385#1*352).#(80#RNB#)2-.*#.)()*0#38#)2*#QSR#03*)7,(8&#0()(.*);#

T*>(6,)#.(6.(5*#1*352).#U*(42VW#

#+(1# 4--9*0#

B2349# IP5# HJ5#@-49)(3,# !"5# X5#B238# "J5# !X5#Y+*<36<##U*;5;Z#C*.)A#%#Z>38*.)A#%#)(.)*#)2*#03>>*+*84*A# &J5# IX5#U[(.*0#-8#(D*+(5*#.67*+<(+9*)#1*352)#7*+#.(6.(5*#+*4-+0*0#38#/--0C(.*#U!V#(77*803K#&V#

T*>(6,)#4--9*0#1*352)#3.#8-)#.)()*0W##HI5#U(D*+(5*#)2349=#)238#(80#7+*<36<V##

;./&'#9"./(.8(*"<2"39(+.(0#%=9>(#

• \3]630.#2(D*#(#.7*43>34#5+(D3)^#.-#!$$<,#3.#8-)#(,1(^.#!$$5#• @-8D*+.3-8#>(4)-+.#<6.)#C*#6.*0#(.#1*352)#<6.)#C*#4-0*0##• Q.*0#>-+#<3,9=#4+*(<=#34*#4+*(<=#*55=#-3,.=#>3__^#0+389.=#!$$E#>+63)#`634*.=#(,4-2-,34#0+389.##• T-#8-)#6.*#>-+#1()*+=#.]6(.2=#)*(=#4->>**###

#

!

! "%"!

!"#$%&'()'%*+,(-./"+.#"/0(1+234(5(1+%/3%#3(6#.+.7.*(8.#(8..3(3"%#4(7.3"/0(29"/0(:"'+6*%/( ( (

!"#$%&!%&!'#

()*#+)#,-.,/.-+0#+10#,)230456)2#7-,+)4#

• 8..#,)230456)2#7-,+)45#-40#-7+04#+10#,)2+02+5#9-:0#62#+10#;<8#;))=#>)4+6)2#<6?05#@))A#B74)C#9-:0#3666DB"D##

• E/.+69.0#3-./0#62#F590,676,#:4-36+GH#,)./C2#@G#+10#C.#:6302##

IJ:J#$%%C.#)7#*1).0#C6.A##### K)230456)2#7-,+)4#!J%"!## # $%%C.#L#!J%"!## # # M#$%'J$:#C6.A###IJ:J#""%C.#,-2#,).-### K)230456)2#7-,+)4#!J%N## # ""%C.#L#!J%N## # # M#"N"J$:#,).-##

O695#

• P7#+10#4059)2=02+#1-5#/50=#-#5C-..#-C)/2+#0J:J#"+@5#)7#C6.A#+10#,)230456)2#7-,+)4#=)05#2)+#200=#+)#@0#/50=##

• O10#*06:1+5#)7#C6.A##7)4#+0-#-2=#,)7700#62#50,+6)2#!J$#3.(/.+#200=#-#,)230456)2#7-,+)4##• P7#+10#-C)/2+#1-5#@002#:6302#62#)/2,05Q#.6+405#)4#962+5#,)2304+#+)#C.#7645+#+102#/50#,)230456)2#

7-,+)4#B3-./05#-40#-+#+10#+)9#)7#+10#5-C0#9-:0#-5#+10#,)230456)2#7-,+)45D#• R06:1+5#)7#:.-5505#-2=#,/95#-40#:6302#62#C.#62#+165#=),/C02+##• <)C0#-.,)1).6,#=462A5#1-30#962+5#-2=#1-.7#962+5#62#+10#=4)9#=)*2Q#+1050#-.40-=G#1-30#+10#

,)230456)2#7-,+)4#5)#6+#=)05#2)+#200=#+)#@0#,-.,/.-+0=#-:-62#• S0#,-407/.#)7#G):/4+#-5#+10#4059)2=02+#5)C0+6C05#9/+5#!$TC.#)4#T%%C.Q#+10G#94)@[email protected]#C0-2#!#

9)+#*16,1#65#!$T:#)4#T%%:#5)#+10#,)230456)2#7-,+)4#=)05#2)+#200=#+)#@0#/50=##• P7#-#3)./C0#)7#7))=#65#:6302#,10,A#+10#7))=#=0256+G#62#;))=#-2=#8:46,/.+/40#B;8UD#;))=5#V0256+G#

V-+-@-50#W0456)2#BNDJ##

<9))2#3)./C05#2)+#.65+0=#62#+10#;))=#<+-2=-4=5#8:02,G#B;<8D#7))=#9)4+6)2#@))A#B"DX#

<04362:#59))2#M#N%C.J#E6265+4G#)7#8:46,/.+/40Q#;65104605#-2=#;))=#BE8;;D#;))=#8+.-5#BTD#

Y-=.0#M#!$%C.X#C0-2#3)./C0#)7#+10#+*)#504362:#59))25#70-+/40=#62#E8;;#;))=#8+.-5#BTD#

P7#-#9-4+6,69-2+#02+045#F59))2H#@/+#=)05#2)+#590,67G#*1-+#56?0Q#+102#-99.G#-994)946-+0#=07-/.+#56?05X#

E0-+#62#5-/,0#&#30:[email protected]#Z#[email protected]))25#

W620:-4#&#104@5#&#596,05#&#5/:-4#[#+0-59))25#

;26<##7)4#+0-#&#,)7700#-2=#)+104#1)+#=462A5#40704#+)#9)4+6)2#-.:)46+1CJ###

P7#-#F,/9H#65#/50=#+)#C0-5/40#.6\/6=5#)+104#+1-2#1)+#=462A5#+-A0#+10#=07-/.+#3)./C0#+)#@0#$T%C.#B]^#,))A5#C0-5/40D#-2=#-99.G#590,676,#:4-36+605#-5#-994)946-+0J####

!

! "%#!

!"#$%&'()'%*+,(-./"+.#"/0(1+234(5(1+%/3%#3(6#.+.7.*(8.#(8..3(3"%#4(7.3"/0(29"/0(:"'+6*%/( ( (

!"#$%&!%&!'#

;+,'#(%#<"+#%#4(='%92#'9>((

()#*)+,-./0#1-#23.45#)+,#67875)9/7:##;1,7#!%%8##

()#*)+,-./0#1-#+.5:#)+,#,347,#-3.45:#;1,7#<%8#

()#:=/):*#1-#14/0#;1,7#>8##

()#,34??/7#1-#:)/),#,37::4+80#@#!%8#A$>5*#=73;7+54/7#-13#:)/),#,37::4+8#:77#37-737+;7#211,9):7#A!B#

?=6'#"%*(='%92#'9C##D1+673:41+#1-#4E=734)/#E7):.37:#51#83)E:#)37#=3164,7,#4+#5*7#2FG#211,#H13541+#F4?7:#911I#A<BJ#=)87#6444K#

L157#5*)5#!;/#@#!%E/J#:1#5*7M#E48*5#:)M#N>;/#1-#O4+7#O*4;*#4:#N>%E/#O*4;*#4:#)#:5)+,)3,#9155/7K##

(

(

(

(

(

(

(

(

(

(

(

(

(

(

(

(

(

!

! "%$!

!"#$%&'()'%*+,(-./"+.#"/0(1+234(5(1+%/3%#3(6#.+.7.*(8.#(8..3(3"%#4(7.3"/0(29"/0(:"'+6*%/( ( (

!"#$%&!%&!'#

;<=(>.#+"./(9"?'(%*0.#"+,@9(

!*0.#"+,@(8.#(6.#+"./(9"?'('9+"@%+"./(5(,.+(3#"/A9(

(

(

(

(

(

(

(

(

(

BCD((

!33"/0(@"*A(+.(,.+(3#"/A9(

()*#&#+,--))#)./)0)1#*2#/3)#+45#&#647#8,946)#:;<)<#!=%#&#$'%69>#/3).#*11#/3)#6;9?<#

!"#"$$%&#$'($)!*$+,)-$*.!/*#!$%,01$

2!*$,3(&4,'3$*.!/*#!$567#$8$4!%,$41,%%!9$%,01$:*4)!&/,4!9$*.!/*#!$;7#$

120%#("/(3#"/A9(

@-#*#5*0/;+;5*./#0)+,012#A!#247*0B#C#)./)0#!#9)8)9#/)*25,,.#:DE7#247*0><#

(

(

(

(

(

(

(

(

(

F*2#/3)#0)25,.1)./#0)-)00)1#/,#*#+45#&#647G#

H)-*49/#/,#647#$'%69#

E26D##-,0#/)*#&#+,--))#*.1#,/3)0#3,/#10;.?2#D#!=%69#8,946)#*.1#*559I#25)+;-;+#70*8;/;)2#*2#0)J4;0)1<#:KL#!=%69#D#/)*#&#+,--))#M;/3,4/#6;9?>#

-20N##-,0#/)*#&#+,--))#*.1#,/3)0#3,/#10;.?2#D#$'%69#8,946)#*.1#*559I#25)+;-;+#70*8;/;)2#*2#0)J4;0)1<#:KL#$'%69#D#/)*#&#+,--))#M;/3,4/#6;9?>#

K,/)N((*11)1#6;9?#42)#*8)0*7)#*6,4./#O%7#-,0#/)*#*.1#EO7#-,0#+,--))(

K,((

I)2#

!

! "%%!

!"#$%&'()'%*+,(-./"+.#"/0(1+234(5(1+%/3%#3(6#.+.7.*(8.#(8..3(3"%#4(7.3"/0(29"/0(:"'+6*%/( ( (

!"#$%&!%&!"#

!*0.#"+,;(8.#(6.#+"./(9"<'('9+";%+"./(5(7.*3(=(%;>"'/+(=(%*7.,.*"7((3#"/?9(3'97#">'3(>4(%(0*%99@(

(

(

(

(

(

(

(

(

(

(

(

(

(

(

(

(

(

(

(

(

(

(

(

(

(

(

'()#*+)#,-(#.()/012(1,#.(3(..(2#,0#+#

)/(45354#60789(#30.#(:+9/7(#;#/51,<#

*+)#,-(#.()/012(1,#.(3(..(2#,0#)9+77#&#9(2589#0.#7+.=(#=7+))<#

>0((

?)(#,-5)#60789(#+12#+//7'#)/(45354#=.+65,5()#+)#.(@85.(2A#

'()#

?)(#2(3+87,#60789()#)-0B1#C(70B#+12#+//7'#)/(45354#=.+65,5()#+)#.(@85.(2A#

$%%97#D101E+740-0754F#

!$G97#DB51(F#

H((.#&#7+=(.#&#452(.I#JK%97#D+6(.+=(#03#/51,#L#;#/51,FA#

A*%99B## M3#,-(.(#5)#10,#+#60789(#&#B(5=-,#51#,-(#NOP#/0.,501#C00Q#DKF#D/+=()#KE"F#30.#,-(#)/(4535(2#75@852#8)(#,-(#30770B51=#2(3+87,#60789()#+12#+//7'#)/(45354#=.+65,5()#+)#.(@85.(2A#

>01E+740-0754I#

R+.=(#S#K%%97#DNOP#C00Q#DKF#60789(#30.#T,89C7(.UF#

P6(.+=(V#9(2589#0.#81)/(4535(2#S#$%%97#60789(#DWPNN#),.+5=-,#=7+))#60789(F#DGF#

O9+77#S#!"%97#DWPNN#)9+77#),.+5=-,#=7+))#60789(F#DGF#

P740-0754I#

X51(I#)9+77#&#9(2589#!$G97#DN002C+)(#$G&G%,-#4(1,57(FA#R+.=(#$G%97#E#!"#$%&"'(&!")%&D"F#

H((.#&#7+=(.#&#452(.I#)9+77#$YJ97#&#7+.=(#G"Y97#

O/5.5,)I#.(3(.#,0#NOP#3002#/0.,501#C00Q#DKF#/+=(#K#Z#"A#

P#T7+.=(U#)/5.5,#5)#,0#,+Q(1#+)#+#208C7(#9(+)8.(##

#

>0((

!

! "%&!

!"#$%&'()'%*+,(-./"+.#"/0(1+234(5(1+%/3%#3(6#.+.7.*(8.#(8..3(3"%#4(7.3"/0(29"/0(:"'+6*%/( ( (

!"#$%&!%&!'#

!*0.#"+,;(8.#(6.#+"./(9"<'('9+";%+"./(5(7.*3(=(%;>"'/+((=(%*7.,.*"7((3#"/?9(3'97#">'3(>4(%(>.++*'(.#(7%/@(

(

(

(

(

(

(

(

(

(

(

(

(

(

(

(

(

(

(

(

(

(

(

(

(

(

()*#+,-#.-*/012-1+#.-3-..-2#)#*/-45354#6.)12#&#+51#*57-8#9:;#()<3#)#4)1#03#.-2#

6=<<8#

>-*#

?*-#+,5*#@0<=A-#)12#)//<>#*/-45354#:.)@5+5-*#)*#.-B=5.-2;#

C0((

()*#+,-#.-*/012-1+#.-3-..-2#+0#*A)<<#&#A-25=A#0.#<).:-#60++<-#0.#4)18#

#

D*#+,-.-#)#@0<=A-#&#E-5:,+#51#+,-#FGH#/0.+501#600I#J/):-*#KL'M#30.#+,-#5+-A;#

>-*#

C0((C0((

?*-#2-3)=<+#@0<=A-#03#)@-.):-#)12#)//<>#*/-45354#:.)@5+5-*#)*#.-B=5.-2;#

#KK%A<#4)1#J101L)<40,0<54M#

KN%A<#4)1#J<):-.O#452-.O#65++-.M#P#A-)1#)@-.):-#03#*A)<<#)12#<).:-#4)1;#

KK%A<#60++<-#J6--.#&#452-.#&#<):-.M#

#

C01L)<40,0<54##L#6)*-2#01#FGH#/0.+501#*57-*JKMQ#

R)1#S#KK%A<#

GA)<<#4).+01#S#$%%A<#

T).:-#60++<-#S#!<+.#

U-25=A#60++<-#S#V%%A<#

GA)<<#60++<-#S#$V%A<#

H<40,0<54Q#

W51-#P#*A)<<#60++<-#S#K"VA<#JFGH#X#60++<-M;#T).:-#60++<-#S#"V%A<#

Y--.#&#<):-.#&#452-.Q#J6)*-2#01#40AA01#*57-*#<5*+-2#)+#G)51*6=.>Z*#&#[-*40#%$&%\&!KMQ##

GA)<<#60++<-#S#$"VA<#

U-25=A#&#.-:=<).#60++<-#S#KK%A<#

T).:-#60++<-#S#!<+#

GA)<<#4)1#S#KK%A<#

U-25=A#4)1#S#KN%A<#

# # # #

>-*#

!

! "%'!

!"#$%&'()'%*+,(-./"+.#"/0(1+234(5(1+%/3%#3(6#.+.7.*(8.#(8..3(3"%#4(7.3"/0(29"/0(:"'+6*%/( ( (

!"#$%&!%&!'#

!*0.#"+,;(8.#(6.#+"./(9"<'('9+";%+"./(5(8..3=(7%//'3(%/3(+"//'3>($,'/(#'7.#3'3(%9(%(6#.6.#+"./(.8(%(7%/(?+"/@(

(

(

(

(

(

(

(

(

(

(

(

(

(

(

(

(

# ()*+,-.#&#-.+/0-#1*234)55#

(

(

1ABC(D(=(3'+'#;"/'(+,'(+"/(9"<'=(

(

(

6/*7-.#4,#8473#24884,#3+,#7+9-#1-*3:)-.#4,##

;*+,7/:)<=7#>#?-724#@-/7+3-#%$&%"&!A#

(

(

B*7#)-7C4,.-,3#)-1-))-.#*#7C-2+1+2#/)*,.#&#3+,#7+9-5#DEFE#B*01#*#B-+,9#G%%F#

3+,#41#/-*,75##

<-7#

H-1-)#34#3I-#8*,:1*23:)-)#4)#)-3*+0-)#@-/7+3-#J-EFE#;*+,7/:)<=7K#?-724#L#34#1+,.#3I-#*CC)4C)+*3-#3+,#@-+FI3E#

M4((

;-0-23#1)48#.-1*:03#3+,#7+9-6N#M4,O.)*+,-.#@-+FI37E#

P+7IN#

?:,*#O#!'%F#

;*084,#O#!%QF#

;*).+,-#R#!$%F#

S-F-3*/0-7#&#C:07-7N#

?48*34-7#R#G%%F#

;@--3#24),#T-),-07#R#!'QF#

U-*7#J,43#8:7I<L#R#A%%F#

V:7I<#C-*7#R#A%%F#

W-*,7#J,43#/*T-.L#R#G%%F#

W*T-.#/-*,7#R#G!QF#

X3I-)N#

;4:C#R#G%%F#

;C*FI-33+#+,#7*:2-#R#G%%F#

P):+3#R#G!QF#

#

#

D,3-)#3I-#Y.)*+,-.#@-+FI3=#&=-.+/0-#C4)3+4,=#41#3I-#C4)3+4,#41#3I-#2*,#-*3-,E#

Z1#<4:#@4:0.#,4)8*00<#24,7:8-#3I-#8-.+:8#+,#@I+2I#3I-#144.#+7#2*,,-.#3I-#.)*+,-.#@-+FI3#@4:0.#[#!%%\E#

-EFE#/*T-.#/-*,7K#3+,,-.#348*34-7K#74:CK#8:7I<#C-*7K#+3-87#3+,,-.#+,#348*34#7*:2-#*,.#3+,,-.#1):+3#@4:0.#[#!%%\E#

Z1#<4:#@4:0.#,43#,4)8*00<#-*3#3I-#8-.+:8#+,#@I+2I#3I-#144.#+7#2*,,-.#J-EFE#4+0#*,.#/)+,-L#<4:#,--.#34#2*02:0*3-#3I-#.)*+,-.#@-+FI3N#

?:,*#&#7*).+,-7#&#7*084,#2*,,-.#+,#4+0#4)#/)+,-#[#"$\#

;@--3#24),#2*,,-.#+,#@*3-)#&#7*03-.#@*3-)#[#"$\##

]I+2T#C-*7#&#3+,,-.#/-*,7#2*,,-.#+,#@*3-)#&#7*03#@*3-)#[#'%\#

D^*8C0-#2*02:0*3+4,N#

Half a 400g can of chick peas

!

! "%(!

!"#$%&'()'%*+,(-./"+.#"/0(1+234(5(1+%/3%#3(6#.+.7.*(8.#(8..3(3"%#4(7.3"/0(29"/0(:"'+6*%/( ( (

!"#$%&!%&!'#

!*0.#"+,;(8.#(6.#+"./(9"<'('9+";%+"./(=>>:(?@>A(7%//'3(B(+"//'3CD(

(

#

#

#

#

#

#

#

#

#

#

#

#

#

#

#

#

#

#

#

#

#

#

#

#

(

()*#+,*-./0,/1#+,2,++,0#)#-)34,1#.+#*56,#.2#)#7+)/0,0#-+.08319#5:,:#)#;)</8;#53,#

3+,);#.+#15/#.2#*.8-9#

=,2,+#1.#1>,#;)/82)318+,+#.+#+,1)5?#@,7*51,#1.#,*15;)1,#-.+15./#*56,#

A,*#

B*#51#)/#5/05C508)?#2..0#51,;9#5:,:#*?53,#.2#7+,)09# A,*#

B*#51#5/#1>,#DEF#

0)1)7)*,9#

B*#1>,#-.+15./#*56,#5/053)1,0#

7G#1>,#-)+1535-)/19#5:,:#;,058;9##

A,*#

F.#F.((

B*#51#?5*1,0#5/#!"#$%&"'(&!")%&H'I9##

#

A,*#

F.#

F.((

(

B*#51#?5*1,0#5/#1>,#*++($"%,#-.+15./#

<850,9#

F.#

B*#1>,+,#)#0,2)8?1#@,5<>1#*1)1,0#5/#1>,#J5+@)C,#3.0,#7..49#

()*#+,*-./0,/1#+,2,++,0#1.#5;)<,*#-)<,#"K!L#.2#1>,#2..0#05)+G9# G,*#

=,2,+#1.#M-.+15./#@,5<>1*#*>,,1N#O#H*,,#/.1,*#7,?.@I#

A,*#

B*#1>,#-.+15./#*56,#5/053)1,0#7G#1>,#-)+1535-)/19#5:,:#

;,058;9##

F.#

F.((

A,*#

P/1,+#1>5*#@,5<>1#

#

P*15;)1,#1>,#-.+15./#*56,#8*5/<#1>,#-.+15./#*56,*#+,3.+0,0#./#.1>,+#0)G*#K#5:,:#@>)1#-.+15./#*56,#@.8?0#7,#3./*5*1,/1#@51>#1>5*#-)+1535-)/1*#5/1)4,:#P/1,+#M;,058;N#.+#M)C,+)<,N#)*#0,2)8?1:#

F.#

P/1,+#1>5*#@,5<>1#

#

A,*#P/1,+#1>5*#@,5<>1#

#

A,*#

P/1,+#1>5*#@,5<>1#

#

F.((

#

P*15;)1,#1>,#-.+15./#*56,#8*5/<#1>,#-.+15./#*56,*#+,3.+0,0#./#.1>,+#0)G*#K#5:,:#@>)1#-.+15./#*56,#@.8?0#7,#3./*5*1,/1#@51>#1>5*#-)+1535-)/1*#5/1)4,:#P/1,+#M;,058;N#.+#M)C,+)<,N#)*#0,2)8?1:#

B*#51#5/#1>,#QRJ#2..0#-.+15./#*56,*#7..49#

#A,*#

P/1,+#1>5*#@,5<>1#

#

R,,4#)0C53,#2+.;#1>,#*8-,+C5*.+#

F.#

!

! "%)!

!"#$%&'()'%*+,(-./"+.#"/0(1+234(5(1+%/3%#3(6#.+.7.*(8.#(8..3(3"%#4(7.3"/0(29"/0(:"'+6*%/( ( (

!"#!"$%"$%&#

'()*+,#*)#*-+#./0123+#4))5#6)0*/)7,#80+4+0#*)#*-+#6)0*/)7#6-)*),#/79:;5+5#/7#*-+#4))5#5/20<#275#*-+#9)00+,6)75/7=#52*2>,+?#

• @-+9A#*-2*#<);#,+:+9*#*-+#6)0*/)7#,/B+#0+:2*/3+#*)#*-+#4))5#*<6+C##D)0#+E2F6:+#6/9*;0+#,+*#%G#0+:2*+5#*)#,2:25#275#>)/:+5#:+24#3+=+*2>:+,H##

+C=C#6-)*)##%G2#I#J%=#4)0#>)/:+5#,6/729-##

# # I#%J=#4)0#:+**;9+#

• K4#2#620*/9/627*#-2,#)7:<#0+4+00+5#*)#)7+#6)0*/)7#,/B+#2=2/7,*#!#4))5,#*-2*#20+#40)F#5/44+0+7*#4))5#=0);6,L#

Eg.##Chicken in curry sauce & boiled rice 8b

M+:+9*#6)0*/)7#N>#4)0#*-+#0/9+#8%OO=?#275#,+:+9*#*-+#P>Q#,/B+#6)0*/)7#4)0#*-+#9-/9A+7#9;00<H#/7#*-/,#+E2F6:+#/*#1);:5#>+#%R>#8%%J=?C#

• K4#2#620*/9/627*#-2,#)7:<#0+4+00+5#*)#)7+#6)0*/)7#,/B+#2=2/7,*#!#4))5,#*-2*#20+#40)F#*-+#,2F+#4))5#=0);6L##

EG. Carrots, broccoli, green beans 13c

M+:+9*#%S9#I#%"O=#5/3/5+#><#S#I#S&=#6+0#*<6+#)4#3+=+*2>:+#

#

(

(

(

(

(

(

(

(

(

(

(

(

(

!

! "&+!

!"#$%&'()'%*+,(-./"+.#"/0(1+234(5(1+%/3%#3(6#.+.7.*(8.#(8..3(3"%#4(7.3"/0(29"/0(:"'+6*%/( ( (

!"#!$%"$%"&#

;<=(>##.#(7,'7?"/0(

• '()*#+,*#(--*--.*/+#• 0,*12#+,*#3455467/8#(8(7/-+#+,*#97(:;#;4<#,()*#7/=<++*9>#

! ?*87-+:(+74/#149*#7-#+,*#-(.*#(-#+,*#149*#4/#+,*#3:4/+#43#+,*#97(:;#;4<#,()*#7/=<++*9#

! @*(5#411(-74/-#,()*#A*#*/+*:*9#7/#+,*#14::*1+#.*(5#/<.A*:#B"CDE#! 049*-#:*=:*-*/+#+,*#3449-#7/#+,*#97(:;# #! F4:+74/#-7G*-#(:*#*/+*:*9#14::*1+5;#! H/;#1,(/8*-#(:*#-()*9#

" '*5*1+#I8:79J##

K,*/#8:79#)7*6#4=*/-#;4<#1(/#94<A5*#1,*12#3449-#(/9#L<(/+7+7*->(#

###############

05712#IL<(/+7+;J#M#+,7-#6755#(::(/8*#(55#7+*.-#A;#L<(/+7+;N##0,*12#946/#+,*#:46-#+4#.(2*#-<:*#(55#L<(/+7+7*-#(:*#(==:4=:7(+*#+4#+,*#3449-#-*5*1+*9N#OPN#0,(/8*-#1(//4+#A*#.(9*#7/#+,7-#)7*6N#(

#

O4+*#+,*#+4+(5#(+#+,*#A4++4.#43#+,*#IQ/*:8;#21(5J#145<./#M#+,7-#1(/#A*#<-*9#+4#94<A5*#1,*12#+,*#.*(/#21(5#7/+(2*#=*:#9(;N##?(/8*#-,4<59#A*#(==:4RN#"S$$#M#T$$$21(5#

!

! "&*!

!"#$%&'()'%*+,(-./"+.#"/0(1+234(5(1+%/3%#3(6#.+.7.*(8.#(8..3(3"%#4(7.3"/0(29"/0(:"'+6*%/( ( (

!!"!#$%#$%&"

;'8'#'/7'9<"%' ())*"+,-.*-/*0"123.45'"())*6-03"!##7"89#'#:'%9;'"1<-=>-6>3"?/)@A"

B,,CA$$DDD'?))*6-03')/2'EF$/30E>,0'CBCG?H/3C)/,H=*I:!'"!' ())*"+,-.*-/*0"123.45'"J4K-.43"-.*"L=**)D0).M0A""NB3"4)@C)0=,=).")?"?))*0'"&,B"

+E@@-/5"3*'"K-@6/=*23A"O)5->"+)4=3,5")?"KB3@=0,/5P"!##!"9' ())*"+,-.*-/*0"123.45'"())*"Q)/,=)."+=R30'"9/*"3*'"S).*).A"N+TP"!#%#'"U' KB-//).*=3/3"OV"W-5,)D=,R"XV"+,-*>@-5/"Y'"(1T$"(TTX+"X3.0=,5"X-,-6-03"Z3/0=)."!'#'"(1TP"

O)@3V"[,->5V"!#%!"7' \3>0)."JV"1,F=.0)."JV"J353/"]A"!"#$%&%'()*$+,"!&-)."%/"0%%1"#%(&+%2"3+45.'"S).*).A"J1((

QE6>=4-,=).0P"%^^_"&' KB353,,3"KV"Y->)>=-"`'"K-/60"-.*"4->0"-.*"C/),3=."-.*"?-,0"%0,"3*'"a.=,3*"b=.2*)@A"KB3>>)"

QE6>=0B=.2P"!#%#'"

"

"

"

((!7=/.$*'30'>'/+9<(

Dietplan6.7 nutritional software programme (Forestfield Software Ltd, Horsham, UK)(

(

(

(

(

(

(

(

(

(

(

(

(

!

! "&"!

!"#$%&'()'%*+,(-./"+.#"/0(1+234(5(1+%/3%#3(6#.+.7.*(8.#(8..3(3"%#4(7.3"/0(29"/0(:"'+6*%/( ( (

!"#!$%&$%&'#

!66'/3";((

()*+,-.#,*/.0#123+#45.#6728*9.#:.*-45#;3<74327</#=4>?@#133?#?7*2@#2.A32?#B123<4#A39.2C#.)*+,-.#2.A32?C#0*+,-.#31#,32473<#07D.#,53430#7<A->?.?C#/.<.2*-#E>.0473<0#7<A->?.?#7<#45.#?7*2@#*<?##>0.?#43#/>7?.#A3?7</#F#2.1.2#43#0.,*2*4.#?3A>+.<4#FG#!"#$%&'()&*$"+),"&,)-.,/)/&#$&0*%)&12324&.5&'()&5../&/#*,6&'.&*#/&-./)&")7)-'#.$8#H#

### #

# #

!!"#$ !

A3.2 Exam

ple of the food diary used to capture dietary intake in the Airw

ave Health M

onitoring Study 2007-2013

!!!

Left hand im

age is an example of three m

eal occasions completed to help guide participants.

The right hand image provides exam

ples of the portion photos included in the food diary. Each im

age relates to foods of a similar

density. Each food im

age corresponds to a potion size (in grams) that is used as part of the coding protocol (refer to A

ppendix A

3.1).

!

!

! "#$!

A3.3 Food groups and descriptions used in the dietary assessment of the Airwave Health Monitoring Study

Food group Items included Items not included

Whole grains1

Amaranth Barley, pearled Barley, whole, Barley meal Barley malt flour

Brown rice, Brown rice flour, Wild rice Bran (all grains)

Buckwheat Corn flour/meal Bulgur wheat Couscous Millett Rice (milled, not whole grain) Oats, Oat flour, Oatmeal Semolina Popcorn (only include plain) Wheat flour (milled, not whole grain) Quinoa Brown flour Whole wheat flour Brown bread Shredded wheat Cornflakes Dark rye bread / pumpernickel Rice krispie Muesli Egg / rice / plain noodles

Ready Brek / quick cook oats Crisp bread

Weetabix, Cheerio’s, Shreddies, All-bran Bran flakes

Whole wheat pasta Nutri-grain Ryvita / rye crackers Special K Oatcakes Granary bread Whole wheat bread Flapjack

Cakes, pastry, biscuits made with wholemeal flour

Low fat dairy skimmed, semi skimmed milk, 1% fat milk, cottage cheese, low fat and fat free yoghurt

Dairy alternatives

Nuts, seeds, legumes

All nuts, peanuts Seed oils Beans Green bean/ French bean Lentils Processed soya bean products

Chickpeas Bean flours Coconut Marzipan

Fruit

Fresh fruit Fruit in cakes and confectionary Dried fruit Jam / chutney Pre prepared fruit including frozen and canned) Fruit juices as cordial

100% pure fruit juices

!

! "#%!

Table A3.3 continued

Vegetables Sweet potato White potato (UK culinary usage classification)

All fresh, and prepared vegetables (frozen, dried, canned) Cassava

Herbs Chutney Tomatoes

Red and processed meat

Beef, Lamb, Pork, Veal, Mutton, Horse,

Poultry

Goat, Venison, Game meat, Offal (from any animal).

Deli meats, such as sliced turkey and bologna, Bacon, Sausages/ Hot dogs/frankfurters, Ham, Pastrami Luncheon meats.

Sugar sweetened beverages1

Cordials and fruit juices with added sugar

Low calorie or artificial sweetened beverages

Energy drinks 100% pure fruit juices

Carbonated sweetened beverages (e.g. colas, lemonade)

1. Definition of whole grains and sugar sweetened beverages taken from Draft Carbohydrates

and Health report. Scientific consultation: 26 June to 1 September 2014. 2014.

!

! "##!

A4.1 Study 1: Included vs. excluded for energy intake misreporting

428 excluded = weight loss diet reported

Excluded (n = 428) Included (n = 5,421) p Age, years (SD) 41.2 9.4 41.5 9.2 0.53 N , % Men 142 33.2 3355 62.0 <0.0001 White 9 2.1 147 2.7 0.45 Marital status 0.010 Cohabiting 68 16.3 885 16.7 Divorced/separated 41 9.8 426 8.1 Married 242 57.9 3387 64.0 Single 67 16.3 596 11.3 Missing* 10 127 Education 0.22 Left school before taking

22 5.1 226 4.2

GCSE or equivalent 112 26.2 1627 30.0 Vocational qualifications 26 6.1 400 7.4 A levels / Highers or

148 34.6 1744 32.2

Bachelor Degree or

85 19.9 1088 20.1 Postgraduate qualifications 35 8.2 335 6.2 Annual household income 0.017 Less than £32,000 88 20.6 852 15.7 £32,000 - £47,999 50 11.7 621 11.5 £48,000 - £57,999 146 34.1 2138 39.5 £58,000- £77,999 89 20.8 1251 23.1 More than £ 78,000 55 12.8 558 10.3 Rank <0.0001 Police staff 178 47.1 1551 32.8 Police Constable/Sergeant 177 46.8 2669 57.1 Inspector/Chief Inspector 15 4.0 309 6.5 Super Intendant or above 3 0.8 63 1.3 Other 5 1.3 107 2.3 Missing 50 692 Region 0.21 England 315 74.3 3835 70.9 Scotland 61 14.4 957 17.7 Wales 48 11.3 614 11.4 Missing 4 1.0 15 0.1 Total hours worked per

0.002

Part time 60 14.0 475 8.8 40 hours or less (including

172 40.4 2061 38.0

41 – 48 hours 114 26.6 1763 32.5 49 – 54 hours 41 9.6 566 10.4 55 hours or more 40 9.4 556 10.3 Physical activity 0.79 Low 56 13.1 650 12.0 Moderate 196 45.8 2490 45.9 High 176 41.1 2281 42.1 Body mass index <0.0001 Healthy (<25kg/m2) 96 21.7 1834 33.8 Over weight (25 - 30kg/m2) 196 45.8 2545 46.9 Obese (>30kg/m2) 139 32.5 1042 19.2

!

! "#&!

A4.2 Sensitivity analyses results tables: predictors of energy intake under reporting

A4.2.1Predictors of under-reporting energy intake by sex* (remove change in appetite) Predictor included in final model OR 95% Confidence interval Women (n 349/911) Body mass index Ref: Healthy (<25kg/m2) 1.00 Over weight (25 - 30kg/m2) 1.99 1.44 2.73 Obese (>30kg/m2) 3.90 2.20 6.91 Physical activity / week Ref: Low 1.00 Moderate 3.47 1.89 6.36 High 8.84 4.77 16.37 Total hours worked per week Part time !"#$ !"%! $"!$ ref: 35 – 40 hours (standard hours) 1.00 41 – 48 hours !"&& !"'( $")* 49 – 54 hours $"!# !"') $"&# >55 hours $"!( !"%$ )"!# Drinking Ref: never 1.00 Former !"'% !")+ $"** Current !"'+ !"(' $"($ Hours sitting per weekday Ref: Low 1.00 Moderate !"+' !"#$ $"($ High $"!) !"#) $"*( Ethnicity Ref: Other 1.00 White 0.18 0.05 0.72 Education Left school before taking GCSE 1.00 GCSE or equivalent 0.55 0.26 1.16 Vocational qualifications 0.71 0.30 1.72 A levels / Highers or equivalent 0.41 0.19 0.88 Bachelor Degree or equivalent 0.38 0.17 0.84 Postgraduate qualifications 0.22 0.09 0.56 Age (per 5 year increase) 0.84 0.77 0.92 Men (n 993/1817) Body mass index Ref: Healthy (<25kg/m2) 1.00 Over weight (25 - 30kg/m2) 2.01 1.59 2.53 Obese (>30kg/m2) 4.40 3.18 6.10 Physical activity / week Ref: Low 1.00 Moderate 3.37 2.25 5.03 High 5.58 3.74 8.34 Total hours worked per week Part time 0.33 0.15 0.74 ref: 35 – 40 hours (standard hours) 1.00 41 – 48 hours 1.11 0.89 1.40 49 – 54 hours 1.13 0.82 1.56 >55 hours 1.81 1.30 2.51 Drinking Ref: never 1.00 Former 2.53 1.09 5.89 Current 1.14 0.58 2.25

!

! "#'!

Table A4.2.1 continued Predictor included in final model OR 95%CI

Hours sitting per weekday Ref: Low 1.00 Moderate 1.04 0.82 1.32 High 0.77 0.60 0.99 Ethnicity Ref: Other 1.00 White !"+) !"%% $"%* Education Left school before taking GCSE 1.00 GCSE or equivalent !"+$ !"%& $"*) Vocational qualifications !"## !"*% $"($ A levels / Highers or equivalent !"## !"*+ $")! Bachelor Degree or equivalent !"#* !"*' $"$+ Postgraduate qualifications !"&! !"*' $"*! Age (per 5 year increase) !"+* !"&+ $"!!

A4.2.2 Predictors of under-reporting energy intake by sex* (remove dieters not weight loss) Predictor included in final model OR 95% Confidence interval Women (n 632/1582) Body mass index Ref: Healthy (<25kg/m2) 1.00 Over weight (25 - 30kg/m2) 1.78 1.40 2.26 Obese (>30kg/m2) 2.94 2.10 4.11 Physical activity / week Ref: Low 1.00 Moderate 2.73 1.87 3.98 High 6.06 4.09 8.96 Total hours worked per week Part time !"#' !"%& $"!! Ref: 35 – 40 hours (standard hours) 1.00 41 – 48 hours $"!* !"&$ $"() 49 – 54 hours $")! !"&! $"&$ >55 hours $"$$ !"#* $"'# Job strain Ref: Low strain (high control, low demand) 1.00 Passive (low control, low demand) 0.70 0.52 0.95 Active (high demand, high control) 0.88 0.65 1.19 High strain (high demand, low control) 1.20 0.89 1.62 Drinking Ref: never 1.00 Former !"'' !"(' $")$ Current !"'+ !"*) $"$) Hours sitting per weekday Ref: Low 1.00 Moderate $"!$ !"&! $")' High $"!* !"&$ $"(( Education Ref: Left school before taking GCSE 1.00 GCSE or equivalent 0.58 0.32 1.06 Vocational qualifications 0.70 0.35 1.39 A levels / Highers or equivalent 0.53 0.29 0.97 Bachelor Degree or equivalent 0.45 0.24 0.85 Postgraduate qualifications 0.30 0.15 0.61 Age (per 5 year increase) 0.89 0.83 0.94

!

! "#(!

Table A4.2.2 continued OR 95%CI

Men (n1476/2637) Body mass index Ref: Healthy (<25kg/m2) 1.00 Over weight (25 - 30kg/m2) 2.29 1.86 2.82 Obese (>30kg/m2) 5.57 4.27 7.26 Physical activity / week Ref: Low 1.00 Moderate 2.89 2.16 3.86 High 4.79 3.56 6.44 Total hours worked per week Part time 0.34 0.16 0.72 Ref: 35 – 40 hours (standard hours) 1.00 41 – 48 hours 0.99 0.82 1.21 49 – 54 hours 0.94 0.72 1.22 >55 hours 1.34 1.02 1.75 Job strain Ref: Low strain (high control, low demand) 1.00 Passive (low control, low demand) !"+& !"#+ $")) Active (high demand, high control) $"!' !"&# $")+ High strain (high demand, low control) !"+# !"#+ $")! Drinking Ref: never 1.00 Former 1.77 0.88 3.53 Current 0.88 0.51 1.53 Hours sitting per weekday Low 1.00 Moderate 1.02 0.84 1.25 High 0.74 0.60 0.92 Education Ref: Left school before taking GCSE 1.00 GCSE or equivalent !"&& !"'! $"(! Vocational qualifications !"#) !"*' $"$) A levels / Highers or equivalent !"&* !"%# $")* Bachelor Degree or equivalent !"#( !"*+ $"!+ Postgraduate qualifications !"#$ !"** $"$* Age (per 5 year increase) 0.93 0.89 0.98

A4.2.3 Predictors of under-reporting energy intake by sex* (S3 - remove chronic disease) Predictor included in final model OR 95% Confidence interval Women (n 615/1528) Body mass index Ref: Healthy (<25kg/m2) 1.00 Over weight (25 - 30kg/m2) 1.74 1.36 2.22 Obese (>30kg/m2) 2.81 2.00 3.95 Physical activity / week Ref: Low 1.00 Moderate 2.71 1.84 3.99 High 6.24 4.19 9.31 Total hours worked per week Part time !"#% !"%# $"!! Ref: 35 – 40 hours (standard hours) 1.00 41 – 48 hours $"!$ !"#+ $"(! 49 – 54 hours $")* !"&) $"&' >55 hours $"!* !"'& $"%+

!

! "&)!

Table A4.2.3 continues OR 95%CI Job strain Ref: Low strain (high control, low demand) 1.00 Passive (low control, low demand) 0.73 0.54 0.99 Active (high demand, high control) 0.95 0.70 1.29 High strain (high demand, low control) 1.20 0.89 1.63 Drinking Ref: never 1.00 Former !"'% !"(% $"$+ Current !"'# !"*$ $"$$ Hours sitting per weekday Low 1.00 Moderate !"+# !"## $")) High !"+# !"#' $")% Education Left school before taking GCSE 1.00 GCSE or equivalent 0.49 0.26 0.92 Vocational qualifications 0.58 0.29 1.19 A levels / Highers or equivalent 0.43 0.23 0.81 Bachelor Degree or equivalent 0.39 0.20 0.75 Postgraduate qualifications 0.24 0.12 0.51 Age (per 5 year increase) 0.90 0.85 0.96 Men (n1476/2637) Body mass index Ref: Healthy (<25kg/m2) 1.00 Over weight (25 - 30kg/m2) 2.23 1.81 2.74 Obese (>30kg/m2) 5.59 4.28 7.31 Physical activity / week Ref: Low 1.00 Moderate 2.77 2.06 3.73 High 4.57 3.39 6.16 Total hours worked per week Part time 0.33 0.15 0.73 Ref: 35 – 40 hours (standard hours) 1.00 41 – 48 hours 1.07 0.88 1.30 49 – 54 hours 0.95 0.73 1.24 >55 hours 1.46 1.12 1.92 Job strain Ref: Low strain (high control, low demand) 1.00 Passive (low control, low demand) $"!) !"&) $")& Active (high demand, high control) $"!% !"&# $")& High strain (high demand, low control) !"+& !"#+ $")$ Drinking Ref: never 1.00 Former "*"" +*+& $*"" Current 0.99 0.57 1.71 Hours sitting per weekday Low 1.00 Moderate 1.00 0.82 1.23 High 0.71 0.58 0.88 Education Ref: Left school before taking GCSE 1.00 GCSE or equivalent !"&* !"%( $"() Vocational qualifications !"&+ !"'! $"() A levels / Highers or equivalent !"## !"%$ $"$' Bachelor Degree or equivalent !"## !"*& $")* Postgraduate qualifications !"+& !"'' $"** Age (per 5 year increase) 0.94 0.89 0.98 Notes to Tables A4.2: Logistic regression model analyses conducted for men and women separately. Variables included in model presented showed significant association (p <0.05) with under-reporting in stepwise logistic regression. Variables in italics were significant predictors for either men or women and are shown to enable comparison across sex. Covariates in the model Food diary 7 days complete (y/n), region, BMI, PAL, police rank, household income, marital status, work hours, education, sitting, TV viewing, ethnicity, smoking status, job strain, drinking status, age. !

!

! "&+!

A5.1 Quintile cut-offs for calculating DASH scores based on individuals from the Airwave Health Monitoring Study

DASH food groups 20% 40% 60% 80% Whole grains g/day Men 8.1 27.7 51.1 89.9 Women 8.4 23.8 41.3 67.3 Low fat dairy g/day Men 88.7 156.6 220.0 302.6 Women 72.1 129.0 190.1 269.9 Nuts, seeds and legumes g/day Men 9.6 19.0 29.4 44.7 Women 7.4 16.7 25.8 41.0 Fruits g/day Men 42.9 102.1 167.7 266.0 Women 47.7 100.2 160.0 251.5

Vegetables (excl. potatoes) g/day Men 72.4 105.2 138.4 185.0 Women 80.2 112.4 148.9 200.1 Processed and red meat g/day Men 43.0 64.2 85.2 110.4 Women 21.9 40.2 57.4 80.4 Sugar sweetened beverages g/day Men 0.0 28.6 98.0 274.0 Women 0.0 29.4 92.9 233.3 Sodium mg/day* Men 2391.6 2798.7 3191.9 3696.0 Women 1906.7 2242.9 2584.2 3008.4 *Sodium is includes only intrinsic sodium (not discretionary salt added to cooking and/or at the table)

!!"#" !

A5.2 D

ietary profile by quintile cut off of DA

SH score for m

en and wom

en in the Airw

ave Health M

onitoring Study

D

ASH

score group based on quintile cut off

Q

1 (least healthy)

Q2

Q3

Q4

Q5

(most healthy)

P*

Men, N

(%)

619 17.7

680 19.5

803 23

710 20.3

685 19.6

D

ASH

score median (range)

17 9 - 19

21 20 -22

24 23 - 25

27 26 - 28

31 29- 39

Mean, S

E

N

utrient and food intake

Mean daily energy intake, kcal

2128 19

2057 18

2083 17

2057 18

2063 18

0.035 %

energy intake – total fat 35.9

0.2 34.5

0.2 34.1

0.2 32.4

0.2 30.9

0.2 <0.0001

% energy intake – saturated fat

13.5 0.1

12.9 0.1

16.7 0.1

11.8 0.1

10.8 0.1

<0.0001 %

energy intake – MU

FA

12.6 0.1

12.0 0.1

11.8 0.1

11.1 0.1

10.5 0.1

<0.0001 %

energy intake – PU

FA

5.9 0.1

5.7 0.1

5.8 0.1

5.6 0.1

5.7 0.1

0.009 %

energy intake - carbohydrate 44.1

0.3 45.5

0.3 46.4

0.2 47.5

0.2 49.8

0.3 <0.0001

% energy intake N

ME

12.8

0.2 11.9

0.2 11.5

0.2 11

0.2 10.8

0.2 <0.0001

% energy intake - protein

16.5 0.1

16.9 0.1

17.0 0.1

17.4 0.1

17.9 0.1

<0.0001 %

energy intake plant protein 9.0

2.7 9.2

2.8 9.5

2.8 9.7

2.8 10.3

3.1 <0.0001

% energy intake anim

al protein 6.0

3.0 5.9

3.0 5.8

2.8 5.7

2.7 5.3

2.4 <0.0001

Energy density food kcal/g

1.83 0.01

1.65 0.01

1.56 0.01

1.43 0.01

1.35 0.01

<0.0001 N

SP

g/1000kcal 4.9

0.1 5.8

0.1 6.7

0.1 7.5

0.1 8.9

0.1 <0.0001

% energy intake – A

lcohol 5.0

9.1 4.6

8.5 4.0

6.9 4.7

7.2 3.6

6.0 0.0003

Vegetables gram

s/1000kcal 48.9

30.4 61.5

35.9 71.5

41 82.3

45.1 101.6

50.8 <0.0001

Fruit grams/1000kcal

22.2 43.1

43.7 59.1

63 68

87.9 77

123.7 93.7

<0.0001 W

holegrain grams/1000kcal

3.7 12.1

10.8 20

18.9 26.3

28.2 31.2

39.4 36.2

<0.0001 Total dairy gram

s/1000kcal 79.6

58.6 104.5

68.3 121.3

73.0 133.8

76.6 160.7

81.2 <0.0001

Fish grams/1000kcal

4.6 12.2

7.5 14.8

7.5 15.4

10.2 16.4

13.1 18.5

<0.0001 Total red and processed m

eat g/1000kcal 47.9

25.3 41.1

26.9 36.9

23.7 32.1

23.1 24.1

20.4 <0.0001

Sugar sw

eetened beverages g/1000kcal 99.5

169.2 49.3

125.7 31.2

94.4 15.3

56.7 0

24.7 <0.0001

Sodium

mg/1000kcal

1570 12

1530 11

1506 10

1452 11

1387 11

<0.0001

!!"#$ !

A5.2 continued (m

en) D

ASH

score group based on quintile cut off

Q

1 (least healthy)

Q2

Q3

Q4

Q5

(most healthy)

P*

N

, %

D

iscretionary salt usage (n = 2883)

<0.0001

None

201 41.4

255 45.9

335 50.4

320 53.6

349 60.2

C

ooking or at the table 137

28.2 157

28.3 185

27.8 159

26.6 144

24.8

Cooking and at the table

109 22.4

99 17.8

94 14.1

68 11.4

42 7.2

S

alt substitute used 39

8.0 44

7.9 51

7.7 50

8.4 45

7.8

Nutritional supplem

ent use~

Mineral (m

ulti or single) 31

5.8 40

6.6 34

4.7 42

6.7 54

8.8 0.05

Vitam


61 11.4

100 16.4

116 16.2

89 14.2

110 17.8

0.029 O

mega (plant and fish)

44 8.2

47 7.7

91 12.7

84 13.4

128 20.8

<0.0001 O

ther / herbal 22

4.1 43

7.0 67

9.3 54

8.6 89

14.4 <0.0001

Alcohol FD

528

17.6 570

19 690

23 631

21 585

19.5 0.09

Units/day* (consum

ers only) 2.2

3.3 2.1

2.9 1.9

2.5 2

2.5 1.6

2.2 <0.0001

Under reporter

292 48.3

273 41.7

338 43.2

291 43

281 44.2

0.17 W

eight loss diet 14

2.3 25

3.7 20

2.5 34

4.8 49

7.2 <0.0001

Other diet

3 0.5

5 0.8

10 1.3

12 1.8

14 2.2

0.047

E

I irregularity score

22.4 0.3

21.5 0.3

20.2 0.3

19.3 0.3

17.6 0.3

<0.0001

!!"#% !

A5.2 continued (w

omen)

DA

SH score group based on quintile cut off

Q

1 (least healthy)

Q2

Q3

Q4

Q5

(most healthy)

P*

Wom

en, N %

421

17.9 475

20.2 515

21.9 491

20.9 450

19.1

DA

SH score m

edian (range) 18

9- 19 21

20- 22 24

23 - 25 27

26 - 28 31

29 -38

Nutrient and food intake

M

ean, SE

Mean daily energy intake, kcal

1711 19

1672 18

1670 17

1648 17

1675 18

0.17 %

energy intake – total fat 35.9

0.3 34.6

0.3 34.2

0.2 32.9

0.2 31.5

0.3 <0.0001

% energy intake – saturated fat

13.4 0.1

12.9 0.1

12.6 0.1

12 0.1

11.1 0.1

<0.0001 %

energy intake – MU

FA

12.5 0.1

11.8 0.1

11.6 0.1

11 0.1

10.5 0.1

<0.0001 %

energy intake – PU

FA

6.1 0.1

5.9 0.1

6.0 0.1

5.9 0.1

5.9 0.1

0.41 %

energy intake - carbohydrate 45.4

0.3 47.3

0.3 47.3

0.3 48.8

0.3 51.3

0.3 <0.0001

% energy intake N

ME

13.7

0.2 12.8

0.2 11.9

0.2 11.5

0.2 11.4

0.2 <0.0001

% energy intake - protein

16.3 0.2

16.6 0.2

17 0.1

17.4 0.2

17.4 0.2

<0.0001 %

energy intake plant protein 9.3

3.0 9.4

3.0 9.5

3.0 9.6

3.0 9.8

2.8 <0.0001

% energy intake anim

al protein 5.7

2.7 5.6

3.1 5.6

3.0 5.6

3.0 5.5

2.9 0.51

Energy density food kcal/g

1.70 0.02

1.58 0.02

1.45 0.02

1.36 0.02

1.24 0.02

<0.0001 N

SP

g/1000kcal 5.3

0.1 6.4

0.1 7.2

0.1 8.3

0.1 9.7

0.1 <0.0001

% energy intake – A

lcohol 4

7.2 3.1

6.8 3.2

6.7 2.9

6.5 2.3

5.1 <0.0001

Vegetables gram

s/1000kcal 60.6

39.2 78

47.7 92.6

54.1 112

60.6 130.9

80.6 <0.0001

Fruit grams/1000kcal

30.6 54.1

59.1 47.6

78.1 80.2

101 89.1

149.3 116.8

<0.0001 W

holegrain grams/1000kcal

4.8 13.4

13 20.4

19 26.5

26.8 27

36.1 30.1

<0.0001 Total dairy gram

s/1000kcal 78.5

64 109

76.3 124.5

83.1 149.1

90.9 184

107.3 <0.0001

Fish grams/1000kcal

6.5 13.1

7.9 16.2

9.4 19.7

12.5 19.8

13.5 22.1

<0.0001 Total red and processed m

eat g/1000kcal 40.3

26.3 35.4

25.8 30.6

26.3 24.7

25.9 16.3

22.8 <0.0001

Sugar sw

eetened beverages g/1000kcal 117.6

176.1 55.2

124.6 31

107.1 18.1

66.7 0

34.7 <0.0001

Sodium

mg/1000kcal

1543 16

1517 15

1466 14

1495 14

1397 15

<0.0001

!!"#& !

A5.2 continued (w

omen)

DA

SH score group based on quintile cut off

Q

1 (least healthy)

Q2

Q3

Q4

Q5

(most healthy)

P*

N

, %

D

iscretionary salt usage (n = 2029)

0.0001

None

160 45.1

203 50.6

228 51.7

228 53.1

250 62.0

C

ooking or at the table 111

31.3 115

28.7 122

27.7 113

26.3 90

22.3

Cooking and at the table

56 15.8

57 14.2

53 12

50 11.7

24 6.0

S

alt substitute used 28

7.9 26

6.5 38

8.6 38

8.9 39

9.7

Nutritional supplem

ent use~

Mineral (m

ulti or single) 30

7.7 40

9.4 35

7.5 56

12.4 73

17.6 <0.0001

Vitam


51 13.1

68 15.9

90 19.3

113 25.1

123 27.6

<0.0001 O

mega (plant and fish)

11 2.8

30 7.0

54 11.6

52 11.6

74 17.8

<0.0001 O

ther / herbal 28

7.2 46

10.8 63

13.5 67

14.9 83

20.0 <0.0001

Alcohol FD

347

18.6 377

20.2 418

22.4 380

20.3 345

18.5 0.16

Units/day* (consum

ers only) median, IQ

R

1.5 2.2

1.3 2.0

1.3 1.7

1.2 1.9

1.0 1.3

<0.0001 U

nder reporter~ 240

61.1 246

57.1 272

59.4 238

56.4 230

63.5 0.23

Weight loss diet~

28 6.7

44 9.3

56 10.9

69 14.2

88 19.8

<0.0001 O

ther diet~ 2

0.5 2

0.5 2

0.7 7

1.9 7

1.9 0.09


22.1

0.4 20.8

0.4 20.4

19.5 0.4

17.9 0.4

<0.0001

Abbreviations: E

I energy intake, FD food diary, M

UFA

mono unsaturated fatty acids, N

SP

non-starch polysaccharides, PU

FA polyunsaturated fatty

acids. *General linear m

odel applied to test linear relationship between dietary intake across fifths of D

AS

H score. M

eans and standard error of the

mean presented. ~ categorical variables, C

hi squared test applied.

!

! "#$!

A5.3 Comparison of key characteristics across ranked and non-ranked police force employees included in the Airwave Health Monitoring Study (n = 5,107)

Non ranking staff and other roles

Police officers (constables and

sergeants)

Higher ranking officers

(Inspector or higher)

P*

N (%) 1841 36.1 2876 56.3 390 7.6 Age, years (SE) 43.3 0.2 39.6 0.2 45.3 0.5 <0.0001 N (%) Male 645 35.0 2065 71.8 333 85.4 <0.0001 White 1798 97.7 2788 97.0 381 97.7 0.35 Relationship status <0.0001 Cohabiting 281 15.7 528 18.7 36 9.3 Divorced/separated 179 10.2 195 6.9 25 6.5 Married 1030 57.6 1824 64.6 312 80.8 Single 297 16.6 276 9.8 13 3.4 Education <0.0001 Left school before taking

132 7.2 87 3.0 5 1.3

GCSE or equivalent 527 28.6 903 31.4 60 15.4 Vocational qualifications 159 8.6 211 7.3 12 3.1 A levels / Highers or

544 29.6 976 33.9 125 32..0

Bachelor Degree or

331 18.0 575 20.0 126 32.3 Postgraduate

148 8.0 123 4.3 62 15.9

Annual household

<0.0001 Less than £32,000 595 32.3 238 8.3 2 0.5 £32,000 - £47,999 225 12.2 355 12.4 4 1.0 £48,000 - £57,999 653 35.5 1252 43.6 93 23.9 £58,000- £77,999 243 13.2 754 26.2 150 38.4 More than £ 78,000 125 6.8 276 9.6 141 36.2 Employment force

<0.0001

England 1317 71.9 1857 64.8 238 61.0 Scotland 241 13.1 665 23.2 109 27.9 Wales 274 15.0 345 12.0 43 11.0 Work environment <0.0001 Mainly office duties 602 32.7 966 33.6 275 14.9 Mainly mobile duties 277 15.0 1699 59.1 87 4.2 Unclassified 962 52.2 211 7.3 28 2.3 Total hours worked

<0.0001

Part time 297 16.1 152 5.3 6 1.5 35-40 hours 1083 58.8 775 26.9 56 14.4 41 – 48 hours 337 18.3 1166 40.5 123 31.5 49 – 54 hours 65 3.5 387 13.5 89 22.8 55 hours or more 59 3.2 396 13.8 116 29.7

!

! "##!

A5.3 continued Non ranking staff and other roles

Police officers (constables and

sergeants)

Higher ranking officers

(Inspector or higher)

P*

Years in police force <0.0001 6 years or less 790 42.9 532 18.5 5 1.3 6 to 12 years 439 23.8 715 24.9 7 1.8 12 to 21 years 294 16.0 879 30.6 111 28.5 21 years or more 318 17.3 750 26.1 267 64.5 Shift work last 30 days <0.0001 Day only 152 45.2 158 10.0 45 34.1 Shift (no nights) 168 50.0 302 19.2 24 18.2 Shift (with night work) 16 4.8 1112 70.8 63 47.8 Job Strain

<0.0001 Low (high control, low

568 30.8 830 28.9 130 33.3

Passive (low control, low

423 23.0 576 20.0 22 5.6 Active (high demand,

501 27.2 712 24.8 198 50.8

High (high demand, low

349 19.0 758 26.4 40 10.3 Physical activity† <0.0001 Low 255 13.8 320 11.1 42 10.8 Moderate 940 51.1 1252 43.5 165 42.3 High 646 35.1 1304 45.3 183 46.9 Smoking status <0.0001 Never smoker 1196 65.2 2031 70.8 306 78.7 Former smoker 478 26.1 596 20.8 72 18.5 Current smoker 159 8.7 241 8.4 11 2.8 Sleep 0.29 5 hours or less 116 6.3 137 4.8 18 4.6 6 hours 477 25.9 786 27.3 116 29.7 7 hours 779 42.3 1201 41.8 164 42.0 8 hours 416 22.6 650 22.6 80 20.5 9 hour or more 53 2.9 101 3.5 12 3.1 Sitting (total

<0.0001

Low (<20 hours) 538 29.2 992 34.5 76 19.5 Moderate (20 – 40

668 36.3 1106 38.5 164 42.0

High (> 40 hours) 635 34.5 778 27.0 150 38.5 Weekly TV viewing 0.10 Low (< 6 hours) 541 29.4 836 29.1 112 28.7 Moderate (6 – 15 hours) 789 42.9 1299 45.2 192 49.2 High (>15 hours) 511 27.8 741 25.8 86 22.1 BMI category <25 kg/m2 722 <0. 875 30.4 85 21.8 <0.0001 25-30 kg/m2 720 39.1 1452 50.5 213 54.6 >30 kg/m2 399 21.7 549 19.1 92 23.6 Abbreviations: SD: standard deviation. GCSE: General certificate of Secondary Education; EI: energy intake. *Student t-test compared mean values. Chi squared test compared differences between men and women across categorical variables; missing data was not included in the analyses. † METs metabolic equivalents, classification by IPAQ guidelines.

!

! "#%!

A5.4 Comparison of key characteristics region of employment for Airwave Health Monitoring Study participants (n = 5,830)

England Scotland Wales p N (%) 4150 71.2 1018 18.3 662 11.9 Age, years (SE) 41.3 0.1 41.7 0.3 41.7 0.4 0.31 N (%) Male 2423 69.5 671 65.9 393 59.4 <0.0001 White 4050 97.6 981 96.4 641 96.8

0.06 Relationship status 0.003 Cohabiting 699 17.3 152 15.3 100 15.4 Divorced/separated 354 8.7 66 6.6 44 6.8 Married 2520 62.2 681 68.4 414 63.8 Single 476 11.8 96 6.7 91 14.0 Education <0.0001 Left school before taking

175 4.2 34 3.3 36 5.4

GCSE or equivalent 1348 32.5 188 18.5 198 29.9 Vocational qualifications 316 7.6 65 6.4 44 6.7 A levels / Highers or

1275 30.7 405 39.8 204 30.8

Bachelor Degree or

798 19.2 245 24.1 128 19.3 Postgraduate

237 5.7 81 8.0 52 7.9

Annual household

<0.0001 Less than £32,000 695 16.8 93 9.1 151 22.8 £32,000 - £47,999 476 11.5 103 10.1 89 13.4 £48,000 - £57,999 1954 38.4 406 39.9 274 41.4 £58,000- £77,999 938 22.6 296 29.1 105 15.9 More than £ 78,000 446 10.8 120 11.8 43 6.5 Rank <0.0001 Police staff /other 1317 38.6 241 23.7 274 41.4 Police Constable/

1857 54.4 681 67.1 350 52.9

Inspector/Chief

238 7.0 93 9.2 38 5.7 Work environment <0.0001 Mainly office duties 1246 36.5 346 34.1 246 37.1 Mainly mobile duties 1316 38.6 491 47.4 248 37.5 Unclassified 850 24.9 178 17.5 168 25.4 Total hours worked

<0.0001

Part time 405 9.8 73 7.2 54 8.2 35-40 hours 1632 39.3 376 36.9 219 33.1 41 – 48 hours 1310 31.6 312 30.7 250 37.8 49 – 54 hours 438 10.5 104 10.2 64 9.7 55 hours or more 365 8.8 153 15.0 75 11.3

!

! "#&!

A5.4 continued England Scotland Wales p

Shift work last 30 days <0.0001 Day only 191 12.3 163 31.0 15 6.2 Shift (no nights) 463 29.8 62 11.8 41 17.0 Shift (with night work) 899 57.9 301 57.2 185 76.8 Job Strain

0.13 Low (high control, low demand)

1217 29.3 331 32.5 183 27.6

Passive (low control, low demand)

858 20.7 217 21.3 129 19.5

Active (high demand, high control)

1121 27.0 258 25.3 201 30.4

High (high demand, low control)

954 23.0 212 20.8 149 22.5

Physical activity† 0.11 Low 492 11.9 121 11.9 89 13.4 Moderate 1887 45.5 502 49.3 290 43.8 High 1771 42.7 395 38.8 283 42.8 Smoking status 0.001 Never smoker 2812 68.0 759 74.8 445 67.5 Former smoker 696 23.4 197 19.4 153 23.2 Current smoker 355 8.6 59 5.8 61 9.3 Sleep 0.001 5 hours or less 221 5.3 58 5.7 37 5.6 6 hours 111 26.8 310 30.4 154 23.3 7 hours 1791 43.2 414 40.7 262 39.6 8 hours 887 21.4 203 19.9 186 28.1 9 hour or more 139 3.4 33 3.2 23 3.5 Sitting (total

0.28

Low (<20 hours) 1268 30.6 326 32.0 228 34.4 Moderate (20 – 40

1594 38.4 389 38.2 249 37.6

High (> 40 hours) 1288 31.0 303 29.8 185 27.9 Weekly TV viewing 0.24 Low (< 6 hours) 1207 29.1 287 28.2 219 33.1 Moderate (6 – 15 hours) 1854 44.7 464 45.6 284 42.9 High (>15 hours) 1089 26.2 267 26.2 159 24.0 BMI category 0.0004 <25 kg/m2 1426 34.4 308 30.3 189 28.6 25-30 kg/m2 1907 45.9 514 50.5 308 46.5 >30 kg/m2 817 19.7 196 17.3 165 24.9 Abbreviations: SD: standard deviation. GCSE: General certificate of Secondary Education; Missing region data n = 19. † METs metabolic equivalents, classification by IPAQ guidelines.

!!"#$!

A 5

.5 C

om

paris

on

of k

ey

ch

ara

cte

ristic

s b

y w

ork

ing

ho

urs

for A

irwav

e H

ealth

Mo

nito

ring

Stu

dy

partic

ipan

ts

A5.5.1 M

en (n = 3,497)*

35

- 40

hrs

/we

ek

4

1 –

48

hrs

/we

ek

4

9 –

54

hrs

/we

ek

>

55

hrs

/we

ek

p

N (%

) 1211

(34.6) 1316

(37.6) 459

(13.1) 453

(12.9)

Ho

urs

wo

rke

d p

er w

ee

k, m

ea

n (S

E)

39.1 (0.1)

43.8 (0.1)

50.5 (0.1)

60.2 (0.1)

<0

.00

01

Ag

e, y

ea

rs m

ea

n (S

E)

44.2 (0.2)

41.5 (0.2)

41.3 (0.4)

40.8 (0.4)

<0

.00

01

N (%

)

Wh

ite

M

arita

l sta

tus

1178

97.4 1267

96.3 441

96.1 441

97.4 0.27

Cohabiting

141 11.9

200 15.3

61 13.4

75 16.8

0.15 D


7.7 74

5.7 33

7.3 32

7.2

Married

874 73.4

943 72.3

332 73.0

309 69.3

S

ingle 84

7.1 88

6.7 29

6.4 30

6.7

Missing

E

du

ca

tion

0

.01

7


SE

79

6.5 46

3.5 19

4.1 13

2.9

GC

SE

or equivalent 383

31.6 398

30.2 143

31.2 129

28.5


85 7.0

96 7.3

34 7.4

35 7.7

A


365 30.1

454 34.5

148 32.2

148 32.7

B


234 19.3

252 19.2

88 19.2

89 19.7

P


5.4 70

5.3 27

5.9 39

8.6

An

nu

al h

ou

se

ho

ld in

co

me

<

0.0

00

1

Less than £32,000 146

12.1 95

7.2 17

3.7 31

6.8

£32,000 - £47,999 172

14.2 145

11.0 46

10.0 39

8.6

£48,000 - £57,999 541

44.7 626

47.6 200

43.6 153

33.8

£58,000- £77,999 250

20.6 346

26.3 129

28.1 155

34.2

More than £ 78,000

102 8.4

104 7.9

67 14.6

75 16.6

!!"#%!

A5.5.1 continued (m

en) 3

5 - 4

0 h

rs/w

ee

k

41

– 4

8 h

rs/w

ee

k

49

– 5

4 h

rs/w

ee

k

>5

5 h

rs/w

ee

k

p

N

(%)

Y

ea

rs in

po

lice

forc

e

6 years or less

237 19.6

213 16.2

64 13.9

72 15.9

<0

.00

01

6 to 12 years

264 21.8

329 25.0

88 19.2

105 23.2

12 – 21 years

273 22.5

373 28.3

149 32.5

119 26.3

21 years or m

ore 437

36.1 401

30.5 158

34.4 157

34.7

Tim

e in

cu

rren

t job

role

0

.04

8

2 years or less 428

35.3 435

33.1 176

38.3 185

40.8

3 to 5 years 352

29.1 422

32.1 138

30.1 128

28.3

6 years or more

431 35.6

459 34.9

145 31.6

140 30.9

S

hift w

ork

30

da

ys

<

0.0

00

1

Day

93 21.4

79 37.4

22 9.5

17 7.7

S

hift no night 118

27.2 125

39.3 43

18.6 32

14.5

Shift w

ith night 223

51.4 498

47.0 166

71.9 172

77.8

Em

plo

ym

en

t (forc

e) c

ou

ntry

<

0.0

00

1

England

889 73.6

892 67.9

320 69.8

279 62.0

S

cotland 214

17.7 248

19.8 83

18.1 117

26.0

Wales

104 8.6

174 13.2

55 12.5

54 12.0

R

an

k

P

olice staff/other 419

40.8 144

12.7

24 5.9

23 5.3

<0

.00

01

P

olice Constable/S

ergeant 562

54.8 889

78.3 298

73.8 307

71.1

Inspector/Chief Inspector or above

45 4.4

102 9.0

82 20.3

102 20.3

J

ob

de

ma

nd

-co

ntro

l

<

0.0

00

1


demand)

500 41.3

386 29.3

128 27.9

96 21.2

P

assive (low control, low

demand)

245 20.2

218 16.6

64 13.8

63 13.9

A


264 21.8

391 29.7

152 33.1

189 41.7

H

igh strain (high demand, low

control) 202

16.7 321

24.4 115

25.1 105

23.2

!!"#"!

A5.5.1 continued (m

en) 35 - 40 hrs/w

eek 41 – 48 hrs/w

eek 49 – 54 hrs/w

eek >55 hrs/w

eek P

N (%

)

Wo

rk e

nv

iron

me

nt

<0

.00

01

M

ainly office duties 340

33.1 362

31.9 170

42.1 180

41.7

Mainly m

obile duties 417

40.6 611

54.8 206

51.0 212

49.1

Unclassified

269 26.2

151 13.3

28 6.9

40 9.3

P

hy

sic

al a

ctiv

ity (M

ET

s)

Low

136

11.2 139

10.6 43

9.4 48

10.6 0.94

Moderate

523 43.2

574 43.6

206 44.8

205 45.2

H

igh 552

45.6 603

45.8 210

45.7 200

44.1

Sm

ok

ing

sta

tus

0

.01

4

Never sm

oker 813

66.4 939

71.7 326

71.2 327

72.4

Former sm

oker 320

26.5 278

21.2 94

20.5 91

20.1

Current sm

oker 74

6.1 92

7.0 38

8.3 34

7.5

Sle

ep

0.06

5 hours or less 51

4.2 61

4.6 28

6.1 33

7.3

6 hours 355

29.3 395

30.0 133

29.0 158

34.9

7 hours 532

43.9 589

44.8 201

43.8 168

37.1

8 hours 235

19.4 233

17.7 84

18.3 87

19.2

9 hour or more

38 3.1

38 2.9

13 2.8

7 1.5

W

ee

kly

ho

urs

sittin

g (w

ee

kd

ay

s)

0.0

15

Low

(<20 hours) 340

28.1 425

32.3 151

32.9 138

30.5

Moderate (20 – 40 hours)

443 36.6

516 39.2

168 36.6

166 36.6

H

igh (>40 hours) 428

35.3 375

28.5 140

30.5 149

32.9

TV

vie

win

g, h

ou

r pe

r we

ek

<0

.00

01

Low

(<6 hours) 306

25.3 307

23.3 124

27.0 143

31.6


501 41.4

643 48.9

218 47.5

196 43.3

H

igh (>15 hours) 404

33.4 366

27.8 117

25.5 114

25.2

Abbreviations: G

CS

E G


econdary Education. M

ETs m

etabolic equivalents, classification by IPA

Q guidelines (250). C

hi squared test to compare differences

across categorical variables, missing data w

as not included in the analyses

!!"#&!

A5.5.2 W

omen (n=2,352)

<

35

h

rs/w

ee

k 3

5 - 4

0

hrs

/we

ek

41

– 4

8

hrs

/we

ek

>4

9

hrs

/we

ek

N (%

) 477

(20.3) 1023

(43.5) 561

(23.8) 291

(12.4)

Ho

urs

wo

rke

d p

er w

ee

k, m

ea

n (S

E)

25.1 0.2

38.2 0.1

43.5 0.1

55.1 0.2

<0

.00

01

Ag

e, y

ea

rs m

ea

n (S

E)

42.0 0.4

40.8 0.3

37.8 0.4

36.8 0.5

<0

.00

01

(not

N (%

)

Wh

ite

471 98.7

1003 98.0

551 98.2

282 96.9

0.35

Ma

rital s

tatu

s

<0

.00

01

Cohabiting

38 8.1

208 21.3

151 27.9

76 27.6

Divorced/separated

40 8.5

108 11.1

57 10.5

31 11.3

Married

366 77.9

454 46.6

205 37.8

102 37.1

Single

26 5.5

205 21.0

129 23.8

66 24.0

Ed

uc

atio

n

<0

.00

01


SE

14

2.9 43

4.2 17

3.0 9

3.1

GC

SE

or equivalent 172

36.1 287

28.1 136

24.2 68

23.4


33 6.9

78 7.6

40 7.1

22 7.6

A levels / H

ighers or equivalent 162

34.0 338

33.0 178

31.7 86

29.6

Bachelor D


15.3 205

20.0 143

25.5 82

28.2

Postgraduate qualifications

23 4.8

72 7.0

47 8.4

24 8.3

An

nu

al h

ou

se

ho

ld in

co

me

<

0.0

00

1

Less than £32,000 123

25.8 292

28.5 144

25.7 65

22.3

£32,000 - £47,999 41

8.6 119

11.6 63

11.2 38

13.1

£48,000 - £57,999 177

37.1 340

33.2 157

28.0 74

25.4

£58,000- £77,999 96

20.1 177

17.3 122

21.7 62

21.3

More than £ 78,000

40 8.4

95 9.3

75 13.4

52 17.9

!!"#'!

A5.5.2 continued (w

omen)

<3

5

hrs

/we

ek

35

- 40

h

rs/w

ee

k 4

1 –

48

h

rs/w

ee

k >

49

h

rs/w

ee

k p

N

(%)

Ye

ars

in p

olic

e fo

rce

<

0.0

00

1

6 years or less 128

26.8 387

37.8 219

39.0 120

41.2

6 to 12 years 122

25.6 251

24.5 156

27.8 61

21.0

12 – 21 years 158

33.1 226

22.1 117

20.9 59

20.3

21 years or more

69 14.5

159 15.5

69 12.3

51 17.5

Tim

e in

cu

rren

t job

role

<

0.0

00

1

2 years or less 150

31.4 405

39.6 242

43.1 148

50.9

3 to 5 years 119

24.9 302

29.5 179

31.9 73

25.1

6 years or more

208 43.6

316 30.9

140 25.0

70 24.1

Sh

ift wo

rk 3

0d

ay

s

<0

.00

01

Day

42 36.5

75 29.5

29 13.0

10 7.6

Shift no night

37 32.2

105 41.3

67 30.0

35 26.5

Shift w

ith night 36

31.3 74

29.1 127

56.9 87

65.9

Em

plo

ym

en

t (forc

e) c

ou

ntry

0

.02

4

England

362 76.4

743 72.8

418 73.7

204 70.1

Scotland

64 13.5

162 15.9

64 11.5

57 19.6

Wales

48 10.3

115 11.3

76 13.6

30 10.3

!!"#(!

A5.5.2 continued (w

omen)

<3

5

hrs

/we

ek

35

- 40

h

rs/w

ee

k 4

1 –

48

h

rs/w

ee

k >

49

h

rs/w

ee

k p

N

(%)

Ra

nk

<

0.0

00

1 P

olice staff/other 262

64.1 664

74.8 193

39.3 77

27.9

Police C

onstable/Sergeant

143 35.0

213 24.0

277 56.4

178 64.5


4 0.9

11 1.2

21 4.3

21 7.6

Jo

b d

em

an

d-c

on

trol

<0

.00

01


demand)

116 24.3

311 30.4

125 22.3

51 17.5

Passive (low

control, low dem

and) 155

32.5 288

28.2 110

19.6 46

15.8

Active (high dem

and, high control) 98

20.5 228

22.3 166

29.6 88

30.2


and, low control)

108 22.6

196 19.2

160 28.5

106 36.4

Wo

rk e

nv

iron

me

nt


163 39.9

364 41.0

158 32.2

94 34.1

<0

.00

01

Mainly m

obile duties 88

21.5 176

19.8 200

40.7 128

46.4

Unclassified

158 38.6

348 39.2

133 27.1

54 19.6

Ph

ys

ica

l ac

tivity

(ME

Ts

)

0

.00

6

Low

65 13.6

167 16.3

69 12.3

33 11.3

Moderate

244 51.2

510 49.8

253 45.1

139 47.8

High

168 35.2

346 33.8

239 42.6

119 40.9

Sm

ok

ing

sta

tus

0.29

Never sm

oker 333

70.0 679

66.5 377

67.4 200

69.2

Former sm

oker 110

23.1 234

22.9 122

21.8 57

19.7

Current sm

oker 33

6.9 108

10.6 60

10.7

32 11.1

!!"#)!

A5.5.2 continued (w

omen)

35 - 40 hrs/week

41 – 48 hrs/week

49 – 54 hrs/week

>55 hrs/week

P

Sle

ep

0

.04

6

5 hours or less 20

4.2 72

7.0 28

5.0 24

8.3

6 hours 99

20.8 226

22.1 117

20.9 77

26.5

7 hours 201

42.1 412

40.3 238

42.4 105

36.1

8 hours 144

30.2 270

26.4 151

26.9 68

23.4

9 hour or more

13 2.7

43 4.2

27 4.8

17 5.8

We

ek

ly h

ou

rs s

itting

(we

ek

da

ys

)

Low (<20 hours)

179 37.5

282 27.6

188 33.5

110 37.8

<0

.00

01


219 45.9

370 36.2

224 39.9

102 35.1

High (>40 hours)

79 16.6

371 36.3

149 26.6

79 27.2

TV

vie

win

g, h

ou

r pe

r we

ek

0

.00

1

Low (<6 hours)

160 33.5

344 33.6

195 34.8

128 44.0


232 48.6

429 41.9

245 43.7

117 40.2

High (>15 hours)

85 17.8

250 24.4

121 21.6

46 15.8

Abbreviations: G

CS

E G



ETS

metabolic equivalents, classification by IP

AQ

guidelines (250). Chi squared test to com

pare differences across categorical variables, m


!!"#*!

A5.6

Partia

l co

rrela

tion

co

effic

ien

ts fo

r die

tary

macro

nu

trien

ts a

nd

foo

d g

rou

p in

takes o

f the A

irwav

e H

ealth

Mo

nito

ring

S

tud

y p

artic

ipan

ts (n = 5,849) 1

DASH score

Mean energy intake

Energy density food

Mean EI irregularity score

% TEI fat

% TEI saturated fat

% TEI protein

% TEI carbohydrate

% TEI NMEs

% TEI alcohol

NSP g/1000kcal

Sodium mg/1000kcal

SSBs g/1000kcal

Full fat Dairy g/1000kcal

Low fat dairy g/1000kcal

Whole grains g/1000kcal

Fruit g/1000kcal

Vegetables g/1000kcal

Legumes g/1000kcal

Fish g/1000kcal

Red Meat g/1000kcal

Mean energy intake

-0.04

E

nergy density food -0

.42

0.22

M

ean EI irregularity score

-0.17 -0.18

0.07

%

TEI fat

-0.29 0.20

0.3

6

0.02

% TE

I saturated fat -0.29

0.20 0

.33

0.00

0.8

0

% TE

I protein 0.13

-0.3

2

-0.26 -0.01

-0.17 -0.20

%

TEI carbohydrate

0.27 -0.07

-0.12 -0.17

-0.5

0 -0

.35

-0.22

% TE

I NM

Es

-0.13 0.20

0.12 -0.04

-0.11 0.01

-0.4

3

0.5

0

%

TEI alcohol

-0.08 0.08

-0.08 0.17

-0.23 -0.21

-0.10 -0

.49

-0.22

NS

P g/1000kcal

0.7

0

-0.17 -0

.40

-0.16

-0.3

4

-0.3

6

0.20 0

.37

-0.20

-0.17

Sodium

mg/1000kcal

-0.20 -0.20

-0.07 0.01

0.07 0.02

0.23 -0.07

-0.24 -0.08

0.08

S

SB

s g/1000kcal -0

.43

0.10

0.11 0.04

-0.06 -0.03

-0.21 0.20

0.5

0

-0.06 -0.22

-0.09

Full fat Dairy g/1000kcal

0.00 0.15

0.05 -0.03

0.20 0.35

-0.10 0.00

0.13 -0.11

-0.07 -0.05

0.01

Low

fat dairy g/1000kcal 0

.47

-0.09

-0.21 -0.14

-0.22 -0.12

0.27 0.22

-0.06 -0.16

0.24 -0.02

-0.15 -0.13

W

hole grains g/1000kcal 0

.54

-0.04

-0.21 -0.11

-0.21 -0.22

0.15 0.18

-0.10 -0.07

0.5

9

0.05 -0.14

0.00 0.22

Fruit g/1000kcal 0

.55

0.01

-0.4

2

-0.15 -0.29

-0.26 0.00

0.3

7

0.13 -0.11

0.4

6

-0.14 -0.07

0.05 0.15

0.23

Vegetables g/1000kcal

0.4

6

-0.17 -0

.44

-0.05

-0.18 -0.22

0.28 0.01

-0.18 0.04

0.4

7

0.10 -0.13

-0.05 0.08

0.17 0.20

Legumes g/1000kcal

0.3

7

-0.08 -0.20

-0.04 -0.10

-0.13 0.07

0.09 -0.10

-0.01 0

.42

0.15

-0.07 -0.07

0.04 0.10

0.08 0.20

Fish g/1000kcal

0.21 -0.06

-0.16 -0.03

-0.08 -0.19

0.26 -0.09

-0.13 0.05

0.15 -0.01

-0.07 -0.04

0.07 0.12

0.11 0.16

0.04

R

ed Meat g/1000kcal

-0.4

1

-0.09 0.01

0.07 0.14

0.15 0

.33

-0

.33

-0.17

0.09 -0.19

0.24 -0.02

-0.13 -0.07

-0.14 -0.17

0.02 -0.03

-0.14

Processed m

eat g/1000kcal -0

.41

-0.03

0.09 0.05

0.14 0.13

0.17 -0.23

-0.11 0.07

-0.21 0

.41

0.04

-0.08 -0.08

-0.12 -0.16

-0.12 -0.11

-0.15 0

.60

Abbreviations: E

I energy intake, NM

E: N


SP

: Non-starch polysaccharides, S

SB

s sugar sweetened beverages.

1Adjusted for sex. S

pearman rank correlation coefficients presented. B

old shaded values signify medium

to large effect, r ± !0.30. All observations ± >/<0.03 are significant (p

<0.05).

!!!

"##!

A5.7

Sen

sitiv

ity a

naly

ses: D

ieta

ry p

rofile

s a

cro

ss w

ork

ing

ho

ur g

rou

ps a

mo

ng

st m

id-ra

nkin

g p

olic

e o

fficers

: men (n =

2,056)*

35

- 40

h

rs/w

ee

k

41

– 4

8

hrs

/we

ek

49

– 5

4

hrs

/we

ek

!

55

h

rs/w

ee

k

P1

P2

N (%

) 562

(27.3) 889

(43.2) 298

(14.5) 307

(14.9)

M

ea

n (S

D)

DA

SH

score 24.5

(4.7) 23.5

(5.0) 23.4

(4.9) 23.3

(5.0) <

0.0

00

1a

0.0

40

a

Mean daily energy, kcal

2080 (471)

2082 (488)

2110 (459)

2018 (464)

0.12 0.07

Energy density of food, kcal/g

1.5 (0.4)

1.6 (0.4)

1.6 (0.4)

1.6 (0.4)

0.0

01

a

0.0

49

a Fat, %

TEI

33.1 (5.6)

33.8 (5.3)

33.5 (5.2)

33.5 (5.4)

0.12 0.10

Saturated fat, %

TEI

12.1 (2.9)

12.4 (2.8)

12.3 (2.6)

12.4 (2.7)

0.15 0.19

Protein, %

TEI

17.4 (3.5)

17.0 (3.4)

17.2 (4.0)

17.3 (3.4)

0.42

0.13 C

arbohydrate, % TE

I 46.7

(6.7) 46.7

(6.8) 46.2

(7.1) 46.9

(7.2) 0.57

0.59 N

ME

, %E

I 11.3

(4.7) 11.8

(4.9) 11.7

(4.9) 12.0

(5.2) 0.09

0.50 Fibre g/1000kcal

6.9 (1.9)

6.5 (1.9)

6.6 (2.0)

6.6 (2.0)

0.0

08

b 0.07

Sodium

mg/1000kcal

1500 (295)

1492 (285)

1505 (304)

1519 (305)

0.55 0.60

M

ed

ian

(IQR

)

A

lcohol % TE

I 4.5

(7.0) 4.6

(7.7) 4.9

(8.3) 3.6

(7.5) 0.32

0.43 S

SB

s g/1000kcal 26.7

(88.7) 36.2

(109.7) 42.2

(111.6) 43.7

(142.2) 0

.00

1a

0.0

25

e


(24.4) 20.0

(25.5) 18.6

(23.7) 19.4

(24.2) 0.44

0.08 Low


(85.1) 88.8

(79.4) 82.3

(74.1) 91.2

(84.3) 0

.03

1c

0.12 W


(31.2) 16.3

(29.8) 19.4

(29.6) 15.5

(28.3) 0

.00

1a

0.0

05

d

Total fruit g/1000kcal 73.0

(85.8) 55.9

(79.8) 65.3

(83.5) 56.3

(78.4) 0

.00

6d

0.08 V


(45.3) 57.4

(40.8) 57.3

(43.3) 55.8

(42.9) 0.30

0.39 Legum

e g/1000kcal 9.9

(13.0) 10.4

(12.4) 9.8

(14.8) 10.7

(14.1) 0.83


8.5 (18.3)

7.3 (17.2)

7.5 (17.7)

6.8 (17.3)

0.56 0.48


36.0 (27.3)

35.6 (25.0)

36.1 (26.3)

36.6 (23.8)

0.94 0.68

Processed m

eat g/1000kcal 19.2

(18.4) 19.0

(16.9) 19.1

(16.4) 21.0

(19.3) 0.77

0.49 E

I irregularity score 19.5

(8.4) 20.6

(8.8) 21.3

(8.7) 21.6

(8.5) 0

.00

1a

0.0

02

f

!!!

"#+!

A5.7 continued (m

en) 3

5 - 4

0

hrs

/we

ek

4

1 –

48

h

rs/w

ee

k

49

– 5

4

hrs

/we

ek

!

55

h

rs/w

ee

k

P

N

(%)


0.81

No salt added

244 (28.2)

122 (28.0)

66 (27.1)

44 (32.3)


375 (43.3)

175 (40.1)

107 (43.9)

55 (40.4)


127 (14.7)

64 (14.7)

39 (16.0)

17 (12.5)


120 (13.9)

75 (17.2)

32 (13.1)

20 (14.7)

Nutritional supplem

ent use~

Mineral (m

ulti or single) 32

(6.2) 45

(5.9) 11

(4.2) 17

(6.4) 0.67

Vitam


71 (13.8)

127 (16.7)

35 (13.4)

43 (15.6)

0.38

Om


(13.4) 83

(10.9)

31 (11.8)

27 (10.1)

0.48

Other / herbal

45 (8.70)

64 (8.4)

20 (7.6)

17 (6.4)

0.68

*n= 2,056 (exc. part time = 9). A

bbreviations: TEI total energy intake, N

ME

: Non-m

ilk extrinsic sugars. NS

P: N

on-starch polysaccharides. SS

BS

: sugar sweetened beverages.

±Salt usage data n = 1,682, ~supplem

ent usage data n =1,804. Chi-squared test to com

pare differences across categorical variables, missing data not included in analyses.

To compare m

eans values between groups one-w

ay AN

OV

A w


ean and standard deviation). If significance indicated (p< 0.05) B

onferroni post hoc test was applied to identify the source of the difference. W


ere conducted for nonparametric data (values presented as

median and inter quartile range). If significance indicated (p <0.05) W


ere then conducted between each group to establish the source of the difference

with B

onferroni post hoc test applied to correct for multiple com


2 General Linear M


ent (n = 2,049 due to m

issing region data). a)

35-40 vs. different to all other groups b)

35-40 vs. 41-48 hrs c)

35-40 vs. 49-54hrs d)

35-40 vs. 41-48 hrs and 35-40 vs. !55hrs e)

35-40 vs. >55hrs f)

35-40 vs. 49-54 and 35-40 vs. !55hrs

!!!

"+$!

A5.8

Sen

sitiv

ity a

naly

ses: D

ieta

ry p

rofile

s a

cro

ss w

ork

ing

ho

ur g

rou

ps a

mo

ng

st ra

nked

po

lice o

fficers

: wo

men (n

= 8

68)*

<

35

h

rs/w

ee

k

35

- 40

h

rs/w

ee

k

41

– 4

8

hrs

/we

ek

!

49

h

rs/w

ee

k

P1

P2

N (%

) 147

(16.9) 224

(25.8) 298

(34.3) 199

(22.9)

M

ea

n (S

D)

DA

SH

score 24.6

(5.0) 24.5

(4.7) 24.1

(5.0) 23.3

(4.9) 0

.03

6a

0.07 M


(353) 1637

(378 1672

(359) 1647

(384) 0

.03

7b

0.0

39

b

Energy density of food (kcal/g)

1.5 (0.4)

1.5 (0.4)

1.5 (0.4)

1.5 (0.4)

0.76 0.41

Fat, % TE

I 35.1

(5.2) 33.7

(5.7) 33.1

(5.6) 33.6

(6.2) 0

.00

6c

0.0

01

c

Saturated fat, %

TEI

12.9 (2.7)

12.4 (3.1)

11.9 (2.6)

12.0 (3.1)

0.0

03

d 0

.00

7d

Protein, %

TEI

16.6 (2.6)

17.1 (3.7)

17.2 (3.0)

17.1 (3.4)

0.25 0.14

Carbohydrate, %

TEI

47.6 (6.0)

47.3 (7.7)

47.4 (7.2)

47.6 (7.2)

0.98 0.91

NM

E, %

EI

11.5 (4.3)

12.0 (5.8)

12.4 (5.6)

12.3 (5.3)

0.40 0.56

Fibre g/1000kcal 7.4

(2.1) 7.4

(2.3) 7.3

(2.2) 7.2

(2.4) 0.58

0.86 S

odium m

g/1000kcal 1506

(326) 1503

(352) 1489

(320) 1514

(297) 0.85

0.74

Me

dia

n (IQ

R)

Alcohol %

TEI

2.6 (5.2)

3.4 (7.3)

4.1 (6.2)

3.1 (7.3)

0.0

31

f 0.07

SS

Bs g/1000kcal

24.4 (107.1)

31.4 (104.5)

46.1 (141.8)

44.3 (122.7)

0.0

13

g 0

.04

0c


(22.3) 19.2

(25.6) 19.2

(22.2) 18.7

(24.6) 0.26

0.48 Low


(88.4) 92.7

(93.2) 87.8

(85.2) 83.6

(86.0) 0.05

0.41 W


(26.0) 20.1

(29.4) 23.0

(28.7) 19.6

(27.9) 0.77

0.80 Total fruit g/1000kcal

76.7 (95.6)

77.7 (95.6)

76.8 (100.2)

64.3 (92.3)

0.41

0.73 V


(52.2) 78.1

(57.8) 80.2

(51.1) 78.4

(56.4) 0.79

0.90 Legum

e g/1000kcal 11.4

(13.4) 12.6

(15.2) 11.2

(14.2) 10.3

(14.1) 0.25


8.6 (18.9)

10.2 (22.4)

11.3 (19.5)

10.2 (19.5)

0.51 0.34


27.8 (28.3)

29.4 (24.3)

30.0 (24.1)

31.2 (29.3)

0.88 0.71

Processed m

eat g/1000kcal

13.9 (15.8)

13.7 (16.8)

13.7 (17.4)

15.3 (19.1)

0.82 0.76


18.0 (11.3)

19.3 (9.4)

19.4 (11.6)

20.9 (12.4)

<0

.00

01

d

0.0

01

d

!!!

"+%!

A5.8 continued (w

omen)

<3

5

hrs

/we

ek

3

5 - 4

0

hrs

/we

ek

4

1 –

48

h

rs/w

ee

k

!4

9

hrs

/we

ek

N

(%)


0.89

No salt added

71 (17.0)

107 (25.7)

143 (34.3)

96 (23.0)


31 (17.7)

42 (24.0)

54 (30.9)

48 (27.4)


13 (14.3)

24 (26.4)

31 (34.1)

23 (25.3)


11 (17.7)

20 (32.3)

20 (32.3)

11 (17.7)

Nutritional supplem

ent use~

Mineral (m

ulti or single) 11

(8.3) 25

(12.4) 25

(9.5) 17

(9.1) 0.59

Vitam


19 (14.4)

48 (23.8)

55 (21.0)

38 (23.7)

0.22

Om


(14.3) 18

(25.6) 20

(31.8) 16

(25.4) 0.89

Other / herbal

16 (12.1)

28 (13.9)

32 (32.3)

23 (23.2)

0.95

Abbreviations: TE


E: N


SP


SB

S: sugar sw

eetened beverages. ±Salt usage n = 745;

~Nutritional supplem

ent usage n = 782. Chi-squared test to com

pare differences across categorical variables, missing data not included in analyses. To com

pare means

values between groups one-w

ay AN

OV

A w


ean and standard deviation). If significance indicated (p< 0.05) Bonferroni post

hoc test was applied to identify the source of the difference. W


ere conducted for nonparametric data (values presented as m

edian and inter quartile range). If significance indicated (p <0.05) W


ere then conducted between each group to establish the source of the difference w

ith Bonferroni post hoc

test applied to correct for multiple com


2 General Linear M


ent (n = 864 due to missing region

data). a)

Negative linear trend

b) <35 hrs vs. 35-40

c) <35 hrs vs. 41-48

d) <35 hrs vs. !49 hrs and <35 vs. 41-48

e) 35-40 hrs vs. >49

f) A

ll except 41-48 vs. !49 hrs and <35 vs. 35-40 g)

<35 hrs vs. 41-48, and 41-48 vs. !49 hrs

!!!

"+"!

A5.9

Su

b c

oh

ort p

rofile

: partic

ipan

t ch

ara

cte

ristic

s a

cro

ss s

hift w

ork

cla

ssific

atio

n (s

hift w

ork

measu

rem

en

t based

on

po

lice ra

dio

calls

mad

e w

ithin

30 d

ay

perio

d p

rior to

date

of h

ealth

scre

en

.

Men

Wom

en

D

ay work

Shift no nights

Shift w

ith nights

p D

ay work

Shift no

nights S

hift with

nights p

N

214

323

1062

156

244

324

Ag

e, y

ea

rs (S

D)

45.4 (8.0)

40.9 (9.1)

38.8 (7.6)

<0

.00

01

42.1

(9.1) 35.7

(8.3) 34.7

(7.5) <

0.0

00

U

sual work hours

43.4 (7.1)

44.5 (7.2)

46.5 (7.2)

<0

.00

01

36.6

(8.3) 40.1

(9.4) 44.4

(9.4) <

0.0

00

M

arita

l sta

tus

(%)

0.0

07

0

.00

4

Cohabiting

19 (9)

46 (14.4)

203 (19.4)

28

(18.3) 67

(28.8) 103

(33.7)

Single

13 (6.1)

26 (8.2)

58 (5.5)

18

(11.8) 27

(11.6) 18

(5.9)

Married

163 (76.9)

223 (69.9)

708 (67.6)

75

(49) 85

(36.5) 117

(38.2)

Divorced/separated

17 (8)

24 (7.5)

78 (7.5)

32

(20.9) 54

(23.2) 68

(22.2)

W

hite

N (%

) 205

(96.2) 317

(98.1) 1026

(96.6) 0.33

152 (97.4)

242 (99.2)

317 (97.8)

0.35 E

du

ca

tion

(%)

0.0

01

0

.01

9

Left school before taking

6 (2.8)

16 (5.0)

34 (3.2)

2

(1.3) 4

(1.6) 8

(2.5)

GC

SE

or equivalent 59

(27.6) 104

(32.2) 312

(29.4)

46 (29.5)

7 (28.7)

73 (22.5)

V


(9.4) 21

(6.5) 82

(7.7)

7 (4.5)

25 (10.2)

25 (7.7)

A

levels / Highers or

68

(31.8) 117

(36.2) 386

(36.4)

60 (38.5)

83 (34)

107 (33.0)

B

achelor Degree or

40

(18.7) 59

(18.3) 212

(20.0)

28 (17.9)

54 (22.1)

97 (29.9)

P


(9.8) 6

(1.9) 35

(3.3)

13 (8.3)

8 (3.3)

14 (4.3)

In

co

me

(ho

us

eh

old

pe

r

<0

.00

01

0

.00

1

Less than £32,000 8

(3.7) 50

(15.5) 72

(6.8)

33 (21.2)

76 (31.1)

62 (19.1)

£32,000 - £47,999

21 (9.8)

35 (10.8)

134 (12.6)

18

(11.5) 31

(12.7) 30

(9.3)

£48,000 - £57,999 74

(34.6) 147

(45.5) 505

(47.6)

48 (30.8)

81 (33.2)

113 (34.9)

£58,000- £77,999

72 (33.6)

64 (19.8)

279 (26.3)

28

(17.9) 36

(14.7) 85

(26.2)

More than £ 78,000

39 (18.2)

27 (8.4)

71 (6.7)

29

(18.6) 20

(8.2) 34

(10.5)

!!!

"+&!

A5.9 continued

Men

Wom

en

D

ay work

Shift no nights

Shift w

ith nights

p D

ay work

Shift no

nights S

hift with

nights p

Em

plo

ym

en

t (forc

e)

<0

.00

01

<

0.0

00

1

England

112 (52.3)

250 (77.3)

692 (65.2)

79

(50.6) 213

(87.3) 207

(63.9)

Scotland

93 (43.4)

47 (14.5)

231 (21.7)

70

(44.9) 15

(6.1) 70

(21.6)

Wales

8 (3.7)

25 (8.2)

138 (13.1)

7

(4.5) 16

(6.1) 47

(14.5)

Ra

nk

(%)

<0

.00

01

<

0.0

00

1

Police staff

56 (27.6)

62 (22.1)

5 (0.5)

87

(57.2) 95

(44.6) 7

(2.6)

Police C

onstable/Sergeant

107 (52.7)

195 (69.4)

856 (93.2)

51

(33.5) 107

(50.2) 256

(93.8)

Inspector/Chief Inspector

38 (18.7)

22 (7.8)

54 (5.9)

7

(4.6) 2

(0.9) 9

(3.3)

Other

2 (1.0)

2 (0.7)

3 (0.3)

7

(4.6) 9

(4.2) 1

(0.4)

Wo

rk e

nv

iron

me

nt

<0

.00

01

<

0.0

00

1


101 (71.6)

61 (23.4)

94 (10.4)

65

(72.2) 26

(13.8) 26

(9.8)

Mainly m

obile duties 40

(28.4) 200

(76.6) 811

(89.6)

25 (27.8)

163 (86.2)

239 (90.2)

T

ota

l ho

urs

wo

rke

d p

er

<

0.0

00

1

<0

.00

01

P

art time

3 (1.4)

5 (1.6)

3 (0.3)

42

(26.9) 37

(15.2) 36

(11.1)

40 hours or less (including

93

(43.5) 118

(36.5) 223

(21.0)

75 (48.1)

105 (43)

74 (22.8)

41 – 48 hours

79 (36.9)

125 (38.7)

498 (46.9)

29

(18.6) 67

(27.5) 127

(39.2)

49 – 54 hours 22

(10.3) 43

(13.3) 166

(15.6)

7 (4.5)

19 (7.8)

33 (10.2)

55 hours or m

ore 17

(7.9) 32

(9.9) 172

(16.2)

3 (1.9)

16 (6.6)

54 (16.7)

Y

ea

rs in

po

lice

forc

e

<

0.0

00

1

6 years or less 15

(7.0) 85

(26.3) 238

(22.4) <

0.0

00

1

42 (26.9)

129 (52.9)

136 (42.0)

6 to 12 years

40 (18.7)

87 (26.9)

300 (28.2)

35

(22.4) 57

(23.4) 99

(30.6)

12 – 21 years 56

(26.2) 71

(22.0) 305

(28.7)

45 (28.8)

48 (19.7)

62 (19.1)

21 years or m

ore 103

(48.1) 80

(24.8) 219

(20.6)

34 (21.8)

10 (4.1)

27 (8.3)

!!!

"+'!

A5.9 continued

Men

Wom

en

D

ay work

Shift no nights

Shift w

ith nights

p D

ay work

Shift no

nights S

hift with

nights p

Jo

b S

train

<

0.0

00

1

0.0

01

Low

(high control, low

99

(46.3) 107

(33.1) 260

(24.5)

48 (30.8)

86 (35.2)

64 (19.8)

P


22 (10.3)

60 (18.6)

226 (21.3)

38

(24.4) 58

(23.8) 83

(25.6)

Active (high dem

and, high

68 (31.8)

83 (25.7)

241 (22.7)

35

(22.4) 41

(16.8) 63

(19.4)

High (high dem

and, low

25

(11.7) 73

(22.6) 335

(31.5)

35 (22.4)

59 (24.2)

114 (35.2)

S

lee

p

0.36

0.20

5 hours or less 8

(3.7) 15

(4.6) 59

(5.6)

9 (5.8)

15 (6.2)

25 (7.7)

6 hours

76 (25.5)

92 (28.5)

31 (29.4)

42

(26.9) 40

(16.4) 66

(20.4)

7 hours 96

(44.9) 137

(42.4) 462

(43.5)

58 (37.2)

103 (42.2)

121 (37.4)

8 hours

29 (13.5)

65 (20.1)

186 (17.5)

44

(28.2) 73

(29.9) 92

(28.4)

9 hour or more

5 (2.3)

14 (4.3)

42 (4.0)

3

(1.9) 13

(5.3) 20

(6.2)

Jo

b s

atis

fac

tion

0.09

0.0

17

V

ery dissatisfied 4

(1.9) 6

(1.9) 32

(3.0)

5 (3.2)

5 (2.0)

4 (1.2)

D

issatisfied 27

(12.6) 46

(14.2) 193

(18.2)

14 (9.0)

28 (11.5)

52 (16.0)

S

atisfied 132

(61.7) 213

(65.9) 644

(60.7)

96 (61.5)

158 (64.8)

220 (67.9)

V

ery satisfied 51

(23.8) 58

(18) 192

(18.1)

41 (26.3)

53 (21.7)

48 (14.8)

M

issing*

P

hy

sic

al a

ctiv

ity

0.23

0.71

Low

24 (11.2)

29 (9.0)

124 (11.7)

17

(10.9) 21

(8.6) 38

(11.7)

Moderate

108 (50.5)

149 (46.1)

461 (43.4)

77

(49.4) 122

(50.0) 148

(45.7)

High

102 (38.3)

145 (44.9)

477 (44.9)

62

(39.7) 101

(41.4) 138

(42.6)

Alc

oh

ol

0.70

0.67

Never drinker

2 (0.9)

7 (2.2)

22 (2.1)

8

(5.1) 7

(2.9) 17

(5.3)

Previous drinker

8 (3.7)

10 (3.1)

45 (4.2)

10

(6.4) 19

(7.8) 22

(6.8)

Current drinker

204 (95.3)

306 (94.7)

995 (93.7)

138

(88.5) 218

(89.3) 285

(88.0)

!!!

"+(!

A5.9 continued

Men

Wom

en

D

ay work

Shift no nights

Shift w

ith nights

p D

ay work

Shift no

nights S

hift with

nights p

S

mo

kin

g s

tatu

s (%

)

0.52

0.14 N

ever smoker

161 (75.6)

222 (69.2)

772 (72.8)

113

(72.4) 169

(69.8) 221

(68.4)

Former sm

oker 38

(17.8) 75

(23.4) 210

(19.8)

36 (23.1)

45 (18.6)

67 (20.7)

C

urrent smoker

14 (6.6)

24 (7.5)

78 (7.4)

7

(4.5) 28

(11.6) 35

(10.8)

Sittin

g (w

ee

kd

ay

s)

<0

.00

01

<

0.0

00

1

Low

50 (23.4)

120 (37.1)

398 (37.5)

41

(26.3) 128

(52.5) 141

(43.5)

Moderate

83 (38.8)

130 (40.2)

412 (38.8)

61

(39.1) 85

(34.8) 138

(42.6)

High

81 (37.8)

73 (22.6)

252 (23.7)

54

(34.6) 31

(12.7) 45

(13.9)

TV

vie

win

g

0.26

0.68

Low

51 (23.8)

77 (23.8)

291 (27.4)

50

(32) 93

(38.1) 118

(36.4)

Moderate

96 (44.9)

145 (44.9)

493 (46.4)

71

(45.5) 103

(42.2) 147

(45.4)

High

67 (31.3)

101 (31.3)

278 (26.2)

35

(22.4) 48

(19.7) 59

(18.2)

Se

lf-rep

ort d

iag

no

se

d

D

yslipidaemia or lipid

lowering m

edication 13

(6.1) 23

(7.1) 58

(5.5) 0.53

3 (1.9)

6 (2.5)

2 (0.6)

0.18

Diagnosed hypertension, or

hypotensive medication

14 (6.5)

16 (4.9)

49 (4.6)

0.49 9

(5.8) 7

(2.9) 7

(2.2) 0.10

Diagnosed diabetes, or

glucose controlling

2 (0.9)

0 (0)

4 (0.4)

0.22 1

(0.6) 0

(0.0) 0

(0.0) 0.16

Diagnosed other chronic

disease 13

(6.1) 10

(3.1) 31

(2.9) 0.06

17 (10.9)

10 (4.1)

11 (3.4)

0.0

02

Abbreviations: G

CS

E G



ETs m


Q guidelines. C

hi squared test to compare differences

across categorical variables, missing data w

as not included in the analyses

!!!

"+)!

A6.1

Airw

av

e H

ealth

Mo

nito

ring

Stu

dy

partic

ipan

t ch

ara

cte

ristic

s a

cro

ss fifth

s o

f DA

SH

sco

re

A6.1.1 C

omparison of participant characteristics and dietary intake across fifths of D

AS

H score: m

en (n = 3,332)*

least healthy Q

uintile of DA

SH

score most healthy

Q1

Q2

Q3

Q4

Q5

p N

(%)

590 (17.7)

650 (19.5)

771 (23.1)

671 (20.1)

650 (19.5)

D

AS

H S

co

re, m

ed

ian

(ran

ge

) 17

(9-19) 21

(20-22) 24

(23-25) 27

(26-28) 31

(29-39)

Ag

e, y

ea

rs (S

E)

39.2 (0.3)

41.1 (0.3)

42.4 (0.3)

44.0 (0.3)

45.0 (0.3)

<0

.00

01

N (%

)

Wh

ite

569 (96.4)

619 (95.4)

750 (97.4)

649 (96.7)

635 (97.7)

0.14 M

arita

l sta

tus

<

0.0

00

1

Cohabiting

111 (19.1)

101 (15.7)

94 (12.4)

80 (12.1)

78 (12.1)

D

ivorced/Separated

40 (6.9)

46 (7.2)

51 (6.7)

45 (6.8)

37 (5.7)

M

arried 367

(63.3) 445

(69.2) 569

(75.0) 501

(75.8) 493

(76.4)

Single

62 (10.7)

51 (7.9)

45 (5.9)

35 (5.3)

37 (5.7)

M

issing 10

(1.7) 7

(1.1) 12

(1.6) 10

(1.5) 10

(1.7)

Ed

uc

atio

n

0.0

27


SE

24

(4.1) 37

(5.7) 29

(3.8) 30

(4.5) 35

(5.4)

GC

SE

or equivalent 189

(32.0) 197

(30.3) 228

(29.6) 205

(30.6) 204

(31.4)


54 (9.2)

50 (7.7)

49 (6.4)

55 (8.6)

32 (4.9)

A


206 (34.9)

215 (33.1)

259 (33.6)

200 (29.8)

199 (30.6)

B


94 (15.9)

113 (17.4)

160 (20.8)

142 (21.2)

128 (19.7)

P


(3.9) 38

(5.8) 45

(5.8) 39

(5.8) 52

(8.0)

An

nu

al h

ou

se

ho

ld in

co

me

0

.00

5

Less than £32,000 60

(10.2) 74

(11.4) 62

(8.0) 56

(8.3) 46

(7.1)

£32,000 - £47,999 76

(12.9) 75

(11.5) 92

(11.9) 58

(8.6) 89

(13.7)

£48,000 - £57,999 283

(48.0) 287

(44.1) 334

(43.4) 295

(44.0) 271

(41.7)

£58,000 - £79,999 126

(21.4) 162

(24.9) 197

(25.6) 187

(27.9) 177

(27.2)

More than £78,000

45 (7.6)

52 (8.0)

85 (11.0)

75 (11.2)

67 (10.3)

!!!

"+*!

A6.1.1 continued (m

en) least healthy Q

uintile of DA

SH

score most healthy

Q1

Q2

Q3

Q4

Q5

P

We

ek

ly w

ork

ing

ho

urs

0

.00

4

Part tim

e (<35 hrs) 7

(1.2) 6

(0.9) 11

(1.4) 15

(2.2) 13

(2.0)

Standard 35 – 40 hrs

166 (28.1)

205 (31.5)

277 (35.9)

245 (36.5)

252 (38.8)

>40 <49 hrs

238 (40.3)

277 (42.6)

274 (35.5)

240 (35.8)

234 (36.0)

>49 <55 hrs

92 (15.6)

78 (12.0)

99 (12.8)

91 (13.6)

81 (12.5)

55 hrs or m

ore 87

(14.7) 84

(12.9) 110

(14.3) 80

(11.9) 70

(10.8)

Tim

e in

cu

rren

t job

role

0.22

2 years or less 226

(38.3) 209

(32.2) 282

(36.6) 249

(37.1) 220

(33.9)

3 to 5 years 185

(31.4) 206

(31.7) 218

(28.3) 195

(29.1) 196

(30.2)

6 years or more

179 (30.3)

235 (36.2)

271 (35.1)

227 (33.8)

234 (36.0)

S

hift w

ork

las

t 30

da

ys

0.16

Day

36 (11.2)

38 (11.8)

38 (10.9)

40 (14.1)

49 (18.1)

S

hift no night 59

(18.3) 62

(19.3) 75

(21.6) 63

(22.3) 54

(19.9)

Shift w

ith night work

227 (70.5)

221 (68.8)

235 (67.5)

180 (63.6)

168 (62.0)

M

issing 268

(45.4) 329

(50.6) 423

(54.9) 388

(57.8) 379

(58.3)

Em

plo

ym

en

t (forc

e) c

ou

ntry

<

0.0

00

1

England

359 (60.8)

446 (69.0)

544 (70.8)

462 (69.0)

488 (75.2)

S

cotland 167

(28.3) 141

(21.8) 145

(18.9) 114

(17.0) 80

(12.3)

Wales

64 (10.8)

59 (9.1)

79 (10.3)

94 (14.0)

81 (12.5)

M

issing 0

(0.0) 4

(0.6) 3

(0.4) 1

(0.1) 1

(0.2)

Ra

nk

P

olice staff 95

(17.9) 94

(16.1) 140

(20.8) 105

(18.5) 134

(24.1) <

0.0

00

1

Police C

onstable/Sergeant

390 (73.6)

428 (73.4)

439 (65.3)

375 (65.9)

351 (63.2)

Inspector/C

hief Inspector 35

(6.6) 41

(7.0) 65

(9.7) 68

(11.9) 46

(8.3)

Super Intendant or above

7 (1.3)

5 (0.9)

16 (2.4)

14 (2.5)

17 (3.1)

O

ther 3

(0.6) 15

(2.6) 12

(1.8) 7

(1.2) 7

(1.3)

Missing

60 (10.2)

67 (10.3)

99 (12.8)

102 (15.2)

95 (14.6)

!!!

"+#!

A6.1.1 continued (m

en) least healthy Q

uintile of DA

SH

score most healthy

Q

1

Q2

Q3

Q4

Q5

p J

ob

stra

in

0.0

18

Low

strain (high control, low dem

and) 155

(26.3) 202

(31.1) 264

(34.2) 232

(34.6) 226

(34.8)

Passive (low

control, low dem

and) 112

(19.0) 112

(17.2) 136

(17.6) 107

(16.0) 118

(18.2)

Active (high dem


(28.3) 185

(28.5) 220

(28.5) 192

(28.6) 188

(28.9)


and, low control)

156 (26.4)

151 (23.2)

151 (19.6)

140 (20.9)

118 (18.2)

W

ork

en

viro

nm

en

t

<

0.0

00

1


145 (27.4)

185 (31.7)

231 (34.4)

234 (41.1)

220 (39.6)

M

ainly mobile duties

315 (59.4)

305 (52.3)

322 (47.9)

247 (43.4)

228 (41.1)

U

nclassified 70

(13.2) 93

(15.9) 119

(17.7) 88

(15.5 107

(19.3)

Missing

60 (10.2)

67 (10.3)

99 (12.8)

102 (15.2)

95 (14.6)

P

hy

sic

al a

ctiv

ity

<0

.00

01

Low

73

(12.4) 90

(13.8) 68

(8.8) 55

(8.2) 60

(9.2)

Moderate

311 (52.7)

314 (48.3)

317 (41.1)

282 (42.0)

231 (35.5)

H

igh 206

(34.9) 246

(37.8) 386

(50.1) 334

(49.8) 359

(55.2)

Sm

ok

ing

sta

tus

<

0.0

00

1

Never sm

oker 406

(69.2) 443

(68.3) 537

(70.0) 474

(71.1) 473

(72.8)

Former sm

oker 114

(19.4) 140

(21.6) 189

(24.6) 157

(23.5) 153

(23.5)

Current sm

oker 67

(11.4) 66

(10.2) 41

(5.4) 36

(5.4) 24

(3.7)

Sle

ep

0.08

5 hours or less 36

(6.1) 35

(5.4) 32

(4.2) 31

(4.6) 31

(4.8)

6 hours 184

(31.2) 217

(33.4) 222

(28.8) 219

(32.6) 163

(25.1)

7 hours 247

(41.9) 271

(41.7) 342

(44.4) 282

(42.0) 306

(47.1)

8 hours 106

(18.0) 106

(16.3) 154

(20.0) 129

(19.2) 129

(19.8)

9 hour or more

17 (2.9)

21 (3.2)

20 (2.6)

10 (1.5)

21 (3.2)

W

ee

kly

ho

urs

sittin

g (w

ee

kd

ay

s)

0.62 Low

(<20 hours) 180

(30.5) 196

(30.2) 238

(30.9) 212

(31.6) 197

(30.3)


242 (41.0)

252 (38.8)

286 (37.1)

235 (35.0)

252 (38.8)

H

igh (>40 hours) 168

(28.5) 202

(31.1) 247

(32.0) 224

(33.4) 201

(30.9)

!!!

"++!

A6.1.1 continued (m

en) least healthy Q

uintile of DA

SH

score most healthy

Q

1

Q2

Q3

Q4

Q5

p T

V v

iew

ing

, ho

ur p

er w

ee

k

0.0

03

Low

(<6 hours) 115

(19.5) 175

(26.9) 195

(25.3) 197

(29.4) 176

(27.1)


286 (48.5)

284 (43.7)

351 (45.5)

276 (41.1)

309 (47.5)

H

igh (>15 hours) 189

(32.0) 191

(29.4) 225

(29.2) 198

(29.5) 165

(25.4)

Se

lf-rep

ort d

iag

no

se

d d

ise

as

e

D

yslipidaemia or on m

edication 38

(6.4) 34

(5.2) 59

(7.7) 57

(8.5) 64

(9.8) 0

.01

9

Diagnosed hypertension, or on m

edication 37

(6.3) 39

(6.0) 46

(6.0) 53

(7.9) 53

(8.1) 0.30

Diagnosed diabetes, or on m

edication 2

(0.3) 5

(0.8) 10

(1.3) 4

(0.6) 9

(1.4) 0.20

Bo

dy

Ma

ss

Ind

ex

0.17

<25 kg/m2

122 (20.7)

128 (19.7)

161 (20.9)

146 (21.8)

164 (25.2)

25-30 kg/m

2 320

(54.2) 363

(55.9) 424

(55.0) 380

(56.6) 358

(55.1)

>30 kg/m2

148 (25.1)

159 (24.5)

186 (24.3)

145 (21.6)

128 (19.7)

W

ais

t circ

um

fere

nc

e

<0

.00

01

Low

risk 282

(47.8) 309

(47.5) 386

(50.1) 370

(55.1) 375

(57.7)

Moderate risk

168 (28.5)

218 (33.5)

247 (32.0)

189 (28.2)

180 (27.7)

H

igh risk 140

(23.7) 123

(18.9) 138

(17.9) 112

(16.7) 95

(14.6)

Hb

A1

c

0.83 N

ormal <5.7 %

367

(62.2) 400

(61.5) 468

(60.7) 408

(60.8) 410

(63.1)

Pre-diabetes >5.7 %

<6.5%

206 (34.9)

224 (34.5)

273 (35.4)

245 (36.5)

221 (34.0)

D

iabetes !6.5 %

17 (2.9)

26 (4.0)

30 (3.9)

18 (2.7)

19 (2.9)

B

loo

d p

res

su

re b

y ris

k c

ate

go

ry

0.42 N

ormal S

BP

<120 + DB

P<80m

mH

g 61

(10.3) 66

(10.2) 64

(8.3) 63

(9.4) 69

(10.6)

Pre-hypertensive S

BP

120-139 ± DB

P 80-89

283 (48.0)

350 (53.9)

407 (52.8)

343 (51.1)

322 (49.5)

H

ypertensive SB

P!140± D

BP!90m

mH

g 246

(41.7) 234

(36.0) 300

(38.9) 265

(39.5) 259

(39.8)

CR

P b

y ris

k c

ate

go

ry

<0

.00

01

Low

risk <1.0mg/L

298 (16.2)

328 (17.8)

407 (22.1)

400 (21.8)

405 (22.0)

M

oderate risk >1.0 <3.0mgl/L

211 (18.5)

247 (21.7)

271 (23.8)

212 (18.6)

198 (5.9)

H

igh risk !3.0mgl/L <10

81 (22.8)

75 (21.1)

93 (26.2)

59 (16.6)

47 (13.2)

A6.1.1 continued (m

en) least healthy Q

uintile of DA

SH

score most healthy

!!!

&$$!

Q

1

Q2

Q3

Q4

Q5

p

H

DL

0

.01

5

Low risk

523 (88.6)

595 (91.5)

724 (93.9)

616 (91.8)

598 (92.0)

E

levated risk 67

(11.4) 55

(8.5) 47

(6.1) 55

(8.2) 52

(8.0)

No

n-H

DL

0

.04

0

Low risk

299 (50.7)

320 (49.2)

417 (54.1)

352 (52.5)

372 (57.2)

E

levated risk 291

(49.3) 330

(50.8) 354

(45.9) 319

(47.5) 278

(42.8)

Nu

mb

er o

f Ca

rdio

me

tab

olic

risk

fac

tors

Classification by num

ber of factors

0.09

None

51 (8.6)

44 (6.8)

69 (9.0)

68 (10.1)

70 (10.8)

1 or 2

279 (47.3)

331 (50.9)

368 (47.7)

334 (49.8)

337 (51.9)

3 or m

ore (high risk) 260

(44.1) 275

(42.3) 334

(43.3) 269

(40.1) 243

(37.4)


ith self-reported chronic disease diagnosis: cancer, diseases of thyroid, chronic liver disease, angina, other heart, stroke and CO

PD

A

bbreviations: SE

standard error, GC

SE

General C

ertificate of Secondary E

ducation. CR

P high sensitivity c-reactive protein, S

BP

systolic blood pressure, D

BP

diastolic blood pressure, HD

L high density lipoprotein. Chi-squared test to com

pare differences across categorical variables, missing data not included in

analyses

!!!

&$%!

A6.1.2 C

omparison of participant characteristics and dietary intake across fifths of D

AS

H score: w

omen (n =2,195)*

least healthy Q

uintile of DA

SH

score most healthy

Q1

Q2

Q3

Q4

Q5

p N

(%)

397 (18.1)

448 (20.4)

480 (21.9)

457 (20.8)

413 (18.8)

D

AS

H S

co

re, m

ed

ian

(ran

ge

) 18

(9-19) 21

(20-22) 24

(23-25) 27

(26-28) 31

(29-38)

Ag

e, y

ea

rs (S

E)

36.1 (0.5)

38.4 (0.4)

39.5 (0.4)

40.9 (0.4)

42.4 (0.4)

N (%

)

W

hite

387

(97.5) 443

(98.9) 467

(97.3) 449

(98.3) 407

(98.5) 0.35

Ma

rital s

tatu

s

<0

.00

01

Cohabiting

98 (26.2)

98 (22.7)

112 (24.0)

83 (18.8)

60 (15.0)

D

ivorced/Separated

29 (7.8)

45 (10.4)

39 (8.4)

58 (13.2)

45 (11.2)

M

arried 154

(41.2) 204

(47.2) 235

(50.4) 235

(53.3) 218

(54.4)

Single

93 (24.9)

85 (19.7)

80 (17.2)

65 (14.7)

78 (19.4)

M

issing 23

(5.8) 16

(3.6) 14

(2.9) 16

(3.5) 12

(2.9)

Ed

uc

atio

n

0.29 Left school before taking G

CS

E

11 (2.8)

15 (3.4)

17 (3.5)

15 (3.3)

16 (3.9)

G

CS

E or equivalent

127 (32.0)

121 (27.0)

126 (6.3)

115 (25.2)

122 (29.5)

V


(7.8) 32

(7.1) 30

(6.2) 38

(8.3) 29

(7.0)

A levels / H


(34.0) 151

(33.7) 157

(32.7) 158

(34.6) 117

(28.3)

Bachelor D


(19.4) 104

(23.2) 109

(22.7) 92

(20.1) 95

(23.0)


16 (4.0)

25 (5.6)

41 (8.5)

39 (8.5)

34 (8.2)

A

nn

ua

l ho

us

eh

old

inc

om

e

0.38 Less than £32,000

115 (29.0)

114 (25.5)

119 (24.8)

110 (24.1)

122 (29.5)

£32,000 - £47,999

44 (11.1)

52 (11.6)

47 (9.8)

54 (11.8)

49 (11.9)

£48,000 - £57,999

128 (32.2)

130 (29.0)

175 (36.5)

144 (31.5)

121 (29.3)

£58,000 - £79,999

77 (19.4)

93 (20.8)

90 (18.8)

94 (20.6)

74 (17.9)

M

ore than £78,000 33

(8.3) 59

(13.2) 49

(10.2) 55

(12.0) 47

(11.4)

!!!

&$"!

A6.1.2 continued (w

omen)


AS

H score m

ost healthy

Q

1

Q2

Q3

Q4

Q5

p W

ee

kly

wo

rkin

g h

ou

rs

0.0

09

P

art time (<35)

54 (13.6)

88 (19.6)

102 (21.2)

99 (21.7)

96 (23.2)

S

tandard 35 – 40 183

(46.1) 181

(40.4) 209

(43.5) 195

(42.7) 180

(43.6)

>40 <49 94

(23.7) 120

(26.8) 112

(23.3) 101

(22.1) 106

(25.7)

>49 <55 32

(8.1) 28

(6.3) 29

(6.0) 37

(8.1) 16

(3.9)

55 or more

34 (8.6)

31 (6.9)

28 (5.8)

25 (5.5)

15 (3.6)

T

ime

in c

urre

nt jo

b ro

le

0.0

05

2 years or less 177

(44.6) 189

(42.2) 204

(42.5) 155

(33.9) 165

(39.9)

3 to 5 years 116

(29.2) 132

(29.5) 124

(25.8) 153

(33.5) 102

(24.7)

6 years or more

104 (26.2)

127 (28.3)

152 (31.7)

149 (32.6)

146 (35.5)

S

hift w

ork

las

t 30

da

ys

0.10

Day

21 (14.2)

31 (21.7)

34 (21.8)

31 (24.8)

22 (19.3)

S

hift no night 55

(37.2) 40

(28.0) 57

(36.5) 34

(27.2) 48

(42.4)

Shift w

ith night work

72 (48.7)

72 (50.3)

65 (41.7)

60 (48.0)

44 (38.6)

M

issing 249

(62.7) 305

(68.1) 324

(67.5) 332

(72.6) 299

(72.4)

Em

plo

ym

en

t (forc

e) c

ou

ntry

0

.00

2

England

279 (70.5)

333 (74.5)

355 (74.1)

339 (74.3)

323 (79.0)

S

cotland 73

(18.4) 66

(14.8) 76

(15.9) 65

(14.2) 30

(7.3)

Wales

44 (11.1)

48 (10.7)

48 (10.0)

52 (11.4)

56 (13.7)

M

issing 1

(0.3) 1

(0.2) 1

(0.2) 1

(0.2) 4

(1.0)

Ra

nk

0.38

Police staff

185 (52.9)

195 (50.8)

235 (56.0)

214 (52.3)

208 (57.8)

P

olice Constable/S

ergeant 149

(42.6) 170

(44.3) 160

(38.1) 163

(39.8) 129

(35.8)


8 (2.3)

7 (1.8)

12 (2.9)

17 (4.2)

9 (2.5)

O

ther 8

(2.3) 12

(3.1) 13

(3.1) 15

(3.7) 14

(3.9)

Missing

47 (11.8)

64 (14.3)

60 (12.5)

48 (10.5)

53 (12.8)

!!!

&$&!

A6.1.2 continued (w

omen)


AS

H score m

ost healthy

Q

1

Q2

Q3

Q4

Q5

p J

ob

stra

in

0.07 Low


and) 98

(24.7) 109

(24.3) 117

(24.4) 123

(26.9) 123

(29.8)

Passive (low

control, low dem

and) 103

(25.9) 111

(24.8) 116

(24.2) 117

(25.6) 108

(26.2)

Active (high dem


(20.2) 115

(25.7) 136

(28.3) 104

(22.8) 103

(24.9)


and, low control)

116 (29.2)

113 (25.2)

111 (23.1)

113 (24.7)

79 (19.1)

J

ob

stra

in

0.07 Low


and) 98

(24.7) 109

(24.3) 117

(24.4) 123

(26.9) 123

(29.8)

Passive (low

control, low dem

and) 103

(25.9) 111

(24.8) 116

(24.2) 117

(25.6) 108

(26.2)

Active (high dem


(20.2) 115

(25.7) 136

(28.3) 104

(22.8) 103

(24.9)


and, low control)

116 (29.2)

113 (25.2)

111 (23.1)

113 (24.7)

79 (19.1)

W

ork

en

viro

nm

en

t

0

.04

2


127 (36.3)

139 (36.2)

154 (36.7)

161 (39.4)

132 (36.7)

M

ainly mobile duties

123 (35.1)

124 (32.3)

122 (29.1)

109 (26.7)

88 (24.4)

U

nclassified 100

(28.6) 121

(31.5) 144

(34.3) 139

(34.0) 140

(38.9)

Missing

47 (11.8)

64 (14.3)

60 (12.5)

48 (10.5)

53 (12.8)

P

hy

sic

al a

ctiv

ity

<0

.00

01

Low

72 (18.1)

69 (15.4)

76 (15.8)

50 (10.9)

41 (11.4)

M

oderate 216

(54.4) 221

(54.4) 229

(47.7) 218

(47.7) 177

(42.9)

High

109 (27.5)

158 (27.5)

175 (36.5)

189 (41.4)

189 (45.8)

S

mo

kin

g s

tatu

s

<0

.00

01

Never sm

oker 222

(56.5) 291

(65.0) 338

(70.7) 333

(73.0) 305

(73.9)

Former sm

oker 87

(22.1) 101

(22.5) 103

(21.6) 92

(20.2) 89

(21.6)

Current sm

oker 84

(21.4) 56

(12.5) 37

(7.7) 31

(6.8) 19

(4.6)

!!!

&$'!

A6.1.2 continued (w

omen)


AS

H score m

ost healthy

Q

1

Q2

Q3

Q4

Q5

p S

lee

p

0.66 5 hours or less

25 (6.3)

31 (6.9)

30 (6.3)

30 (6.6)

24 (5.8)

6 hours

83 (20.9)

90 (20.1)

103 (21.5)

107 (23.4)

88 (21.3)

7 hours

158 (39.8)

187 (41.7)

199 (41.5)

184 (40.3)

175 (42.4)

8 hours

103 (25.9)

125 (27.9)

134 (27.9)

117 (25.6)

107 (25.9)

9 hour or m

ore 28

(7.0) 15

(3.3) 14

(2.9) 19

(4.2) 19

(4.6)

We

ek

ly h

ou

rs s

itting

(we

ek

da

ys

)

0.94

Low (<20 hours)

129 (32.5)

145 (32.3)

161 (33.5)

147 (32.2)

125 (30.3)

M


(38.0) 179

(40.0) 193

(40.2) 177

(38.7) 162

(39.2)

High (>40 hours)

117 (29.5)

124 (27.7)

126 (26.3)

133 (29.1)

126 (30.5)

W

ee

kly

ho

urs

sittin

g (w

ee

kd

ay

s)

0.94 Low

(<20 hours) 129

(32.5) 145

(32.3) 161

(33.5) 147

(32.2) 125

(30.3)


151 (38.0)

179 (40.0)

193 (40.2)

177 (38.7)

162 (39.2)

H

igh (>40 hours) 117

(29.5) 124

(27.7) 126

(26.3) 133

(29.1) 126

(30.5)

TV

vie

win

g, h

ou

r pe

r we

ek

0.42

Low (<6 hours)

135 (34.0)

161 (35.9)

182 (37.9)

144 (31.5)

145 (35.1)

M


(42.6) 186

(41.5) 206

(42.9) 220

(48.1) 182

(44.1)

High (>15 hours)

93 (23.4)

101 (22.5)

92 (19.2)

93 (20.4)

86 (20.8)

M

en

op

au

se

<

0.0

00

1

No

350 (88.4)

382 (85.3)

410 (85.6)

363 (79.4)

309 (75.0)

Y

es 23

(5.8) 38

(8.5) 42

(8.8) 62

(13.6) 68

(16.5)

Don't know

23

(5.8) 28

(6.2) 27

(5.6) 32

(7.0) 35

(8.5)

Se

lf-rep

ort d

iag

no

se

d d

ise

as

e

D

yslipidaemia or on m

edication 9

(2.3) 11

(2.5) 13

(2.7) 10

(2.2) 21

(5.1) 0.07

Diagnosed hypertension, or on m

edication 19

(4.8) 27

(6.0) 30

(6.3) 32

(7.0) 28

(6.8) 0.71

Diagnosed diabetes, or on m

edication 2

(0.5) 2

(0.4) 3

(0.6) 2

(0.4) 6

(1.4) 0.33

!!!

&$(!

A6.1.2 continued (w

omen)


AS

H score m

ost healthy

Q

1

Q2

Q3

Q4

Q5

p B

od

y M

as

s In

de

x

0.40 <25 kg/m

2 205

(51.6) 208

(46.4) 235

(49.0) 243

(53.2) 219

(53.0)

25-30 kg/m2

132 (33.3)

162 (36.2)

177 (36.9)

143 (31.3)

138 (33.4)

>30 kg/m

2 60

(15.1) 78

(17.4) 68

(14.2) 71

(15.5) 56

(13.6)

Wa

ist c

ircu

mfe

ren

ce

0.25

Low risk

196 (49.4)

204 (45.5)

220 (45.8)

225 (49.2)

220 (53.3)

M

oderate risk 98

(24.7) 112

(25.0) 134

(27.9) 124

(27.1) 99

(24.0)

High risk

103 (25.9)

132 (29.5)

126 (26.3)

108 (23.6)

94 (22.8)

H

bA

1c

0.41

Norm

al <5.7 %

232 (58.4)

254 (56.7)

282 (58.7)

248 (54.3)

218 (52.8)

P

re-diabetes >5.7 % <6.5%

155

(39.0) 178

(39.7) 182

(37.9) 195

(42.7) 174

(42.1)

Diabetes !6.5 %

10

(2.5) 16

(3.6) 16

(3.3) 14

(3.1) 21

(5.1)

Blo

od

pre

ss

ure

by

risk

ca

teg

ory

0.14

Norm

al SB

P <120 + D

BP

<80mm

Hg

188 (47.4)

199 (44.4)

206 (42.9)

186 (40.7)

180 (43.6)

P

re-hypertensive SB

P120-139 ± D

BP

80-89 157

(39.6) 195

(43.5) 198

(41.3) 186

(40.7) 179

(43.3)

Hypertensive S

BP!140± D

BP!90m

mH

g 52

(13.1) 54

(12.0) 76

(15.8) 85

(18.6) 54

(13.1)

CR

P b

y ris

k c

ate

go

ry

0.0

03

Low

risk <1.0mg/L

192 (48.4)

224 (50.0)

250 (52.1)

249 (54.5)

241 (58.4)

M

oderate risk >1.0 <3.0mgl/L

125 (31.5)

158 (35.3)

134 (27.9)

142 (31.1)

123 (29.8)

H

igh risk !3.0mgl/L <10

80 (20.2)

66 (14.7)

96 (20.0)

66 (14.4)

49 (11.9)

H

DL

0

.03

0

Low risk

392 (98.7)

435 (97.1)

477 (99.4)

453 (99.1)

408 (98.8)

E

levated risk 5

(1.3) 13

(2.9) 3

(0.6) 4

(0.9) 5

(1.2)

!!!

&$)!

A6.1.2 continued (w

omen)


AS

H score m

ost healthy

Q

1

Q2

Q3

Q4

Q5

p N

on

-HD

L

0.87 Low

risk 318

(80.1) 351

(78.3) 377

(78.5) 354

(77.5) 330

(79.9)

Elevated risk

80 (19.9)

97 (21.7)

103 (21.5)

103 (22.5)

83 (20.1)

N

um

be

r of C

ard

iom

eta

bo

lic ris

k fa

cto

rs

C

lassification by number of factors

0.63 N

one 80

(20.2) 78

(17.4) 87

(18.3) 74

(16.2) 76

(18.4)

1 or 2 214

(53.9) 256

(57.1) 266

(55.4) 271

(59.3) 247

(59.8)

3 or more (high risk - m

etabolic syndrome)

103 (25.9)

114 (25.4)

127 (26.5)

112 (24.5)

90 (21.8)

*E

xcluding participants with self-reported chronic disease diagnosis: cancer, diseases of thyroid, chronic liver disease, angina, other heart, stroke and C

OP

D

Abbreviations: S

E standard error, G

CS

E G


econdary Education. C

RP

high sensitivity c-reactive protein, SB

P systolic blood pressure,

DB

P diastolic blood pressure, H

DL high density lipoprotein. C

hi-squared test to compare differences across categorical variables, m

issing data not included in analyses

!!!

&$*!

A6.2

Partia

l co

rrela

tion

co

effic

ien

ts fo

r die

tary

varia

ble

s, m

ark

ers

of c

ard

iom

eta

bo

lic ris

k, w

ork

ing

ho

urs

an

d a

ctiv

ity*

!

,-./!01234!

533467893:;<!01234!

=>436<!:>;9?4!

=>436<!@4>0:;<!!

A=5!B9;!

A=5!09;739;4@!B9;!

A=5!C32;4:>!!

A=5!193D2E<@39;4!

A=5!FG=!

A=5!-812E28!

H:D34!6I%$$$?198!

F9!J6I%$$$?198!

KG5!

L9:0;!1:31M!

K2@<!B9;N!A!

F2M!J4;9D28:1!B91;230!

,:90;28:1!KO!

.<0;28:1!KO!

PQO!

/D-%1!

/,R!

>2>!/,R!

SPT!/,R!

L23?:>6!E2730!

OE<0:198!91;:U:;<V!

L44?@9<!0:;;:>6!

=>436<!:>;9?4!W$M$'!

!!

!!

!!

!!

!!

!!

!!

!!

!!

!!

!!

!!

!533467893:;<!01234!

W$M%(!!

!!

!!

!!

!!

!!

!!

!!

!!

!!

!!

!!

!!

=>436<!@4>0:;<!!!"#$%&

$M$(!$M"&!

!!

!!

!!

!!

!!

!!

!!

!!

!!

!!

!!

!A=5!B9;!

!"#$'&$M$"!

$M%#!$M&*!

!!

!!

!!

!!

!!

!!

!!

!!

!!

!!

!!

A=5!09;739;4@!B9;!

!"#$(&$M$$!

$M%#!$M&(!

"#)(&!

!!

!!

!!

!!

!!

!!

!!

!!

!!

!!

A=5!C32;4:>!!

$M%&!W$M$%!

W$M")!W$M"'!

W$M%#!W$M"$!

!!

!!

!!

!!

!!

!!

!!

!!

!!

!!

A=5!193D2E<@39;4!

"#$(&W$M%#!

W$M$(!W$M%&!

!"#*'&!"#$)&

W$M"*!!

!!

!!

!!

!!

!!

!!

!!

!!

!!

A=5!F

G=!

W$M%"!W$M$)!

$M%#!$M$+!

W$M%(!W$M$"!

!"#+(&"#*'&

!!

!!

!!

!!

!!

!!

!!

!!

!!

A=5!-812E28!

W$M%"!$M%#!

$M$'!W$M$#!

W$M"#!W$M"(!

W$M%"!!"#*(&

W$M"&!!

!!

!!

!!

!!

!!

!!

!!

!!

H:D34!6I%$$$?198!"#,-&

W$M%(!W$M%#!

W"#$)!!"#$-&

!"#$%&$M%(!

"#+(&W$M%+!

W$M"%!!

!!

!!

!!

!!

!!

!!

!!

!F9!J

6I%$$$?198!W$M%#!

$M$$!W$M""!

W$M%$!$M$)!

$M$%!$M%#!

W$M$)!W$M"(!

W$M%$!$M%"!

!!

!!

!!

!!

!!

!!

!!

!KG

5!W$M%$!

$M$#!W$M$'!

$M$$!$M$%!

W$M$%!$M$#!

W$M$#!W$M%$!

$M$&!W$M$(!

$M$+!!

!!

!!

!!

!!

!!

!!

!L9:0;!1:31M!

W$M%'!$M$+!

W$M$&!$M$&!

$M$'!$M$"!

$M$'!W$M%$!

W$M$+!$M$(!

W$M$*!$M$#!

"#)*&!

!!

!!

!!

!!

!!

!!

K2@<!B9;N!A!

W$M%'!$M$)!

W$M%$!$M$"!

$M$&!$M$"!

$M$&!W$M$+!

W$M$#!$M$(!

W$M$*!$M%%!

"#)"&"#-%&

!!

!!

!!

!!

!!

!!

F2M!J

4;9D28:1!B91;230!!W$M%'!

$M$&!W$M$(!

$M$"!$M$(!

$M$'!$M$%!

W$M$*!W$M$'!

$M$&!W$M$+!

$M$)!"#*$&

"#*)&"#*(&

!!

!!

!!

!!

!!

!,:90;28:1!KO!

W$M$+!$M$(!

W$M$)!$M$$!

W$M$%!W$M$"!

W$M$"!W$M$#!

W$M$(!$M%"!

W$M$(!$M$'!

"#$'&"#$"&

"#$$&"#+$&

&!

!!

!!

!!

!!

.<0;28:1!KO!W$M$&!

$M$%!W$M$"!

W$M$&!W$M$'!

W$M$&!$M$%!

W$M$(!W$M$'!

$M%$!W$M$"!

$M$"!$M"*!

$M"&!$M""!

"#+(&"#-'&

!!

!!

!!

!!

!PQO!

W$M%%!$M$(!

W$M$'!$M$(!

$M$'!$M$'!

$M$%!W$M$'!

W$M$"!$M$%!

W$M$)!$M$"!

"#$$&"#$(&

"#$'&"#++&

$M%)!$M%&!

!!

!!

!!

!!

/D-%1!W$M$'!

W$M$&!$M$$!

$M$&!$M$*!

$M$)!W$M$"!

$M$%!W$M$%!

W$M$#!W$M$"!

$M$&!$M%(!

$M%)!$M%%!

"#$%&$M$&!

$M$&!$M%%!

!!

!!

!!

!/,R!

$M$&!$M$$!

$M$*!W$M$%!

W$M$"!W$M$%!

W$M$)!W$M%&!

W$M$)!$M""!

W$M$"!W$M$'!

!"#$"&!"#$$&

W$M"*!W$M")!

W$M$(!$M$%!

W$M%"!W$M%$!

!!

!!

!!

F2>!/,R!

W$M%%!$M$%!

W$M$'!$M$"!

$M$*!$M$#!

W$M$%!W$M$*!

$M$"!$M$&!

W$M%%!$M$$!

$M%#!$M"$!

$M"(!"#+'&

$M"$!$M%"!

$M$+!$M$(!

W$M%)!!

!!

!!

SPT!/,R!W$M%%!

$M$%!W$M$*!

$M$"!$M$(!

$M$(!$M$"!

$M$&!$M$(!

W$M%$!W$M$)!

$M$&!$M"#!

"#$(&"#$"&

"#+$&$M%)!

$M$#!$M%'!

$M%$!!"#,+&

"#-'&!

!!

!L23?:>6!E2730!

W$M$&!$M$#!

W$M$'!W$M$"!

W$M$'!W$M$'!

$M$"!W$M$'!

$M$$!$M$*!

W$M$&!$M$$!

$M$+!$M$)!

$M$)!$M$&!

$M$(!$M$$!

$M$&!W$M$%!

$M$"!$M$'!

$M$%!!

!!

OE<0:198!91;:U:;<V!$M$"!

W$M$%!W$M$%!

W$M$&!W$M$'!

W$M$&!$M$&!

$M$&!$M$'!

W$M$"!$M$"!

W$M$%!W$M$)!

W$M%%!W$M%)!

W$M$'!W$M$'!

$M$&!W$M$"!

$M$%!$M$#!

W$M$(!W$M$#!

$M$'!!

!L44?@9<!0:;;:>6!!

W$M$%!W$M$"!

$M$)!$M$$!

$M$"!$M$"!

$M$$!W$M$"!

$M$$!W$M$%!

$M$$!$M$"!

$M$+!$M%$!

$M%%!$M$)!

$M$'!$M$%!

$M$(!W$M$%!

W$M$(!$M$&!

$M$)!$M$&!

W$M""!!

L44?8<!SX!!

W$M$)!W$M$&!

$M$%!$M$&!

$M$&!$M$'!

$M$%!W$M$"!

$M$%!W$M$%!

W$M$(!$M$&!

$M$*!$M$#!

$M%"!$M$+!

$M$(!$M$&!

$M$)!W$M$%!

W$M$"!$M$'!

$M$'!W$M$'!

W$M$#!$M"*!

n= 5,281, missing data = 246 participants (C

RP

and/or body fat). Spearm

an rank correlation coefficients, adjusted for age and sex. B

old shaded values signify medium

to large effect, r ! ±0.30. All

observations >/< ±0.03 are significant (p <0.05). Abbreviations: E

I energy intake, Na S

odium, N

ME

: Non-m

ilk extrinsic sugars. NS

P: N

on-starch polysaccharides, SS

Bs sugar sw

eetened beverages. C

RP

C-reactive protein, B

P blood pressure, H

DL high-density lipoprotein, TC

total cholesterol. ~Physical activity based on m

ean metabolic equivalents

!

! ! "#$!

A6.3 Sensitivity analyses Chapter 6: diet and cardiometabolic risk (energy intake under-reporters excluded)

A6.3.1 Association of DASH score with markers of cardiometabolic health in the Airwave Health Monitoring Study: Men (energy intake under-reporters and chronic disease diagnosis excluded)* ! (change in marker per

1 unit change in DASH) SE p

N (%) Body mass index, kg/m2 Model 1 0.055 0.018 0.002 Model 2 -0.041 0.018 0.026 Model 3 -0.039 0.018 0.033

Waist circumference, cm Model 1 -0.230 0.045 <0.0001 Model 2 -0.164 0.046 0.001 Model 3 -0.158 0.045 0.001 Model 4 -0.076 0.025 0.002 Body fat, % Model 1 -0.193 0.029 <0.0001 Model 2 -0.137 0.030 <0.0001 Model 3 -0.134 0.029 <0.0001 Model 4 -0.085 0.019 <0.0001 HbA1c, % Model 1 -0.006 0.003 0.014 Model 2 -0.005 0.003 0.07 Model 3 -0.004 0.003 0.08 Model 4 -0.004 0.003 0.16 HDL mmol/L Model 1 -0.001 0.002 0.76 Model 2 -0.002 0.002 0.35 Model 3 -0.002 0.002 0.34 Model 4 -0.003 0.002 0.10 Non HDL mmol/L Model 1 -0.024 0.005 <0.0001 Model 2 -0.018 0.005 0.001 Model 3 -0.017 0.005 0.001 Model 4 -0.015 0.005 0.002 TC:HDL ratio Model 1 -0.018 0.005 0.0009 Model 2 -0.011 0.005 0.05 Model 3 -0.010 0.005 0.06 Model 4 -0.006 0.005 0.21 Diastolic blood pressure mmHg Model 1 -0.225 0.047 <0.0001 Model 2 -0.178 0.049 0.001 Model 3 -0.172 0.048 0.001 Model 4 -0.143 0.046 0.002

!

! ! "#%!

! (change in marker per 1 unit change in DASH)

SE p

Systolic blood pressure mmHg Model 1 -0.128 0.068 0.06 Model 2 -0.087 0.071 0.22 Model 3 -0.078 0.069 0.26 Model 4 -0.048 0.068 0.48 CRPmg/L± Model 1 -0.022 0.004 <0.0001 Model 2 -0.018 0.004 <0.0001 Model 3 -0.018 0.004 <0.0001 Model 4 -0.015 0.004 0.001

Abbreviations HDL High density lipoprotein, TC total cholesterol, CRP high sensitivity c reactive protein. SE standard error. ±CRP log transformed to allow parametric testing, untransformed values presented. Linear regression conducted to test for association between DASH score and markers of cardiometabolic risk Model 1 adjusted for age, Model 2 + physical activity, smoking, education, TV viewing, job strain, mean energy intake, alcohol; Model 3 + diagnosed ± treatment for diabetes, lipids or blood pressure. Model 4 + body mass index (continuous).

!

! ! "&#!

A6.3.2 Association of DASH score with markers of cardiometabolic health in

the Airwave Health Monitoring Study: Women (energy intake under-reporters

excluded)*


SE p

N (%) Body mass index, kg/m2 Model 1 -0.096 0.024 <0.0001 Model 2 -0.077 0.025 0.002 Model 3 -0.072 0.025 0.004 Waist circumference, cm Model 1 -0.262 0.056 <0.0001 Model 2 -0.196 0.058 0.001 Model 3 -0.186 0.058 0.001 Model 4 -0.038 0.029 0.18 Body fat, % Model 1 -0.182 0.040 <0.0001 Model 2 -0.132 0.041 0.001 Model 3 -0.128 0.041 0.002 Model 4 -0.042 0.022 0.05 HbA1c, % Model 1 0.001 0.003 0.85 Model 2 0.001 0.003 0.85 Model 3 0.001 0.003 0.74 Model 4 0.002 0.003 0.54 HDL mmol/L Model 1 0.000 0.002 0.87 Model 2 -0.001 0.002 0.72 Model 3 -0.001 0.002 0.69 Model 4 -0.003 0.002 0.15 Non HDL mmol/L Model 1 -0.016 0.005 0.001 Model 2 -0.012 0.005 0.015 Model 3 -0.011 0.005 0.020 Model 4 -0.008 0.005 0.09 TC:HDL ratio Model 1 -0.011 0.004 0.014 Model 2 -0.007 0.004 0.10 Model 3 -0.007 0.004 0.12 Model 4 -0.002 0.004 0.60 Diastolic blood pressure mmHg Model 1 -0.222 0.050 <0.0001 Model 2 -0.179 0.053 0.001 Model 3 -0.169 0.052 0.001 Model 4 -0.120 0.050 0.016

!

! ! "&&!


SE p

Systolic blood pressure mmHg Model 1 -0.158 0.072 0.029 Model 2 -0.101 0.076 0.18 Model 3 -0.083 0.074 0.26 Model 4 -0.021 0.071 0.77 CRPmg/L± Model 1 0.387 0.144 0.007 Model 2 -0.024 0.005 <0.0001 Model 3 -0.023 0.005 <0.0001 Model 4 -0.018 0.005 0.001

Abbreviations HDL High density lipoprotein, TC total cholesterol, CRP high sensitivity c reactive protein. SE standard error. ±CRP log transformed to allow parametric testing, untransformed values presented. Linear regression conducted to test for association between DASH score and markers of cardiometabolic risk Model 1 adjusted for age, Model 2 + physical activity, smoking, education, TV viewing, job strain, menopause status, mean energy intake, alcohol; Model 3 + diagnosed ± treatment for diabetes, lipids or blood pressure. Model 4 + body mass index (continuous).

!!!

"#$!

A6.3.3 O

dds ratio of having three or more m

arkers of metabolic risk per quartile of D

AS

H score: m

en (energy intake under-reporters excluded)*

F

ully adjusted§ F

ully adjusted§ +BM

I

Cases/N

O

R

(95%C

I) O

R

(95%C

I)

Ref: Q


1.00

1.00

Q

4 265/658

1.02 0.69

1.52 0.91

0.59 1.41

Q3

330/763 1.31

0.90 1.91

1.07 0.71

1.62 Q

2 271/640

1.32 0.90

2.00 1.01

0.65 1.56

Q1 (unhealthiest)

256/581 1.52

1.01 2.29

1.31 0.83

2.10 p-trend

0.021

0.24

A

bbreviations: BM

I body mass index. C

I confidence intervals. OR

Odds R

atio. §Fully adjusted: age +physical activity, sm

oking status, education and TV

view

ing, job strain, continuous variables: mean energy intake and m

ean alcohol g/day A

6.3.4 Odds ratio of having three or m

ore markers of m


SH

score: wom

en (energy intake under-reporters excluded)*

F

ully adjusted§ F

ully adjusted§ +BM

I

C

ases/N

OR

(95%

CI)

OR

(95%

CI)

Ref: Q


1.00

1.00

Q4

107/444 1.50

0.87 2.60

1.64 0.88

3.06

Q3

126/473 2.72

1.62 4.56

2.84 1.58

5.12 Q

2 108/438

2.19 1.26

3.82 1.89

1.01 3.52

Q1 (unhealthiest)

92/378 2.28

1.29 4.02

2.05 1.07

3.93 p-trend

0.002

0.033 A

bbreviations: BM

I body mass index. C

I confidence intervals. OR

Odds R

atio. §Fully adjusted: age + physical activity, sm

oking status, education and TV

view

ing, job strain, menopause; continuous variables: m

ean energy intake and mean alcohol g/day

!!!

"#"!

A7.1 C

omparison of participants reporting part-tim

e work and/or chronic disease diagnosis* w

ith the rest those who w

ork full tim

e without chronic disease diagnosis across socio-dem

ographic and occupational characteristics

*E

xcluded from

Study 5

Included in Study

5 p

N (%

) 843

(13.9) 5036

(86.1)

Age, years (S

D)

43.8 (9.4)

41.1 (9.2)

<0.0001

N (%

)

Male

217 (26.7)

3280 (65.1)

C

hronic disease diagnosis (self report) 322

(38.2)

Part-tim

e work

535 (63.8)

M

arital status

<0.0001

Cohabiting

78 (9.9)

875 (17.8)

D


(8.5) 400

(8.1)

Married

588 (74.3)

3041 (61.8)

S

ingle 58

(7.3) 605

(12.3)

Income (household per anum

)

<0.0001

Less than £32,000 195

(24.0) 745

(12.7)

£32,000 - £47,999 82

(10.1) 589

(10.1)

£48,000 - £57,999 294

(36.2) 1990

(34.0)

£58,000- £77,999 155

(19.1) 1185

(20.3)

More than £ 78,000

86 (10.6)

527 (9.0)

R

ank

<0.0001

Police staff

368 (53.1)

1361 (26.6)

P

olice Constable/S

ergeant 265

(38.2) 2611

(51.1)


27 (4.1)

363 (7.1)

O

ther 33

(4.8) 79

(1.5)

Em

ployment (force) country

0.24 E

ngland 596

(73.7) 3554

(70.8)

Wales

127 (15.7)

891 (17.7)

S

cotland 86

(10.6) 576

(11.5)

!

! ! "#$!

Table A7.2 Comparison of demographic, lifestyle and occupational characteristics across men

and women for Study 4, Chapter 7 (n = 5,036) Women Men P* N (%) 1756 35.0 3280 65.0 Mean age, years, SD 38.9 9.4 42.2 8.1 <0.0001 Weekly working hours (exc. overtime) mean, SD

40.5 5.2 42.2 5.4 <0.0001

Weekly overtime hours, mean, SD

2.0 3.2 3.1 4.2 <0.0001

N(%) White 1720 98.0 3171 96.7 0.014 Relationship status Cohabiting 414 24.6 461 14.2 Divorced/separated 182 10.8 218 6.7 Married 704 41.9 2337 72.1 Single 381 22.7 224 6.9 Missing 75 4.3 40 1.2 Education 0.0001 Left school before taking GCSE 63 3.6 147 4.5 GCSE or equivalent 454 25.8 1004 30.6 Vocational qualifications 129 7.4 238 7.3 A levels / Highers or equivalent 567 32.3 1067 32.5 Bachelor Degree or equivalent 410 39.4 630 19.2 Postgraduate qualifications 133 7.6 194 5.9 Annual household income <0.0001 Less than £32,000 741 26.8 274 8.4 £32,000 - £47,999 206 11.7 383 11.7 £48,000 - £57,999 531 30.4 1459 44.4 £58,000- £77,999 339 19.3 846 25.8 More than £ 78,000 206 11.7 321 9.8 Employment force, country <0.0001 England 1293 73.9 2261 69.1 Scotland 253 14.5 638 19.5 Wales 204 11.7 372 11.4 Missing 6 9 Rank <0.0001 Police staff 817 52.8 544 19.0 Police Constable/ Sergeant 637 41.2 1974 68.8 Inspector/Chief Inspector or

50 3.2 313 10.9

Other 42 2.7 37 1.3 Missing 210 11.2 412 12.6 Work environment <0.0001 Mainly office duties 565 36.6 1005 35.0 Mainly mobile duties 482 31.2 1405 49.0 Unclassified 499 32.3 458 16.0 Missing

!"#$ ""%!$ &"!$ "!%'$ Total hours worked per week <0.0001 35-40 hours 948 54.0 1145 34.9 41 – 48 hours 533 30.3 1263 38.5 49 – 54 hours 412 8.1 441 13.4 55 hours or more 133 7.6 431 13.1

!

! ! "#%!

Table A7.2 continued

Women Men P* Shift work last 30 days <0.0001 Day only 101 17.5 198 12.9 Shift (no nights) 197 34.2 309 20.1 Shift (with night work) 278 48.3

1028 67.0 Missing 1180 67.2 1745 53.5 Job Strain

<0.0001 Low (high control, low demand) 461 26.2 1057 32.2 Passive (low control, low

413 23.5 570 17.4

Active (high demand, high

447 25.5 943 28.8 High (high demand, low control) 435 24.8 710 21.7 Physical activity† <0.0001 Low 255 14.5 340 10.4 Moderate 840 47.8 1428 43.5 High 661 37.6 1512 46.1 Smoking status <0.0001 Never smoker 1184 67.7 2302 70.4 Former smoker 371 21.2 737 22.5 Current smoker 195 11.1 229 7.0 Missing 6 12 Sleep <0.0001 5 hours or less 121 6.7 164 5.0 6 hours 384 21.9 991 30.2 7 hours 717 40.8 1424 43.4 8 hours 452 25.7 612 57.5 9 hour or more 82 4.7 89 52.0 Sitting (total weekdays) 0.97 Low (<20 hours) 544 31.0 1010 30.8 Moderate (20 – 40 hours) 657 37.4 1238 37.7 High (> 40 hours) 555 31.6 1032 31.5 Weekly TV viewing <0.0001 Low (< 6 hours) 624 35.5 845 25.8 Moderate (6 – 15 hours) 744 42.4 1485 45.2 High (>15 hours) 388 22.1 950 29.0

Abbreviations: SD: standard deviation. GCSE: General certificate of Secondary Education; *Student t-test compared mean age between male and female participants. Chi squared test compared differences between men and women across categorical variables; missing data were not included in the analyses. †METs metabolic equivalents, classification by IPAQ guidelines !

!

! ! "#&!

A7.3 Prevalence of cardiometabolic biomarkers and anthropometric measurements against risk categories across men and women in the Airwave Health Monitoring Study with dietary data (n =5,036)

Total Women Men P* N, (%) 5,036 (100.0) 1,756 (35.0) 3,280 (65.0) Body mass index by risk category <0.0001 Healthy (<25kg/m2) #%'$! ("#)*+! 886 (50.5) 708 (21.6) Over weight (25 - 30kg/m2) ,$#&! ($-).+! 597 (34.0) 1819 (55.5) Obese (!30kg/m2) #.,&! (,.)$+! 273 (15.5) 753 (23.0) Waist circumference risk by risk category1 ! ! <0.0001 Low risk <94 (men)<80cm (women) ,%%%! (%.)*+! 858 (48.9) 1697 (51.7) Increased risk 940-102 80-88(women) #$,*! (,-)"+! 442 (21.2) 985 (30.0) High risk !102(men), !88 (women) #.%$! (,.)'+! 456 (26.0) 598 (18.2) HDL by risk category ! ! <0.0001 Low risk >1.0mmol/L (men) >1.3mmol/L (women) $*$,! ('$),+! 1734 (98.7) 3008 (91.7) High risk <1.0mmol/L (men) <1.3mmol/L (men) ,'$! (%)-+! 22 (1.3) 272 (8.3) Non HDL by risk category ! ! <0.0001 Low risk <4.0mmol/L "##-! (&#)'+! 1384 (78.8) 1734 (52.9) High risk !4.0mmol/L #'#-! ("-)#+! 372 (21.2) 1546 (47.1) HbA1c by risk category ! ! 0.006 Normal <5.7 % ".",! (&.),+! 1004 (57.2) 2028 (61.8) Pre-diabetes >5.7 % <6.5% #-"&! ("&)%+! 688 (39.2) 1148 (35.0) Diabetes !6.5 % #&-! (")"+! 64 (3.6) 104 (3.2) Blood pressure by risk category ! ! <0.0001 Normal SBP <120 + DBP<80mmHg #.*&! (,#)$+! 758 (43.2) 318 (9.7) Pre-hypertensive SBP120-139 ± DBP 80-89 ,$"#! ($-)"+! 745 (42.4) 1686 (51.4) Hypertensive SBP!140± DBP!90mmHg #%,'! (".)$+! 253 (14.4) 1276 (38.9) HsC-reactive by risk category ! ! <0.0001 Low risk <1.0mg/L ,*#-! (%$).+! 904 (51.5) 1814 (55.3) Moderate risk >1.0 <3.0mgl/L #&&&! ("")#+! 546 (31.2) 1120 (34.1) High risk !3.0mgl/L <10 &%,! (#,)'+! 306 (17.4) 346 (10.6) Self-report diagnosed disease ! ! Dyslipidaemia or lipid lowering medication ,'#! (%)-+! 50 (2.9) 241 (7.4) <0.0001 Diagnosed hypertension, or on medication ",-! (&)%+! 109 (6.2) 219 (6.7) 0.52 Diagnosed diabetes, or on medication $#! (.)-+! 12 (0.7) 29 (0.9) 0.45 Classification by number of factors6 ! ! <0.0001 None &,-! (#,)%+! 328 (18.7) 300 (9.2) 1 or 2 ,&#"! (%#)'+! 988 (56.3) 1625 (49.5) 3 or more (high risk - metabolic syndrome) #*'%! ("%)&+! 440 (25.1) 1355 (41.3)

Abbreviations: HDL High density lipoprotein, HbA1c glycated haemoglobin, BP blood pressure. . Chi-squared test compared differences across categorical variables. 1Waist risk based on WHO European values. 6Elevated risk classification for waist, HbA1c (or T2DM diagnosis/treatment), blood pressure (or hypertension diagnoses/treatment), HDL, non HDL risk (or dyslipidaemia diagnosis/treatment) and CRP

!!!

#*'!

A7.4 Sam

ple characteristics by group of weekly w

orking hours

A7.4.1 C

omparison of dem

ographic, lifestyle and occupational characteristics across working hour groups: m

en (n= 3,280)

35 - 40 hrs/week

41 – 48 hrs/week

49 – 54 hrs/week

>55 hrs/week

p N

(%)

1145 (34.9)

1263 (38.5)

441 (13.5)

431 (13.4)

Hours w

orked per week, m

ean (SD)

39.1 (1.4)

43.8 (1.9)

50.1 (1.2)

60.2 (6.1)

Age, years m

ean (SD)

44.0 (9.4)

41.4 (8.3)

41.2 (8.0)

40.8 (8.0)

<0.0001

N

(%)

White

1114 (97.5)

1215 (96.2)

423 (95.9)

419 (97.2)

0.23

Marital status

0.30

Cohabiting

137 (12.2)

192 (15.3)

60 (13.7)

72 (16.9)

Divorced/separated

87 (7.7)

73 (5.8)

30 (6.9)

28 (6.6)

Married

822 (73.1)

900 (71.8)

319 (73.0)

296 (69.7)

Single

79 (7.0)

88 (7.0)

28 (6.4)

29 (6.8)

M

issing 20

(1.7) 10

(0.8) 4

(0.9) 6

(1.4)

Education

0.031


SE

74

(6.5) 44

(3.5) 16

(3.6) 13

(3.0)

GC

SE

or equivalent 358

(31.3) 384

(30.4) 138

(31.3) 124

(28.8)


81 (7.1)

93 (7.4)

31 (7.0)

33 (7.7)

A levels / H


(30.5) 435

(34.4) 143

(32.4) 140

(32.5)

Bachelor D


(19.3) 239

(28.9) 87

(19.7) 83

(19.3)


62 (5.4)

68 (5.4)

26 (5.9)

38 (8.8)

!!!

#*(!

A7.4.1 continued (m

en) 35 - 40 hrs/w

eek 41 – 48 hrs/w

eek 49 – 54 hrs/w

eek >55 hrs/w

eek p

N

(%)

A

nnual household income

<0.0001 Less than £32,000

136 (11.9)

92 (7.3)

17 (3.9)

29 (6.7)

£32,000 - £47,999

160 (14.0)

144 (11.4)

44 (10.0)

35 (8.1)

£48,000 - £57,999

518 (45.2)

600 (47.5)

191 (43.3)

147 (34.1)

£58,000- £77,999

237 (20.7)

335 (26.4)

125 (28.6)

149 (34.6)

M

ore than £ 78,000 94

(8.2) 92

(7.3) 64

(14.5) 71

(16.5)

Years in police force

6 years or less 229

(20.0) 208

(16.5) 64

(14.5) 66

(15.3) 0.0001

6 to 12 years 256

(22.4) 314

(24.9) 86

(19.5) 103

(23.9)

12 – 21 years 256

(22.4) 360

(28.5) 143

(32.4) 114

(26.5)

21 years or more

404 (35.3)

381 (30.2)

148 (33.6)

148 (34.3)

Tim

e in current job role

0.08

2 years or less 408

(35.6) 420

(33.2) 169

(38.3) 176

(40.8)

3 to 5 years 332

(29.0) 403

(31.9) 132

(29.9) 118

(27.4)

6 years or more

405 (35.4)

440 (34.8)

140 (31.8)

137 (31.8)

Shift w

ork 30days

<0.0001

Day

84 (20.1)

78 (11.5)

21 (9.3)

15 (7.1)

S

hift no night 114

(27.2) 121

(17.8) 43

(19.0) 31

(14.7)

Shift w

ith night 221

(52.7) 480

(70.7) 162

(71.7) 165

(78.2)

Employm

ent (force) country

<0.0001

England

846 (73.9)

850 (67.4)

306 (69.5)

262 (61.1)

S

cotland 200

(17.5) 243

(19.3) 80

(18.2) 115

(26.8)

Wales

98 (8.6)

168 (13.2)

54 (12.3)

52 (12.1)

!!!

#*)!

A7.4.1 continued (m

en) 35 - 40 hrs/w

eek 41 – 48 hrs/w

eek 49 – 54 hrs/w

eek >55 hrs/w

eek p

N

(%)

R

ank

Police staff

377 (38.7)

128 (11.7)

22 (5.6)

17 (4.1)

<0.0001 P

olice Constable/S

ergeant 538

(55.2) 855

(78.4) 287

(73.6) 294

(71.4)


40 (4.1)

97 (8.9)

79 (20.3)

97 (23.5)

O

ther 20

(2.0) 11

(1.0) 2

(0.5) 4

(1.0)

Missing

170 (14.8)

172 (13.6)

51 (13.1)

19 (4.4)

Job dem

and-control

<0.0001


demand)

467 (40.8)

373 (29.5)

122 (27.7)

95 (22.0)

P


demand)

237

(20.7) 211

(16.7) 62

(14.1) 60

(13.9)

Active (high dem


(21.8) 371

(29.4) 144

(32.7) 179

(41.5)


and, low control)

192 (16.8)

308 (24.4)

113 (25.6)

97 (22.5)

W

ork environment

<0.0001 M

ainly office duties 321

(32.9) 349

(32.0) 162

(41.5) 173

(42.0)

Mainly m

obile duties 404

(41.4) 599

(54.9) 200

(51.3) 202

(49.0)

Unclassified

250 (25.6)

143 (13.1)

28 (7.2)

37 (9.0)

M

issing 170

(14.8) 172

(13.6) 51

(13.1) 19

(4.4)

Physical activity (METs)

0.69 Low

125

(10.9) 134

(10.6) 39

(8.8) 42

(9.7)

Moderate

486 (42.5)

541 (42.8)

201 (45.6)

200 (46.4)

H

igh 534

(46.6) 588

(46.6) 201

(45.6) 189

(43.9)

Smoking status

0.023 N

ever smoker

773 (67.8)

901 (71.7)

317 (72.0)

311 (72.2)

Form

er smoker

298 (26.1)

266 (21.2)

87 (19.8)

86 (19.9)

C

urrent smoker

70 (6.1)

89 (7.1)

36 (8.2)

34 (7.9)

M

issing 4

(0.0) 7

(0.6) 1

(0.0) 0

(0.0)

!!!

#"+!

A7.4.1 continued (m

en) 35 - 40 hrs/w

eek 41 – 48 hrs/w

eek 49 – 54 hrs/w

eek >55 hrs/w

eek p

N

(%)

Sleep

0.11

5 hours or less 47

(4.1) 60

(4.8) 27

(6.1) 30

(7.0)

6 hours 337

(29.4) 372

(29.5) 130

(29.5) 152

(35.3)

7 hours 505

(44.1) 568

(45.0) 191

(43.3) 160

(37.1)

8 hours 222

(19.4) 227

(18.0) 81

(18.4) 82

(19.0)

9 hour or more

34 (3.0)

36 (2.8)

12 (2.7)

7 (1.6)

W

eekly hours sitting (weekdays)

<0.0001 Low

(<20 hours) 293

(25.6) 297

(23.5) 119

(27.0) 136

(31.6)


473 (41.3)

618 (48.9)

209 (47.4)

185 (42.9)

H

igh (>40 hours) 379

(33.1) 348

(27.6) 113

(25.6) 110

(25.5)

TV viewing, hour per w

eek

0.013

Low (<6 hours)

323 (28.2)

411 (32.5)

146 (33.1)

130 (30.2)

M


(36.7) 498

(39.4) 161

(36.5) 159

(36.7)

High (>15 hours)

402 (35.1)

354 (28.0)

134 (30.4)

142 (32.9)

Q

uintile daily EI irregularity score

<0.0001

Q1 (m

ost regular EI)

264 (23.1)

241 (19.1)

83 (18.2)

68 (15.8)

Q

2 258

(22.5)

234 (18.5)

82 (18.6)

82 (19.0)

Q

3 233

(20.3) 253

(20.0) 88

(19.5) 82

(19.0)

Q4

205 (17.9)

273 (21.6)

84 (19.0)

94 (21.8)

Q

5 (most irregular E

I) 185

(16.2) 262

(20.7) 104

(23.6) 105

(24.4)

Quintile D

ASH

score

0.004

Q1 (unhealthiest)

164 (14.3)

236 (18.7)

90 (20.4)

88 (20.4)

Q

2 207

(18.1) 276

(21.8) 80

(18.1) 82

(19.0)

Q3

269 (24.5)

274 (21.7)

97 (22.0)

110 (25.5)

Q

4 256

(23.4) 242

(19.2) 93

(21.1) 80

(18.6)

Q5 (healthiest)

249 (21.8)

235 (18.6)

81 (18.4)

71 (16.5)

!!!

#"*!

A7.4.1 continued (m

en) 35 - 40 hrs/w

eek 41 – 48 hrs/w

eek 49 – 54 hrs/w

eek >55 hrs/w

eek p

N

(%)

H

ealth status

Diagnosed / on treatm

ent blood

102 (8.9)

61 (4.8)

27 (6.1)

29 (6.7)

0.001 D

iagnosed / on treatment diabetes

10 (0.9)

11 (0.9)

4 (0.9)

4 (0.9)

0.99 D

iagnosed / on treatment lipid

107

(9.4) 76

(6.0) 28

(6.3) 30

(7.0) 0.013


ber of risk

0.69 N

one 96

(8.4) 121

(9.6) 41

(9.3) 42

(9.7)

1 or 2 (increased risk) 558

(47.8) 643

(50.9) 212

(48.1) 212

(49.2)

3 or more (high risk)

491 (42.9)

499 (39.5)

188 (42.6)

177 (41.1)

A

bbreviations: GC

SE

General C

ertificate of Secondary E

ducation. ME

Ts metabolic equivalents, classification by IP

AQ

guidelines (250). Chi squared test to com

pare differences across categorical variables, m


!!!

#""!

A7.4.2 C

omparison of dem

ographic, lifestyle and occupational characteristics across working hour groups: w

omen (n= 1,756)

35 - 40 hrs/w

eek 41 – 48 hrs/w

eek >49 hrs/w

eek p

N (%

) 948

(54.0) 533

(30.3) 275

(15.7)

Hours w

orked per week, m

ean (SD)

38.3 (1.6)

43.5 (2.0)

55.1 (5.9)

A

ge, years mean (SD

) 40.3

(10.1) 37.6

(8.7) 36.6

(9.0) <0.0001

N

(%)

W

hite 929

(98.0) 525

(98.5) 275

(96.7) 0.24

Marital status

0.011 C

ohabiting 195

(21.6) 147

(28.6) 72

(27.3)

Divorced/separated

99 (11.0)

53 (10.3)

30 (11.4)

M

arried 415

(46.0) 191

(37.2) 98

(37.1)

Single

194 (21.5)

123 (23.9)

64 (24.2)

M

issing 45

(4.5) 19

(3.6) 10

(3.6)

Education

0.32


SE

37

(3.9) 17

(3.2) 9

(3.3)

GC

SE

or equivalent 260

(27.4) 131

(24.6) 63

(22.9)


70 (7.4)

38 (7.1)

21 (7.6)

A


317 (33.4)

169 (31.7)

81 (29.5)

B


198 (20.9)

133 (24.9)

79 (28.7)

P


(7.0) 45

(8.4) 22

(8.0)

Annual household incom

e

0.0004

Less than £32,000 273

(28.8) 138

(25.9) 60

(21.8)

£32,000 - £47,999 109

(11.5) 59

(11.1) 38

(13.8)

£48,000 - £57,999 314

(33.1) 150

(28.1) 70

(25.5)

£58,000- £77,999 164

(17.3) 116

(21.8) 59

(21.5)

More than £ 78,000

88 (9.3)

70 (13.1)

48 (17.4)

!!!

#"#!

A

7.4.2 continued (wom

en) 35 - 40 hrs/w

eek 41 – 48 hrs/w

eek >49 hrs/w

eek p

N

(%)

Years in police force

0.12 6 years or less

368 (38.8)

210 (39.4)

115 (41.8)

6 to 12 years

233 (24.6)

150 (28.1)

57 (20.7)

12 – 21 years

203 (21.4)

110 (20.6)

54 (19.6)

21 years or m

ore 144

(15.2) 63

(11.8) 49

(17.8)

Time in current job role

0.003 2 years or less

382 (40.3)

230 (43.2)

143 (52.0)

3 to 5 years

282 (29.5)

171 (32.1)

67 (24.4)

6 years or m

ore 286

(30.2) 132

(24.8) 65

(23.6)

Shift work 30 days

<0.0001 D

ay 63

(26.9) 28

(13.0) 10

(7.9)

Shift no night

100 (42.7)

64 (29.8)

33 (26.0)

S

hift with night

71 (30.3)

123 (57.2)

84 (66.1)

M

issing* 714

(75.3) 318

(59.6) 147

(53.6)

Employm

ent (force) country

0.08

England

697 (73.8)

401 (75.7)

195 (70.9)

S

cotland 143

(15.1) 60

(11.3) 50

(18.2)

Wales

105 (11.1)

69 (13.0)

30 (10.9)

M

issing* 3

(0.1) 3

(0.1) 0

(0.0)

Rank

<0.0001 P

olice staff 584

(71.3) 174

(37.4) 59

(22.5)

Police C

onstable/Sergeant

200 (24.4)

267 (57.4)

170 (64.9)

Inspector/C

hief Inspector or above 10

(1.2) 19

(4.1) 21

(8.0)

Other

25 (3.0)

5 (1.1)

12 (4.6)

M

issing* 129

(15.8) 68

(12.8) 12

(4.4)

!!!

#"$!

A7.4.2 continued (w

omen)

35 - 40 hrs/week

41 – 48 hrs/week

>49 hrs/week

p

N (%

)

Job strain

<0.0001


demand)

293 (30.9)

120 (22.5)

48 (17.5)

P


demand)

267 (28.2)

103 (19.3)

43 (15.6)

A


207 (21.8)

157 (29.5)

83 (30.2)

H

igh strain (high demand, low

control) 181

(19.1) 153

(28.7) 101

(36.7)

Work environm

ent

<0.0001


328 (40.1)

148 (31.8)

89 (34.0)

M

ainly mobile duties

167 (20.4)

191 (41.1)

124 (47.3)

U

nclassified 324

(39.6) 126

(27.1) 49

(18.7)

Missing

129 (15.8)

68 (12.8)

12 (4.4)

Physical activity (M

ETs)

0.002

Low

157 (16.6)

66 (12.4)

32 (11.6)

M

oderate 471

(49.7) 237

(44.5) 132

(48.0)

High

320 (33.8)

230 (43.2)

111 (40.4)

Sm

oking status

0.79

Never sm

oker 631

(66.7) 363

(68.4) 190

(69.6)

Former sm

oker 209

(22.1) 111

(20.9) 51

(18.7)

Current sm

oker 106

(11.2) 57

(10.7) 32

(11.7)

Missing

2 (0.1)

2 (0.4)

1 (0.4)

Sleep

0.12 5 hours or less

71 (7.5)

26 (4.9)

24 (8.7)

6 hours

201 (21.2)

110 (20.6)

73 (26.6)

7 hours

388 (40.9)

230 (43.2)

99 (36.0)

8 hours

248 (26.2)

141 (26.5)

63 (22.9)

9 hour or m

ore 40

(4.2) 26

(4.9) 16

(5.8)

!!!

#"%!

A7.4.2 continued (w

omen)

35 - 40 hrs/week

41 – 48 hrs/week

>49 hrs/week

p

N (%

)

Weekly hours sitting (w

eekdays)

0.004

Low (<20 hours)

313 (33.0)

189 (35.5)

122 (44.4)

M


(42.8) 228

(42.8) 110

(40.0)

High (>40 hours)

229 (24.2)

116 (21.8)

43 (15.6)

TV view

ing, hour per week

<0.0001 Low

(<6 hours) 260

(27.4) 179

(33.6) 105

(38.2)


344 (36.3)

216 (40.5)

97 (35.3)

H

igh (>15 hours) 344

(36.3) 138

(25.9) 73

(26.6)

Quintile daily EI irregularity score

Q

1 (most regular E

I) 199

(21.0) 103

(19.3) 49

(17.8) 0.29

Q2

189 (19.9)

104 (19.5)

58 (21.1)

Q

3 203

(21.4) 105

(19.7) 44

(12.5)

Q4

184 (19.4)

111 (20.8)

16 (20.4)

Q

5 (most irregular E

I) 173

(18.2) 110

(20.6) 68

(24.7)

Quintile D

ASH

score

0.012

Q1 (unhealthiest)

175 (18.5)

89 (16.7)

64 (23.3)

Q

2 174

(18.3) 117

(21.9) 57

(20.7)

Q3

212 (22.4)

115 (21.6)

55 (20.0)

Q

4 199

(21.0) 100

(18.8) 66

(24.0)

Q5 (healthiest)

188 (19.8)

112 (21.0)

33 (12.0)

H

ealth status


ent blood pressure 66

(7.0) 31

(5.8) 4.4

(11.0) 0.26


ent diabetes 9

(1.0) 2

(0.4) 1

(0.4) 0.34


ent lipid managem

ent 34

(3.6) 13

(2.4) 3

(1.1) 0.07


ber of risk factors

<0.0001

None

150 (15.8)

117 22.0

61 (22.2)

1 or 2 (increased risk)

510 (53.8)

308 57.8

170 (61.8)

3 or m

ore (high risk) 288

(30.4) 108

20.3 44

(16.0)

Abbreviations: G

CS

E G



ETs m


Q guidelines (250). C

hi squared test to com

pare differences across categorical variables, missing data w

as not included in the analyses.

!!!

#"&!

A7.5 Sensitivity analyses Study 4: non-m

id rank employees excluded - relationship betw

een working hours and m

arkers of cardiom

etabolic risk: men

35 - 40

hrs/week

41 – 48 hrs/w

eek 49 – 54

hrs/week

>55 hrs/w

eek P

trend

N (%

) 538

27.3 855

43.3 287

14.5 294

14.9

B


2 A

djusted mean (SE)

M

odel 1 27.7

0.1 27.6

0.1 27.8

0.2 28.4

0.2 0.003

M

odel 2 27.9

0.3 27.7

0.2 27.9

0.3 28.5

0.3 0.006

M

odel 3 27.9

0.3 27.7

0.2 27.9

0.3 28.4

0.3 0.017

M

odel 4 29.3

0.6 29.2

0.6 29.4

0.6 29.9

0.6 0.017

W


M

odel 1 93.7

0.4 93.3

0.3 93.6

0.5 95.1

0.5 0.030

M

odel 2 94.8

0.7 94.3

0.6 94.5

0.7 96.0

0.7 0.047

M

odel 3 94.8

0.7 94.2

0.6 94.4

0.7 95.8

0.7 0.16

M

odel 4 99.4

1.6 98.9

1.5 98.9

1.6 100.3

1.6 0.17

M

odel 5 96.0

0.9 95.8

0.9 95.5

0.9 95.7

0.9 0.29

Percentage body fat

M

odel 1 22.1

0.2 21.8

0.2 22.2

0.3 23.4

0.3 0.001

M

odel 2 22.8

0.4 22.4

0.4 22.8

0.4 24.0

0.4 0.002

M

odel 3 22.8

0.4 22.3

0.4 22.7

0.4 23.7

0.4 0.010

M

odel 4 25.1

0.9 24.7

0.9 25.1

0.9 26.1

0.9 0.011

M

odel 5 23.3

0.6 23.0

0.6 23.2

0.6 23.5

0.6 0.38

TC

:HD

L

M

odel 1 4.1

0.0 4.1

0.0 4.1

0.1 4.1

0.1 0.36

M

odel 2 4.3

0.1 4.2

0.1 4.2

0.1 4.3

0.1 0.70

M

odel 3 4.3

0.1 4.2

0.1 4.2

0.1 4.3

0.1 0.98

M

odel 4 4.1

0.2 4.1

0.2 4.1

0.2 4.1

0.2 0.94

M

odel 5 4.0

0.2 3.9

0.2 3.9

0.2 3.9

0.2 0.54

!!!

#"'!

A7.5 m

en

35 - 40 41 – 48

49 – 54 >55

P

trend P

interaction

HD

L mm

ol/L

M

odel 1 1.38

0.01 0.38

0.01 1.37

0.02 1.35

0.02 0.15

M

odel 2 1.34

0.02 1.34

0.02 1.33

0.03 1.31

0.03 0.26

M

odel 3 1.33

0.02 1.34

0.02 1.33

0.03 1.32

0.03 0.42

M

odel 4 1.29

0.05 1.29

0.05 1.29

0.06 1.27

0.05 0.42

M

odel 5 1.32

0.05 1.33

0.05 1.33

0.05 1.32

0.05 0.87

N

on HD

L mm

ol/L

M

odel 1 3.96

0.04 3.98

0.03 3.98

0.06 3.99

0.06 0.69

M

odel 2 4.09

0.07 4.09

0.07 4.09

0.08 4.08

0.08 0.94

M

odel 3 4.09

0.07 4.08

0.07 4.07

0.08 4.06

0.08 0.70

M

odel 4 3.71

0.17 3.71

0.17 3.71

0.17 3.70

0.17 0.83

M

odel 5 3.63

0.17 3.63

0.17 3.63

0.17 3.58

0.17 0.51

H

bA1c, %

M

odel 1 5.57

0.02 5.57

0.02 5.62

0.03 5.60

0.03 0.23

M

odel 2 5.58

0.04 5.58

0.04 5.63

0.04 5.61

0.04 0.19

M

odel 3 5.58

0.04 5.58

0.04 5.63

0.04 5.61

0.04 0.25

M

odel 4 6.90

0.08 6.90

0.08 6.93

0.09 6.90

0.08 0.67

M

odel 5 6.86

0.08 6.87

0.08 6.90

0.08 6.86

0.08 0.93

D

iastolic blood pressure, mm

Hg

M

odel 1 81.0

0.4 81.5

0.3 81.1

0.5 82.1

0.5 0.15

M

odel 2 81.9

0.7 82.3

0.6 81.8

0.8 82.9

0.8 0.27

M

odel 3 81.8

0.7 82.1

0.6 81.6

0.8 82.5

0.8 0.46

M

odel 4 86.3

1.6 86.8

1.6 86.2

1.6 86.2

1.6 0.46

M

odel 5 85.0

1.5 85.6

1.5 84.9

1.5 85.3

1.5 0.53

Systolic blood pressure, m

mH

g

Model 1

135.1 0.6

135.1 0.4

134.9 0.8

136.2 0.8

0.30

Model 2

134.6 1.0

134.5 0.9

134.2 1.1

135.6 1.1

0.37

Model 3

134.5 1.0

134.4 0.9

134.1 1.1

135.5 1.1

0.38

Model 4

141.1 2.3

141.3 2.2

140.8 2.3

142.1 2.3

0.41

Model 5

139.5 2.2

139.9 2.2

139.2 2.2

140.0 2.2

0.52

!!!

#"(!

CR

P, mg/L

±

Model 1

0.88 0.94

0.95 0.93

0.98 0.95

1.02 0.95

0.026

Model 2

0.96 0.96

1.01 0.96

1.02 0.97

1.10 0.97

0.05

M

odel 3 0.96

0.96 0.99

0.96 1.01

0.97 1.07

0.97 0.10

M

odel 4 0.95

1.05 1.00

1.05 1.02

1.06 1.02

1.06 0.08

M

odel 5 0.83

1.05 0.88

1.04 0.89

1.05 0.89

1.05 0.28

A

bbreviations HD




±CR

P log transform

ed to allow param






AS

H score m




!!#")!

A7.6 Sensitivity analyses Study 4: energy intake under-reporters excluded - relationship betw


arkers of cardiom

etabolic risk A

7.6.1 men

35 - 40 hrs/w

eek 41 – 48

hrs/week

49 – 54 hrs/w

eek >55

hrs/week

Ptrend

N (%

) 494

36.1 544

39.7 188

13.7 144

10.51

B


2 A

djusted mean (SE)

M

odel 1 26.5

0.1 26.6

0.1 26.8

0.2 27.5

0.3 0.001

M

odel 2 26.5

0.2 26.7

0.2 26.8

0.3 27.5

0.3 0.001

M

odel 3 26.7

0.2 26.7

0.2 26.9

0.3 27.5

0.3 0.003

M

odel 4 27.6

0.6 27.7

0.6 27.9

0.6 28.5

0.6 0.002

W


M

odel 1 90.8

0.4 91.2

0.4 91.5

0.6 93.2

0.7 0.003

M

odel 2 91.0

0.6 91.4

0.6 91.7

0.7 93.1

0.8 0.007

M

odel 3 91.4

0.6 91.4

0.6 92.0

0.7 93.1

0.8 0.017

M

odel 4 95.1

1.4 95.3

1.4 95.7

1.5 96.9

1.5 0.014

M

odel 5 93.4

0.8 93.3

0.8 93.4

0.9 93.3

0.9 0.99

Percentage body fat

M

odel 1 20.5

0.3 20.5

0.2 21.1

0.4 22.2

0.5 0.001

M

odel 2 20.6

0.4 20.6

0.4 21.3

0.5 22.3

0.5 0.001

M

odel 3 20.7

0.4 20.6

0.4 21.3

0.5 22.2

0.5 0.002

M

odel 4 21.6

0.9 21.6

1.0 22.1

1.0 23.1

1.0 0.002

M

odel 5 20.5

0.7 20.5

0.7 20.8

0.7 20.9

0.7 0.18

TC

:HD

L

M

odel 1 3.9

0.0 3.9

0.0 4.0

0.1 4.0

0.1 0.48

M

odel 2 4.1

0.1 4.1

0.1 4.1

0.1 4.1

0.1 0.87

M

odel 3 4.1

0.1 4.0

0.1 4.1

0.1 4.1

0.1 0.98

M

odel 4 3.7

0.2 3.6

0.2 3.6

0.2 3.6

0.2 0.95

M

odel 5 3.6

0.2 3.5

0.2 3.5

0.2 3.5

0.2 0.34

!!!

##+!

A

7.6.1 men

continued

35 - 40

hrs/week

41 – 48

hrs/week

49 – 54

hrs/week

>55

hrs/week

HD

L mm

ol/L

M

odel 1 1.40

0.01 1.41

0.01 1.42

0.02 0.40

0.03 0.82

M

odel 2 1.37

0.02 1.38

0.02 1.39

0.03 1.38

0.03 0.70

M

odel 3 1.37

0.02 1.37

0.02 1.39

0.03 1.38

0.03 0.66

M

odel 4 1.36

0.06 1.37

0.06 1.38

0.06 1.37

0.06 0.67

M

odel 5 1.38

0.05 1.39

0.05 1.41

0.06 1.41

0.06 0.21

N

on HD

L mm

ol/L

M

odel 1 3.93

0.04 3.90

0.04 3.99

0.07 4.02

0.08 0.20

M

odel 2 4.01

0.07 3.96

0.07 4.04

0.09 4.04

0.10 0.55

M

odel 3 4.01

0.07 3.95

0.07 3.95

0.07 4.02

0.10 0.69

M

odel 4 3.42

0.17 3.36

0.17 3.46

0.18 3.42

0.18 0.73

M

odel 5 3.37

0.17 3.31

0.17 3.39

0.17 3.32

0.18 0.83

H

bA1c, %

M

odel 1 5.60

0.02 5.57

0.02 5.62

0.04 5.57

0.04 0.83

M

odel 2 5.60

0.04 5.57

0.04 5.61

0.05 5.61

0.05 0.60

M

odel 3 5.61

0.04 5.56

0.04 5.61

0.05 5.55

0.05 0.45

M

odel 4 6.45

0.09 6.41

0.09 6.44

0.09 6.39

0.09 0.37

M

odel 5 6.42

0.09 6.39

0.09 6.41

0.09 6.35

0.09 0.16

!!!

##*!

A

7.6.1 men

continued

35 - 40 hrs/w

eek 41 – 48

hrs/week

49 – 54 hrs/w

eek >55

hrs/week

Ptrend


mH

g

Model 1

80.5 0.4

80.5 0.4

80.5 0.7

80.7 0.8

0.82

Model 2

80.8 0.7

80.7 0.6

80.7 0.8

80.8 0.9

0.99

Model 3

80.8 0.7

80.6 0.6

80.7 0.8

80.6 0.9

0.85

Model 4

81.1 1.7

81.1 1.7

81.1 1.7

81.1 1.8

0.99

Model 5

80.4 1.6

80.3 1.6

80.1 1.7

79.6 1.7

0.37


Hg

M

odel 1 135.1

0.6 134.4

0.6 134.7

1.0 134.8

1.1 0.85

M

odel 2 135.2

0.9 134.6

0.9 134.9

1.2 135.1

1.3 1.00

M

odel 3 135.3

1.0 134.6

0.9 135.0

1.2 135.0

1.4 0.90

M

odel 4 141.1

2.4 140.9

2.4 141.0

2.5 141.3

2.6 0.89

M

odel 5 140.3

2.3 140.0

2.3 139.9

2.4 139.7

2.5 0.60

C

RP, m

g/L±

M

odel 1 0.82

0.94 0.85

0.94 0.85

0.96 0.89

0.97 0.38

M

odel 2 0.89

0.96 0.91

0.96 0.93

0.98 0.97

0.99 0.32

M

odel 3 0.89

0.96 0.90

0.96 0.92

0.98 0.94

0.99 0.53

M

odel 4 0.77

1.07 0.78

1.07 0.78

1.07 0.82

1.08 0.49

M

odel 5 0.71

1.06 0.71

1.06 0.72

1.07 0.69

1.08 0.82

A

bbreviations HD




±CR

P log transform

ed to allow param






AS

H score m




!

! ! ""#!

A7.6.2 Women

35 - 40 hrs/week

41 – 48 hrs/week

>49 hrs/week

Ptrend

N (%) 499 56.1 268 30.1 122 13.7

Body mass index, kg/m2 Adjusted mean (SE)

Model 1 25.3 0.2 24.8 0.3 25.2 0.4 0.82

Model 2 25.3 0.3 24.7 0.4 25.4 0.5 0.85

Model 3 25.3 0.3 24.7 0.4 25.2 0.5 0.75

Model 4 27.0 0.9 26.4 1.0 26.9 1.0 0.76

Waist circumference, cm

Model 1 80.6 0.5 78.8 0.6 80.0 0.9 0.55

Model 2 80.7 0.8 78.6 1.0 80.4 1.1 0.80

Model 3 80.7 0.8 78.7 0.9 79.9 1.1 0.43

Model 4 84.3 2.3 82.4 2.3 83.5 2.4 0.43

Model 5 80.4 1.2 79.6 1.2 79.9 1.3 0.31

Percentage body fat

Model 1 31.6 0.3 30.8 0.4 31.0 0.6 0.36

Model 2 31.6 0.6 30.7 0.7 31.2 0.8 0.54

Model 3 31.5 0.6 30.7 0.6 30.8 0.8 0.28

Model 4 33.5 1.6 32.7 1.6 32.8 1.7 0.29

Model 5 30.7 0.9 30.6 0.9 30.1 0.9 0.12

HDL, mmol/L

Model 1 1.77 0.02 1.79 0.02 1.79 0.02 0.57 Model 2 1.77 0.03 1.80 0.04 1.80 0.05 0.52 Model 3 1.77 0.03 1.81 0.04 1.80 0.05 0.38

Model 4 1.68 0.10 1.72 0.10 1.71 0.10 0.39 Model 5 1.73 0.09 1.76 0.09 1.77 0.10 0.42 TC:HDL

Model 1 3.0 0.0 2.9 0.0 2.9 0.1 0.07 Model 2 3.1 0.1 3.0 0.1 3.0 0.1 0.12 Model 3 3.1 0.1 3.0 0.1 3.0 0.1 0.08

Model 4 3.4 0.2 3.4 0.2 3.3 0.2 0.09 Model 5 3.3 0.2 3.3 0.2 3.2 0.2 0.09 Non HDL, mmol/l

Model 1 3.35 0.04 3.29 0.05 3.23 0.08 0.19

Model 2 3.46 0.07 3.41 0.08 3.39 0.10 0.38

Model 3 3.46 0.07 3.41 0.08 3.38 0.10 0.32

Model 4 3.76 0.20 3.71 0.21 3.68 0.21 0.36

Model 5 3.35 0.04 3.29 0.05 3.23 0.08 0.19

HbA1c, %

Model 1 5.70 0.03 5.63 0.03 5.62 0.05 0.15 Model 2 5.64 0.05 5.58 0.05 5.56 0.06 0.18

Model 3 5.63 0.05 5.58 0.05 5.56 0.07 0.24

Model 4 7.23 0.12 7.20 0.12 7.17 0.13 0.22

Model 5 3.69 0.20 3.66 0.20 3.61 0.21 0.38

!

! ! """!

A7.6.2 continued (women) 35 – 40 hrs/week

41 – 48 hrs/week

>49 hrs/week

Ptrend

Diastolic blood pressure, mmHg

Model 1 76.3 0.4 76.2 0.6 76.2 0.6 0.61

Model 2 76.5 0.8 76.7 0.9 77.3 1.1 0.38

Model 3 76.4 0.8 76.7 0.9 77.1 1.1 0.49

Model 4 77.8 2.2 78.1 2.3 78.4 2.3 0.50

Model 5

Systolic blood pressure, mmHg

Model 1 123.1 0.6 123.1 0.6 123.4 1.2 0.83

Model 2 122.9 1.1 122.5 1.3 123.8 1.5 0.54

Model 3 122.7 1.1 122.7 1.3 123.6 1.5 0.50

Model 4 126.5 3.1 126.5 3.2 126.5 3.2 0.51

Model 5 124.8 3.0 125.3 3.1 125.8 3.2 0.42

CRP, mg/L

Model 1 1.07 0.94 0.87 0.96 0.87 0.99 0.047

Model 2 1.17 0.98 0.94 0.99 0.99 1.01 0.10

Model 3 1.15 0.98 0.93 0.99 0.95 1.01 0.07

Model 4 1.09 1.16 0.89 1.16 0.91 1.17 0.07

Model 5 0.97 1.14 0.83 1.15 0.82 1.16 0.07

Abbreviations HDL High density lipoprotein, TC total cholesterol, CRP high sensitivity c reactive protein.

P1, P-for-trend across groups of working hours.

±CRP log transformed to allow parametric testing, untransformed values presented.

Model 1 adjusted for age, Model 2 + physical activity, smoking, education, TV viewing, household

income, job strain, menopause status. Model 3 + DASH score mean energy intake, alcohol (continuous

variables). Model 4 + diagnosed ± treatment for diabetes, lipids or blood pressure. Model 5 + body mass

index (continuous)

!!!

""#!

A7.7 Sensitivity analyses Study 4: include part tim

e workers - relationship betw


arkers of cardiom

etabolic risk: wom

en

<35

hrs/week

35 - 40 hrs/w

eek 41 – 49

hrs/week

>49 hrs/w

eek P

trend

N (%

) 439

20 948

43.2 533

24.3 275

12.5

Body m

ass index, kg/m2

Adjusted m

ean (SE)

M

odel 1 25.2

0.2 26.2

0.1 25.7

0.2 26.1

0.3 0.05

Model 2

25.3 0.3

26.3 0.3

25.8 0.3

26.3 0.3

0.030 M

odel 3 25.3

0.3 26.2

0.3 25.7

0.3 26.2

0.3 0.05*

Model 4

27.2 0.7

28.1 0.6

27.6 0.7

28.1 0.7

0.05* W


Model 1

80.9 0.5

82.7 0.3

81.0 0.5

81.7 0.7

0.82 M

odel 2 81.3

0.7 83.0

0.6 81.3

0.7 82.3

0.8 0.63

Model 3

81.3 0.7

82.7 0.6

81.1 0.7

81.9 0.8

0.91 M

odel 4 86.2

1.6 87.6

1.5 85.9

1.6 86.8

1.6 0.92

Model 5

83.3 0.8

82.9 0.8

82.2 0.8

82.2 0.8

0.001 Percentage body fat

Model 1

31.7 0.3

33.1 0.2

32.2 0.3

32.7 0.4

0.23 M

odel 2 32.1

0.5 33.3

0.4 32.5

0.5 33.1

0.5 0.17

Model 3

32.0 0.5

33.2 0.4

32.4 0.5

32.9 0.5

0.30 M

odel 4 34.0

1.1 35.1

1.0 34.4

1.1 34.9

1.1 0.31

Model 5

32.2 0.6

32.2 0.6

32.0 0.6

31.9 0.6

0.17

!!!

""$!

A7.7 continued

<35 hrs/w

eek 35 - 40

hrs/week

41 – 49 hrs/w

eek >49

hrs/week

Ptrend

TC:H

DL

M

odel 1 3.1

0.0 3.2

0.0 3.0

0.0 3.0

0.1 0.05

Model 2

3.2 0.1

3.2 0.1

3.1 0.1

3.1 0.1

0.036 M

odel 3 3.2

0.1 3.2

0.1 3.1

0.1 3.1

0.1 0.06

Model 4

3.3 0.1

3.4 0.1

3.2 0.1

3.2 0.1

0.06 M

odel 5 3.2

0.1 3.2

0.1 3.1

0.1 3.1

0.1 0.012

HD

L mm

ol/L

Model 1

1.69 0.02

1.70 0.01

1.75 0.02

1.77 0.02

0.002 M

odel 2 1.69

0.03 1.71

0.02 1.76

0.03 1.77

0.03 0.004

Model 3

1.70 0.03

1.70 0.02

1.74 0.02

1.76 0.03

0.014 M

odel 4 1.60

0.06 1.60

0.06 1.64

0.06 1.66

0.06 0.013

Model 5

1.63 0.06

1.66 0.05

1.69 0.06

1.72 0.06

0.001 N

on HD

L mm

ol/L

Model 1

3.3 0.0

3.4 0.0

3.3 0.0

3.3 0.1

0.83 M

odel 2 3.4

0.1 3.5

0.0 3.4

0.1 3.5

0.1 0.93

Model 3

3.4 0.1

3.5 0.0

3.4 0.1

3.4 0.1

0.90 M

odel 4 3.5

0.1 3.6

0.1 3.5

0.1 3.5

0.1 0.95

Model 5

3.4 0.1

3.5 0.1

3.4 0.1

3.4 0.1

0.64 H

bA1c, %

Model 1

5.7 0.0

5.7 0.0

5.6 0.0

5.6 0.0

0.09 M

odel 2 5.7

0.0 5.7

0.0 5.6

0.0 5.6

0.0 0.20

Model 3

5.7 0.0

5.7 0.0

5.7 0.0

5.6 0.0

0.40 M

odel 4 6.7

0.1 6.7

0.1 6.7

0.1 6.7

0.1 0.52

Model 5

6.7 0.1

6.7 0.1

6.7 0.1

6.7 0.1

0.41

!!!

""%!

A7.7 continued

<35 hrs/w

eek 35 - 40

hrs/week

41 – 49 hrs/w

eek >49

hrs/week

Ptrend


mH

g

M

odel 1 75.4

0.4 77.0

0.3 76.8

0.4 77.1

0.6 0.032

Model 2

76.3 0.7

77.7 0.5

77.7 0.6

78.0 0.7

0.021 M

odel 3 76.3

0.7 77.4

0.5 77.4

0.6 77.6

0.7 0.08

Model 4

76.7 1.4

77.9 1.3

77.8 1.4

78.0 1.4

0.09 M

odel 5 75.8

1.3 76.4

1.3 76.6

1.3 76.6

1.4 0.26


Hg

Model 1

121.9 0.6

123.9 -0.4

123.0 -0.6

122.9 -0.8

0.28 M

odel 2 123.6

0.9 125.8

0.8 125.2

0.8 125.4

1.0 0.15

Model 3

123.7 0.9

125.5 0.8

124.9 0.8

125.1 1.0

0.28 M

odel 4 126.9

2.0 128.7

1.9 128.0

2.0 128.2

2.1 0.33

Model 5

125.7 1.9

126.8 1.8

126.4 1.9

126.3 2.0

0.66 C

RP, m

g/L±

Model 1

0.9 0.9

3.4 0.3

2.9 0.3

3.1 0.2

0.54 M

odel 2 1.1

1.0 1.3

1.0 1.1

1.0 1.2

1.0 0.48

Model 3

1.0 1.0

1.2 1.0

1.1 1.0

1.1 1.0

0.78 M

odel 4 1.2

1.1 1.5

1.0 1.3

1.1 1.3

1.1 0.78

Model 5

1.1 1.0

1.2 1.0

1.1 1.0

1.1 1.0

0.53 A

bbreviations HD




P-for-trend across groups of w

orking hours. ±CR

P log transform

ed to allow param

etric testing, untransformed values presented. *B

onferroni post hoc test; significant difference betw

een part-time w

orkers and workers >49hours per w

eek. M

odel 1 adjusted for age, Model 2 + physical activity, sm


ing, household income, job strain, m

enopause status. Model 3 + D

AS

H score m




!

! ! ""#!

A7.8 Association between shift wok and cardiometabolic health markers*

Day Shift Night P n 370 567 1387 Body mass index, kg/m2 Adjusted mean (SE) Model 1 27.1 (0.3) 27.6 (0.2) 27.7 (0.1) 0.10

Model 2 26.7 (0.3) 26.5 (0.2) 26.7 (0.2) 0.61 Model 3 26.6 (0.3) 26.4 (0.2) 26.6 (0.2) 0.66

Model 4 27.4 (0.8) 27.2 (0.8) 27.4 (0.8) 0.65 Waist Circumference, cm Model 1 87.7 (0.5) 86.4 (0.4) 86.6 (0.3) 0.18 Model 2 88.4 (0.7) 87.1 (0.6) 87.3 (0.5) 0.13 Model 3 88.1 (0.7) 86.9 (0.6) 87.1 (0.5) 0.18 Model 4 90.4 (1.9) 89.2 (1.9) 89.5 (1.9) 0.19 Model 5 89.6 (1.1) 88.7 (1.1) 88.5 (1.0) 0.001§ Body fat, % Model 1 26.8 (0.4) 26.9 (0.3) 26.9 (0.2) 0.95 Model 2 27.3 (0.4) 27.3 (0.4) 27.1 (0.3) 0.88 Model 3 27.0 (0.4) 27.1 (0.4) 27.1 (0.3) 0.91 Model 4 28.7 (1.2) 28.8 (1.2) 28.9 (1.2) 0.90 Model 5 28.1 (0.8) 28.4 (0.8) 28.3 (0.7) 0.37 HDL, mmol/L Model 1 1.53 (0.02) 1.58 (0.01) 1.55 (0.01) 0.08 Model 2 1.50 (0.02) 1.55 (0.02) 1.52 (0.02) 0.11 Model 3 1.50 (0.02) 1.55 (0.02) 1.52 (0.02) 0.10 Model 4 1.56 (0.07) 1.61 (0.07) 1.59 (0.07) 0.11 Model 5 1.57 (0.07) 1.62 (0.06) 1.60 (0.06) 0.13 TC:HDL Model 1 4.0 (0.1) 4.0 (0.1) 4.1 (0.0) 0.36 Model 2 3.6 (0.1) 3.6 (0.1) 3.7 (0.1) 0.34 Model 3 3.6 (0.1) 3.6 (0.1) 3.7 (0.1) 0.52 Model 4 3.1 (0.2) 3.1 (0.2) 3.1 (0.2) 0.41 Model 5 3.1 (0.2) 3.0 (0.2) 3.1 (0.2) 0.53 Non HDL, mmol/l Model 1 3.54 (0.06) 3.58 (0.04) 3.62 (0.03) 0.37 Model 2 3.61 (0.07) 3.65 (0.06) 3.69 (0.05) 0.35 Model 3 3.57 (0.07) 3.63 (0.06) 3.66 (0.05) 0.33 Model 4 3.04 (0.20) 3.07 (0.19) 3.12 (0.19) 0.35 Model 5 3.03 (0.19) 3.06 (0.19) 3.10 (0.19) 0.39 HbA1c, % Model 1 5.60 (0.03) 5.59 (0.02) 5.58 (0.01) 0.86 Model 2 5.59 (0.03) 5.59 (0.03) 5.58 (0.03) 0.76 Model 3 5.60 (0.03) 5.60 (0.03) 5.57 (0.03) 0.63 Model 4 6.52 (0.10) 6.53 (0.10) 6.50 (0.10) 0.52 Model 5 6.51 (0.10) 6.52 (0.10) 6.49 (0.10) 0.44

!

! ! ""$!

A 8.7 continued Day Shift Night P Diastolic blood pressure, mmHg Model 1 77.7 (0.5) 78.6 (0.4) 78.2 (0.3) 0.42 Model 2 78.6 (0.7) 79.5 (0.6) 79.0 (0.5) 0.39 Model 3 78.1 (0.7) 79.2 (0.6) 78.7 (0.5) 0.30 Model 4 81.0 (1.9) 82.1 (1.9) 81.7 (1.9) 0.29 Model 5 80.7 (1.8) 81.9 (1.8) 81.4 (1.8) 0.18 Systolic blood pressure, mmHg Model 1 127.4 (0.8) 129.0 (0.6) 128.0 (0.4) 0.18 Model 2 127.7 (0.9) 129.2 (0.8) 128.1 (0.7)

0.23 Model 3 127.4 (0.9) 129.0 (0.8) 128.0 (0.7) 0.22 Model 4 130.5 (2.7) 132.1 (2.7) 131.3 (2.6) 0.22 Model 5 130.1 (2.6) 131.9 (2.6) 130.9 (2.5) 0.12 CRP, mg/L Model 1 0.99 (0.95) 0.94 (0.94) 0.95 (0.93) 0.77 Model 2 1.05 (0.96) 0.97 (0.95) 0.99 (0.95) 0.53 Model 3 1.01 (0.96) 0.95 (0.95) 0.96 (0.95) 0.62 Model 4 0.96 (0.95) 0.98 (1.10) 1.00 (1.09) 0.60 Model 5 1.02 (1.09) 0.96 (1.09) 0.97 (1.08) 0.63 *Pooled analyses men and women, shift work determined by police radio records 30days prior to health screen. Abbreviations HDL High density lipoprotein, TC total cholesterol, CRP high sensitivity C-reactive protein. ±CRP log transformed to allow parametric testing, untransformed values presented. Differences across shift types tested using general linear models. Model 1 adjusted for age and sex, Model 2 + physical activity, smoking, education, TV viewing, household income, job strain, menopause status. Model 3 + DASH score mean energy intake, alcohol (continuous variables). Model 4 + diagnosed ± treatment for diabetes, lipids or blood pressure. Model 5 + body mass index (continuous). § Bonferroni post hoc test, significant difference between all groups.

!

! ! ""%!

A8.1 Redesigned by time food diary used in the Airwave Health Monitoring Study 2014 onwards

Example of a completed diary record to help guide participants in the completion of the food diary.

!

!

! ! "&'!

A8.2 Amended food diary: pilot

Piloting of the amended food diary Methods: From the participants enrolled in the Health Monitoring Study

between 2004 and 2012 (who completed the initial food diary) 100 were

selected by random proportional stratified (gender and rank) sampling to be

included in the pilot study. Participants were contacted by telephone to

consent to undertake an additional food diary record. The amended 7-day

food diary was posted to participants with written instructions on how to

complete for the 24-hour period. Additionally participants were asked to

complete a short questionnaire about their current body weight and if they are

on a weight loss or other special diet. A stamped addressed envelope was

provided for diary and questionnaire return. The standard operating protocol

for food diary coding was adapted to include initial instructions of how to code

the amended food diaries by time (Figure A8.2.1).

Results: 67% consented to complete an additional food diary. From this

sample 40 (60%) returned the completed food diary. The mean difference

between initial health screen (proxy date for first food diary completion) was

19 months (range 17 – 23 months). All participants completed the food diary

for 7-days and recorded time of food intake. Only one participant who

returned the diary did not complete the shift work section.

The baseline food diary assessments were compared where available to the

new food diary assessments (n = 26). Mean eating occasions recorded in the

new food diary 4.3 (SD 0.8) compared to original food diary 3.7 (SD 0.6),

Wilcoxon Rank p <0.001. A Bland Altman plot was constructed to determine

the agreement between the two measures for energy intake reporting, Figure

A8.2.1. The mean difference between the two measures was 27.8kcal. From

the sample 88% of participants were within 2±SD.

341

Figure A8.2.1 Amendments made to standard coding protocol to code by time food diaries

Replaces section 2.0 in A3.1 (Standard coding procedure document)

5

Codingeatingoccasionsbytime

Eachmeal=1xeatingoccasion

Meal1=thefirsteatingoccasion.

E.g.The examplediaryon the leftwouldbecodedasfollows:

Meal 1: 05:00 Tea / semi skimmed milk /water

Meal2:08:00yoghurt

Meal3:13:00…etc…

!

! ! "&)!

!

!

!

!

!

!

!

!

!

!

!

!

!

!

!

!

!

!

!

!

!

!

!

!

!

!

!

!

! ! !

"!!

#$%$&'(!)*+,!'%-!+&./$-0&$,1!

! 2%!$')*%3!4!-&*%5*%3!.//',*.%!*,!/(',,*6*$-!',!'%7)8*%3!/.%,09$-!:*)8!')!($',)!'!;<!9*%0)$!3'+1!$=3=!!;>1;[email protected]$$!:*)8!9*(5!A!-*3$,)*B$!C*,/0*)!!;>1D<!-*3$,)*B$!C*,/0*)!!" E8*,!:.0(-!C$!/.-$-!',!.%$!F9$'(G!')!;>1;?!

!

;>1;[email protected]$$!:*)8!9*(5!A!-*3$,)*B$!C*,/0*)!!;>1H<!-*3$,)*B$!C*,/0*)!

" E8*,!:.0(-!C$!/.-$-!',!):.!!F9$'(,G!')!;>1;?!'%-!;>1H<!

!

! @0)!'%-!+',)$!60%/)*.%!:*((!/8'%3$!)8$!)*9$!).!-$6'0()!*%!)8$!9$'(!)8')!7.0!+',)$!*%).=!

! I.0!/'%!/8'%3$!)8$!)*9$!.6!'!9$'(!97!$%)$&*%3!'%.)8$&!/.-$!!:*)8!)8$!/.&&$/)!)*9$!J'%-!-$($)*%3!*)!*6!&$K0*&$-L!

! M$9$9C$&!).!,'B$!7.0&!:.&5!',!7.0!3.!).!'B.*-!'%7!-')'!(.,,=!! @.-*%3!%$$-,!).!C$!6.&!)8$!DN!8&!+$&*.-1!??1??!).!DH1<O!P!*6!'!+'&)*/*+'%)!&$/.&-,!

6..-,!'6)$&!??1??!J*=$=!6.&!)8$!%$Q)!-'7L!)8$7!90,)!C$!/.-$-!*%!)8$!6.((.:*%3!-'7=!! R)$9,!'&$!(*,)$-!*%!)8$!F'%7)8*%3!$(,$G!,$/)*.%!')!)8$!$%-!.6!$'/8!-'71!

o R6!/.%,09$-!')!)8$!,'9$!)*9$!',!'!9$'(!'(&$'-7!&$/.&-$-!P!'-!)8$!*)$9J,L!).!)8$!'++&.+&*')$!9$'(!.//',*.%!!

o R6!*)!*,!/.%,09$-!')!'!-*66$&$%)!)*9$!).!)8.,$!'(&$'-7!&$/.&-$-!)8$%!,$)!0+!'!%$:!9$'(!*%!)8$!%$Q)!'B'*('C($=!!2,!(.%3!',!)8$!9$'(!)*9$!*,!/.&&$/)!*)!:*((!$Q+.&)!*%!)8$!/.&&$/)!.&-$&=!

!!

!

!

!

!

!

!

!

!

! ! "&"!

A8.2.2 Bland Altman plot mean daily energy intake (kcal) measured by

baseline food diary and new (by time) food diary

!"###

!$##

#

$##

"###

"$##

# $## "%### "%$## &%### &%$## '%###

!"##$

%$&'$()$*+$$&(,$*-./0(1,$2&(3'245/267

8$2&(.#().*-(,$*-./0( 1,$2&(3'245/267

()*+,!&-./

0".12,34,/'#

!".12,34,!/'#

!!!

"##!

A8.3 D

ecision tree classification to determine com

mon shift type reported in a sub-group of participants from

the Airw

ave H

ealth Monitoring Study (n = 240). B

ranch terminated at 10 or less participants.

!

! ! #$%!

A8.4 Summary of frequency and mean length of shift categories recorded included for analyses Shift classification*

Length of shift (hours)

Number of participants recorded

Mean (SD) n (%)

Late 9.7 (2.2) 106 (44)

Early 9.9 (2.3) 24 (10)

Day 8.2 (1.0) 193 (80)

Rest day n/a n/a 233 (97)

Night –Night± 13.3 (2.1) 33 (14)

Abbreviations: SD standard deviation. *Refer to Table 8.1 (main text) for classification criteria. ±

Night shift length refers to mean start time less mean finish time for the complete shift.

!

! ! #$&!

A8.5 New shift work questionnaire

!

! ! #$'!

!

! ! #$(!

!

! ! #$)!

!

! ! #%+!

!

! ! #%*!

!

! ! #%"!

!

! ! #%#!

!

! ! #%$!

COPYRIGHT PERMISSIONS APPENDIX

Type of work

Name of work Source of work Copyright holder and contact

Permission requested on

Permission granted

Figure Figure 1.1 Schematic of the hypothesized causal pathway between visceral adiposity, type 2 diabetes and cardiovascular diseases

Matsuzawa Y. Therapy Insight: adipocytokines in metabolic syndrome and related cardiovascular disease. Nat Clin Pract Cardiovasc Med. 2006 Jan;3(1):35–42. (Adapted from Figure 6)

Copyright © 2006, Rights Managed by Nature Publishing Group

28.11.16 via RightsLink (no fee)

Yes – see agreement below, includes adaptation permission.

Image Figure 1.2 Proposed pathophysiology of the metabolic syndrome including inflammatory pathways

Eckel RH, Grundy SM, Zimmet PZ, Cameron A, Shaw J, Zimmet P, et al. The metabolic syndrome. Lancet , 2004;365(9468):1415–28. (Figure 2 panel A)

Copyright © 2005 Elsevier Ltd. All rights reserved.


Yes – see agreement below.

Figure Figure 1.4 Framework for the determinants of eating behaviour

Booth SL, Sallis JF, Ritenbaugh C, Hill JO, Birch LL, Frank LD, et al. Environmental and Societal Factors Affect Food Choice and Physical Activity: Rationale, Influences, and Leverage Points. Nutr Rev. 2001 Apr 27;59(3):S21–36. (Figure 1a)

Oxford University Press

© 2001 International Life Sciences Institute


Yes – see agreement below

Image and Table

Figure 2.1 Enrolment in the Airwave Health Monitoring Study per region by the end of 2012

Elliott P, Vergnaud A-C, Singh D, Neasham D, Spear J, Heard A. The Airwave Health Monitoring Study of police officers and staff in Great Britain: Rationale, design and methods. Environ Res. 2014 Sep 3;134C:280–5. (Figure 2b)

Copyright © 2014 The Authors. Published by Elsevier Inc.


Yes – see agreement below

NATURE PUBLISHING GROUP LICENSETERMS AND CONDITIONS

Mar 26, 2017

This Agreement between R Gibson ("You") and Nature Publishing Group ("NaturePublishing Group") consists of your license details and the terms and conditions provided byNature Publishing Group and Copyright Clearance Center.

License Number 3997660127013

License date

Licensed Content Publisher Nature Publishing Group

Licensed Content Publication Nature Reviews Cardiology

Licensed Content Title Therapy Insight: adipocytokines in metabolic syndrome and relatedcardiovascular disease

Licensed Content Author Yuji Matsuzawa

Licensed Content Date Jan 1, 2006

Licensed Content Volume 3

Licensed Content Issue 1

Type of Use reuse in a dissertation / thesis

Requestor type academic/educational

Format print and electronic

Portion figures/tables/illustrations

Number offigures/tables/illustrations

1

High-res required no

Figures Figure 6

Author of this NPG article no

Your reference number

Title of your thesis /dissertation

THE ASSOCIATION BETWEEN DIET AND WORKING HOURS WITHMARKERS OF CARDIOMETABOLIC HEALTH IN THE BRITISH POLICEFORCE

Expected completion date Dec 2016

Estimated size (number ofpages)

200

Requestor Location R GibsonImperial College

London, W12United KingdomAttn: R Gibson

Billing Type Invoice

Billing Address R GibsonImperial College

London, United Kingdom W12

Attn: R Gibson

Total 0.00 GBP

Terms and Conditions

Terms and Conditions for PermissionsNature Publishing Group hereby grants you a non-exclusive license to reproduce thismaterial for this purpose, and for no other use,subject to the conditions below:

1. NPG warrants that it has, to the best of its knowledge, the rights to license reuse of thismaterial. However, you should ensure that the material you are requesting is original toNature Publishing Group and does not carry the copyright of another entity (as credited inthe published version). If the credit line on any part of the material you have requestedindicates that it was reprinted or adapted by NPG with permission from another source,then you should also seek permission from that source to reuse the material.

2. Permission granted free of charge for material in print is also usually granted for anyelectronic version of that work, provided that the material is incidental to the work as awhole and that the electronic version is essentially equivalent to, or substitutes for, theprint version.Where print permission has been granted for a fee, separate permission mustbe obtained for any additional, electronic re-use (unless, as in the case of a full paper, thishas already been accounted for during your initial request in the calculation of a printrun).NB: In all cases, web-based use of full-text articles must be authorized separatelythrough the 'Use on a Web Site' option when requesting permission.

3. Permission granted for a first edition does not apply to second and subsequent editions andfor editions in other languages (except for signatories to the STM Permissions Guidelines,or where the first edition permission was granted for free).

4. Nature Publishing Group's permission must be acknowledged next to the figure, table orabstract in print. In electronic form, this acknowledgement must be visible at the sametime as the figure/table/abstract, and must be hyperlinked to the journal's homepage.

5. The credit line should read:Reprinted by permission from Macmillan Publishers Ltd: [JOURNAL NAME] (referencecitation), copyright (year of publication)For AOP papers, the credit line should read:Reprinted by permission from Macmillan Publishers Ltd: [JOURNAL NAME], advance onlinepublication, day month year (doi: 10.1038/sj.[JOURNAL ACRONYM].XXXXX)

Note: For republication from the British Journal of Cancer, the following creditlines apply.Reprinted by permission from Macmillan Publishers Ltd on behalf of Cancer Research UK:[JOURNAL NAME] (reference citation), copyright (year of publication)For AOP papers, thecredit line should read:Reprinted by permission from Macmillan Publishers Ltd on behalf of Cancer Research UK:[JOURNAL NAME], advance online publication, day month year (doi: 10.1038/sj.[JOURNALACRONYM].XXXXX)

6. Adaptations of single figures do not require NPG approval. However, the adaptation should

be credited as follows:

Adapted by permission from Macmillan Publishers Ltd: [JOURNAL NAME] (referencecitation), copyright (year of publication)

Note: For adaptation from the British Journal of Cancer, the following credit lineapplies.Adapted by permission from Macmillan Publishers Ltd on behalf of Cancer Research UK:[JOURNAL NAME] (reference citation), copyright (year of publication)

7. Translations of 401 words up to a whole article require NPG approval. Please visithttp://www.macmillanmedicalcommunications.com for more information.Translations of up

http://www.macmillanmedicalcommunications.com/

to a 400 words do not require NPG approval. The translation should be credited as follows:

Translated by permission from Macmillan Publishers Ltd: [JOURNAL NAME] (referencecitation), copyright (year of publication).

Note: For translation from the British Journal of Cancer, the following credit lineapplies.Translated by permission from Macmillan Publishers Ltd on behalf of Cancer Research UK:[JOURNAL NAME] (reference citation), copyright (year of publication)

We are certain that all parties will benefit from this agreement and wish you the best in theuse of this material. Thank you.Special Terms:v1.1Questions? [email protected] or +1-855-239-3415 (toll free in the US) or+1-978-646-2777.

mailto:[email protected]

ELSEVIER LICENSETERMS AND CONDITIONS

Mar 26, 2017

This Agreement between R Gibson ("You") and Elsevier ("Elsevier") consists of yourlicense details and the terms and conditions provided by Elsevier and Copyright ClearanceCenter.


License date

Licensed Content Publisher Elsevier

Licensed Content Publication The Lancet

Licensed Content Title The metabolic syndrome

Licensed Content Author Robert H Eckel,Scott M Grundy,Paul Z Zimmet

Licensed Content Date 16–22 April 2005


Licensed Content Issue 9468

Licensed Content Pages 14

Start Page 1415

End Page 1428

Type of Use reuse in a thesis/dissertation



1

Format both print and electronic

Are you the author of thisElsevier article?

No

Will you be translating? No

Order reference number

Original figure numbers Figure 2

Title of yourthesis/dissertation




200

Elsevier VAT number GB 494 6272 12



Total 0.00 GBP


INTRODUCTION1. The publisher for this copyrighted material is Elsevier. By clicking "accept" inconnection with completing this licensing transaction, you agree that the following termsand conditions apply to this transaction (along with the Billing and Payment terms andconditions established by Copyright Clearance Center, Inc. ("CCC"), at the time that youopened your Rightslink account and that are available at any time athttp://myaccount.copyright.com).

GENERAL TERMS2. Elsevier hereby grants you permission to reproduce the aforementioned material subject tothe terms and conditions indicated.3. Acknowledgement: If any part of the material to be used (for example, figures) hasappeared in our publication with credit or acknowledgement to another source, permissionmust also be sought from that source. If such permission is not obtained then that materialmay not be included in your publication/copies. Suitable acknowledgement to the sourcemust be made, either as a footnote or in a reference list at the end of your publication, asfollows:"Reprinted from Publication title, Vol /edition number, Author(s), Title of article / title ofchapter, Pages No., Copyright (Year), with permission from Elsevier [OR APPLICABLESOCIETY COPYRIGHT OWNER]." Also Lancet special credit - "Reprinted from TheLancet, Vol. number, Author(s), Title of article, Pages No., Copyright (Year), withpermission from Elsevier."4. Reproduction of this material is confined to the purpose and/or media for whichpermission is hereby given.5. Altering/Modifying Material: Not Permitted. However figures and illustrations may bealtered/adapted minimally to serve your work. Any other abbreviations, additions, deletionsand/or any other alterations shall be made only with prior written authorization of ElsevierLtd. (Please contact Elsevier at [email protected])6. If the permission fee for the requested use of our material is waived in this instance,please be advised that your future requests for Elsevier materials may attract a fee.7. Reservation of Rights: Publisher reserves all rights not specifically granted in thecombination of (i) the license details provided by you and accepted in the course of thislicensing transaction, (ii) these terms and conditions and (iii) CCC's Billing and Paymentterms and conditions.8. License Contingent Upon Payment: While you may exercise the rights licensedimmediately upon issuance of the license at the end of the licensing process for thetransaction, provided that you have disclosed complete and accurate details of your proposeduse, no license is finally effective unless and until full payment is received from you (eitherby publisher or by CCC) as provided in CCC's Billing and Payment terms and conditions. Iffull payment is not received on a timely basis, then any license preliminarily granted shall bedeemed automatically revoked and shall be void as if never granted. Further, in the eventthat you breach any of these terms and conditions or any of CCC's Billing and Paymentterms and conditions, the license is automatically revoked and shall be void as if nevergranted. Use of materials as described in a revoked license, as well as any use of thematerials beyond the scope of an unrevoked license, may constitute copyright infringementand publisher reserves the right to take any and all action to protect its copyright in thematerials.9. Warranties: Publisher makes no representations or warranties with respect to the licensedmaterial.10. Indemnity: You hereby indemnify and agree to hold harmless publisher and CCC, andtheir respective officers, directors, employees and agents, from and against any and allclaims arising out of your use of the licensed material other than as specifically authorizedpursuant to this license.

http://myaccount.copyright.com/

11. No Transfer of License: This license is personal to you and may not be sublicensed,assigned, or transferred by you to any other person without publisher's written permission.12. No Amendment Except in Writing: This license may not be amended except in a writingsigned by both parties (or, in the case of publisher, by CCC on publisher's behalf).13. Objection to Contrary Terms: Publisher hereby objects to any terms contained in anypurchase order, acknowledgment, check endorsement or other writing prepared by you,which terms are inconsistent with these terms and conditions or CCC's Billing and Paymentterms and conditions. These terms and conditions, together with CCC's Billing and Paymentterms and conditions (which are incorporated herein), comprise the entire agreementbetween you and publisher (and CCC) concerning this licensing transaction. In the event ofany conflict between your obligations established by these terms and conditions and thoseestablished by CCC's Billing and Payment terms and conditions, these terms and conditionsshall control.14. Revocation: Elsevier or Copyright Clearance Center may deny the permissions describedin this License at their sole discretion, for any reason or no reason, with a full refund payableto you. Notice of such denial will be made using the contact information provided by you. Failure to receive such notice will not alter or invalidate the denial. In no event will Elsevieror Copyright Clearance Center be responsible or liable for any costs, expenses or damageincurred by you as a result of a denial of your permission request, other than a refund of theamount(s) paid by you to Elsevier and/or Copyright Clearance Center for deniedpermissions.

LIMITED LICENSEThe following terms and conditions apply only to specific license types:15. Translation: This permission is granted for non-exclusive world English rights onlyunless your license was granted for translation rights. If you licensed translation rights youmay only translate this content into the languages you requested. A professional translatormust perform all translations and reproduce the content word for word preserving theintegrity of the article.16. Posting licensed content on any Website: The following terms and conditions apply asfollows: Licensing material from an Elsevier journal: All content posted to the web site mustmaintain the copyright information line on the bottom of each image; A hyper-text must beincluded to the Homepage of the journal from which you are licensing athttp://www.sciencedirect.com/science/journal/xxxxx or the Elsevier homepage for books athttp://www.elsevier.com; Central Storage: This license does not include permission for ascanned version of the material to be stored in a central repository such as that provided byHeron/XanEdu.Licensing material from an Elsevier book: A hyper-text link must be included to the Elsevierhomepage at http://www.elsevier.com . All content posted to the web site must maintain thecopyright information line on the bottom of each image.

Posting licensed content on Electronic reserve: In addition to the above the followingclauses are applicable: The web site must be password-protected and made available only tobona fide students registered on a relevant course. This permission is granted for 1 year only.You may obtain a new license for future website posting.17. For journal authors: the following clauses are applicable in addition to the above:Preprints:A preprint is an author's own write-up of research results and analysis, it has not been peer-reviewed, nor has it had any other value added to it by a publisher (such as formatting,copyright, technical enhancement etc.).Authors can share their preprints anywhere at any time. Preprints should not be added to orenhanced in any way in order to appear more like, or to substitute for, the final versions ofarticles however authors can update their preprints on arXiv or RePEc with their AcceptedAuthor Manuscript (see below).If accepted for publication, we encourage authors to link from the preprint to their formal

http://www.sciencedirect.com/science/journal/xxxxx

http://www.elsevier.com/


OXFORD UNIVERSITY PRESS LICENSETERMS AND CONDITIONS

Mar 26, 2017

This Agreement between R Gibson ("You") and Oxford University Press ("OxfordUniversity Press") consists of your license details and the terms and conditions provided byOxford University Press and Copyright Clearance Center.


License date

Licensed content publisher Oxford University Press

Licensed content publication Nutrition Reviews

Licensed content title Environmental and Societal Factors Affect Food Choice and PhysicalActivity: Rationale, Influences, and Leverage Points

Licensed content author Sarah L. Booth,James F. Sallis,Cheryl Ritenbaugh,James O.Hill,Leann L. Birch,Lawrence D. Frank,Karen Glanz,David A.Himmelgreen,Michael Mudd,Barry M. Popkin,Karyl A.Rickard,Sachiko St. Jeor,Nicholas P. Hays

Licensed content date 2001-03-01

Type of Use Thesis/Dissertation

Institution name

Title of your work THE ASSOCIATION BETWEEN DIET AND WORKING HOURS WITHMARKERS OF CARDIOMETABOLIC HEALTH IN THE BRITISH POLICEFORCE

Publisher of your work n/a

Expected publication date Dec 2016

Permissions cost 0.00 GBP

Value added tax 0.00 GBP

Total 0.00 GBP



Publisher Tax ID GB125506730

Billing Type Invoice

Billing Address R GibsonImperial College

London, United Kingdom W12Attn: R Gibson

Total 0.00 GBP


STANDARD TERMS AND CONDITIONS FOR REPRODUCTION OF MATERIAL

FROM AN OXFORD UNIVERSITY PRESS JOURNAL1. Use of the material is restricted to the type of use specified in your order details.2. This permission covers the use of the material in the English language in the followingterritory: world. If you have requested additional permission to translate this material, theterms and conditions of this reuse will be set out in clause 12.3. This permission is limited to the particular use authorized in (1) above and does not allowyou to sanction its use elsewhere in any other format other than specified above, nor does itapply to quotations, images, artistic works etc that have been reproduced from other sourceswhich may be part of the material to be used.4. No alteration, omission or addition is made to the material without our written consent.Permission must be re-cleared with Oxford University Press if/when you decide to reprint.5. The following credit line appears wherever the material is used: author, title, journal, year,volume, issue number, pagination, by permission of Oxford University Press or thesponsoring society if the journal is a society journal. Where a journal is being published onbehalf of a learned society, the details of that society must be included in the credit line.6. For the reproduction of a full article from an Oxford University Press journal for whateverpurpose, the corresponding author of the material concerned should be informed of theproposed use. Contact details for the corresponding authors of all Oxford University Pressjournal contact can be found alongside either the abstract or full text of the article concerned,accessible from www.oxfordjournals.org Should there be a problem clearing these rights,please contact [email protected]. If the credit line or acknowledgement in our publication indicates that any of the figures,images or photos was reproduced, drawn or modified from an earlier source it will benecessary for you to clear this permission with the original publisher as well. If thispermission has not been obtained, please note that this material cannot be included in yourpublication/photocopies.8. While you may exercise the rights licensed immediately upon issuance of the license atthe end of the licensing process for the transaction, provided that you have disclosedcomplete and accurate details of your proposed use, no license is finally effective unless anduntil full payment is received from you (either by Oxford University Press or by CopyrightClearance Center (CCC)) as provided in CCC's Billing and Payment terms and conditions. Iffull payment is not received on a timely basis, then any license preliminarily granted shall bedeemed automatically revoked and shall be void as if never granted. Further, in the eventthat you breach any of these terms and conditions or any of CCC's Billing and Paymentterms and conditions, the license is automatically revoked and shall be void as if nevergranted. Use of materials as described in a revoked license, as well as any use of thematerials beyond the scope of an unrevoked license, may constitute copyright infringementand Oxford University Press reserves the right to take any and all action to protect itscopyright in the materials.9. This license is personal to you and may not be sublicensed, assigned or transferred by youto any other person without Oxford University Press’s written permission.10. Oxford University Press reserves all rights not specifically granted in the combination of(i) the license details provided by you and accepted in the course of this licensingtransaction, (ii) these terms and conditions and (iii) CCC’s Billing and Payment terms andconditions.11. You hereby indemnify and agree to hold harmless Oxford University Press and CCC, andtheir respective officers, directors, employs and agents, from and against any and all claimsarising out of your use of the licensed material other than as specifically authorized pursuantto this license.12. Other Terms and Conditions:v1.4Questions? [email protected] or +1-855-239-3415 (toll free in the US) or+1-978-646-2777.

mailto:[email protected]

ELSEVIER LICENSETERMS AND CONDITIONS

Mar 26, 2017

This Agreement between R Gibson ("You") and Elsevier ("Elsevier") consists of yourlicense details and the terms and conditions provided by Elsevier and Copyright ClearanceCenter.


License date

Licensed Content Publisher Elsevier

Licensed Content Publication Environmental Research

Licensed Content Title The Airwave Health Monitoring Study of police officers and staff inGreat Britain: Rationale, design and methods

Licensed Content Author Paul Elliott,Anne-Claire Vergnaud,Deepa Singh,DavidNeasham,Jeanette Spear,Andy Heard

Licensed Content Date October 2014


Licensed Content Issue n/a

Licensed Content Pages 6

Start Page 280

End Page 285

Type of Use reuse in a thesis/dissertation



2

Format both print and electronic

Are you the author of thisElsevier article?

No

Will you be translating? No

Order reference number

Original figure numbers Figure 2(b) and Supp Table 6

Title of yourthesis/dissertation




200

Elsevier VAT number GB 494 6272 12



Total 0.00 GBP


INTRODUCTION1. The publisher for this copyrighted material is Elsevier. By clicking "accept" inconnection with completing this licensing transaction, you agree that the following termsand conditions apply to this transaction (along with the Billing and Payment terms andconditions established by Copyright Clearance Center, Inc. ("CCC"), at the time that youopened your Rightslink account and that are available at any time athttp://myaccount.copyright.com).

GENERAL TERMS2. Elsevier hereby grants you permission to reproduce the aforementioned material subject tothe terms and conditions indicated.3. Acknowledgement: If any part of the material to be used (for example, figures) hasappeared in our publication with credit or acknowledgement to another source, permissionmust also be sought from that source. If such permission is not obtained then that materialmay not be included in your publication/copies. Suitable acknowledgement to the sourcemust be made, either as a footnote or in a reference list at the end of your publication, asfollows:"Reprinted from Publication title, Vol /edition number, Author(s), Title of article / title ofchapter, Pages No., Copyright (Year), with permission from Elsevier [OR APPLICABLESOCIETY COPYRIGHT OWNER]." Also Lancet special credit - "Reprinted from TheLancet, Vol. number, Author(s), Title of article, Pages No., Copyright (Year), withpermission from Elsevier."4. Reproduction of this material is confined to the purpose and/or media for whichpermission is hereby given.5. Altering/Modifying Material: Not Permitted. However figures and illustrations may bealtered/adapted minimally to serve your work. Any other abbreviations, additions, deletionsand/or any other alterations shall be made only with prior written authorization of ElsevierLtd. (Please contact Elsevier at [email protected])6. If the permission fee for the requested use of our material is waived in this instance,please be advised that your future requests for Elsevier materials may attract a fee.7. Reservation of Rights: Publisher reserves all rights not specifically granted in thecombination of (i) the license details provided by you and accepted in the course of thislicensing transaction, (ii) these terms and conditions and (iii) CCC's Billing and Paymentterms and conditions.8. License Contingent Upon Payment: While you may exercise the rights licensedimmediately upon issuance of the license at the end of the licensing process for thetransaction, provided that you have disclosed complete and accurate details of your proposeduse, no license is finally effective unless and until full payment is received from you (eitherby publisher or by CCC) as provided in CCC's Billing and Payment terms and conditions. Iffull payment is not received on a timely basis, then any license preliminarily granted shall bedeemed automatically revoked and shall be void as if never granted. Further, in the eventthat you breach any of these terms and conditions or any of CCC's Billing and Paymentterms and conditions, the license is automatically revoked and shall be void as if nevergranted. Use of materials as described in a revoked license, as well as any use of thematerials beyond the scope of an unrevoked license, may constitute copyright infringementand publisher reserves the right to take any and all action to protect its copyright in thematerials.9. Warranties: Publisher makes no representations or warranties with respect to the licensedmaterial.10. Indemnity: You hereby indemnify and agree to hold harmless publisher and CCC, andtheir respective officers, directors, employees and agents, from and against any and allclaims arising out of your use of the licensed material other than as specifically authorized

http://myaccount.copyright.com/

pursuant to this license.11. No Transfer of License: This license is personal to you and may not be sublicensed,assigned, or transferred by you to any other person without publisher's written permission.12. No Amendment Except in Writing: This license may not be amended except in a writingsigned by both parties (or, in the case of publisher, by CCC on publisher's behalf).13. Objection to Contrary Terms: Publisher hereby objects to any terms contained in anypurchase order, acknowledgment, check endorsement or other writing prepared by you,which terms are inconsistent with these terms and conditions or CCC's Billing and Paymentterms and conditions. These terms and conditions, together with CCC's Billing and Paymentterms and conditions (which are incorporated herein), comprise the entire agreementbetween you and publisher (and CCC) concerning this licensing transaction. In the event ofany conflict between your obligations established by these terms and conditions and thoseestablished by CCC's Billing and Payment terms and conditions, these terms and conditionsshall control.14. Revocation: Elsevier or Copyright Clearance Center may deny the permissions describedin this License at their sole discretion, for any reason or no reason, with a full refund payableto you. Notice of such denial will be made using the contact information provided by you. Failure to receive such notice will not alter or invalidate the denial. In no event will Elsevieror Copyright Clearance Center be responsible or liable for any costs, expenses or damageincurred by you as a result of a denial of your permission request, other than a refund of theamount(s) paid by you to Elsevier and/or Copyright Clearance Center for deniedpermissions.

LIMITED LICENSEThe following terms and conditions apply only to specific license types:15. Translation: This permission is granted for non-exclusive world English rights onlyunless your license was granted for translation rights. If you licensed translation rights youmay only translate this content into the languages you requested. A professional translatormust perform all translations and reproduce the content word for word preserving theintegrity of the article.16. Posting licensed content on any Website: The following terms and conditions apply asfollows: Licensing material from an Elsevier journal: All content posted to the web site mustmaintain the copyright information line on the bottom of each image; A hyper-text must beincluded to the Homepage of the journal from which you are licensing athttp://www.sciencedirect.com/science/journal/xxxxx or the Elsevier homepage for books athttp://www.elsevier.com; Central Storage: This license does not include permission for ascanned version of the material to be stored in a central repository such as that provided byHeron/XanEdu.Licensing material from an Elsevier book: A hyper-text link must be included to the Elsevierhomepage at http://www.elsevier.com . All content posted to the web site must maintain thecopyright information line on the bottom of each image.

Posting licensed content on Electronic reserve: In addition to the above the followingclauses are applicable: The web site must be password-protected and made available only tobona fide students registered on a relevant course. This permission is granted for 1 year only.You may obtain a new license for future website posting.17. For journal authors: the following clauses are applicable in addition to the above:Preprints:A preprint is an author's own write-up of research results and analysis, it has not been peer-reviewed, nor has it had any other value added to it by a publisher (such as formatting,copyright, technical enhancement etc.).Authors can share their preprints anywhere at any time. Preprints should not be added to orenhanced in any way in order to appear more like, or to substitute for, the final versions ofarticles however authors can update their preprints on arXiv or RePEc with their AcceptedAuthor Manuscript (see below).

http://www.sciencedirect.com/science/journal/xxxxx



allowed to download and post the published electronic version of your chapter, nor may youscan the printed edition to create an electronic version. Posting to a repository: Authors arepermitted to post a summary of their chapter only in their institution's repository.19. Thesis/Dissertation: If your license is for use in a thesis/dissertation your thesis may besubmitted to your institution in either print or electronic form. Should your thesis bepublished commercially, please reapply for permission. These requirements includepermission for the Library and Archives of Canada to supply single copies, on demand, ofthe complete thesis and include permission for Proquest/UMI to supply single copies, ondemand, of the complete thesis. Should your thesis be published commercially, pleasereapply for permission. Theses and dissertations which contain embedded PJAs as part ofthe formal submission can be posted publicly by the awarding institution with DOI linksback to the formal publications on ScienceDirect. Elsevier Open Access Terms and ConditionsYou can publish open access with Elsevier in hundreds of open access journals or in nearly2000 established subscription journals that support open access publishing. Permitted thirdparty re-use of these open access articles is defined by the author's choice of CreativeCommons user license. See our open access license policy for more information.Terms & Conditions applicable to all Open Access articles published with Elsevier:Any reuse of the article must not represent the author as endorsing the adaptation of thearticle nor should the article be modified in such a way as to damage the author's honour orreputation. If any changes have been made, such changes must be clearly indicated.The author(s) must be appropriately credited and we ask that you include the end userlicense and a DOI link to the formal publication on ScienceDirect.If any part of the material to be used (for example, figures) has appeared in our publicationwith credit or acknowledgement to another source it is the responsibility of the user toensure their reuse complies with the terms and conditions determined by the rights holder.Additional Terms & Conditions applicable to each Creative Commons user license:CC BY: The CC-BY license allows users to copy, to create extracts, abstracts and newworks from the Article, to alter and revise the Article and to make commercial use of theArticle (including reuse and/or resale of the Article by commercial entities), provided theuser gives appropriate credit (with a link to the formal publication through the relevantDOI), provides a link to the license, indicates if changes were made and the licensor is notrepresented as endorsing the use made of the work. The full details of the license areavailable at http://creativecommons.org/licenses/by/4.0.CC BY NC SA: The CC BY-NC-SA license allows users to copy, to create extracts,abstracts and new works from the Article, to alter and revise the Article, provided this is notdone for commercial purposes, and that the user gives appropriate credit (with a link to theformal publication through the relevant DOI), provides a link to the license, indicates ifchanges were made and the licensor is not represented as endorsing the use made of thework. Further, any new works must be made available on the same conditions. The fulldetails of the license are available at http://creativecommons.org/licenses/by-nc-sa/4.0.CC BY NC ND: The CC BY-NC-ND license allows users to copy and distribute the Article,provided this is not done for commercial purposes and further does not permit distribution ofthe Article if it is changed or edited in any way, and provided the user gives appropriatecredit (with a link to the formal publication through the relevant DOI), provides a link to thelicense, and that the licensor is not represented as endorsing the use made of the work. Thefull details of the license are available at http://creativecommons.org/licenses/by-nc-nd/4.0.Any commercial reuse of Open Access articles published with a CC BY NC SA or CC BYNC ND license requires permission from Elsevier and will be subject to a fee.Commercial reuse includes:

- Associating advertising with the full text of the Article- Charging fees for document delivery or access

http://www.elsevier.com/about/open-access/open-access-policies/oa-license-policy

http://creativecommons.org/licenses/by/4.0

http://creativecommons.org/licenses/by-nc-sa/4.0

http://creativecommons.org/licenses/by-nc-nd/4.0

the association between diet and working hours with

Documents