The Epidemiology of Diabetes among Immigrants to Ontario

by

Maria Isabella Creatore

A thesis submitted in conformity with the requirements for the degree of Doctor of Philosophy

Institute of Medical Science University of Toronto

© Copyright by Maria I. Creatore (2013)

ii

The Epidemiology of Diabetes Among Immigrants to Ontario

Maria I. Creatore

Doctor of Philosophy (PhD) in Epidemiology

Institute of Medical Sciences University of Toronto

2013

Abstract

Type 2 Diabetes Mellitus (T2DM) prevalence is increasing globally with roughly 2.4

million people currently living with this condition in Canada. T2DM occurs more

commonly in non-European ethnoracial groups, however the distribution of risk by age,

sex, ethnicity and immigration status in Canada are not completely understood.

The purpose of this thesis is to investigate the epidemiology of diabetes in an

immigrant, multi-ethnic population using linked immigration and health data for the

province of Ontario. The ultimate goal of this work is to generate information that can be

used to design appropriate and effective targeted programs for diabetes prevention,

management and control in order to reduce inequities in health.

The principal findings of this work indicate that:

1) South Asians had a three-fold higher risk for developing diabetes as compared with

people of European ethnicity and this disparity in risk was evident at a very young age;

2) The young age at diabetes onset experienced by many of our high-risk ethnic groups,

including South Asians and people of African and Middle Eastern descent, suggest that

iii

in order to capture an equivalent risk of disease, screening may be recommended up to

15 years earlier in these groups – which is not reflected in current screening guidelines;

3) Contrary to patterns seen in Western European populations, women belonging to

many high–risk ethnicities had equivalent or, in some cases, higher risk than men;

4) Risk varied substantially across country and region of birth making broad definitions

of race or ethnicity (eg. ‘Asian’ or ‘Black’) inappropriate.

These findings emphasize the heterogeneity of risk experienced by different ethnoracial

populations in Canada and suggest that targeted primary prevention programs aimed at

young adults and adolescents belonging to high-risk ethnic groups may be warranted. In

addition, screening guidelines may need to be updated to reflect the younger age at

onset in these populations. Further research is necessary to identify culturally

appropriate and effective programs to reduce diabetes risk and associated health

problems in these populations.

iv

Acknowledgments

There are many individuals whom I would like to acknowledge for their role in supporting

me in the creation and completion of this thesis.

I would like to thank my supervisor, Dr. Richard Glazier for his constant support and

mentorship and his patience and faith in me over these many years. Throughout my

PhD journey, juggling work, school, family and children, he was always flexible and

ready to suggest creative solutions for how I could manage to continue with my studies

while accommodating the rest of life. I would also like to thank my other committee

members – Drs. Gillian Booth, Doug Manuel and Rahim Moineddin - for their thoughtful

comments, guidance and support throughout this process. In particular I wish to thank

Dr. Rahim Moineddin who patiently supervised and advised me on all my analyses, was

always available and full of encouragement and from whom I have learned so much.

I would also like to thank my friends and colleagues at the Centre for Research on Inner

City Health (CRICH) at St.Michael’s Hospital and the Institute for Clinical Evaluative

Sciences (ICES) for their encouragement and frequent morale boosts. In particular I

would like to thank Flora Matheson, Peter Gozdyra, Jonathan Weyman, Jim Dunn,

Chaim Bell, Joel Ray, Mohammad Agha, Anne-Marie Tynan, Donna Hoppenheim, Jane

Polsky, Marcelo Urquia and Hadas Fischer.

There are many people whose love, support and encouragement made this work

possible. First and foremost I owe my gratitude to my wonderful husband whose

unconditional love and support makes all things possible. My children, Jacob and Anna

Sofia inspired me daily to work hard and do my best to make them proud. Eternal

gratitude to my parents, Cheryl and Giuseppe Creatore, who have always had

unwavering faith in me and have supported me in everything I have ever attempted. And

loving thanks to the rest of my family: Karina, Bryan, Bill, Christine, Myra and Frank.

v

Contributions

I was responsible for the design, analysis, manuscript writing and preparation however

a number of people contributed to the completion of this thesis.

Dr. Richard Glazier was my supervisor and provided feedback on study design,

analysis, and interpretation of findings as well as feedback on the chapter and

manuscript drafts.

Committee members Dr. Doug Manuel and Dr. Gillian Booth provided feedback on

study design, analysis, and interpretation of findings as well as feedback on the chapter

and manuscript drafts. In addition to providing feedback on study design, interpretation

and writing, committee member Dr. Rahim Moineddin supervised all statistical analyses.

Alexander Kopp and Nadia Gunraj at the Institute for Clinical Evaluative Sciences

prepared the datasets.

Marie DesMeules and Sarah McDermott contributed to the acquisition of data and

creation of the linked datasets.

Drs. Jeffrey Johnson, Lorraine Lipscombe and Jan Hux acted as external reviewers of

this work, and Dr. Johnson provided valuable written comments and suggestions.

I would also like to acknowledge the Public Health Agency of Canada (PHAC) and

Citizenship and Immigration Canada (CIC) for their contribution of the data and support

of this study.

vi

Table of Contents

Table of Contents .................................................................................................... v

List of Tables ......................................................................................................... ix

List of Figures ........................................................................................................ xi

List of Appendices .................................................................................................. xii

Chapter 1 Introduction ........................................................................................... 1

1.1 Thesis Overview .............................................................................................. 2

1.2 Background ...................................................................................................... 3

1.3 Theoretical Framework .................................................................................... 19

1.4 Research Questions and Rationale for the Objectives ..................................... 22

Chapter 2 Age and Sex Patterns of Diabetes Among Immigrants to Ontario ........ 27

Abstract .................................................................................................................. 28

2.1 Introduction ...................................................................................................... 30

2.2 Research Design and Methods ........................................................................ 31

2.2.1 Data Sources and Study Population ............................................. 31

2.2.2 Statistical Analysis ........................................................................ 33

2.3 Results ............................................................................................................. 34

2.3.1 Characteristics of the Study Population ........................................ 34

vii

2.3.2 Trends in Diabetes Prevalence ..................................................... 36

2.4 Discussion ........................................................................................................ 44

Chapter 3 Diabetes Screening Among Immigrants: A Population-Based Urban

Cohort Study .......................................................................................................... 49

Abstract ................................................................................................................ 50

3.1 Introduction ...................................................................................................... 52

3.2 Research Design and Methods ........................................................................ 54

3.2.1 Study Population ........................................................................... 54

3.2.2 Study Outcomes ........................................................................... 55

3.2.3 Statistical Analyses ....................................................................... 57

3.3 Results ............................................................................................................. 59

3.3.1 Baseline Study Characteristics ..................................................... 59

3.3.2 Diabetes Screening ....................................................................... 61

3.3.3 Screening Efficiency ..................................................................... 61

3.3.4 Predictors of Diabetes Screening .................................................. 64

3.3.5 Undiagnosed Diabetes .................................................................. 67

3.4 Discussion ........................................................................................................ 69

Chapter 4 A Population-Based Study of Diabetes Incidence by Ethnicity and Age:

Support for the Development of Ethnic-Specific Age Guidelines for Screening ..... 75

viii

Abstract ................................................................................................................ 76

4.1 Introduction ...................................................................................................... 78

4.2 Research Design and Methods ........................................................................ 80

4.2.1 Study Design ................................................................................ 80

4.2.2 Study Population ........................................................................... 80

4.2.3 Measures ...................................................................................... 82

4.2.4 Study Outcomes ........................................................................... 83

4.2.5 Analysis ........................................................................................ 84

4.3 Results ............................................................................................................. 87

4.4 Discussion ........................................................................................................ 96

Chapter 5 Discussion ............................................................................................ 104

5.1 Main Findings ................................................................................................... 105

5.2 Research Implications for Policy and Practice ................................................. 106

5.3 Interpretation of Findings in the Context of Our Theoretical Framework .......... 112

5.4 Limitations ........................................................................................................ 118

5.5 Unanswered Questions and Future Research ................................................. 123

5.6 Conclusions ..................................................................................................... 127

References ............................................................................................................ 130

Appendices ............................................................................................................ 157

ix

List of Tables

Table 2.1 Baseline characteristics of the Ontario long-term resident and recent

immigrant study populations, 2005 ........................................................................ 35

Table 3.1 Baseline characteristics of the urban Ontario general population

(excluding immigrant cohort) and immigrant study populations†, aged 40 and up

and diabetes-free on April 1, 2004 ......................................................................... 60

Table 3.2 Measures of screening uptake and efficiency: The number and percent

of the study population (overall, and by sex) with no previous diagnosis of diabetes,

having a laboratory test to screen for diabetes in the 3-year period, 2004-2007;

the proportion of those screened with newly diagnosed diabetes (screening

efficiency); and the number needed to screen to identify one new case ................ 62

Table 3.3 Predictors of receiving a diabetes screen test during the 3-year study

period (April 1, 2004 - March 31, 2007): results of regression analyses. Study

population limited to immigrants without prior diagnosed diabetes, aged 40 and

over ................................................................................................................ 65

Table 3.4 Estimated number and percentage of 'undiagnosed' diabetes cases by

world region and immigration status, 2004-2007, among those aged 40 and up

with no prior diabetes diagnosis on April 1, 2004 ................................................... 68

Table 4.1 Baseline characteristics of the Ontario long-term residence and recent

immigrant diabetes-free study populations, by world region of birth ...................... 88

Table 4.2 Diabetes incidence rates over a 5-year follow-up period by primary

covariates, age and sex ......................................................................................... 90

Table 4.3 Age of equivalent diabetes risk by ethnicity. The risk experience by

men aged 40 of Western European ethnicity was used as the standard for

comparison ............................................................................................................ 95

x

Tables 5.1 Summary of the elements of the Theoretical Model and how they relate

to the current work ................................................................................................. 128

xi

List of Figures

Figure 1.1 World Health Organization’s Commission on the Social Determinants

of Health Conceptual Framework ........................................................................... 21

Figure 2.1 Age-adjusted, sex-specific diabetes prevalence rates (2005) by world

region of origin for immigrants (1985-2000) and long-term Ontario residents ........ 37

Figure 2.2 Age-adjusted diabetes prevalence (per 100) with 95% confidence

intervals among immigrants to Ontario (1985-2000), comparing the 15 countries

with the highest prevalence in 2005 with Ontario’s prevalence in 2005/06 ............ 38

Figure 2.3 Age-specific diabetes prevalence rates for immigrants and long-term

residents by sex, 2005 ........................................................................................... 39

Figure 2.4 Age-specific diabetes prevalence rates by WRO (males), 2005 ........... 41

Figure 2.5 Age-specific diabetes prevalence rates by WRO (females), 2005 ........ 42

Figure 2.6 Risk factors for diabetes (2005) among immigrants to Ontario (1985-

2000) by sex .......................................................................................................... 43

Figure 4.1 Adjusted average diabetes incidence rates by ethnicity and age, men

(1994-2008) ........................................................................................................... 93

Figure 4.2 Adjusted average diabetes incidence rates by ethnicity and age,

women (1994-2008) ............................................................................................... 94

Figure 5.1 The World Health Organization’s Commission on the Social

Determinants of Health conceptual framework ...................................................... 113

xii

List of Appendices

Appendix A. Data sources ................................................................................... 158

Appendix B. ODD inclusion schema .................................................................... 163

Appendix C. Countries included in the Citizenship and Immigration Canada

database and the assigned World Region of Origin ............................................... 164

Appendix D. Incidence cohort creation schema ................................................... 169

Appendix E. Age-Period-Cohort effects ............................................................... 170

Appendix F. Detailed Methodology of the Cox Proportional Hazard model ......... 172

Appendix G. Characteristics of the Ontario long-term resident and recent

immigrant study populations, as well as those excluded due to prior diabetes

diagnosis, 2010 ...................................................................................................... 183

Appendix H. Diabetes incidence rates before and after restriction to those who

have received a diabetes test ................................................................................ 184

Appendix I. Cox Proportional Hazard model sensitivity analyses ........................ 185

Chapter 1

Introduction

1

1.1 Thesis Overview

The purpose of this thesis is to investigate the epidemiology of diabetes in an

immigrant, multi-ethnic population through an equity lens. The ultimate goal of this work

is to generate information that can be used to design appropriate and effective targeted

programs for diabetes prevention, management and control in order to reduce inequities

in health. To this end, the thesis has been divided into three major sections dealing with

distinct yet interconnected objectives. Each objective is dealt with in a separate chapter.

Objective 1: To investigate disparities in diabetes burden among immigrant ethnic

groups. (Chapter 2)

Objective 2: To determine rates of screening for diabetes in immigrant ethnic groups

and identify whether disparities exist by immigration status, ethnicity and sex.

(Chapter 3)

Objective 3: To quantify diabetes risk by ethnicity and sex and explore whether the age

of diabetes onset differs between ethnic groups. (Chapter 4)

2

1.2 Background

Type 2 Diabetes Mellitus (T2DM) prevalence is increasing globally, (King & Rewers,

1993) largely in response to rising rates of obesity (Mokdad et al., 2000b; Katzmarzyk,

2002). It was estimated in 2003 that 194 million adults worldwide were living with

diabetes and an additional 314 million people had impaired glucose tolerance (putting

them at a high risk of subsequent diabetes) (Sicree et al., 2006). One American study

recently estimated that the lifetime risk of developing diabetes for Americans born in the

year 2000 is one in three (Narayan et al., 2003). In Canada, roughly 2.4 million

Canadians are currently living with this condition (Public Health Agency of Canada,

2011). Age and sex-adjusted diabetes prevalence in Ontario increased by 69% in the 10

years between 1995 and 2005 to 8.8% (Lipscombe & Hux, 2007) - an increase that

surpassed predictions made by the World Health Organization for prevalence in the

year 2030 (Wild et al., 2004). Since diabetes is associated with increased rates of

morbidity, mortality and disability (Public Health Agency of Canada, 2009), these global

trends have serious implications for population health, the economic sustainability of

Canada’s universal health system and the economy in general.

Type 2 diabetes, which accounts for roughly 90-95% of all diabetes (Zimmet et al.,

2001) and is the focus of this dissertation, results from a complex interplay between

genetic and environmental factors. Throughout this thesis the abbreviated terms T2DM

and diabetes will be used interchangeably to refer to Type 2 Diabetes Mellitus. T2DM is

a chronic metabolic disorder resulting from increased insulin resistance and defective

insulin secretion. This underlying insulin resistance, together with hyperglycemia and

3

other metabolic derangements, impacts the vascular system and other organs. Diabetes

is a leading cause of cardiovascular disease (Kapral et al., 2003; Booth et al., 2003),

end-stage renal failure leading to dialysis (Oliver et al., 2003), amputations (Hux et al.,

2003) and blindness (Klein & Klein, 1995). One-third of all admissions for myocardial

infarction and stroke and two-thirds of all non-traumatic amputations in Ontario occur in

persons with diabetes (Hux et al., 2003; Kapral et al., 2003; Booth et al., 2003). People

with diabetes are at a 50-60% greater risk of depression and experience higher rates of

physical disability due to circulatory, nervous and immune system problems (Manuel &

Schultz, 2003). Diabetes is associated with a 13 year reduction in life expectancy and

mortality rates in adults with the condition are twice as high as among those without

diabetes (Manuel & Schultz, 2004; Public Health Agency of Canada, 2009). Diabetes is

clearly a clinically important disease, which affects not only general health but also

physical functioning, mental health and life expectancy. In 2010 the direct and indirect

costs of diabetes in Canada were estimated at $11.7 billion (Canadian Diabetes

Association Clinical Practice Guidelines Expert Committee & Diabete Quebec, 2011).

Diabetes Risk Factors and their Distribution Across Populations

Obesity is the most significant risk factor for the development of type 2 diabetes. The

likelihood of developing diabetes among individuals who are classified as obese

(defined as a body mass index, or BMI, >=30) is more than seven times higher than

among those with normal body weight (Abdullah et al., 2010). Increasing age is also

associated with an increased likelihood of transitioning from insulin resistance to

diabetes, and the highest incidence of diabetes is found in those over age 65 (Public

4

Health Agency of Canada, 2009). In addition to the increased risk associated with

obesity and advanced age, T2DM occurs more commonly in non-European ethnoracial

groups, largely due to ethnic differences in genetic susceptibility that are not completely

understood. This ethnic disparity in risk is the focus of this dissertation.

The epidemiology of diabetes also varies by sex and gender. The prevalence of type 2

diabetes has traditionally been found to be higher in men than in women (Gourdy et al.,

2001; Public Health Agency of Canada, 2009; Wild et al., 2004). There is evidence,

however, that this may not be the case in some high-risk ethnic groups including

Aboriginal, ‘Black’ and Mexican-American populations, in which women have been

found to have equally high, or higher risk than men in some studies (Young et al., 2000;

Chiu et al., 2010; Cowie et al., 2006; Dyck et al., 2010). High rates of overweight and

obesity have also been observed in women in these communities, which may increase

their susceptibility to developing disease and be the cause for this departure from

traditionally observed sex-specific differences in risk (Flegal et al., 2002; McDermott et

al., 2010). Such gender differences in the prevalence of risk factors for the development

of diabetes, such as obesity and physical inactivity have been described previously

(Matheson et al., 2008; Chiu et al., 2010; Meisinger et al., 2002). Furthermore, women

have higher rates of contact with the health care system and are therefore more likely to

be screened for diabetes (Wilson et al., 2009). As a result of these gender differences in

prevalence, distribution of risk factors and patterns of health services utilization, all

analyses in this thesis were stratified by sex.

5

Finally, diabetes risk is unequally distributed across socioeconomic groups, with a larger

burden of disease in low income and less educated populations, an association that

may be stronger for women (Ross et al., 2010). Due to the higher prevalence of this

condition in specific ethno-racial groups and among people of lower socioeconomic

status, as well as the possible interaction between socioeconomic status and gender,

population health disparities and health equity are underlying themes of this research.

Diabetes and Ethnicity: Genetic Origins and Pathophysiology

A strong genetic component to the risk for type 2 diabetes was established through

early epidemiologic studies showing extremely high concordance in monozygotic twins,

and a very high prevalence among close family relations (Kumar & Clark, 1999; Pincus

& White, 1933). A genetic role in susceptibility to diabetes was further supported by the

observation that a high level of variation existed between ethnicities living in the same

environment with respect to diabetes prevalence and insulin responses to oral glucose

tests in nondiabetic individuals (Rimoin, 1969; Ali et al., 1993). Conversely, other

landmark studies have looked at genetically similar populations in different settings,

including Japanese migrants living in Hawaii and Los Angeles (Hara et al., 1994), and

West African migrants to the Caribbean and Britain (Mbanya et al., 1999), and have

found high variability in diabetes prevalence depending on where people live. Therefore

taken together, these results support the theory that ethnic differences in diabetes risk

have a genetic origin which impacts the predisposition for developing disease; however

expression of these differences is dependent on gene-environment interactions. The

exact mechanism for the expression of these differences is still not entirely clear.

6

The predominant theories for the higher incidence of diabetes and insulin resistance in

certain ethnic groups are the ‘thrifty genotype’ and ‘thrifty phenotype’ hypotheses. The

‘thrifty genotype’ hypothesis states that frequent exposure to periods of starvation or

insufficient nutrient intake resulted in a survival advantage for an adaptive genotype that

was very efficient at storing nutrients as abdominal fat. Presented as evidence for this is

the higher prevalence of abdominal fat in Asians even in persons of normal BMI

(McKeigue et al., 1992; Abate et al., 2004; Raji et al., 2001). The ‘thrifty phenotype’

hypothesis postulates a fetal and maternal origin of disease. This theory, supported by

recent research, hypothesizes that an adverse intrauterine environment, due to

maternal undernutrition, resulting in low birthweight and rapid post-natal growth are

associated with later development of obesity and higher diabetes risk (Yajnik &

Deshmukh, 2008; Ma & Chan, 2009). This pattern of fetal/post-natal growth is then

propagated in subsequent generations. In addition, there is evidence from the Child

Heart and Health Study in England (CHASE) that children (=<10 yrs) of South Asian

descent already are more likely to show metabolic tendencies and precursors of

diabetes as compared with same-age children of Western European descent (Whincup

et al., 2010). These theories are not mutually exclusive and both theories postulate that

in an era of food abundance, these ‘survival adaptations’ result in a mismatch between

metabolic phenotype and the current energy-dense food environment. This mismatch

creates an increased susceptibility of developing insulin resistance in response to

increased body mass index and sedentary lifestyles.

7

Diabetes and ethnicity: Epidemiology

The highest rates of diabetes worldwide have been reported in Aboriginal populations in

Australia, the Pacific Islands, the United States and Canada (Pavkov et al., 2007;

Young et al., 2000; Dyck et al., 2010; Daniel et al., 1999; O'Dea et al., 1993; Zimmet et

al., 1977; Khambalia et al., 2011). In these populations, the prevalence of diabetes may

be as high as 20 to 50 per cent (Pavkov et al., 2007; Shah et al., 2003; Harris et al.,

1997). Other groups found to have a higher prevalence of diabetes include people of

South Asian (Chiu et al., 2010; Ramachandran et al., 2008; Anand et al., 2001; Gupta et

al., 2011), African (Cowie et al., 2009), African-Caribbean (Mbanya et al., 1999;

Zaninotto et al., 2007; Hennis et al., 2002) and Hispanic (Narayan et al., 2003; Cowie

et al., 2009) ethnic background. In Canada, T2DM has been found to be higher in

South Asian (Chiu et al., 2010; Anand et al., 2001; Manuel & Schultz, 2003; Khan et al.,

2011) and Black (Chiu et al., 2010) populations. The majority of Canadian studies on

diabetes have been cross-sectional prevalence studies, or have relied on self-report

survey data. To date there have been two diabetes incidence studies conducted in

Canada that included information on ethnicity (but not immigration). Both these studies,

however, looked at a small number of ethnic groups due to data availability or quality.

The first study relied on self-report survey data and, due to sample size restrictions and

the under-representation of ethnic minorities typical in survey data, was limited to

looking at South Asians, Black, Chinese, and White ethnic groups (Chiu et al., 2011).

This study found that the risk of diabetes, adjusted for age, sex, socioeconomic status

and BMI was higher in all three visible minority groups as compared to the White

population. In the one population-based, longitudinal study conducted in Canada,

8

researchers in British Columbia and Alberta used administrative health data to

determine diabetes incidence in individuals aged 35 and over. However, they were

required to rely on surname algorithms to assign ethnicity and were therefore limited to

looking at people of South Asian and Chinese ancestry (Khan et al., 2011). This study

also found that South Asians had significantly higher (10.8 per 1,000) incidence of

diabetes as compared to the general population (8.5 per 1,000), but in contrast to the

findings by Chiu et al (2011) these researchers found Chinese participants to have

significantly lower incidence (6.8 per 1,000) (Khan et al., 2011).

A large number of studies conducted in indigenous people (Ramachandran et al.,

2001), rural to urban internal migrants (Ebrahim et al., 2010) and external migrants

(Misra & Ganda, 2007) show that South Asians in particular (who comprised 7% of the

total Ontario population in 2006) (Statistics Canada, 2007b) have very high rates of

diabetes. Among this latter group, persons from India, Pakistan and Bangladesh have

the highest rates (Abate & Chandalia, 2001). In the UK, where the majority of research

on ethnicity-based risk for diabetes originates, people of South Asian descent (Indian,

Pakistani and Bangladeshi) have prevalence rates of diabetes that are 3-6 times that of

the white, British population (Dhawan et al., 1994; Mather & Keen, 1985; McKeigue et

al., 1992; Cruickshank et al., 1991). In Canada, South Asian populations have a

prevalence of diabetes that is roughly double that of the White population (Anand et al.,

2000; Manuel & Schultz, 2003; Chiu et al., 2010; Khan et al., 2011). In addition,

research from the U.S. suggests that diabetes prevalence in South Asian populations is

rising faster than in any other ethnic group (Mokdad et al., 2000a). Finally, this

increased risk of developing diabetes among South Asians tends to begin at an earlier

9

age, at a lower body mass index (Chiu et al., 2011) and in subjects with a smaller waist

circumference (Gray et al., 2011).

This tendency to develop diabetes at an earlier age further increases the burden of

disease in South Asian populations by increasing the lifelong risk of complications

related to this disease (Tunis et al., 1993; Brancati et al., 1996; Brancati et al., 1992).

British research has shown that South Asians with T2DM are more likely than their

white counterparts to develop related complications including premature coronary heart

disease (McKeigue, 1992), microalbuminuria (Allawi et al., 1988) and end-stage renal

failure (Roderick et al., 1994; Burden et al., 1992). This younger age of onset has been

mostly documented in South Asian populations; however, some recent research

suggests that this trend may extend to other populations (Tseng et al., 2006; Jimenez-

Corona et al., 2010; Dabelea et al., 2007). In many developed countries, including

Canada, the incidence of diabetes has increased most dramatically in young adults

(defined variously as ages 30-49, 20-49, or <35) (Lipscombe & Hux, 2007; Engelgau et

al., 2004; Tseng et al., 2006), and in the U.S. young adults have experienced an

approximate doubling in incidence rates in the past 10 years (Engelgau et al., 2004).

Detailed information, however, is lacking of age-specific diabetes incidence rates in

younger adults and how these may differ by ethnicity. More research is required to

better understand demographic changes in diabetes onset, particularly in high risk and

vulnerable populations, and the significance that this may have for prevention and

screening programs.

10

Diabetes & Migration

Roughly 80% of people with diabetes in the world are in developing countries, with

China and India being the largest contributors (Sicree et al., 2006). Canada receives

roughly 200,000 immigrants annually, and in current years, persons from China and

South Asia combined comprise the largest group (Citizenship and Immigration Canada,

2009). In 2009, 69% of all immigrants to Canada came from Asia, Africa or the Middle

East (Citizenship and Immigration Canada, 2009). The global rise in obesity and

obesity-related chronic conditions is causing a gradual shift in focus away from

infectious diseases (such as tuberculosis, HIV/AIDS, hepatitis B) and onto chronic

conditions (such as diabetes, hypertension and cardiovascular disease) – a shift that is

occurring among immigrants as well. Lifestyle changes affecting levels of physical

activity and food availability associated with the move from more agrarian societies to

increasingly urbanized environments are identified as significant factors driving this

global increase in obesity (Ebrahim et al., 2010; Allender et al., 2011).

Not only may there be a genetic predisposition in some immigrant groups towards

central adiposity and insulin resistance, but there are socio-environmental factors that

may exacerbate this tendency. A higher risk in these groups is likely amplified by

lifestyle changes co-occurring with urbanization and modernization in their home

countries as mentioned above, as well as lifestyle changes occurring with migration to

more industrialized countries. Urbanization, rural to urban migration within the same

country, and migration from less industrialized and urbanized countries to those that are

more so have all been shown to be associated with decreased levels of physical

11

activity, and increases in BMI and upper-body adiposity (Ramachandran et al., 2008;

Misra & Ganda, 2007; Ramachandran et al., 2004). The observed change over time

post-migration of cultural, language and lifestyle characteristics and behaviours to more

closely resemble those of the adopted country is often referred to as ‘acculturation’. It is

thought that the adoption of more ‘Westernized’ behaviours including ‘nutrition

transition’ (i.e. a move from diet rich in fruits/vegetables to a ‘Western’ diet rich in fats,

meat, processed foods and salt) lower levels of daily physical activity and increasingly

sedentary lifestyles resulting in weight gain, occurs through this acculturation process

(Misra & Ganda, 2007; Iacoviello et al., 2001). Migration studies show that the body

weight of many immigrants increases after only 10 years of residence in the new host

country (McDonald & Kennedy, 2005; Goel et al., 2004). This increase in body weight

over time may accelerate the development of insulin resistance and diabetes. Some

research also shows the psychological stress of settlement can lead to unhealthy eating

habits (Misra & Ganda, 2007; Misra & Khurana, 2009) and may even directly increase

the risk for developing diabetes (Maty et al., 2010). Although many chronic conditions

occur less frequently in recent immigrants (a phenomenon referred to as the ‘healthy

immigrant effect’ (McDonald & Kennedy, 2004), the prevalence of diabetes may be

higher in immigrant groups from certain world regions due to the ethnic composition of

these populations. Furthermore, the health of recent immigrants has been shown to

decline over time until it approximates that of the receiving population (Chen et al.,

1996a; Newbold & Danforth, 2003). This phenomenon has been observed both in

Canada (Chen et al., 1996a; Newbold & Danforth, 2003) and elsewhere (Young, 1992;

Stephen et al., 1994).

12

Changes in socioeconomic status (SES) may further exacerbate the risk associated

with migration (Misra & Ganda, 2007). Although migrants may move from less affluent

to more affluent countries, recent immigrants and visible minorities in Canada tend to

have lower incomes than Canadian-born individuals of European descent and this may

further increase health disparities (Urquia et al., 2007; Chen et al., 1996b; Statistics

Canada, 2007a). In studies looking at race and socioeconomic variables, a strong

interaction has been found between race and SES for diabetes (Brancati et al., 1996;

Cowie et al., 1993). Racial disparity in T2DM rates in the U.S. has been shown to be

greatest at lower levels of education and income, especially among women (Cowie et

al., 1993). Post-migration stress precipitated by low SES, poor working conditions and

loss of social support has been associated with diabetes in South Asian immigrants in

the United Kingdom (Misra & Ganda, 2007; Patel et al., 2006). Immigrants may be more

likely to experience prolonged periods of low income, underemployment, lack of stable

housing, food insecurity, and competing priorities that may impede the pursuit of healthy

eating habits and regular physical activity.

To sum up, immigrants to Canada may be at particular risk for developing diabetes due

to a triple combination of risk factors: they are increasingly composed of populations

with an elevated genetic susceptibility to developing insulin resistance and diabetes;

they are undergoing potentially difficult transition periods associated with the migration

process itself which may impact the ability to focus on healthy lifestyle; and they are

arriving from countries that are undergoing rapid changes in diet and lifestyle associated

with urbanization. Researchers often focus on the health risks of transitioning to a

‘Westernized’ (and unhealthy) lifestyle post-migration, but it is likely that many

13

immigrants are being exposed to environments that may be diabetogenic well before

migration.

The experiences associated with migration inevitably have an effect on the health of the

individual undergoing such a significant transition; however, at the macro level migration

also impacts the health care system and the health of the receiving country. As a result

of increasing demographic diversity in Canada and elsewhere, health systems need to

adapt to attempt to meet the health needs of many diverse populations and effective

policies must be designed and implemented to help promote and maintain good health

in these populations.

Canadian Patterns of Immigration and Immigration Policies

Since 1989 Canada has received roughly 200,000 – 250,000 immigrants annually. The

majority of immigrants settle in one of 3 provinces: Ontario, British Columbia and

Quebec, accounting in 2008 for 81% of all immigrants (Citizenship and Immigration

Canada, 2009). Ontario generally is the most popular destination (receiving roughly

50% of all immigrants to Canada) and Toronto receives the majority of provincial

migration (Citizenship and Immigration Canada, 2009).

Immigration policies influence the demographics of our society by impacting age-

distribution and ethnic composition. Canada’s expansionist immigration policy, which

began in the late 19th Century in order to attempt to populate the vast amounts of

unsettled lands in the Western provinces, is now largely intended to counter-act the

demographic and economic effects of a declining birth rate and an aging population.

14

Historically, immigration to Canada was almost entirely from Europe and the United

States, a pattern that gradually changed after new immigration legislation was put

forward in 1962 to eliminate discrimination based on race, religion or national origin.

The intention of immigration as primarily a source of human resources became evident

with the introduction of the point system in 1967 which allowed for a preference to be

given to skilled workers and business immigrants. The current Immigration and Refugee

Protection Act (IRPA) has been in place since 2002 and similar to previous acts it

upholds the anti-discriminatory (on the basis of race, religion or nationality) selection

criteria for entry to Canada while increasing the power of the immigration authorities to

deny admission to those deemed inadmissible for health, legal or other reasons.

Citizenship and Immigration Canada (CIC) is responsible for administering the

Citizenship Act and the IRPA.

The IRPA requires that all applicants for permanent residence and some visitors

undergo a medical exam and that those with certain conditions be referred for ‘medical

surveillance’. Although the purpose of the medical exam is to identify individuals with

certain health condition in order to refer them for medical surveillance in order to

“help(s) people with certain conditions maintain their own health, and protect(s)

…people in Canada” it is also in place to screen out people with conditions that make

them a threat to public health or that will place “excessive demand on health or social

services” such as active tuberculosis, HIV, and certain psychiatric illnesses (Citizenship

and Immigration Canada, 2012). These people are consequently considered “medically

inadmissible”. This medical screening process contributes to the ‘healthy immigrant’

phenomenon described above. Family Class sponsored spouses, common-law partners

15

and dependent children, as well as some refugees (approximately 30% and 10% of all

immigrants, respectively) are exempted from this ‘Excessive demand’ legislation

(Citizenship and Immigration Canada, 2012). From 1994 to July of 2012, CIC was also

responsible for providing essential health services to refugee claimants, and refugees

not yet covered by provincial health plans, under the ‘Interim Federal Health Program’

(Gagnon, 2002). This provision of essential services to refugees has recently been

changed and new, more restrictive guidelines for health services eligibility have been

established.

Once a person has been given Permanent Residence Status, they are eligible to apply

for provincial health insurance. Five provinces/territories, Ontario, British Columbia,

Quebec, Nunavut and the Yukon, have 3 month waiting periods before coverage takes

effect (Gagnon, 2002). In the time between arrival and provincial Medicare coverage,

immigrants (not including refugee claimants) must either privately buy insurance, or

remain uninsured.

Immigration, ethnicity and health care utilization and access

Diabetes is a controllable disease but control requires good access to health care,

educational programs, healthy food and opportunities to be physically active. Having

good access to a continuous source of care and, when necessary, being referred to

specialists and allied health professionals is protective of developing diabetes

complications (Jacobson et al., 1997). As proposed by Gelberg and Anderson (2000) in

their adapted model for predicting the health services utilization of vulnerable

populations – the health services utilization of recent immigrants, and non-European

16

recent immigrants in particular, is likely to be impacted by characteristics that can be

broken down into predisposing, enabling and need domains (Gelberg et al., 2000).

Within the predisposing domain of the model, recent immigrants may face barriers of

language, culture, and unfamiliarity with the Canadian health care system. Within the

enabling domain health seeking may be impacted by factors such as lack of

transportation, child care and social support networks (Kliewer & Kazanjian, 2000a).

New immigrants may be less able to negotiate the health care system or advocate for

their health needs, which may result in poorer access to diabetes prevention and

management programs (Shah, 2008; Quan et al., 2006; Glazier et al., 2004). A number

of studies have shown people of South Asian origin to be at higher risk for developing

diabetes-related complications, cardiovascular disease and associated mortality as

compared with European populations (Anand et al., 2000; Anand et al., 2000; Webb et

al., 2011; Bhopal et al., 1999; Fischbacher et al., 2003), which may be a reflection of

late diagnosis or sub-optimal disease management. In addition, as mentioned above,

the SES of many recent immigrants is complex, as they tend to have high educational

attainment but low income when first arriving in Canada and often reside in low income

neighbourhoods with few resources (Booth et al., 2007). Finally, in terms of the health

need domain, as discussed earlier, although initially experiencing a health advantage,

the health of recent immigrants tends to decline over time to levels that are equivalent,

or below that, of the general population. Despite declining health status over time, and

possibly contributing to this decline, research suggests that recent immigrants tend to

have fewer visits to physicians (Kliewer & Kazanjian, 2000b; Laroche, 2000b), use

fewer preventive services (Laroche, 2000a; Matuk, 1996; Glazier et al., 2004; Goel,

17

1994; Katz & Gagnon, 2002) and report encountering difficulties accessing and

navigating the health system (Desmeules et al., 2004; Matuk, 1996).

It has been suggested that up to one-third of all diabetes remains undiagnosed in the

general population (Cowie et al., 2006; Young & Mustard, 2001). Identifying people with

undiagnosed diabetes reduces the risk of preventable complications related to this

disease such as retinopathy, coronary heart disease, stroke and peripheral vascular

disease. Given that many recent immigrants face barriers accessing health care it begs

the question of whether some immigrant and ethnic-minority populations are less likely

to be screened for diabetes as compared with the general population. However the

answer to this question is unclear since most diabetes-related research to date that

includes information on ethnicity or race has been focused on diabetes management

(Shah, 2008; Mukhopadhyay et al., 2006; Vimalananda et al., 2011; Chatterji et al.,

2012) and very little attention has been paid to rates of screening in different ethnic

groups. Interestingly, the one previous Canadian study looking at screening that

incorporated some measures of immigration and ethnicity using a large survey sample

found that along with increasing age, hypertension, and number of physician visits,

immigrant status and non-white ethnicity were associated with a significantly increased

likelihood of having a serum blood glucose (SBG) test (Wilson et al., 2010). Given the

higher risk of diabetes among certain ethnic groups, particularly South Asians, and the

importance of early diagnosis and management of the disease, more information is

needed about diabetes screening in specific immigrant and ethnic populations.

18

As a result of changing immigration patterns over the past 30-40 years, Canada now

has a very diverse ethnic and cultural composition. Along with the benefits and cultural

wealth that this multi-ethnic environment provides comes the challenge of meeting the

health needs of this diverse population. Meeting these needs requires greater

understanding of the health status, health behaviours and risk factors prevalent in

different groups within the population. This thesis attempts to contribute to the scientific

literature on diabetes, as well as to contribute practical information that may assist in

health planning for ethnic groups that may be at higher risk for developing diabetes and,

if not promptly diagnosed or adequately controlled, may be at elevated risk of related

complications. The health, social and economic costs of the diabetes epidemic poses a

tremendous challenge to health services providers, planners and policy-makers.

Canada is one of the most multi-ethnic countries in the world and understanding the

distribution of diabetes risk among ethnic groups is an important step towards planning

effective and equitable health programs and policies.

1.3 Theoretical Framework

A large body of literature exists around the social determinants of health and health

inequities. Many theoretical frameworks have been criticized for over-simplifying the

complex and multi-level relationships between the social, political, economic, cultural

and physical environments that form the context in which health is determined. The

migration and resettlement processes would be expected to add a further layer of

19

complexity to these relationships. The World Health Organization’s Commission on

Social Determinants of Health proposed a theoretical framework in 2007 that recognizes

the multi-layered effects on health and health inequities from the macro environment

(political and socioeconomic context) down to the individual’s biology and behaviours

(See figure 1 below) (Solar & Irwin, 2010). The first two columns depicted in the model

below refer to the structural determinants of health inequities which encompass, from

left to right, both the socio-economic and political context of a country or society as well

as the social and class structure of the society. These latter determinants of health

inequities include socially defined roles around gender and race/ethnicity. In the case of

our study, these determinants include both factors experienced in the home country, as

well as those experienced after migration to Canada.

These structural causes of inequities precede and lay the foundation for what the

Commission referred to as the intermediary determinants of health which encompass

the more proximal influences on health such as material wealth, social circumstances,

behaviours, biological factors and the health system itself (depicted in the third column).

Having recently immigrated to Canada, or having a genetic predisposition to disease

would fall into this intermediary determinant of health section along with behaviours

related to healthy lifestyle (i.e. diet and exercise) and health-seeking behaviours. Again,

both pre- and post- migration circumstances would exert an effect.

20

Figure 1.1 : The World Health Organization’s Commission on the Social Determinants of Health conceptual framework. Reprinted with permission from Solar, O. & Irwin, A. (2010). A conceptual framework for action on the social determinants of health Geneva: World Health Organization. © World Health Organization 2010.

This model provides a framework for understanding the context of the social

determinants of health and how they may contribute to inequities as well as suggesting

areas for intervention. The framework identifies four key areas for intervention: 1) at the

level of the structural determinants of inequalities such as reducing income and

educational inequalities; 2) at the level of intermediary determinants by reducing

exposures to factors that might damage health in people in disadvantaged groups; 3) at

21

the level of the health care system such as developing targeted screening programs (a

key focus of this thesis); or finally, 4) interventions intended to reduce inequities in how

health status in turn impacts socio-economic status - for example protecting people with

poor health or disability from unemployment or poverty.

For the purpose of this thesis the WHO framework was used to help understand the

context of health and health inequities in the population. This framework is useful both

for interpreting the findings and for helping us identify appropriate interventions and

recommendations with a goal of reducing health disparities and inequities.

1.4 Research Questions and Rationale for the Objectives

Objective 1: Diabetes burden in immigrant ethnic groups

Research Questions:

i) Is the prevalence of diabetes higher in immigrants than in the general

population?

ii) How does the prevalence of diabetes vary by ethnicity and sex?

iii) How do income, education, visa category and time since arrival impact

diabetes prevalence?

Apart from evidence that diabetes prevalence is higher among aboriginal populations

and in persons of non-European ethnicity, little is known about the ethnic distribution of

22

diabetes in Canada. The first objective of this thesis is to describe the distribution of

diabetes among immigrants to Ontario and evaluate how the burden varies by ethnicity

and sex. Previous population estimates of diabetes have relied on cross-sectional

surveys, within which diabetes is often significantly under-reported and which tend to

under-sample immigrants and persons who do not speak English ( Mackenbach et al.,

1996). In addition, due to small or limited samples, most previous studies have focused

either on a small number of ethnic groups, or have been forced to combine ethnically

and culturally heterogeneous populations into single groups (eg. White, Black, Asian or

Other). In the absence of detailed prevalence data on the wide number of ethnic groups

living in Ontario, the purpose of this chapter is to lay the foundation for understanding

which ethnic groups bear the highest burden of disease and how might this risk vary by

gender, income, education and time in Canada. The large sample size further allows us

to identify not just which ethnic groups are at highest risk, but to identify which countries

of birth are associated with the highest risk which can then be used to further inform

targeted diabetes screening or prevention programs, including language and cultural

services. Ontario is an ideal setting for this research as it receives roughly 50% of the

more than 200,000 - 250,000 immigrants to Canada annually (Citizenship and

Immigration Canada, 2009).

Objective 2: Rates of screening for diabetes in immigrant ethnic groups

Research Questions:

i) Do immigrants get screened for diabetes more or less than the general

population?

23

ii) Do screening rates vary by ethnicity and sex?

iii) What is the yield of screen-detected, new diabetes cases by ethnicity? Does it

support targeted screening programs in these populations?

iv) What is the burden of undiagnosed diabetes in ethnic groups in Ontario?

The second objective of the thesis is to determine the level of screening that is being

achieved in immigrant populations and to evaluate whether screening for diabetes is

occurring equitably across ethnic groups. Early identification and optimal management

of diabetes are critical in order to avoid diabetes-related complications. There is

evidence that immigrants and ethnic minorities may face barriers accessing the health

care system, including for preventive services (Lofters et al., 2010; Glazier et al., 2004)

but to date there have been very few studies evaluating levels of diabetes screening in

ethnic minority and immigrant populations (Gray et al., 2010; Cowie et al., 1994; Wilson

et al., 2010) and there have been no studies looking at diabetes screening among

specific ethnic groups. As part of this investigation we will also determine the

effectiveness of screening in different ethnic populations that have varying underlying

disease risks and, using this information, we will attempt to estimate the underlying

burden of undiagnosed diabetes. In combination with the results from the first objective,

this work can help inform policy around targeted screening programs.

This chapter also has a practical methodological secondary objective. The current

algorithm used to identify individuals with diabetes for the Ontario Diabetes Database

(ODD), the diabetes registry used in this thesis, relies on administrative health services

data and therefore captures only physician diagnosed diabetes (See Appendix A, Data

24

Sources for more information). Although the algorithm used to populate the ODD has

been validated and found to have high sensitivity and specificity (Hux et al., 2002),

information is not currently available regarding how well the ODD captures people with

diabetes among population sub-groups, including by ethnicity and immigrant status.

Therefore, in addition to the population health question above regarding rates of

screening and undiagnosed disease in this vulnerable population, this chapter also

attempts to answer a practical question: Are immigrants and people of minority ethnic

populations as likely to be captured in the Ontario diabetes registry as are people in the

general population?

Objective 3: Diabetes risk and age of onset by ethnicity and sex

Research Questions:

i) What is the incidence of diabetes among ethnic groups in Ontario?

ii) How does the age-specific risk compare across ethnic groups?

iii) What age should we begin to screen high-risk ethnic minority groups?

Robust, ethnic-specific diabetes incidence estimates by age are lacking. There is some

evidence that risk for diabetes may begin at a younger age in those ethnic populations

that are at highest risk. If this is the case, appropriate diabetes screening and diabetes

prevention programs should be designed accordingly and information about differences

in age at onset is required to do so. The few diabetes incidence studies to date have

looked at broad age groups and have generally not looked at incidence in younger ages

(i.e. < age 40). Linking immigration data with the Ontario diabetes registry allows us to

follow roughly 2 million individuals over a period of 20 years to identify new cases of

25

diabetes and will give us the opportunity to derive robust, population-based incidence

rates by ethnicity, country of origin and gender. In addition, this work will allow us to look

at incidence by fine age categories, beginning in very early adulthood. This chapter will

present the first population-based, incident cohort study looking at virtually all ethnic

groups in a single health system and socio-political setting in order to better quantify the

level of risk experienced by ethnic groups and determine the influence of age and sex

on this risk.

26

Chapter 2

Age and Sex Patterns of Diabetes Among Immigrants to Ontario

Credits: This chapter represents a prepublication version of the following article: Creatore MI, Moineddin R, Booth G, Manuel D, Glazier RH. Age- and sex-related prevalence of diabetes mellitus among immigrants to Ontario, Canada. Canadian Medical Association Journal 2010; 182:781-789.

27

Abstract

Background: The majority of immigrants to Canada originate from the

developing world, where the most rapid increase in prevalence of diabetes

mellitus is occurring. We undertook a population-based study involving

immigrants to Ontario, Canada, to evaluate the distribution of risk for diabetes in

this population.

Methods: We used linked administrative health and immigration records to

calculate age-specific and age-adjusted prevalence rates among men and

women aged 20 years or older in 2005. We compared rates among 1,122,771

immigrants to Ontario by country and region of birth to rates among long-term

residents of the province. We used logistic regression to identify and quantify

risk factors for diabetes in the immigrant population.

Results: After controlling for age, immigration category, level of education,

income and time since arrival, we found that, as compared with immigrants from

western Europe and North America, risk for diabetes was elevated among

immigrants from South Asia (odds ratio [OR] for men 4.01, 95% CI 3.82-4.21;

OR for women 3.22, 95% CI 3.07-3.37), Latin America and Caribbean (OR for

men 2.18, 95% CI 2.08-2.30; OR for women 2.40, 95% CI 2.29-2.52), and sub-

Saharan Africa (OR for men 2.31, 95% CI 2.17 – 2.45; OR for women 1.83,

95% CI 1.72-1.95). Increased risk became evident at an early age (35-49 years)

and was equally high, or higher, among women as compared with men. Lower

socio-economic status and greater time living in Canada were also associated

with increased risk for diabetes.

28

Conclusions: Recent immigrants, particularly women and immigrants of South

Asian and African origin, are at high risk for diabetes compared with long-term

residents of Ontario. This risk becomes evident at an early age, suggesting that

effective programs for prevention of diabetes should be developed and targeted

to immigrants in all age groups.

29

2.1 Introduction

Although the prevalence of type 2 diabetes is generally higher in developed

countries, it is increasing most rapidly in developing countries (International

Diabetes Federation (IDF), 2003; Ramachandran et al., 2008; Wild et al., 2004).

The greatest relative increases in diabetes in the next 25 years is predicted to

occur in the Middle Eastern crescent, sub-Saharan Africa and India (Wild et al.,

2004). Approximately 250,000 individuals immigrate to Canada annually, the

largest percentages from Asia, Africa and the Middle East (Citizenship and

Immigration Canada, 2007).

Apart from conventional risk factors for diabetes such as age and obesity, risk

also does not appear to be evenly distributed across ethno-racial groups.

However, apart from evidence that diabetes prevalence is higher among

persons of non-European ethnicity (McBean et al., 2004; Mokdad et al., 2000;

Mather & Keen, 1985; Abate & Chandalia, 2001; Tillin et al., 2005; Rotimi et al.,

1999; Dowse et al., 1990; Fujimoto et al., 1994; Franco, 1996; Hara et al.,

1994), who comprise the vast majority of all immigrants to Canada, little is

known about the epidemiology of diabetes in immigrants to Western countries.

In ethno-racially and culturally heterogeneous countries such as Canada,

diabetes prevention and control programs can benefit from a greater

understanding of the distribution of risk among different ethno-racial groups and

newcomers from various world regions.

30

The purpose of this study was to determine the prevalence of diabetes among

over one million immigrants to Ontario, from various world regions, make

comparisons to the long-term Ontario population, and examine the effect on

prevalence of gender, age, country of birth, time since arrival, and

socioeconomic characteristics.

2.2 Methods

2.2.1 Data Sources and Study Population

The Registered Persons Database, an electronic registry of all individuals who

are eligible for health coverage in Ontario in a given year, was probabilistically

linked to the Canadian Landed Immigrant Database (LIDS), maintained by

Citizenship and Immigration Canada (linkage rate of 84%). From this dataset,

all adults (age 20 years or older) who were eligible for coverage under the

province's universal health insurance program on March 31, 2005 were included

in the study if they had a valid health card number and date of birth. Individuals

in the study population who were granted permanent residency status in

Canada between 1985 and 2000 were flagged as ”recent immigrants”. The

LIDS database includes information, collected at the time of application for

immigration status, on education level, intended occupation, language ability,

immigration category, as well as sex and date of birth. Feasibility of linkage

between the LIDS and health administrative datasets was tested in pilot projects

(Kliewer & Kazanjian, 2000), which showed that differences in linkage by

31

immigration class, date of immigration, education and country of birth were not

likely sufficient to produce significant bias in any study results. The dataset has

been previously used in Ontario to look at immunizations in children of

immigrant mothers (Guttmann et al., 2008), and perinatal outcomes (Urquia et

al., 2009).

Individuals who were diagnosed with diabetes on or before March 31, 2005

were identified using the Ontario Diabetes Database, a validated administrative

data registry created from hospital records and physician services claims. This

database uses an algorithm of two primary care visits or one hospitalization for

diabetes within a 2 year period to identify diagnosed cases of diabetes

(excluding gestational diabetes). This algorithm has a sensitivity of 86% and a

specificity of over 97% in identifying patients with confirmed diabetes (Hux et

al., 2002).

Due to the absence of any individual-level income information in the immigration

or health datasets, 2005 residential postal codes were linked to area-level

income from the 2006 Canadian Census using the postal code conversion file

(Wilkins, 2008). The conversion file assigns relative income quintiles based on

the smallest geographical unit for which Census data is available, and is

adjusted for household and community size. Since 98% of all immigrants to

Canada between 1985 and 2000 settled in urban areas based on our data, the

study excluded rural populations, defined as those with rural residential postal

codes. Cities are also home to the majority (84%) of the general Ontario

32

population, which in the 2006 Census was 12,160,282 (Statistics Canada,

2006).

2.2.2 Statistical Analyses

Age-adjusted diabetes point prevalence rates were calculated by sex for recent

immigrants and long-term residents on March 31st 2005. For each population,

sex-specific rates were also generated by age group (20-34, 35-49, 50-64, 65-

74, and ≥75) and for the immigrant population, by country of birth and world

region (based on the World Bank schema, available at

(http://go.worldbank.org/FFZ0CTE2V0). The top 15 countries that experienced the

highest prevalence of diabetes were also identified. Direct age-standardization

to the 1991 Canada Census population was used to adjust for differences in

population distribution across different world regions.

Logistic regression was used to estimate the association of risk factors with

diabetes prevalence, including age, world region of birth, pre-migration

educational attainment, post-migration area-level income and time since arrival

in Canada. All analyses were stratified by sex. Educational attainment at time

of immigration categorized as no education, secondary (high school) or less;

non-university qualifications (including diplomas or certificates from institutions

not qualifying as universities, apprenticeships and other non-university post-

secondary education), some university, or university education or higher. To

examine risk by region of birth, immigrants from North America and Western

Europe were used as a comparator. Since point prevalence was measured in

33

2005 and immigration records were only available up until 2000, we were

unable to evaluate risk in those living in Canada for less than 5 years.

2.3 Results

2.3.1 Characteristics of the Study Population

The study included 1,122,771 recent immigrants and 7,503,085 long-term

residents meeting the study eligibility requirements. The characteristics of both

the recent immigrant and long-term resident population are shown in table 2.1.

Compared with long-term residents, immigrants tended to be younger and were

more likely to live in lower income neighbourhoods. Recent immigrants

originated predominantly from East-Asia (27.5%), South Asia (19.4%) and Latin

American and the Caribbean (LAC) (15.8%).

34

Table 2.1 Baseline characteristics of the Ontario long-term resident and recent immigrant study populations*, 2005.

Long-term

Residents Recent Immigrants

Study Population Characteristics (Urban population aged 20+)

Population 7,503,085 1,122,771 Median Age † 47 43 % > age 65† 18.2 10.2 % Male 49.3 49.5 Income quintile§ of neighbourhood of settlement (%):

Q1 (lowest income) 18.2 26.4 Q2 19.6 21.7 Q3 19.7 19.7 Q4 20.4 18.8 Q5 21.2 13.0

World Region of Birth East Asia & the Pacific - 309,043 (27.5%)

South Asia - 217,367 (19.4%) Latin America & the Caribbean - 177,191 (15.8%) Eastern Europe & Central Asia - 167,456 (14.9%)

Western Europe & North America - 98,931 (8.8%) North Africa & the Middle East - 87,610 (7.8%)

Sub-Saharan Africa - 64,367 (5.7%) Unknown/Stateless - 795 (0.07%)

None specified 11 (0.001%) Immigration Visa Category

Family - 448,142 (39.9%) Economic-Skilled-Independent - 285,322 (25.4%)

Refugee - 184,588 (16.4%) Economic-Skilled-Family - 124,465 (11.1%)

Economic-Business - 54,056 (4.8%) Other - 26,185 (2.3%)

None specified 13 (0.001%) Educational Qualifications at Landing (%)

No Education - 47,604 (4.2%) Secondary or Less - 608,925 (54.2%)

Non-University Qualifications - 171,106 (15.2%) Some University - 56,021 (5.0%)

University Degree or Higher - 239,031 (21.3%) None specified 84 (0.007%)

Years Since Arrival (using 2005 as year of reference) (%) 5 - 9 years - 322,047 (28.7%)

10 - 14 years - 384,608 (34.2%) 15 years or more - 416,116 (37.1%)

*Population eligible for provincial health care based on administrative databases. Recent immigrant population includes those who obtained legal landed status between 1985 and 2000. **Urban areas identified from first three characters of the postal code of residence (the Forward Sortation Area(FSA)).

† Based on age as of March 31st, 2005. § 2001 Census income information was applied based on the individual's postal code of residence in 2005.

35

2.3.2 Trends in Diabetes Prevalence

Immigrants from South Asia, LAC, Sub-Saharan Africa and North Africa and the

Middle East all experienced significantly higher diabetes rates than Ontario

long-term residents (Figure 2.1). The fifteen countries of birth (out of 239

countries and geopolitical regions) that were associated with the highest rates

of diabetes among both sexes were found in South Asia, the Pacific Islands,

Latin America, the Caribbean and Africa (Figure 2.2). In the general Ontario

population men have higher rates than women (6.5 % versus 6.2 %,

respectively) but women who were recent immigrants had rates equal to or

higher than immigrant men from the same regions, with the exception of women

from Sub-Saharan Africa.

Overall, immigrants of both sexes had statistically higher rates of diabetes than

the Ontario long-term population at all ages (except for males aged 75+), with a

large disparity between women who were recent immigrants and women who

were long-term residents (Figure 2.3).

36

37

38

39

Diabetes prevalence increased sharply with age until age 75 in both sexes and among

both recent immigrants and long-term residents (Figures 2.3-2.5). An increase in

diabetes prevalence appeared at a young age by immigrants from the highest risk

regions and this risk was sustained across all age groups. Men from South Asia had the

highest prevalence rates across all age groups (reaching 36.7% in the 65-74 age group)

followed by men from Latin America and the Caribbean and Sub-Saharan Africa (35.0%

and 33.2%, respectively, in the 65-74 age group) (see Figure 2.4). Women from Latin

America and the Caribbean and South Asia had the highest prevalence rates by age

(reaching a maximum of 37.0% and 34.8%, respectively, in the 65-74 age group)

(Figure 2.5). The lowest rates were found among men and women from Europe, North

America and Central Asia.

The multivariate logistic regression analysis results (Figure 2.6) show that after

controlling for age, immigrant category, education, income and time since arrival, men

and women from South Asia had significantly higher diabetes rates (OR = 4.01, 95% CI

3.82 – 4.21 and OR=3.22, 95% CI 3.07-3.37, respectively) compared to immigrants

from Western Europe and North America. The next highest risk was experiences by

men and women from Latin America and the Caribbean (OR= 2.18, 95% CI 2.08 – 2.30

and OR=2.40, 95% CI 2.29-2.52, respectively) and Sub-Saharan Africa (OR= 2.31, 95%

CI 2.17 – 2.45 and OR=1.83, 95% CI 1.72-1.95, respectively).

40

41

42

Fig. 2.6 Risk factors for diabetes (2005) among immigrants to Ontario (1985-2000) by sex.

43

An income gradient was observed, whereby lower income was associated with

higher risk (OR 1.31, 95% CI 1.26-1.36 for men and OR 1.38, 95% CI 1.33-1.44 for

women in the lowest, relative to the highest, income quintile neighbourhoods).

Women with secondary education or less had the highest risk as compared with

those with a university degree or higher (OR = 1.32, 95% CI 1.28 – 1.37). Men with

no education had the lowest diabetes risk (OR = 0.56, 95% CI 0.53 – 0.60).

The multiple logistic regression model also displayed a gradient for time since arrival

with males and females living in Canada for 15 years or more having the highest

diabetes prevalence (OR= 1.52, 95% CI 1.48 – 1.56 and OR=1.40, 95% CI 1.36-

1.44, respectively) compared with those living in Canada 5-9 years.

2.4 Discussion

This study used a unique population-based dataset in a setting with high rates of

immigration to describe the epidemiology of diabetes among a heterogeneous

immigrant population. The increased relative rates of diabetes among immigrants

from South Asia, Latin America and the Caribbean, Africa and the Middle East are

particularly striking given that the population of Ontario is itself highly diverse ethno-

racially. We also found that the risk for South Asians was at least triple that of

immigrants from Western Europe and North America, and the risk for people from

Latin America and the Caribbean and Sub-Saharan Africa was roughly double, even

after controlling for age, sex, time since arrival, income and immigration-related

variables.

44

Our findings support those of three previous Canadian studies that found South

Asians had 2-3 times the rate of diabetes as compared with the overall Ontario

population, the ”white” Ontario population; or a sample of Canadians of European

heritage (Manuel & Schultz, 2003; Shah, 2008; Anand et al., 2000). These studies

were all either based on health survey data (Manuel & Schultz, 2003; Shah, 2008),

or a relatively small population sample (Anand et al., 2000). Furthermore, all three

were focused on ethnic differences in diabetes in the population overall and did not

look at immigration status. Research conducted in the UK found people of South

Asian descent had diabetes prevalence rates that were 3-6 times than the white

British population, which is consistent with our findings (Mather & Keen, 1985). Our

prevalence estimates are higher than those generated by the World Health

Organization (WHO) in 2004, however those estimates used data derived from a

small number of often out-dated studies and were based on extrapolations and

assumptions that were likely to lead to underestimations of risk (Wild et al., 2004).

A detailed description of age and sex patterns of diabetes risk among immigrants to

Western countries has previously been lacking. Although diabetes prevalence is

generally higher in men than women (Wild et al., 2004; Public Health Agency of

Canada, 2008), recent immigrant women in our study had prevalence rates roughly

equivalent to or higher than men from the same countries. Women from Latin

America and the Caribbean experienced particularly high rates relative to men. This

pattern of elevated risk in women relative to men has been previously described only

in areas with high proportions of Aboriginal populations (Young et al., 2000). We also

found that the largest disparity in risk existed between recent immigrant and long-

term resident women. These findings suggest that recent immigrant women may be

at particularly high risk for diabetes. Combined with the social isolation and barriers

45

to access of services experienced by many recently immigrated women (Bierman et

al., 2010), this higher risk may raise important health-related issues for immigrant

women and should be of concern to health providers and planners.

We also found an age-related disparity in diabetes prevalence between the highest

and lowest risk immigrant groups that became apparent in young adulthood (by age

35 for South Asians) and increased with age. By age 65, more than a third of men

and women from these regions had diabetes. Above age 75, a plateau or decrease

in risk was seen in people from all regions, which has been described previously in

the general Canadian population (Public Health Agency of Canada, 2008).

We found that diabetes risk increased with time since arrival. An increase in

prevalence of chronic disease and declining health status over time since

immigration has been previously described in the Canadian literature based on

health survey data (Perez, 2002; Newbold, 2005; Dunn & Dyck, 2000). Many

possible causes of this observed deterioration in health status over time have been

suggested including uptake of unhealthy behaviours, acculturation stress, decreased

social, economic and political status, barriers to accessing preventative services, and

competing priorities resulting in reduced self-care (Newbold, 2009).

Socioeconomic status is known to have an inverse relationship with diabetes risk

(Brancati et al., 1996; Robbins et al., 2001), and low education has been previously

identified as a correlate of higher diabetes rates in Canada (Manuel & Schultz,

2003). In the current study, we found a gender-education interaction, whereby low

educational attainment (less than high school diploma) was a significant independent

risk factor for diabetes in immigrant women, but little or no formal education was

46

protective for immigrant men. The latter could reflect more physically demanding

employment (e.g. manual labour) among men in this group. We also observed an

inverse relationship between income and diabetes risk in women, as has been

described before in Canada (Manuel & Schultz, 2003). One explanation for this may

be that higher rates of obesity are reported among women with low socioeconomic

status than among men (Matheson et al., 2008).

High rates of diabetes in specific ethnic migrant groups are likely to be due to a

complex interplay of genetic and environmental factors, including acculturation,

stress, social isolation as well as employment and economic challenges (Misra &

Ganda, 2007). Despite the limitations of our data in addressing social and

experiential exposures, it must be noted that strong ethnic differences continue to be

apparent even after controlling for certain pre-migration (country of birth, immigrant

category, and education) and post-migration (neighbourhood income, time since

arrival) factors.

Additional limitations of this study are related to the use of administrative data.

Although we were unable to differentiate type 1 from type 2 diabetes in the

administrative data, the former represents a very small proportion of all diabetes (5-

10%) and is therefore unlikely to bias our results (International Diabetes Federation,

2009). Administrative data may also underestimate the prevalence of diabetes.

Studies in other jurisdictions suggest that up to 30% of diabetes in the population

may be undiagnosed (Harris & Eastman, 2000) and it is possible that the probability

of diagnosis may differ by immigration status and country of birth. Finally, a small

proportion (< 5%) of health services that are not billed on a fee-for-service basis will

not be captured in our data (Williams & Young W, 1996).

47

We found that recent immigrants from South Asia, the Caribbean, South America

and Africa, have much higher risk of diabetes than both long-term residents of

Ontario and recent immigrants from Europe and Central Asia. This risk begins in

young adulthood and continues throughout their life course. The largest disparity in

risk between immigrants and the general population was observed among women,

the etiology of which should be further explored. Risk increases over time since

immigration, but ethnic differences persist even after controlling for this variable,

suggesting that acculturation and transition to a ”westernized” diet and lifestyle

contributes to, and may exacerbate, but does not explain the differences. Although a

few studies have shown promising results, lifestyle interventions aimed at recent

immigrants should be explored further (Renzaho et al., 2010). Our findings should

also aid policy-makers and planners to develop specific screening guidelines and

community-level targeted diabetes educational programs. Finally, this study

highlights the critical importance of routine collection of data on immigration status

and ethnicity for population health research.

48

Chapter 3

Diabetes Screening Among Immigrants: A Population-Based Urban Cohort Study

Credits

This chapter represents a prepublication version of the following article:

Creatore MI, Booth G, Manuel D, Moineddin R, Glazier RH. Diabetes screening

among immigrants: A population-based urban cohort study. Diabetes Care 2012;

35:754-761.

49

Abstract

Background: To examine diabetes screening, predictors of screening, and the

burden of undiagnosed diabetes in the immigrant population and whether these

estimates differ by ethnicity.

Methods: A population-based retrospective cohort linking administrative health

data to immigration files was used to follow the entire diabetes-free population

aged 40 and up in Ontario, Canada (N = 3,484,222) for three years (2004 to

2007) to determine whether individuals were screened for diabetes. Multivariate

regression was used to determine predictors of having a diabetes test.

Results: Screening rates were slightly higher in the immigrant versus the general

population (76.0% and 74.4%, respectively, p<0.001), with the highest rates in

people born in South Asia, Mexico, Latin America and the Caribbean. Immigrant

seniors (>= age 65) were screened less than non-immigrant seniors. Percent

yield of new diabetes cases among those screened was high for certain countries

of birth (South Asia, 13.0%; Mexico and Latin America, 12.1%; Caribbean, 9.5%)

and low among others (Europe, Central Asia, U.S., 5.1-5.2%). The number of

physician visits was the single-most important predictor of screening and many

high-risk ethnic groups required numerous visits before a test was administered.

The proportion of diabetes that remained undiagnosed was estimated to be 9.7%

in the general population and 9.0% in immigrants.

Conclusions: Overall diabetes screening rates are high in Canada’s universal

health care setting, including among high-risk ethnic groups. Despite this finding,

50

disparities in screening rates between immigrant sub-groups persist and multiple

physician visits are often required to achieve recommended screening levels.

51

3.1 Introduction

Diabetes is a serious chronic disease that is associated with substantial

increases in morbidity and mortality, and imposes a huge economic burden on

society. Although screening for diabetes is increasing in Canada (Wilson et al.,

2009), up to one-third of all diabetes cases are thought to be undiagnosed in the

general population in Canada and the U.S., an estimate that may now be out-of-

date (Cowie et al., 2006; Young, 2001). One significant factor that is likely

contributing to increased screening is the rising prevalence of obesity in the

population.

Early detection and control of diabetes can potentially reduce the heightened risk

of cardiovascular morbidity and mortality associated with this disease. People

with screen-detected diabetes have an increased risk of heart disease as

compared to the general population, and this risk is modifiable with treatment

(Sandbaek et al., 2008; Janssen et al., 2009; Griffin et al., 2011). In addition,

timely screening can prevent the onset of common diabetes-related

complications that could be avoided through early detection and treatment (eg.

retinopathy, peripheral neuropathy, peripheral vascular disease) (Colagiuri &

Davies, 2009).

National guidelines in both the U.S. and Canada recommend that diabetes

screening should be performed on those age 45 (U.S.) or 40 (Canada) and over

every 3 years with more frequent or earlier screening for those with additional

52

risk factors, including belonging to a high-risk ethnic group (Ur et al., 2008;

American Diabetes Association, 2010). Ethnic groups that have been shown to

display an elevated risk for diabetes include people of South Asian (Dhawan et

al., 1994; Creatore et al., 2010; Cruickshank et al., 1991), Aboriginal (Public

Health Agency of Canada, 2009) and African-Caribbean descent (Cowie et al.,

2006; Creatore et al., 2010). Many of the 250,000 immigrants to Canada every

year (Research and Evaluation Branch Citizenship and Immigration Canada,

2009), belong to ethnicities that experience higher rates of diabetes (Creatore et

al., 2010), and who therefore should be screened regularly and beginning at a

younger age. There is evidence, however, that immigrants may have lower

health care utilization (Kliewer & Kazanjian, 2000), which may predispose this

group to have lower rates of screening than the Canadian-born population. An

important and currently unanswered question therefore, is whether some ethnic

or migrant groups are more likely to be ‘under-diagnosed’ than others. In this

study, we describe the pattern of diabetes screening among recent immigrants to

Ontario by looking at screening rates, screening efficiency/yield, predictors of

screening and the burden of undiagnosed diabetes in this population by region of

origin.

53

3.2 Research Design and Methods

3.2.1 Study Population

We conducted a retrospective cohort study to examine rates of screening for

diabetes among immigrants to Canada compared to those in the general

population during the 3-year period from April 1, 2004 (the baseline date for this

study) to March 31, 2007. To do so, all adults aged 40 or older (based on

Canadian screening recommendations) who were living in Ontario during the 3-

year period prior to baseline (from April 1, 2001) were identified from the

Registered Persons Database (RPDB), an electronic registry of all individuals

who are eligible for health coverage in Ontario. In order to identify immigrants to

Canada, RPDB records were linked to immigration data from Citizenship and

Immigration Canada (CIC), which contains information on all individuals having

been granted permanent residency in Canada between 1985 and 2000

(N=1,377,816). This database includes demographic and socioeconomic

information collected at the time of application for immigration status. Eighty-four

percent of CIC records were linked to the RPDB using probabilistic linkage

techniques. Feasibility of linkage between the CIC and health administrative

datasets was tested in pilot projects (Kliewer & Kazanjian, 2000), and differences

in linkage by immigration variables in these previous studies were found to be

small and unlikely to produce significant bias in study results. For the purpose of

this study, the general population comprised those who did not have a record of

immigration between 1985 and 2000, so individuals having immigrated prior to

54

1985 were included in this group. Furthermore, in order to avoid misclassifying

immigrants who were not captured in the CIC data linkage as non-immigrants,

individuals in the general population were excluded from the study if they first

became eligible for provincial health insurance after 1991. Nineteen-ninety-one

is the first date for which administrative data on health insurance eligibility in

Ontario is available. The majority of these excluded adults are likely to be

external migrants not captured by the CIC data with a small proportion comprised

of internal migrants arriving from another province.

Individuals with a diagnosis of diabetes at baseline, which accounted for roughly

11% (422,878 individuals) and 12% (59,766 individuals) of our general

population and immigrant cohorts, respectively, were excluded from the study.

Those who had no health care contact between April 1, 1999 (5 years before

baseline) and March 31, 2007 (end of 3-year observation period) were also

excluded. Since 98% of all immigrants in our database settled in urban areas, we

excluded rural populations using a Statistics Canada algorithm based on postal

codes. This resulted in the further exclusion of 2.0% (12,092) of immigrants and

17.3% (922,028) of long-term residents from the study.

3.2.2 Study Outcomes

Screening rates

Diagnosis of diabetes prior to, or during the study period was established by

linking the study population to the Ontario Diabetes Database (ODD), a validated

55

population-based, cumulative, diabetes registry based on physician visits and

hospitalizations for diabetes, excluding gestational diabetes (Hux et al., 2002).

We determined the percentage of people without prior diabetes diagnosis, who

were screened within the 3-year follow-up, along with 95% confidence intervals.

Under the universal health insurance program in Ontario, over 95% of health

services provided are captured in provincial, administrative data (Williams &

Young W, 1996), allowing us to identify what services, including laboratory tests,

were billed and when with the exception of a very small proportion of tests

conducted in hospitals. Provincial health services data were linked to our study

population by encrypted individual health card number. In the 3 year study follow-

up, individuals were considered to be screened for diabetes if they had one or

more physician or laboratory billing for a serum blood glucose, hemoglobin A1c

or a non-pregnancy related oral glucose tolerance test. Due to our use of

administrative data, we could not differentiate whether the test was for screening

(in asymptomatic individuals) or diagnosis (in symptomatic individuals).

Screening efficiency

Screening efficiency (defined as the percent positive of the total screened with

previously undetected diabetes) was measured. We also calculated the

reciprocal of screening efficiency, the number needed to screen (NNS) within

each risk group to identify one previously undiagnosed case of diabetes (NNS =

Total number screened / Total number of newly diagnosed cases).

56

Burden of undiagnosed diabetes

Finally, based on the yield of new diabetes cases among the screened

population, we estimated the number of people with undiagnosed diabetes we

would expect to find in the unscreened population on March 31st, 2007 using the

formula:.

Undiagnosed cases = Total unscreened population X screening efficiency

(Wilson et al., 2010).

The proportion of all diabetes in the population that is undiagnosed was then

estimated by dividing the number of undiagnosed cases by the total number of

people with diabetes. Total cases of diabetes was calculated as the sum of all

diagnosed (both prevalent at baseline as well as newly diagnosed during the

study period) and undiagnosed cases.

3.2.3 Statistical analysis

All analyses were performed by world region of origin and were stratified by sex

since there is evidence supporting a larger proportion of undiagnosed diabetes in

men than in women (Wilson et al., 2010). Comparisons across sub-groups for the

descriptive analyses above were conducted using chi-square tests.

Along with the descriptive analyses described above, multivariate log-binomial

regression was used to determine the association between receiving a diabetes

test within the recommended time frame and the covariates of interest. Three

57

different models were fit: 1) adjusted model to determine characteristics of those

having a diabetes test within the recommended period; 2) same as model 1 but

including number of primary care physician visits during the study period to

adjust for patterns of utilization; 3) adjusted model to determine the predictors of

being tested in any one visit (as opposed to being tested at any point in the 3

years of the study observation period, as with models 1 and 2). The latter model

was generated using the number of visits up to the first diabetes test as an offset

in the model and a Poisson distribution.

Covariates included in the model were age (40-49, 50-59, 60+), world region of

birth, immigration visa category, educational qualifications at time of immigration,

time in Canada (as of April 1st, 2004), income (based on residential postal code)

and number of physician visits (derived from physician billing data and excluding

specialist visits), during the study period. Due to the absence of individual-level

income information in provincial health administrative databases, residential

postal codes were linked to 2006 Canada Census data at the dissemination-area

level (an area containing roughly 400-600 people) using a Postal Code

Conversion File (PCCF+ v. 5D). Relative income quintiles adjusted for household

and community size were then generated and assigned to individuals.

All analyses were performed using SAS (version 9.2). This protocol received

ethical approval from the Institutional Review Board at Sunnybrook Health

Sciences Centre and the University of Toronto.

58

3.3 Results

3.3.1 Baseline Study Characteristics

A total of 3,927,059 individuals were observed for the 3-year period. Compared

with the general Ontario population, immigrants were younger, more likely to be

male and more likely to live in low income neighbourhoods (Table 3.1). The

largest proportion of immigrants was from Asia and Eastern Europe. The majority

of people immigrated under the Economic (including investors, entrepreneurs,

skilled workers) and Family (predominantly family reunification and sponsorship)

visa categories. Over the 3-year period 212,137 new cases of diabetes were

identified.

59

Table 3.1 Baseline characteristics of the urban* Ontario general population (excluding immigrant cohort) and immigrant study populations†, aged 40 and up and diabetes-free on April 1, 2004. General

Population Immigrant

Cohort Study Population Characteristics Population 3,484,222 442,837 Median Age‡ 54 48 % Male 46.5 48.9 Income quintile§ of neighbourhood of settlement:

Q1 (lowest income) 17.6 27.6 Q2 19.1 23.1 Q3 19.1 19.7 Q4 20.4 16.8 Q5 23.6 12.6

World Region of Birth East Asia & the Pacific - 133,360 (30.1%)

Eastern Europe & Central Asia - 78,098 (17.6%) South Asia - 73,212 (16.5%)

Western Europe & U.S.A. - 37,183 (8.4%) Mexico & Latin America - 35,009 (7.9%)

North Africa & the Middle East - 32,596 (7.4%) Caribbean - 29.758 (6.7%)

Sub-Saharan Africa - 23,246 (5.2%) Unknown/Stateless - 375 (0.1%)

Immigration Visa Category Economic - 194,584 (43.9%)

Family - 158,652 (35.8%) Refugee - 77,680 (17.5%)

Other - 11,915 (2.7%) Educational Qualifications at Landing (%)

No Education - 12,469 (2.8%) Secondary or Less - 204,833 (46.3%)

Non-University Qualifications - 90,288 (20.4%) Some University - 22,277 (5.0%)

University Degree or Higher - 112,933 (25.5%)

Years Since Arrival (using 2004 as year of reference) (%) 4 - 9 years - 137,339 (31.0%)

10 - 14 years - 156,663 (35.4% 15 years or more - 148,835 (33.6%)

*Urban areas identified from first three characters of the postal code of residence (the Forward Sortation Area (FSA)).

†Urban population eligible for provincial health care between April 1, 2004 and March 31, 2007, based on administrative databases.

‡ Based on age as of April 1st, 2004. § 2006 Census income information was applied based on the individual's postal code of residence on April 1, 2004.

60

3.3.2 Diabetes screening

Diabetes testing rates were high. Although statistically significant, the difference

in screening rates between immigrants overall and the general population were

small (76.0% vs. 74.4%, p<0.0001) (Table 3.2). There were differences by region

of birth whereby people born in East and South Asia, North Africa, the

Caribbean, Mexico, Latin America, and the Middle East were screened more

than the general population (all differences, p<0.0001). Screening rates

increased with age in the general population; however, the increase was minimal

for immigrant men and rates decreased with age among immigrant women. Over

the age of 65, immigrants were screened less than the general population

(75.9% vs. 83.2% and 77.7% vs. 84.8% in males and females, respectively, both

p<0.0001). Women, both in the immigrant cohort and in the general population,

were screened more than men (p<0.0001).

3.3.3 Screening Efficiency

Screening efficiency was similar although statistically higher in immigrants (with

8.1% diagnosed with diabetes) than in the general population (7.1%) (p<0.0001),

and it was higher in men than in women (p<0.0001) (Table 3.2). Screening

61

Tabl

e 3.

2 M

easu

res

of s

cree

ning

upt

ake

and

effic

ienc

y: T

he n

umbe

r and

per

cent

of t

he s

tudy

pop

ulat

ion

(ove

rall,

and

by

sex)

with

no

prev

ious

dia

gnos

is o

f dia

bete

s, h

avin

g a

labo

rato

ry te

st* t

o sc

reen

for d

iabe

tes

in

the

3-ye

ar p

erio

d, 2

004-

2007

; the

pro

port

ion

of th

ose

scre

ened

with

new

ly d

iagn

osed

dia

bete

s (s

cree

ning

ef

ficie

ncy)

; and

the

num

ber n

eede

d to

scr

een

to id

entif

y on

e ne

w c

ase.

Gen

eral

Po

pula

tion

Imm

igra

nt

Coh

ort

Wor

ld R

egio

n of

Orig

in

Ea

st A

sia

& P

acifi

c

E.

Euro

pe &

C

entr

al

Asi

a

Mex

. &

Latin

A

mer

ica

Car

ibbe

an

N. A

fric

a &

Mid

dle

East

So

uth

Asi

a Su

b-Sa

hara

n A

fric

a

W.

Euro

pe

& U

.S.

% S

cree

ned

(N)

Mal

e, A

ge 4

0+

69.7

71

.8

73.4

67

.5

72.6

69

.6

71.9

78

.3

69.8

63

.8

(1

,128

,581

) (1

54,8

79)

(43,

999)

(2

5,80

2)

(12,

093)

(9

,147

) (1

3,08

7)

(29,

970)

(9

,086

) (1

1,69

5)

40-

64

66.1

71

.2

73.0

66

.9

72.0

69

.4

71.3

78

.3

69.3

63

.4

(8

50,7

76)

(136

,987

) (3

6,51

4)

(23,

904)

(1

0,99

1)

(8,4

85)

(11,

737)

(2

6,13

9)

(8,5

90)

(10,

627)

65+

83.2

75

.9

75.4

75

.6

78.8

71

.9

77.4

78

.7

78.7

68

.2

(2

27,8

05)

(17,

892)

(7

,845

) (1

,898

) (1

,102

) (6

62)

(1,3

50)

(3,8

31)

(496

) (1

,068

) Fe

mal

e, A

ge 4

0+

78.5

80

.1

79.7

78

.5

83.8

83

.5

81.5

83

.4

80.3

71

.1

(1

,464

,315

) (1

81,0

04)

(58,

363)

(3

1,21

9)

(15,

323)

(1

3,82

7)

(11,

682)

(2

9,08

7)

(8,1

444)

(1

3,35

9)

40-

64

76.3

80

.6

80.4

78

.4

84.1

84

.5

82.0

84

.4

80.4

71

.1

(1

,049

,932

) (1

53,2

07)

(47,

770)

(2

6,98

0)

(13,

278)

(1

2,32

4)

(10,

096)

(2

4,25

0)

(7,1

42)

(11,

367)

65+

84

.8

77.7

77

.0

79.3

82

.1

76.6

78

.2

78.8

79

.4

70.9

(414

,383

) (2

7,79

7)

(10,

593)

(4

,239

) (2

,045

) (1

,503

) (1

,586

) (4

,837

) (1

,002

) (1

,992

) B

oth

sexe

s, a

ge 4

0+

74.4

76

.0

76.9

73

.1

78.5

77

.3

76.1

80

.7

74.4

67

.5

(2

,592

,896

) (3

35,8

83)

(102

,362

) (5

7,02

1)

(27,

416)

(2

2,97

4)

(24,

769)

(5

9,05

7)

(17,

230)

(2

5,05

4)

%

of T

otal

Sc

reen

ed w

ith

New

ly D

iagn

osed

D

iabe

tes

Mal

e, A

ge 4

0+

8.5

9.3

8.6

6.1

9.8

10.3

8.

2 14

.3

9.0

6.4

40-

64

7.6

8.8

7.8

5.6

9.3

9.8

7.8

14.0

8.

8 5.

9

6

5+

11.1

13

.2

12.1

11

.9

14.7

17

.1

11.9

16

.2

11.7

11

.0

Fem

ale,

Age

40+

6.

1 7.

1 6.

4 4.

6 8.

1 9.

0 6.

5 11

.7

6.8

4.0

40-6

4 5.

2 6.

3 5.

4 3.

7 7.

4 8.

2 5.

8 11

.1

6.4

3.4

65+

8.4

11.7

11

.2

9.7

13.2

15

.2

10.9

14

.8

9.7

7.8

62

Bot

h se

xes,

ag

e 40

+ 7.

1 8.

1 7.

3 5.

2 8.

9 9.

5 7.

4 13

.0

7.9

5.1

N

umbe

r Nee

ded

to

Scre

en †

Mal

e, A

ge 4

0+

12

11

12

17

10

10

12

7 11

16

40-

64

13

11

13

18

11

10

13

7 11

17

6

5+

9 8

8 8

7 6

8 6

9 9

Fem

ale,

Age

40+

16

14

16

22

12

11

15

9

15

25

4

0-64

19

16

19

27

14

12

17

9

16

30

65+

12

9

9 10

8

7 9

7 10

13

B

oth

sexe

s, a

ge

40+

14

12

14

19

11

11

14

8 13

20

* Tes

ts in

clud

e S

erum

Blo

od G

luco

se (G

002,

L11

L11

2), H

bAIC

(L09

3) o

r OG

TT (L

104)

. Pre

gnan

cy-r

elat

ed te

sts

are

excl

uded

†

Cal

cula

ted

as th

e to

tal n

umbe

r of i

ndiv

idua

ls s

cree

ned

by th

e nu

mbe

r of n

ew c

ases

det

ecte

d.

63

efficiency was highest in people from South Asia (13.0% of people screened had

undiagnosed diabetes) followed by the Caribbean (9.5%) and Mexico and Latin

America (8.9%) particularly among seniors from these regions. The lowest

screening efficiency was in immigrants from Europe, the United States and

Central Asia (5.1-5.2%).

The number needed to screen (NNS) to identify one new case was lowest in men

and women from South Asia (NNS=8), followed by the Caribbean (NNS=11) and

Mexico and Latin America (NNS=11).

3.3.4 Predictors of diabetes screening

Model 1 showed that: male gender, age under 50, living in the lowest income

neighbourhoods, being born in Western Europe or the U.S., immigrating under

the family reunification visa category and living in Canada for less than 15 years

were all associated with lower rates of diabetes screening (Table 3.3). When

number of physician visits was added to the model (model 2), it was by far the

strongest predictor of whether or not a person received a diabetes test and all

other effects were attenuated. Although attenuated, being born in a non-Western

European country and female gender were still predictive of receiving a diabetes

test. Conversely, living in the lowest income neighbourhoods, having no formal

education and being less than 50 years of age were still associated with not

being screened, even after all other variables were controlled for.

64

Table 3.3 Predictors of receiving a diabetes screen test during the 3-year study period (April 1, 2004 - March 31, 2007): results of regression analyses. Study population limited to immigrants without prior diagnosed diabetes, aged 40 and over (N = 442,837). Adjusted Rate Ratio (95% confidence interval)

Model 1: Adjusted model

Model 2: Adjusted model including

utilization measure*

Model 3: Adjusted model, probability of screening per visit

Sex F Reference Reference Reference M 0.902 (0.899-

0.905) * 0.986 (0.899-

0.988) * 1.106 (1.099-

1.1114) * Age Group

40-49 0.973 (0.969-0.978) *

0.978 (0.975-0.980) *

1.135 (1.124-1.146) *

50-59 1.019 (1.014-1.024) *

0.999 (0.996-1.002)

1.144 (1.131-1.157)*

60+ Reference Reference Reference # physician visits† during study period

0-1 - Reference - 2-5 - 6.275 (6.132-

6.421) * -

6-10 - 8.029 (7.851-8.213) *

-

11+ - 8.731 (8.538-8.93) *

-

Income quintile ‡ of residential neighbourhood (%)

Q1(lowest income)

0.989 (0.983-0.995) §

0.988 (0.985-0.991) *

0.874 (0.863-0.884) *

Q2 1.018 (1.012-1.024) *

0.997 (0.993-1.000)

0.924 (0.913-0.935) *

Q3 1.034 (1.028-1.04) *

0.999 (0.996-1.002)

0.940 (0.929-0.952) *

Q4 1.037 (1.031-1.043) *

1.004 (1.001-1.007)

0.961 (0.949-0.973) *

Q5 Reference Reference Reference World Region of Birth

Western Europe & U.S.

Reference Reference Reference

East Asia & the Pacific

1.145 (1.137-1.154) *

1.055 (1.049-1.060) *

1.032 (1.018-1.047) *

South Asia 1.223 (1.213-1.233) *

1.058 (1.053-1.064) *

0.940 (0.926-0.955) *

Mexico and Latin America

1.158 (1.147-1.168) *

1.047 (1.041-1.053) *

0.984 (0.967-1.001)

The Caribbean 1.143 (1.132-1.153) *

1.046 (1.039-1.052) *

1.018 (1.000-1.037)

Eastern Europe 1.101 (1.092- 1.036 (1.031- 1.063 (1.047-

65

& Central Asia 1.11) * 1.042) * 1.079) * North Africa &

the Middle East 1.157 (1.146-

1.167) * 1.058 (1.052-

1.064) * 0.989 (0.972-

1.007) Sub-Saharan

Africa 1.122 (1.110-

1.133) * 1.039 (1.033-

1.046) * 0.945 (0.927-

0.964) * Immigration Visa Category

Economic Reference Reference Reference Family 0.986 (0.982-

0.990) * 0.984 (0.982-

0.986) * 0.917 (0.909-

0.925)* Refugee 1.005 (0.969-

0.978) 0.981 (0.978-

0.984) * 0.904 (0.895-

0.913)* Other 0.991 (0.982-

1.001) 0.983 (0.978-

0.989) * 0.943 (0.923-

0.963)* Educational Qualifications at Landing (%)

No Education 1.000 (0.990-1.010)

0.990 (0.985-0.996) §

0.860 (0.842-0.880) *

Secondary or Less

1.051 (1.047-1.056) *

1.005 (1.002-1.007) *

0.916 (0.908-0.925) *

Non-University Qualifications

1.039 (1.034-1.044) *

1.003 (1.000-1.006)

0.957 (0.947-0.967) *

Some University 1.022 (1.014-1.031) *

0.999 (0.995-1.004)

0.963 (0.947-0.979) *

University Degree or

Higher

Reference Reference Reference

Time in Canada

4-9 years 0.931 (0.927-0.935) *

1.005 (1.003-1.007) *

1.062 (1.054-1.072) *

10-15 years 0.978 (0.974-0.981)*

1.006 (1.004-1.008) *

1.037 (1.029-1.046) *

>15 years Reference Reference Reference *p <0.0001 † Sum of physician visits (excluding specialists) during the 3-year observation period ‡ Dissemination area (DA)-level income quintile derived from residential postal code and adjusted for family size and community size § p<0.001

66

When the probability of being tested per visit was modeled (Model 3), we found

that although women were more likely to be tested overall, in any given visit, men

were more likely to be tested. Similarly, per visit, adults aged 40-59 were more

likely to be tested than seniors. Immigrants from all regions of the world except

Eastern Europe, Central and East Asia were less likely to be tested per visit than

people from Western Europe and the U.S., our lowest diabetes risk group.

Compared with the highest income quintile and highest education category, all

other income and education categories had a lower probability of being tested

per visit, with the lowest probability in the lowest income and education groups.

3.3.5 Undiagnosed Diabetes

Despite high rates of screening, there was still a large number of people with

undiagnosed diabetes estimated among the newcomer South Asian population

(1,832 undiagnosed cases), due to the high diabetes prevalence in this

population. The highest burden of undiagnosed cases among immigrants,

however, was estimated to be in people from East Asia and the Pacific (2,259

undiagnosed cases) primarily due to the large number of newcomers from that

region (Table 3.4). When we estimated the percent of total diabetes cases that

was undiagnosed, we found the percent ranged from 5.3% in women from South

Asia, to 16.6-16.7% in men from Europe, the U.S. and Central Asia. Overall,

immigrants and the general Ontario population had a similar proportion of

undiagnosed cases, and both had a higher proportion undiagnosed among men

(11.2% vs. 7.1% among immigrant males and females, respectively, p<0.0001;

67

Table 3.4 Estimated number and percentage of 'undiagnosed' diabetes cases by world region and immigration status, 2004-2007, among those aged 40 and up with no prior diabetes diagnosis on April 1, 2004. Number of diagnosed

cases* in 2007, persons aged 40+

Estimated Number of ‘undiagnosed’ cases†

Percent (%) Undiagnosed‡

Population All Male Female All§ Male Female All Male Female General Population

607,742 315,010 292,732 65,391 42,428 25,193 9.7 11.9 7.9

Immigrant Cohort

86,923 44,774 42,149 8,597 5,652 3,204 9.0 11.2 7.1

By World Region of Birth

East Asia & Pacific

22,528 10,833 11,695 2,259 1,365 951 9.1 11.2 7.5

E. Europe & C. Asia

7,321 3,773 3,548 1,098 754 389 13.0 16.7 9.9

Mex. & Latin

America

8,129 3,876 4,253 666 446 241 7.6 10.3 5.4

Caribbean 7,707 3,075 4,632 640 413 245 7.7 11.8 5.0 N. Africa &

Middle East 5,822 3,388 2,434 575 419 173 9.0 11.0 6.6

South Asia 26,947 15,004 11,943 1,832 1,185 676 6.4 7.3 5.4 Sub-

Saharan Africa

4,473 2,712 1,761 471 353 136 9.5 11.5 7.2

W. Europe & U.S.

3,996 2,113 1,883 618 421 219 13.4 16.6 10.4

*Includes cases diagnosed during the 3-year study observation (2004-2007) period and prevalent cases at baseline (2004) from the Ontario Diabetes Database. † Calculated as the Total unscreened population multiplied by the screening efficiency ( # unscreened * (Newly diagnosed cases/Total screened)) ‡ As a proportion of all true diabetes cases (diagnosed + undiagnosed) in the population. § Male and female estimates do not sum to the total since the estimates were modeled for each group separately.

68

11.9% vs. 8.9% among general Ontario population males and females,

respectively, p<0.0001).

3.4 Discussion

We found a high rate of diabetes screening in our immigrant study population,

particularly among the groups with the highest risk for diabetes as shown in

previous work (Creatore et al., 2010). In addition, we found that in this highly

screened population, the number needed to screen to identify one new case of

diabetes was still low and the screening efficiency was very high. These results

are consistent with the recent ADDITION-Leicester trial that found among the

screened population, South Asians had a 2-fold higher risk of presenting with a

previously undiagnosed glucose disorder as compared with white Europeans

(Webb et al., 2011). These findings are important because early screening has

the potential to reduce the long-term risk of diabetes complications through timely

control of blood glucose and early initiation of cardiovascular risk reduction

therapy (Colagiuri & Davies, 2009; Holman et al., 2008; Kahn et al., 2010;

Sandbaek et al., 2008). In particular, targeted screening of high-risk ethnic

groups, including South Asians, may provide an opportunity for considerable

population health gains through cardiovascular risk reduction interventions. In

addition, periods of poor glycemic control have been shown to have long-lasting

69

effects, further emphasizing the importance of identifying people with diabetes

and asymptomatic hyperglycemia (Chalmers & Cooper, 2008).

The most significant driver of an individual being screened for diabetes was

frequent contact with a physician. This finding is unsurprising if the majority of

tests were due to opportunistic screening of patients as part of routine care.

Targeted and step-wise screening programs have typically reported lower

screening rates (van den Donk et al., 2011). Although they achieved high rates of

screening, many of our high risk groups required multiple visits to do so, and their

likelihood of getting tested in any one particular visit was actually lower than in

other groups. Of possible concern was the relatively low screening rate observed

among immigrant seniors as compared to seniors in the general population. This

finding suggests that the oldest immigrants may face additional barriers to care

that need to be addressed in order to reduce this disparity. Another important

finding is that in our universal health care setting, with no overt financial barriers

to screening, the percent of all diabetes that was undiagnosed was lower than

reported in past literature (Cowie et al., 2006; Leiter et al., 2001). This suggests

that in Canada, recommendations from clinical practice guidelines on screening

have been adopted into practice and a low proportion of patients, particularly

among high risk groups, are undiagnosed. This is supported by results of two

earlier studies that found that the majority of adults in Ontario aged 40+ are being

screened for diabetes and that non-white ethnicity and immigrant status were

associated with an increased likelihood of being screened (Wilson et al., 2010).

Both the above findings are dependent on good access to primary care which

70

may have serious implications in settings and for populations that experience

barriers to care – whether due to low physician supply, geographic or insurance-

based issues.

The prevalence of undiagnosed diabetes is related to the overall prevalence of

diabetes and diabetes-related risk factors in the population, as well as the use of

health services by the population at risk and the screening policies in the

jurisdiction. In the U.S. the 2005-2006 National Health and Nutrition Examination

Survey found that the percentage of undiagnosed diabetes, although high at

nearly 40%, has remained relatively stable over the last 10-15 years overall but

has decreased significantly in Mexican Americans (Cowie et al., 2009). These

results, in combination with our study, suggest that the proportion of undiagnosed

cases in the U.S and Canada has remained stable or even decreased in the past

10 years despite an increasing prevalence of diabetes, and that higher risk

groups are being screened. We found that roughly two-thirds of all undiagnosed

cases of diabetes are men, a finding supported by previous research (Wilson et

al., 2010).

One limitation of this study is that our estimate of the number of people with

undiagnosed diabetes assumes that the prevalence among those that are

screened is equivalent to the prevalence among those unscreened. This

assumption may not hold if some people with diabetes related health issues or

risk factors (such as high BMI or family history) may be more likely to be in

contact with the health care system and more likely to be tested for diabetes. In

71

that case our proportion of undiagnosed diabetes would actually be a “worst case

scenario” and the true proportion would actually be lower. However it is possible

that the opposite may also be true and prevalence may be higher in the untested

group since diabetes and health care access share some risk factors. Due to our

reliance on administrative data, we were also subject to the restrictions of the

available data. For instance, the administrative data do not contain clinical

information on risk factors such as body mass or family history, so we were

unable to adjust for these factors in our analyses. The immigration data was also

limited to individuals immigrating to Canada between 1985 and 2000, thus our

study population did not include the most recent immigrants (those arriving

between 2001 and 2004) who may experience the greatest barriers to accessing

health services. We also cannot ascertain from the administrative data whether

an individual was screened as part of routine medical care, or based on

symptoms, nor do we differentiate between type of test used. We do not feel that

the latter is a major limitation since the objectives of this study were not to

investigate reasons for screening or type of test used; we were interested in

overall screening rates and to identify the screening patterns in immigrant

populations and by ethnicity. A further limitation is that individuals with no health

system contact in the 5 years prior to baseline and the 3 years of follow-up (a

total of 8 years) were excluded since we could not ascertain their continued

residency in the province. Although these individuals could be under-utilizing the

health system, they comprise a very small proportion of the total population (81%

of Canadians see a healthcare provider annually) (Statistics Canada, 2006) and

72

many of these individuals may have emigrated from the province. Finally,

whether an individual is screened for diabetes depends on individual, physician,

social and system factors that are not all available in our data. By adjusting for

utilization of primary care, we have attempted to disentangle some of the effects

of overall access.

Population-wide screening for diabetes is still controversial. However

opportunistic screening of high-risk groups is increasingly recommended

(Sandbaek et al., 2008; Janssen et al., 2009; Colagiuri & Davies, 2009) and

recent evidence suggests that aggressive treatment of screen-detected patients

with diabetes results in a significant improvement in their cardiovascular risk

profile (Griffin et al., 2011). Recommendations state that individuals belonging to

high-risk ethnic groups should be screened regularly and beginning at a younger

age, which may pose a challenge if immigrant groups have less contact with the

health care system or face barriers accessing care as suggested by some

studies (Kliewer & Kazanjian, 2000; Webb et al., 2011). In our universal health

care system we found no evidence of lower screening in immigrants, nor did we

find disparities in screening by region of birth (i.e. the highest risk groups were

being screened more than the lowest risk groups) and we found a fairly low

proportion of undiagnosed cases. A diabetes screening rate of 76% as found in

our study compares favorably with other screening programs in the same setting

such as cervical and breast cancer screening, which are currently reported as

occurring in 61% and 59% of the recommended population, respectively (Lofters

et al., 2010; Swanson & Kaczorowski, 2007). Ideally, in a universal health care

73

setting, 100% of screening guidelines/targets would be met, but this is rarely the

case and special efforts have to be made to reach at-risk populations. One

possible concern was that many immigrant groups required frequent physician

visits to achieve the observed high rates of screening which may have serious

implications for settings where there is poor, or inequitable, access to health

care. These results also suggest that in addition to universal access to physician

services, there are other important factors that must be identified and addressed

in order to achieve high rates of diabetes screening.

74

Chapter 4

A Population-Based Study of Diabetes Incidence by Ethnicity and Age: Support for the Development of

Ethnic-Specific Age Guidelines for Screening

75

Abstract

Background: Diabetes incidence rates vary significantly by ethnicity. Not only is the

prevalence of type 2 diabetes higher in certain ethnic groups, but there is evidence that

the onset of disease may occur at younger ages in these populations, particularly

among South Asians. Currently, National diabetes clinical care guidelines suggest

initiating screening in the general population at age 40 while recommending that a

younger threshold be considered for individuals belonging to high risk ethnic groups.

However, the optimal age to initiate screening among these individuals remains unclear.

Methods: We conducted a longitudinal, population-based, retrospective rolling cohort

study using linked administrative health and immigration records for 592,376 immigrants

to Ontario, Canada. We used Cox-Proportional Hazard models to generate adjusted

incidence rates by ethnicity, sex and age and determined the age cut-offs at which

different ethnicities experienced equivalent risk of developing diabetes.

Results: Individuals from South Asia had the highest overall incidence rates with an

age-sex adjusted rate of 15.7 per 1,000 person-years, which was 3.3 times higher than

in the Western European group (p < 0.05). Individuals from the Caribbean, Latin

America, Mexico, Africa, the Middle East, East Asia and the Pacific all had a higher

overall incidence rate as compared with the Western European population (for all p <

0.05). Men from all regions consistently had higher incidence rates than women with the

exception of the Caribbean where women experienced the highest risk. The risk for

developing diabetes among 40-44 year-old Western European men was 3.7 per 1,000

76

person-years, and was roughly equivalent to the risk experienced by South Asian men

between age 25 and 29. For women, the diabetes risk in those of Western European

origin at age 40-44 (1.7 per 1,000 person years) was lower than observed at any age in

South Asian women. These differences in risk persisted despite controlling for income,

education, immigration category and time in Canada.

Conclusions: The risk of diabetes in South Asians and other high risk ethnicities rose

rapidly after age 20, resulting in a significant divergence in risk between ethnic groups

by age 30. By age 40, there appeared to be a marked disparity in incidence between

South Asian and Western European populations – in the order of 5-fold for men and 9-

fold for women. In order to target a similar level of risk for diabetes as is seen in men of

Western European ethnicity at age 40 one would have to initiate screening among

South Asians at age 25. For all other non-European ethnic groups, in order to target an

equivalent risk, screening would have to be initiated between age 30 and 35.

77

4.1 Introduction

Diabetes is a serious chronic disease that increases mortality, morbidity and disability in

the population. In Canada, as in many areas of the world, diabetes prevalence has

been increasing dramatically over the past 20 years (Lipscombe & Hux, 2007; Finucane

et al., 2011; Wild et al., 2004; Danaei et al., 2011). Diabetes incidence rates vary

significantly by ethnicity, and people of Aboriginal, South Asian, African, Hispanic and

Middle Eastern descent have all been shown to have increased risk for developing

diabetes as compared with individuals of Western European/Caucasian ancestry (Khan

et al., 2011; Oza-Frank et al., 2009; Chiu et al., 2010).

Not only is the prevalence of type 2 diabetes higher in certain ethnic groups, but there is

evidence that the onset of disease may occur at younger ages in these populations,

particularly among South Asians (Chiu et al., 2011; Khan et al., 2011; Qiao et al., 2003;

Feltbower et al., 2003). Currently, National diabetes clinical care guidelines suggest

initiating screening in the general population at age 40 in Canada and the UK, and at

age 45 in the U.S., while recommending that a younger threshold should be considered

for individuals with additional risk factors (such as belonging to a high-risk ethnic group)

(Ur et al., 2008; American Diabetes Association, 2010; NHS, 2012). However, current

age guidelines are largely based on expert opinion and the optimal age to initiate

screening among high-risk ethnic groups remains unclear.

Data on ethnic differences in age at diagnosis come predominantly from cross-sectional

prevalence studies, and incidence data have generally come from surveys or relatively

78

small samples that have allowed for inclusion of only a limited number of ethnic groups

and age ranges. To date, no longitudinal, multi-ethnic, population-based studies have

been conducted looking at age-specific diabetes incidence across adulthood.

Understanding ethnic differences in age of onset and age-related risk has practical

implications in terms of designing and implementing primary prevention programs and

developing screening guidelines. Due to the microvascular and macrovascular damage

associated with untreated or inadequately controlled disease, it also has serious

implications for prevention of cardiovascular disease and other diabetes-related

complications.

The primary objectives of this population-based cohort study were: 1) to examine the

population age distribution of diabetes incidence by ethnicity; 2) to determine the age

cut-offs at which different ethnicities experienced equivalent risk of developing diabetes.

Since age recommendations have largely been established in Western European and

North American populations, we used risk at age 40 among men of Western European

origin as a baseline for comparison. By determining ethnic-specific risk-equivalent ages

for diabetes incidence we hope to help inform the development of ethnic-appropriate

age guidelines for diabetes screening.

79

4.2 Research Design and Methods

4.2.1 Study design

We conducted a longitudinal, population-based, retrospective rolling cohort study linking

administrative health claims to immigration records in Ontario, Canada. In the province

of Ontario, over 95% of health services provided are captured in provincial,

administrative data under the universal health insurance program (Williams & Young W,

1996).

This protocol received ethical approval from the Institutional Review Board at

Sunnybrook Health Sciences Centre and the University of Toronto.

4.2.2 Study Population

Since administrative health records in Canada do not contain information on ethnicity,

we identified a cohort of individuals through immigration records from Citizenship and

Immigration Canada (CIC). This database contains information on all individuals

granted permanent residency in the province of Ontario between 1985 and 2000

(N=1,377,816) and includes demographic and socioeconomic information collected at

the time of application for immigration status including country of birth. These individuals

were then linked to the Registered Persons Database (RPDB), an electronic registry of

all individuals who are eligible for health coverage in Ontario. Health care eligibility is

extended to virtually all Canadian citizens, permanent residents or landed immigrants

who have Ontario as their primary place of residence. Feasibility of linkage between the

80

CIC and health administrative datasets was tested in pilot projects, and differences in

linkage by immigration variables in these previous studies were found to be small and

unlikely to produce significant bias in study results (Kliewer & Kazanjian, 2000). Eighty-

four percent of CIC records were linked to the RPDB using probabilistic linkage

techniques.

As the main outcome of this study was type 2 diabetes incidence, individuals with a

diagnosis of diabetes at baseline were excluded from the study. Furthermore, we

restricted our cohort to those aged 20 or over at baseline. A three year look-back

window was used to ensure that individuals were diabetes free at cohort entry.

Therefore, for the study cohort, baseline was defined as three years after first enrolling

in the provincial health insurance program (which normally occurs at immigration or

shortly thereafter). Diabetes incidence was only available in our data from 1994 so we

limited our cohort to those arriving on or after January 1st, 1991. Thus, accounting for

the 3-year look-back period, people continually entered the cohort from 1994 to 2003.

Any individual who had no contact with the health care sector for the duration of the

study follow-up and the three-year look back window was excluded from the study. This

was done since there is evidence that most Canadians (81%) have annual contact with

the health care system and there is a high probability that those with no contact over a

prolonged period did not actually reside in the province (Statistics Canada, 2006).

Finally, since 98% of all immigrants in our database settled in urban areas, we excluded

rural populations based on postal codes. Please see Appendix D for more information

on cohort creation.

81

4.2.3 Measures

Our primary covariates were age and country of birth which we used as a proxy for

ethnicity. Since our study included individuals born in 235 different countries, we

grouped countries into ethnically meaningful world regions. World regions were based

on the World Bank schema and then customized to better reflect ethnic composition as

a primary criteria (and not income or level of industrialization). Where there was

uncertainty as to the most appropriate grouping, countries were assessed individually

using local government websites, Wikipedia and other resources to determine the

dominant ethnic compositions. A list of countries by world region are available in

Appendix C. Other covariates included sex, education, visa category, year of arrival and

income. Income and education have been shown to be associated with diabetes risk

(Dalstra et al., 2005; Ross et al., 2010; Maty et al., 2010) and immigration visa category

may also be a potential confounder related to social status and pre-migration life

experiences that could impact health.

Finally, due to the nature of our study cohort we were faced with the issue of possible

age-period-cohort effects. Our study population consists of multiple waves of

immigrants arriving in different years, born in different time periods, and who arrive at

different ages. In order to address this potential issue, in addition to stratifying by age,

we also grouped individuals by their year of arrival to create three distinct periods of

immigration - those arriving between 1991 and 1993, 1994 and 1996 and 1997 and

2000 - to include as a covariate in the model. Please see appendix E for a more in-

depth discussion of age-period-cohort effects and how we addressed them in this study.

82

Age and sex were derived directly from health administrative records. Highest level of

education attained, visa category and year of arrival in Canada were based on

information collected at time of immigration. As individual-level information on income

was not available for our ethnic-specific study or control population, we used area-level

level income as a surrogate. To do this, the most updated residential postal code (which

corresponds roughly to a city block face) recorded for an individual was linked to

dissemination-area level data collected for the 2006 Canadian census. This Census

data was then used to generate relative income quintiles adjusted for household and

community size. This is a recognized method for attributing relative income and poverty

in our setting (Wilkins, 2008; Krieger, 1992).

4.2.4 Study outcomes

The study population was followed forward in time from baseline (i.e. 3 years after their

arrival in Ontario) for 5 years for the development of diabetes. Since people were

entering the study continuously between 1994 and 2003, the observation window

extended up to March 31st, 2008. Individuals were censored upon death, or loss of

health insurance eligibility (normally due to emigration).

A five-year period of follow-up was used in order to attempt to control for bias that may

be introduced by differences in the composition of earlier versus newer waves of

immigration cohorts and differential follow-up times. Simple age-standardization or

adjustments may not adequately adjust for these differences.

83

Individuals with incident diabetes were identified using the Ontario Diabetes Database

(ODD). The ODD is a validated electronic database that uses the following algorithm to

identify persons with diagnosed diabetes: individuals with at least one hospitalization or

at least two claims for physicians' services (within two years) bearing a diagnosis of

diabetes. Individuals with gestational diabetes are excluded. This algorithm was found

to be highly sensitive (86%) and specific (97%) for identifying patients in whom diabetes

was recorded in primary care charts (Hux et al., 2002). The same algorithm was used

to exclude individuals who had pre-existing diabetes at baseline from our study

population.

4.2.5 Analysis

All analyses were performed by world region of birth and were stratified by sex since

there is evidence supporting: A larger proportion of undiagnosed diabetes in men

(Rathmann et al., 2003; Leiter et al., 2001); sex differences in diabetes prevalence by

ethnicity (Creatore et al., 2010; Chiu et al., 2010; Jenum et al., 2005); and sex

differences in the prevalence of risk factors for the development of diabetes (Matheson

et al., 2008; Chiu et al., 2010; Meisinger et al., 2002).

All analyses were performed using SAS (version 9.2).

Incidence rates

Our first objective was to examine the population distribution of diabetes risk by age and

ethnicity. As a first step we used descriptive analyses to evaluate the pattern of

84

incidence rates in the population by age, sex and ethnicity, as well as other important

covariates and possible confounders. Incidence rates were calculated by dividing the

total number of new cases observed over a 5-year period from baseline within an age-

sex group, by the total number of person-years observed in this period. Individuals were

grouped into age categories based on age at baseline. Age-standardized and age-sex

standardized incidence rates were generated using the 1991 Canada census population

as a standard.

In a previous study we have shown that screening rates across all population groups

(age 40 and up) are very high, with a three-year uptake ranging from two-thirds to four-

fifths of the population (Creatore et al., 2012). However, in order to address the potential

bias caused by differential screening and diagnosis across our study groups, as a

sensitivity analysis we also generated age-standardized rates among those who had a

diabetes test within the study observation period. Ninety-five percent confidence

intervals were calculated for all rates.

Survival Analysis

In order to look at whether the risk of developing diabetes that is associated with age

differed by ethnicity, we used Cox Proportional Hazard models to perform strata specific

analyses. As with the descriptive analyses, diabetes-free individuals were followed up

from baseline to diabetes diagnosis, death, loss of eligibility or the end of the five-year

observation period. A diabetes diagnosis between the beginning and end of the

observation window constituted an event. Individuals with no diabetes diagnosis prior to

the end of the observation window, those who died in that period, or moved out of the

85

province were all censored. For a more detailed discussion of the methods used to

construct the Cox Proportional Hazard model, please refer to Appendix F.

The population was divided into 5-year age groups beginning at age 20 (i.e. 20-24, 25-

29, 30-34) up to age 65 and a model was run to estimate the risk for developing

diabetes for each age group over a 5-year observation period. From these models we

estimated the incidence of diabetes by sex and ethnicity within each age category

adjusted for income, education, immigration visa category and year of arrival. These

estimates were then plotted on a single graph for each sex to display incidence by age

and ethnicity. Both unadjusted and adjusted Cox models were run and results were

compared. In addition, because this method makes the assumption that the risk is

relatively stable within 5-year age groups, as a sensitivity analysis we also looked at 1-

year risk of developing diabetes for each age category and compared these results to

the 5-year model.

Proportional hazards assumptions were assessed graphically and tested for each

covariate using Schoenfeld residuals (Hosmer & Lemeshow, 1999). As the only

continuous variable in our model, we tested the linearity assumption for age within the

five year age categories graphically and by examining the Martingale residuals.

Determining Ethnic-specific Age Equivalency for Diabetes Risk

Our second objective was to determine the ethnic-specific ages by sex at which

diabetes risk would be equivalent to that experienced by men of Western European

ancestry at age 40 (the age when screening is recommended in Canada and the UK).

86

These numbers were derived from the Cox-Proportional Hazard model adjusted

average incidence estimates. Where estimates of equivalent risk fell between two

different 5-year age categories, the risk was rounded up to the lowest age of the next

highest age category. For example, if the equivalent risk fell between age categories 25-

29 and 30-34, then the estimated age would be recorded as 30.

4.3 Results

A total of 592,376 individuals representing 235 different countries of birth were included

in our study population. As compared with long-term residents in the adult (age 20 and

over) general population (which excludes the study population), the immigrants

comprising our study cohort were younger and more likely to live in low income

neighbourhoods (Table 4.1). The regions of birth accounting for the largest proportion of

our study population were East and South Asia. The majority of people immigrated

either under the Economic (including investors, entrepreneurs, skilled workers) or

Family (predominantly family reunification and sponsorship) visa categories. A total of

22,903 individuals were excluded due to a prior diabetes diagnosis and the excluded

population was older, more likely to be male, live in low income neighbourhoods and to

be from South Asia as compared with the study population (see Appendix G for

characteristics of excluded population).

87

Tabl

e 4.

1 B

asel

ine

char

acte

ristic

s of

the

Ont

ario

long

-term

resi

dent

and

rece

nt im

mig

rant

dia

bete

s-fr

ee s

tudy

pop

ulat

ions

*, by

wor

ld re

gion

of

birt

h.

Long

-te

rm

Res

iden

ts

Rec

ent I

mm

igra

nts

All

Sout

h A

sia

The

Car

ibbe

an

Latin

A

mer

ica

&

Mex

ico

Sub-

Saha

ran

Afr

ica

Nor

th

Afr

ica

&

the

Mid

dle

East

East

A

sia

&

the

Paci

fic

East

ern

Euro

pe

&

Cen

tral

A

sia

Wes

tern

Eu

rope

&

U.S

.

Popu

latio

n (N

) 5,

421,

654

592,

376

141,

638

30,0

62

32,9

68

30,6

06

46,5

61

179,

534

97,8

40

32,7

51

Mea

n A

ge a

t bas

elin

e 42

38

37

36

37

34

37

41

38

39

%

age

d 65

+ 15

.5

7.0

5.6

5.5

5.3

4.0

5.0

10.1

5.

7 8.

4 %

mal

e 47

.5

47.2

49

.0

47.3

44

.7

46.6

51

.9

45.4

47

.1

45.9

In

com

e qu

intil

e§ o

f nei

ghbo

urho

od o

f res

iden

ce (%

):

Q1

(low

est i

ncom

e)

18.4

29

.1

32.9

37

.3

34.9

46

.6

29.3

24

.8

26.2

15

.5

Q2

19.6

23

.3

26.3

24

.7

25.2

19

.3

19.0

25

.4

19.0

18

.2

Q3

19.5

19

.8

21.3

20

.5

19.2

13

.6

19.3

20

.1

19.7

18

.5

Q4

20.1

16

.3

12.9

11

.4

12.6

11

.4

18.7

17

.3

21.0

20

.3

Q5

22.2

11

.3

6.4

5.8

7.9

8.6

13.6

12

.2

14.0

27

.2

Imm

igra

tion

Visa

Cat

egor

y (%

)

Fa

mily

-

43.6

48

.3

79.1

67

.4

36.3

29

.0

43.0

25

.5

52.1

E

cono

mic

-

41.0

37

.7

16.8

21

.6

22.8

46

.9

48.3

48

.1

44.0

R

efug

ee

- 12

.9

13.5

1.

9 9.

4 39

.4

23.4

2.

5 25

.8

1.8

Oth

er

- 2.

6 0.

6 2.

2 1.

6 1.

4 0.

7 6.

2 0.

6 2.

2 Ed

ucat

iona

l Qua

lific

atio

ns a

t Lan

ding

(%)

0-9

yrs

scho

olin

g -

19.8

18

.2

26.2

32

.3

19.9

18

.1

23.7

10

.0

18.7

10

yrs

up

to s

econ

dary

-

29.4

33

.5

48.2

33

.7

40.7

28

.8

25.0

22

.8

23.7

N

on-U

nive

rsity

Qua

lific

atio

ns/

Som

e un

iver

sity

-

21.1

14

.1

18.9

19

.7

22.1

17

.2

21.2

30

.8

29.7

Uni

vers

ity D

egre

e or

Hig

her

- 29

.7

34.2

6.

6 14

.2

17.3

35

.9

30.1

36

.4

27.9

Ye

ar o

f arr

ival

(%)

1991

-199

3 -

31.8

24

.0

48.0

40

.9

39.4

26

.2

31.6

33

.1

39.8

19

94-1

996

- 30

.2

28.9

28

.1

31.7

30

.5

30.9

32

.9

27.1

28

.1

1997

-200

0 -

38.0

47

.1

23.9

27

.4

30.1

42

.9

35.4

39

.8

32.0

*I

n or

der t

o be

incl

uded

in th

e st

udy,

at b

asel

ine

indi

vidu

als

had

to b

e: a

live,

elig

ible

for p

rovi

ncia

l hea

lth c

are

for a

min

imum

of 3

yea

rs, u

rban

dw

ellin

g, a

ge 2

0 or

ove

r an

d di

abet

es fr

ee.

§ M

ost r

ecen

t pos

tal c

ode

of re

side

nce

was

use

d an

d lin

ked

to th

e m

ost r

elev

ant C

ensu

s in

com

e in

form

atio

n fo

r tha

t yea

r.

88

Over the study observation period 60,029 individuals were diagnosed with diabetes.

Individuals from South Asia had the highest overall incidence rates with an age-sex

adjusted rate of 15.7 per 1,000 person-years (17.3 and 15.4 per 1,000 person-years for

males and females, respectively), which was 3.3 times higher than in the Western

European and U.S. group (p < 0.05) (Table 4.2). Individuals from the Caribbean, Latin

America, Mexico, Africa, the Middle East, East Asia and the Pacific all had a higher

overall incidence rate as compared with the Western European and U.S. population (for

all p < 0.05). Men from all regions consistently had higher incidence rates than women

with the exception of the Caribbean where women experienced the highest risk. The

observed incidence rates indicated that although the high risk ethnic groups

experienced higher rates at every age, the elevated risk as compared with the lower risk

populations (i.e West and East Europe, Central Asia) was most pronounced in the

youngest age groups.

All rates were increased by restricting the denominator to only those who had

undergone a diabetes test over the study observation period. Rates increased slightly

more for males than for females, but the relative differences between ethnic groups

remained the same (see appendix H). Therefore, the results presented here are only for

the principal analysis.

For all ethnicities diabetes rates were higher in the lowest income and lowest education

groups (Table 4.2). Low income men consistently had the highest risk, with the

exception of Caribbean populations in which low income women experienced the

greatest risk (14.5 vs. 11.2 per 1,000 in women and men, respectively) and men and

89

Tabl

e 4.

2 D

iabe

tes

inci

denc

e ra

tes*

ove

r a 5

-yea

r fol

low

-up

perio

d by

prim

ary

cova

riate

s, a

ge†

and

sex.

W

orld

regi

on o

f birt

h

Sout

h A

sia

The

Car

ibbe

an

Latin

A

mer

ica

&

Mex

ico

Sub-

Saha

ran

Afr

ica

Nor

th

Afr

ica

& th

e M

iddl

e Ea

st

East

Asi

a &

th

e Pa

cific

Ea

ster

n Eu

rope

&

Cen

tral

Asi

a

Wes

tern

Eu

rope

&

U.S

. O

vera

ll (b

oth

sexe

s)

15.7

(1

5.4,

16.1

) 12

.1

(11.

4,13

.0)

10.0

(9

.4,1

0.7)

9.

4 (8

.6,1

0.2)

8.

8 (8

.3,9

.4)

7.3

(7.1

,7.5

) 6.

2 (5

.9,6

.6)

4.8

(4.3

,5.2

)

Mal

e

20-3

4 5.

4 (5

.1,5

.8)

2.4

(1.9

,2.9

) 2.

4 (1

.9,2

.9)

3.2

(2.6

,3.7

) 1.

7 (1

.4,2

.1)

1.7

(1.5

,1.9

) 1.

0 (0

.8,1

.2)

0.9

(0.6

,1.2

)

35-4

9 18

.9

(18.

1,19

.7)

9.7

(8.5

,11.

0)

8.3

(7.2

,9.5

) 10

.0

(8.8

,11.

3)

8.7

(7.9

,9.6

) 7.

1 (6

.7,7

.5)

4.4

(4.0

,4.8

) 3.

4 (2

.7,4

.1)

50-6

4 30

.6

(28.

9,32

.3)

22.1

(1

8.1,

26.1

) 23

.7

(20.

1,27

.3)

23.4

(1

8.6,

28.2

) 17

.1

(14.

7,19

.4)

16.1

(1

5.0,

17.3

) 13

.9

(12.

2,15

.6)

9.2

(6.9

,11.

4)

65+

27.2

(2

4.8,

29.6

) 23

.2

(17.

6,28

.9)

15.9

(1

1.6,

20.2

) 13

.6

(8.5

,18.

7)

25.4

(2

1.2,

29.7

) 15

.1

(13.

9,16

.4)

19.0

(1

6.1,

21.9

) 15

.2

(11.

4,19

.0)

Ove

rall

(20+

) 17

.3

(16.

8,17

.9)

11.2

(1

0.1,

12.4

) 10

.1

(9.1

,11.

1)

10.5

(9

.3,1

1.8)

10

.0

(9.2

,10.

8)

7.9

(7.6

,8.2

) 6.

9 (6

.4,7

.5)

5.2

(4.5

,6.0

)

Fem

ale

20-3

4 4.

5 (4

.2,4

.8)

3.5

(2.9

,4.0

) 2.

8 (2

.3,3

.3)

2.9

(2.4

,3.3

) 1.

7 (1

.4,2

.1)

1.9

(1.7

,2.1

) 1.

2 (1

.0,1

.4)

1.5

(1.1

,1.8

)

35-4

9 15

.0

(14.

2,15

.7)

9.6

(8.3

,10.

8)

7.7

(6.6

,8.7

) 7.

6 (6

.4,8

.8)

5.5

(4.8

,6.3

) 4.

4 (4

.1,4

.7)

2.7

(2.4

,3.0

) 2.

4 (1

.8,3

.0)

50-6

4 29

.1

(27.

6,30

.7)

26.2

(2

2.7,

29.6

) 23

.7

(20.

8,26

.6)

16.7

(1

3.3,

20.1

) 17

.4

(15.

0,19

.7)

15.1

(1

4.1,

16.1

) 11

.3

(10.

0,12

.6)

8.0

(6.0

,10.

0)

65+

22.6

(2

0.4,

24.8

) 23

.5

(19.

3,27

.8)

18.5

(1

4.8,

22.3

) 15

.4

(11.

4,19

.4)

18.0

(1

4.5,

21.5

) 15

.1

(14.

0,16

.2)

17.6

(1

5.6,

19.6

) 13

.2

(10.

8,15

.7)

Ove

rall

(20+

) 15

.4

(14.

8,15

.9)

13.0

(1

2.0,

14.2

) 10

.9

(10.

0,11

.9)

9.0

(8.0

,10.

1)

8.6

(7.8

,9.5

) 7.

4 (7

.2,7

.7)

6.5

(6.0

,6.9

) 5.

0 (4

.5,5

.7)

In

com

e qu

intil

e§ o

f nei

ghbo

urho

od o

f res

iden

ce (%

):

Low

inco

me

(bot

h se

xes)

17

.2

(16.

5,17

.9)

12.9

(1

1.6,

14.4

) 11

.9

(10.

7,13

.1)

11.8

(1

0.3,

13.3

) 11

.1

(9.9

,12.

3)

8.5

(8.0

,8.9

) 7.

9 (7

.2,8

.7)

6.1

(4.9

,7.4

)

Mal

e 17

.8

(16.

9,18

.8)

11.2

(9.3

,13.

4)

11.1

(9

.5,1

2.9)

13

.5

(11.

1,16

.1)

10.8

(9

.3,1

2.4)

8.

9 (8

.2,9

.6)

7.8

(6.8

,8.9

) 5.

6 (3

.9,7

.6)

Fem

ale

16.6

(1

5.6,

17.7

) 14

.5

(12.

6,16

.6)

12.6

(1

1.0,

14.3

) 10

.1

(8.5

,11.

9)

11.4

(9

.6,1

3.3)

8.

1 (7

.4,8

.7)

8.1

(7.2

,9.1

) 6.

6 (5

.0,8

.4)

Hig

h in

com

e (b

oth

sexe

s)

12.6

(1

1.4,

13.8

) 11

.9 (9

.1,1

5.1)

6.

1 (4

.3,8

.4)

6.5

(4.8

,8.5

) 7.

1 (6

.0,8

.4)

6.5

(6.0

,7.1

) 5.

1 (4

.4,5

.9)

3.5

(2.8

,4.4

)

Mal

e 13

.5

(11.

8,15

.4)

12.3

(8.2

,17.

7)

6.0

(3.3

,9.6

) 7.

9 (5

.2,1

1.3)

8.

9 (7

.1,1

0.9)

6.

8 (6

.0,7

.6)

4.9

(3.8

,6.1

) 4.

0 (2

.8,5

.4)

Fem

ale

11

.7

(10.

1,13

.4)

11.4

(8.0

,15.

8)

6.3

(4.0

,9.4

) 5.

1 (3

.1,7

.7)

5.5

(4.0

,7.2

) 6.

3 (5

.6,7

.2)

5.4

(4.4

,6.5

) 3.

1 (2

.2,4

.2)

90

Educ

atio

nal Q

ualif

icat

ions

at L

andi

ng

Less

than

pos

t-sec

onda

ry (b

oth

sexe

s)

16.6

(1

6.1,

17.1

) 12

.8

(11.

9,13

.7)

12.1

(1

1.3,

12.9

) 11

.0

(9.8

,12.

2)

10.8

(1

0.0,

11.7

) 7.

6 (7

.3,7

.9)

7.4

(6.8

,8.0

) 6.

4 (5

.7,7

.2)

Mal

e 16

.7

(16.

0,17

.6)

11.5

(1

0.2,

12.9

) 11

.6

(10.

4,12

.9)

12.3

(1

0.4,

14.4

) 12

.1

(10.

8,13

.5)

7.6

(7.2

,8.1

) 7.

7 (6

.8,8

.7)

6.4

(5.3

,7.7

)

Fem

ale

16.4

(1

5.8,

17.1

) 14

.0

(12.

8,15

.2)

12.5

(1

1.5,

13.7

) 9.

7 (8

.5,1

1.0)

9.

6 (8

.6,1

0.6)

7.

6 (7

.3,8

.0)

7.1

(6.5

,7.7

) 6.

4 (5

.6,7

.4)

Pos

t-sec

onda

ry (b

oth

sexe

s)

15.9

(1

4.9,

16.9

) 9.

8 (7

.6,1

2.1)

5.

8 (4

.6,7

.1)

7.7

(6.4

,9.1

) 7.

9 (6

.8,9

.1)

7.5

(7.1

,7.9

) 6.

5 (6

.0,7

.0)

3.5

(2.8

,4.2

)

Mal

e 18

.1

(17.

2,19

.1)

10.4

(7.7

,13.

4)

6.4

(4.8

,8.2

) 9.

2 (7

.6,1

0.9)

8.

7 (7

.6,9

.8)

8.1

(7.6

,8.7

) 6.

9 (6

.2,7

.6)

4.4

(3.4

,5.6

)

Fem

ale

13.8

(1

2.1,

15.6

) 9.

2 (6

.0,1

2.9)

5.

2 (3

.4,7

.3)

6.3

(4.3

,8.4

) 7.

2 (5

.2,9

.4)

6.9

(6.2

,7.5

) 6.

1 (5

.4,6

.9)

2.6

(1.8

,3.6

)

Imm

igra

tion

Visa

Cat

egor

y

Fa

mily

15

.3

(14.

8,15

.9)

12.3

(1

1.4,

13.2

) 11

.8

(11.

0,12

.6)

9.3

(8.1

,10.

5)

8.9

(8.0

,9.8

) 8.

0 (7

.7,8

.3)

6.5

(6.0

,7.0

) 5.

9 (5

.3,6

.6)

Eco

nom

ic

16.6

(1

4.1,

19.2

) 12

.7 (7

.1,1

8.9)

8.

9 (4

.1,1

4.1)

9.

2 (3

.7,1

5.4)

8.

9 (5

.9,1

2.0)

7.

2 (5

.9,8

.5)

6.3

(4.8

,8.0

) 4.

0 (2

.7,5

.3)

Ref

ugee

20

.1

(18.

3,22

.0)

8.7

(1.8

,18.

9)

5.1

(2.8

,8.0

) 13

.4

(11.

5,15

.5)

15.0

(1

2.6,

17.5

) 8.

9 (5

.1,1

3.5)

7.

8 (6

.8,8

.8)

3.5

(0.6

,8.6

)

Year

of a

rriv

al

1991

-199

3 13

.4

(12.

7,14

.1)

11.2

(1

0.2,

12.3

) 8.

9 (8

.0,9

.9)

9.3

(8.1

,10.

6)

8.9

(8.0

,10.

0)

6.8

(6.5

,7.2

) 5.

1 (4

.6,5

.7)

4.9

(4.3

,5.7

)

1994

-199

6 16

.0

(15.

3,16

.7)

12.7

(1

1.2,

14.4

) 12

.0

(10.

8,13

.3)

11.1

(9

.6,1

2.7)

8.

8 (7

.9,9

.9)

7.3

(7.0

,7.7

) 6.

5 (5

.9,7

.2)

5.2

(4.3

,6.2

)

1997

-200

0 18

.5

(17.

8,19

.2)

14.2

(1

2.0,

16.6

) 11

.8

(10.

1,13

.5)

8.9

(7.5

,10.

5)

10.1

(9

.1,1

1.2)

9.

2 (8

.7,9

.7)

8.1

(7.5

,8.8

) 5.

2 (4

.4,6

.2)

* A

ll ra

tes

have

age

-sta

ndar

dize

d ap

art f

rom

the

age-

sex

spec

ific

rate

s.

†

Age

bas

ed o

n ag

e at

bas

elin

e

§

Low

and

hig

h in

com

e de

fined

as

the

low

est a

nd h

ighe

st 2

inco

me

quin

tiles

, res

pect

ivel

y

91

women from North Africa and the Middle East who did not statistically differ from each

other. The relative increased risk associated with low education was most pronounced

in populations from Latin America, Mexico, Western Europe and the U.S.

Figures 4.1 and 4.2 show the adjusted estimates from the Cox Proportional hazard

models of diabetes incidence rate per 1,000 person-years by age and ethnicity for men

and women, respectively. For both sexes the adjusted incidence for South Asians was

notably higher than for all other ethnic groups until age 55 at which point the risk

merged with that of other high risk ethnic groups. For men over age 55, Sub-Saharan

African, Latin American, and Mexican origin all conferred similar risk to that for South

Asians. For women, only Caribbean origin was associated with a comparable risk to that

experienced by South Asians, and only over age 50. For South Asians, the increased

risk as compared with European populations appeared by age 20 at which point a rapid

increase in risk with age was seen over the next 30 years. A similar yet attenuated

pattern was seen for people from Africa, the Caribbean, Latin America and Mexico. In

contrast, people of Western European, Eastern European and Central Asian ethnicity

exhibited a relatively flat risk over time until roughly age 35. Women of Western

European origin had the lowest risk which remained relatively flat until after age 40. The

difference in risk at age 40 between South Asians (highest risk) and Western

Europeans (lowest risk) was 5-fold for men and 9-fold for women.

92

93

94

Table 4.3 Age of equivalent diabetes risk by ethnicity. The risk experienced by men aged 40 of Western European ethnicity was used as the standard for comparison*.

Ethnicity Men Women

South Asian 25 25

Caribbean 30 30

Hispanic 30 30

African (Sub-Saharan) 30 35

North African & Middle Eastern 35 35

East Asian & Pacific 35 40

Eastern European & Central Asian 40 40

Western Europe 40 – standard 45

*This does not take into account individual risk factors such as BMI, family history or other clinical factors.

95

The risk for developing diabetes seen in Western European men in the 40-44 age group

was 3.7 per 1,000 person-years, and was roughly equivalent to the risk experienced by

South Asian men between age 25-29, a fifteen year shift in age-equivalent risk (Figures

4.1, 4.2 & Table 4.3). For women, the diabetes risk in those of Western European origin

at age 40-44 (1.7 per 1,000 person years) was lower than observed at any age in South

Asian women. These differences in risk persisted despite controlling for income,

education, immigration category and time in Canada. Ethnic-specific ages that

correspond to the equivalent diabetes risk experienced by men of Western European

ancestry at age 40-44 are presented in Table 4.3.

The results of the unadjusted Cox models were very similar to the adjusted models so

only the adjusted results are presented here (see Appendix I for unadjusted results). In

addition, when we generated risks based on one-year follow-up times the pattern was

very similar although the standard errors were much larger (see Appendix I). Therefore

we decided to present the results derived from the model using 5-year follow-up times

and averaged to generate annual estimates of risk.

4.4 Discussion

Our study found an extraordinarily high incidence of diabetes in young adults of South

Asian descent compared to Western Europeans of the same age. The risk of diabetes

in South Asians rose rapidly after age 20, resulting in a significant divergence in risk

from other ethnic groups by age 30. The risk in most other ethnic groups rose more

96

gradually over time, with the slowest rise evident in people of Western European

descent. By age 40, there appeared to be a marked disparity in incidence between

South Asian and Western European populations – in the order of 5-fold for men and 9-

fold for women. In order to target a similar level of risk for diabetes as is seen in men of

Western European ethnicity at age 40 one would have to initiate screening among

South Asians at age 25. For all other non-European ethnic groups, in order to target an

equivalent risk, screening would have to be initiated between age 30 and 35.

Our findings are consistent with other research describing the particularly high diabetes

risk in people of South Asian ethnicity, seen in indigenous, migrant and second

generation migrant populations (Ramachandran et al., 2001; Creatore et al., 2010;

Misra & Ganda, 2007; Ebrahim et al., 2010). This increased risk of developing diabetes

in South Asians has been previously shown to begin at an earlier age as compared with

white, European populations (Ramachandran et al., 2001; Khan et al., 2011; Chiu et al.,

2011; UK Prospective Diabetes Study Group, 1994). Younger age of onset has been

reported in other high-risk populations including in Mexico (Jimenez-Corona et al., 2010;

Aguilar-Salinas et al., 2001; Rull et al., 2005), and among Native, African and Asian

Americans (Dabelea et al., 2007). Most evidence of higher rates of diabetes among the

young in high-risk ethnicities has come from cross-sectional prevalence studies which

make it difficult to determine the age at onset or control for possible confounders related

to development of disease. The few prospective studies looking at diabetes incidence

have been largely based on questionnaire/survey data and thus suffer from the

limitations of possible sampling bias and insufficient sample size to examine ethnicity

beyond large groupings of heterogeneous populations, i.e. ‘Black’ or ‘Asian’.

97

Additionally, most studies looking at age at diagnosis were either focused on the

screen-eligible population (i.e aged 40 and up) or on pediatric populations (Dabelea et

al., 2007; American Diabetes Association, 2000) and thus do not capture the 20-40 age

group. Our findings suggest that it is in this age group where disparities in risk appear.

As far as we are aware, ours is the first study to look at age-specific incidence data

based on the entire adult population in a multi-ethnic setting.

Diabetes screening has the potential to prevent diabetes-related complications and

cardiovascular disease if the disease is brought under control at an early stage.

Establishment of screening age cut-offs in Canada, the U.S. and the UK has been

largely based on expert consensus, although existing thresholds are supported by

research showing: 1) a relatively high prevalence of diabetes, pre-diabetes or impaired

glucose tolerance (IGT) in those over age 40; and 2) economic models indicating that

screening in this group is cost effective (Leiter et al., 2001; Waugh et al., 2007). Our

research suggests that there is a large divergence in risk by ethnicity for the

development of diabetes beginning in the 20-30 age group. If screening practices fail to

take these differences into account, certain ethnic groups may experience a greater

delay between disease onset and diagnosis which may, in turn, increase the likelihood

of avoidable complications. Therefore, understanding differences in age at onset is

crucial to the design of timely, effective and equitable screening policies and programs.

Ethnicity affects the risk of diabetes through genetic factors that are not completely

understood. Our findings indicate that South Asians and other high risk ethnic groups

exhibit an earlier age of onset, but there is also evidence that these groups tend to

98

present at a later stage of disease and may have a more accelerated disease course –

further underscoring the importance of timely identification and control of disease.

South Asians tend to have a high prevalence of chronic comorbidities related to

diabetes such as coronary heart disease (CHD), retinopathy and nephropathy at the

time of diagnosis (Joshi et al., 2007; Mather et al., 1998; Raymond et al., 2009). There

is also evidence that individuals presenting with diabetes at an earlier age often show

similar adverse cardiovascular risk profiles as older patients (Hillier & Pedula, 2003), an

observation found to be particularly true for South Asians (Joshi et al., 2007; Song &

Hardisty, 2009). This latter finding has even prompted some to suggest that earlier

onset may represent a more aggressive form of disease (Benhalima et al., 2011),

further supporting the benefits of early identification.

In Canada, rates of diabetes screening in those aged forty and over are high overall and

immigrant status and non-white ethnicity are associated with increased rates of

screening (Wilson et al., 2009; Creatore et al., 2012). As far as we know, however, few

have looked in detail at the effectiveness or benefits of screening in populations under

the age of 40. A recent U.S. study using mathematical models to determine the most

cost effective age to initiate screening found that initiating screening between the ages

of 30 and 45 years and repeating every 3-5 years was optimal (Kahn et al., 2010). The

ADDITION-Leicester study on diabetes screening in the UK applied a lower age

threshold for South Asian populations of 25, however this group was excluded from

published analyses so data is not available to evaluate the effectiveness or feasibility of

targeting this age group (Webb et al., 2010). Very recently in the UK, an update to

current screening guidelines has been proposed recommending testing all those aged

99

25 and over of South Asian or Chinese descent whose BMI is greater than 23 kg/m2. At

the time of writing, this had yet to be adopted into national guidelines (National Institute

for Health and Clinical Excellence (NICE), 2011). Our study findings would support this

latter recommendation for South Asians, however, we found that the risk in people of

Chinese origin was considerably lower and an older age (i.e. 35-40) to initiate screening

may be recommended for this group.

We also found that risk varied by socio-economic characteristics in our study and that

this association differed somewhat by ethnicity. With few exceptions, however, low

income and low education was associated with higher risk for all ethnic groups.

Interestingly, the results from the unadjusted and adjusted multivariate models were

virtually identical suggesting that although diabetes incidence varied by socioeconomic

status, age and ethnicity were by far the largest drivers of risk in our study.

This study had numerous strengths including being a large, population-based,

longitudinal cohort with a study population of over half a million individuals

encompassing virtually all ethnicities. The study was conducted in a single geographic,

sociopolitical and health care setting allowing for robust comparisons across groups.

There are, however, several limitations to be noted. First, our primary covariate was

country of birth which we used as a proxy for ethnicity which may have resulted in some

misclassification. This measure allows us only to look at the impact of ethnicity on risk

for diabetes in first generation immigrants. It is possible that immigrants have higher

risk (due to stress of migration) or lower risk (due to healthy immigrant effect) as

compared to non-immigrants of the same ethnicity living in Canada. It should be noted,

100

however, that since the entire study cohort was composed of immigrants, even if

immigrants as a whole differ from the non-immigrant population and this impacts the

absolute incidence rates, the relative differences by ethnicity are likely to persist. A

second limitation is that diabetes incidence was based on physician diagnosed diabetes

so it likely underestimates ‘true’ diabetes in the population. In a previous study we have

shown, however, that diabetes screening is very high in Canada and reasonably

equitable across population groups, reaching roughly three-quarters of the population

over age forty within a three year period, and over 80% of the recommended population

within five years (Creatore et al., 2012). In addition, our sensitivity analysis restricting

the analysis to those who were screened within the study period did not identify any

significant differences in relative incidence rates or rates by age. Another limitation

relating to our definition of incidence is that we only required recent immigrants to have

a three-year “wash-in” period of observation in order to identify incident (versus

prevalent) cases. It is possible that some individuals with prevalent diabetes who had no

health care contact related to their condition within the three-year observation period

could be incorrectly included as incident cases (and therefore false-positives). A fourth

limitation is that the administrative data do not allow us to differentiate between type 1

and type 2 diabetes. Although studies suggest that the majority of individuals with

diabetes (90-95%) are likely to have type 2 diabetes, this estimate may be lower for

young adults. We may therefore be capturing a higher number of false-positives (i.e.

type 1 cases misclassified as type 2) in our younger as compared with our older age

groups in our study. There is evidence from recent studies looking at children and youth

(<20 years of age), however, that among high-risk ethnic groups the proportion of

101

diabetes that is comprised of type 2 is high (ranging from 46% in Hispanic to 86% in

Aboriginal youth) and is rising (Ogawa et al., 2007; Dabelea et al., 2007). Therefore,

although this limitation will likely result in an overestimation of the incidence of type 2

diabetes particularly in our youngest age group, it is likely that the overestimation

disproportionately impacts our estimates for our lowest risk population, thus biasing our

findings towards the null. Finally, clinical risk factors such as BMI, family history and

lifestyle measures were not available in our data sources so we were unable to account

for these in our models. Relative ethnic differences in diabetes risk appear to persist

after adjustment for adiposity, physical activity and other individual risk factors (Jenum

et al., 2005), however these factors play an important role in determining individual-level

risk. It is also important to note that the purpose of this paper was not to supplant

existing individual risk algorithms, but to ascertain and the population level what are

equivalent risk age cut-offs to help inform universal guidelines.

The purpose of screening is to provide the opportunity to intervene at an earlier stage of

disease and to improve health outcomes. Clinical Practice Guidelines currently

recommend screening earlier and more often in high risk ethnic groups. Our study

presents age thresholds that could be applied to various ethnic groups based on the

age at which their risk of developing diabetes is equivalent to men of western European

ancestry at age 40. The results of our study have several implications. First, if age

recommendations for diabetes screening are based on risk of developing disease, then

age to initiate screening should differ significantly by ethnicity and is up to fifteen years

younger in South Asians than in people of Western European ancestry. From a health

care effectiveness and equity perspective this implies that ethnic-specific age cut-offs to

102

initiate screening are required. Secondly we have shown that the risk of developing

diabetes increases very rapidly beginning after age twenty for high risk groups,

particularly South Asians. Given the increased cardiovascular risk associated with

diabetes, appropriate interventions have to be evaluated and tailored to a young-adult

population for these groups. A recent study found that younger adults with diabetes

were less likely to receive appropriate diabetes management and less likely to achieve

outcome targets as compared with older patients (Guthrie et al., 2009). Thus, more

research is needed to determine the achievable risk reduction (primarily cardiovascular)

in younger populations with screen-detected diabetes and how such programs can best

be implemented. Finally, our results suggest that diabetes primary prevention programs

aimed at promoting healthy diet and physical activity in South Asian and other high-risk

populations should begin prior to age twenty where significant disparities in diabetes

risk already appear. We believe our findings can significantly add to the understanding

around ethnic differences in risk of disease throughout adulthood and has practical

implications for policy, medical practice and future diabetes research.

103

Chapter 5

Discussion

104

5.1 Main Findings

This thesis presents three distinct studies looking at the epidemiology of diabetes in a

large, population-based immigrant cohort in Ontario, Canada. Although the chapters are

intended to be able to stand alone as manuscripts looking at specific research

questions, together they form a complementary body of work attempting to further our

understanding of the population burden and risk of diabetes by age and sex among

different ethnic groups. We hope that this information may help inform policies that

address, and reduce, health disparities. The principle findings of this thesis are:

1) South Asians had a very high risk (three-fold higher) for developing diabetes as

compared with people of European ethnicity and a significant disparity in risk was

already evident by age 30;

2) Although all non-European populations had higher risk of diabetes as compared with

individuals of European ancestry, this risk varied substantially across country and region

of birth making broad definitions of race or ethnicity (eg. ‘Asian’ or ‘Black’) inappropriate;

3) Contrary to patterns seen in Western European populations, women belonging to

many high–risk ethnicities had equivalent or, in the case of Caribbean women, higher

risk than men;

4) In order to capture an equivalent risk to that of Western European populations at age

40, which is the recommended age to initiate screening in the general population,

105

people of South Asian ancestry should be considered for screening as early as age 25.

In other non-European ethnicities screening should be considered between ages 30-35;

5) In Canada, high risk ethnic groups are generally being screened as recommended

(every 3 years by age 40), but these recommendations may fall short of what is required

to effectively prevent diabetes-related complications in high risk groups (see # 4 above);

6) The burden of ‘undiagnosed’ diabetes as a percentage of all diabetes in the

population varied by ethnicity, age and sex; however it was estimated to be significantly

lower than the previously reported 33%, at roughly 6-13%.

5.2 Research Implications for Policy and Practice

One of the most important implications of this research is that it suggests that the

current diabetes screening recommendations in Canada, as well as those in the U.S.

and the UK, may not adequately address the early age of onset experienced by high

risk ethnic groups. The primary benefit of diabetes screening comes from the ability to

intervene early in the disease progression and to prevent diabetes-related

complications. The current Canadian guidelines recommend initiating screening at age

40, and although earlier screening is recommended for individuals belonging to high risk

ethnic groups, the appropriate age to begin screening in these populations is unclear.

We found that South Asians exhibit an equivalent risk by age 25 to that observed in

Western Europeans at age 40. This has significant public health implications since if

106

individuals of South Asian and other high-risk ethnic groups develop diabetes in their

20’s and 30’s, then initiating screening programs at age 40 may be too late to prevent

future morbidity and early mortality associated with years of unmanaged disease. Our

results suggest that we may be justified in initiating diabetes screening programs up to

15 years earlier in South Asian populations and 5 to 10 years earlier in other high risk

populations (i.e. Caribbean, African, Middle Eastern).

Although to our knowledge no other population-based longitudinal study has been

conducted looking at age-specific incidence by ethnicity, our results are supported by

other research finding earlier age of diabetes onset among people of South Asian ethnic

ancestry (Chiu et al., 2011; Khan et al., 2011; Qiao et al., 2003; Feltbower et al., 2003).

In addition, recent research conducted in Canada, the U.S., the UK, Mexico and Asia

have found diabetes prevalence to be increasing in younger age groups (Lipscombe &

Hux, 2007; Engelgau et al., 2004; Tseng et al., 2006; Jimenez-Corona et al., 2010;

Dabelea et al., 2007) which has largely been attributed to rising rates of obesity in

children and youth (Dabelea et al., 2007). The SEARCH for Diabetes in Youth study in

the U.S. recently found ethnic differences in the prevalence of type 2 diabetes in

children as young as age 10-14 (although the number of cases of type 2 diabetes in this

age group was very small and in most cases differences did not achieve statistical

significance) (Dabelea et al., 2007). These studies, in combination with our findings,

suggest that diabetes prevention programs aimed at establishing healthy body weights

may be a key public health priority for school-age children and teenagers, particularly in

neighbourhoods with a high-density of South Asian, Afro-Caribbean, African or Middle

Eastern ethnic populations.

107

Our findings in chapter 3 suggested that in Ontario, high risk ethnic groups including

South Asians are generally meeting the recommended screening guidelines. These

findings are in-line with recent Canadian research that found an absence of disparities

in the quality of primary diabetes care among ethnic minority groups (Shah et al., 2012).

Our results are promising since other researchers have found low response rates in

screening programs among ethnic minority populations such as in the South Asian

community in England (van den et al., 2011). However, although we are achieving high

rates of screening in Ontario among those aged 40 and up, if diabetes incidence

actually occurs in younger age groups for many ethnic groups we may still be falling

short of what is required to identify diabetes in a timely way in order to prevent adverse

health outcomes in these populations. Whether screening in younger adults results in

the same health gains achieved in older individuals through the control of

cardiovascular risk factors and the prevention of related complications requires further

investigation.

We also found a lower than previously recorded estimate of undiagnosed diabetes,

likely reflecting higher rates of screening in the past decade, as previously observed by

Wilson and colleagues (Wilson et al., 2009). Despite high levels of screening and less

undiagnosed disease in the general population, men were nearly twice as likely as

women to remain ‘undiagnosed’ and older immigrants (> age 65) were significantly less

likely to be screened than immigrants in the younger age group (40-64). Therefore,

notwithstanding promising trends, significant disparities remain. It must also be noted

that undiagnosed diabetes was estimated only for the population within the current age

range for initiating screening. The number of people under age 40 with undiagnosed

108

diabetes as a proportion of all ‘true’ diabetes cases in this age group may, in fact, be

higher than what we estimated for older adults. In order to answer this question more

research is needed looking at screening in younger (< age 40) adults in our setting.

Although all non-European populations had higher risk of diabetes as compared with

individuals of Western European ancestry, this risk varied substantially across ethnic

groups making broad definitions of race or ethnicity (eg. ‘Asian’ or ‘Black’) inappropriate.

For instance, although South Asians had a very high risk (3-fold), East Asians had only

a slightly elevated (i.e. 40% higher) incidence as compared with people of Western

European ancestry (see chapter 4). Similarly, there were significant differences between

people from Sub-Saharan Africa and the Caribbean which would be masked by

grouping people from these regions into a category defined as ‘Black’. Even within our

world region groupings, in chapter 2 we described significant differences by country. For

instance, among South Asians, Sri Lankan immigrants had the highest rates of diabetes

with a prevalence that was 60% higher than that of Indians. We also observed sex

differences in risk that varied by world region even among individuals of similar ethno-

racial background, that may reflect cultural, religious or economic differences. For

example, for most sub-Saharan African countries men exhibited higher diabetes rates,

however, among individuals from the Caribbean women had higher rates. Again, these

differences would be masked if these individuals were all defined as ‘Black’.

Some important sex differences were observed in this thesis. Contrary to traditional

patterns of diabetes risk where men have higher incidence than women (Gourdy et al.,

2001; Public Health Agency of Canada, 2009; Wild et al., 2004), women from many

109

ethnic groups had equivalent or higher risk than men including women from the

Caribbean, Mexico, Latin America, North Africa, and the Middle East. Women from the

Caribbean had particularly high rates of diabetes which is supported by previous

research in Canada and the U.S. finding higher rates of diabetes in ‘Black’ women as

compared to men (Chiu et al., 2011; Cowie et al., 2006). In addition to women of Afro-

Caribbean ancestry, higher rates of diabetes in women relative to men have also been

reported in Aboriginal communities (Young et al., 2000) and among Mexican-Americans

(Cowie et al., 2006). Interestingly, even among immigrants from Western Europe and

the U.S. gender differences in prevalence and incidence were small.

There are various theories that may help to explain our findings some of which will be

discussed in the following section in the context of our Theoretical Framework. It has

been suggested that the observed increase in diabetes rates among women in certain

communities is a reflection of higher rates of obesity in these populations. Some

research shows that there are differences in culturally accepted body weight for women

within some ethnic groups, including South Asian, African and Caribbean populations,

where a higher BMI is perceived as acceptable (Altabe, 1998; Rubin et al., 2003). There

may also be interactions between culture, gender and physical activity whereby women

of some ethnicities may have lower average levels of physical activity (Fischbacher et

al., 2004; Pomerleau et al., 1999; Bryan et al., 2006; Tremblay et al., 2006). Although in

the past 20 years increases in BMI among women have generally outpaced those

among men in Canada (Shields et al., 2010), increasing obesity trends have been

observed in particular among marginalized women, including those of low

socioeconomic status (McLaren, 2007), and those belonging to ethno-racial minorities

110

(Ng et al., 2011). In addition, the effects of psychological stress on immigrant women

may increase their risk for obesity and diabetes, more so than for men. This

phenomenon of increasing BMI in reaction to psycho-social stress among women has

been documented (Jeffery et al., 1991) and may help explain why we see a reduced

gender gap in risk across many immigrant groups, including those from Western

Europe.

Our results in chapter 3 indicated that the single-most important predictor of being

screened for diabetes was frequency of contact with a physician. Not surprising then,

due to their higher use of health services, women of all ethnicities were found to be

more likely to be screened for diabetes than men. Due to these differences in screening

behaviours, our data may under-capture some disease in men, an issue that will be

further discussed in the limitations section of this chapter. However, when we looked

only at incidence among those who had a diabetes test, patterns of risk by ethnicity

persisted, although some of the gender differences were attenuated. Despite this

attenuation, diabetes rates in women from the Caribbean, Mexico, Latin American,

North Africa and the Middle East persisted in remaining higher or equivalent to men

from the same regions.

Finally, it is important to emphasize that ethnicity is not a modifiable risk factor.

However, the genetic risk for diabetes associated with different ethnic origins interacts

with environmental exposures resulting in differing levels of expressed risk. Our

opportunities for intervention lie in attempting to modify the individual’s exposure to risk

factors (such as poor diet, low levels of physical activity), or the level of impact that the

111

exposure ultimately has on health (through early identification of diabetes or related

complications and access to effective interventions). In order to design such

interventions a better understanding of the epidemiology of diabetes in high-risk and

vulnerable groups is required and we hope that the findings from this research can

contribute to this greater understanding and can assist in developing appropriate

policies and programs.

5.3 Interpretation of Findings in the Context of our Theoretical

Framework

Our proposed theoretical framework, as described in chapter 1, places health status in

the context of both structural and intermediary social determinants of health and health

inequities. Socio-economic position is a strong predictor of health status and a

socioeconomic gradient in health has been shown to exist for almost all health

outcomes (Lynch et al., 2000; Singh-Manoux et al., 2002; Marmot et al., 1991). In this

model, the structural determinants of health inequities, composed of the socio-political

and social context and the social stratification within the society, work together to shape

the distribution of the intermediary/social determinants of health in the same society

(see below, figure 1). Cutting across all dimensions of this model is the concept of social

capital and social cohesion that impacts health through the mechanism of social

supports and relationships. One of the key advantages of this framework is that it

emphasizes that in order to tackle health inequities, interventions should not focus

112

solely on the ‘social determinants of health but must address the ‘structural

determinants’ that cause the systematic differential distribution of health determinants in

society, and thus form the root of inequities.

Figure 5.1. The World Health Organization’s Commission on the Social Determinants of Health conceptual framework. Reprinted with permission from Solar, O. & Irwin, A. (2010). A conceptual framework for action on the social determinants of health Geneva: World Health Organization. © World Health Organization 2010.

113

As discussed in the thesis introduction, I propose that immigration influences both the

structural determinants as well as the intermediary determinants in this model. For

immigrants, the social determinants of health and health inequities are also operating in

the context of the migration experience and exposures pre- and post-migration. The

socioeconomic and political environment, the social structure of, and distribution of

power within, the source countries, and previous exposures to discrimination based on

gender or race will all impact the individual’s pre-migration socio-economic position

(SEP). Current Canadian immigration policies select immigrants based largely on their

pre-migration SEP by selecting highly trained, highly educated applicants; however,

there is not necessarily a high correlation between pre- and post-migration socio-

economic position. The Canadian socioeconomic and political context, along with

gender roles, perceptions about immigrants, labour markets and regulations,

discrimination based on race and the availability of social supports and networks all

further contribute to establish the individuals’ socio-economic position in Canadian

society. These post-migration influences often result in a decline in SEP in the host

country (Chen et al., 1996a; Kinnon, 1999), a trend that also holds true for health status

(Ng et al., 2005). The model discussed here offers a context and mechanisms by which

this downward trend in health may occur.

Based on this model, observed differences in diabetes risk in our research reflect a

combination of structural determinants as well as intermediary factors such as biological

and genetic differences in risk, health behaviours (such as diet, physical activity and

health services utilization), material circumstances (such as access to resources and

the physical and occupational environment) and access to appropriate health care (see

114

table 5.1 for further examples). In addition, immigrants likely have fewer social networks

and social supports on which they can rely (Salinero-Fort et al., 2011; Hedemalm et al.,

2010). Studies looking at the health of recent immigrants have shown that immigrants’

health deteriorates over time (Chen et al., 1996b; Newbold & Danforth, 2003; Perez,

2002), suggesting that newcomers may be particularly vulnerable to social, economic

and environmental exposures that negatively impact health.

Material circumstances were measured in our analyses using proxies such as

neighbourhood level income and individual education (proxies for the financial means to

buy healthy food, and other important health-related resources), as well as English

language proficiency (as a measure of how well they can access resources). The

primary biological factor that was measured in our research was genetic susceptibility to

diabetes related to ethnicity and our findings suggest that different ethnic groups

experience large differences in risk. According to our theoretical framework, the

expression of this differential genetic risk will be linked to exposures (i.e. obesity, poor

diet, physical inactivity) that are unequally distributed across socio-economic, gender,

and racial groups. Our adopted framework suggests that in addition to addressing the

structural determinants of inequities, appropriate interventions would target the social

and environmental exposures that contribute to higher rates of overweight and obesity,

lower levels of physical activity and unhealthy diet choices in our vulnerable groups.

We were unable to measure health behaviours related to lifestyle such as diet and

physical activity; however, in chapter 3 we addressed health-seeking behaviours in

relation to diabetes screening by immigrant groups. We began with the hypothesis that

115

due to access barriers, lack of education around the benefits of screening, and

competing priorities due to stress and low socio-economic position in society,

immigrants would have lower screening rates. However, we found that this was not the

case and in fact found a relatively equitable distribution (i.e. highest rates were found in

the highest risk groups) of a health-enhancing exposure (i.e. diabetes screening). This

finding has positive implications both from the perspective of health behaviours, eg.

high-risk immigrant populations seeking out health care and adopting health-promoting

practices (such as screening); and from the perspective of the health system that does

seem to be screening high-risk groups.

Psychosocial factors were not directly measured in our study. However, many

immigrants report high levels of stress, lack of social supports, discrimination, and

periods of unemployment or under-employment (employed in a position that

underutilizes your training and education) (Block & Galabuzi, 2011; De Maio & Kemp,

2010). Although we could not measure it, this may have contributed to our findings that

some immigrant women in our study seem to have lost the ‘protective’ factor against

developing diabetes normally associated with female sex. The health of immigrant

women may be particularly susceptible to these structural and intermediary

determinants of health inequities as, in addition to the immigration experience, they also

experience social and economic marginalization, and face discrimination related to race,

culture and gender. There is evidence that minority ethnic migrant women may

experience greater levels of disadvantage, discrimination, violence, and stress than

men and that these experiences result in higher levels of physical and psychological

illness (Cooper, 2002). Previous research has shown that the effect of material

116

deprivation and socio-economic inequality on diabetes risk is greater for women, a

relationship that persists after controlling for obesity (Imkampe & Gulliford, 2011; Coeli

et al., 2009; Tang et al., 2003). This suggests that other variables such as psycho-

social factors may contribute differentially to the risk for diabetes among marginalized

women as compared to men. More research is required to disentangle the effects of

immigration and ethnicity, socio-economic position, gender and the risk for developing

obesity and diabetes.

In our theoretical model the health system acts as an intermediary social determinant of

health. As mentioned earlier, current diabetes screening guidelines, although

acknowledging the need for earlier screening in high risk ethnic groups, do not provide

adequate information about the age to initiate screening in these groups. If unclear

guidelines result in a delay in diabetes diagnosis among individuals of certain ethnic

groups, the benefits of screening may be concentrated in lower risk ethnic groups such

as people of European ancestry, further exacerbating ethno-racial health disparities.

Our findings in chapter 4 suggest that age differences in risk by ethnicity are large

enough that this is likely the case. This finding is an example of how the health system

can negatively affect the distribution of health determinants in society; however, it also

presents an opportunity for the health system to promote program or policy changes

that could improve the health status of vulnerable groups. Other possible roles for the

health system may include providing educational materials around diabetes prevention

in various languages, ensuring access to translators and providing programs that are

culturally-sensitive.

117

Our theoretical framework highlights the complexity of the relationship between the

social determinants of health, structural causes of health inequities, the health system,

and health outcomes and we have further added the dimension of immigration,

including pre- and post-migration experiences. The structural determinants, through

creating unequal socio-economic positions, give rise to differences in the distribution of

the social determinants of health in society based on social hierarchy. Immigrants

traditionally occupy a lower position in a society’s social hierarchy; therefore our results

may reflect an elevated risk associated with ethnicity that is further compounded by the

effects of immigration, the social determinants of health and structural determinants of

health inequities.

5.4 Limitations

There are several limitations that have been discussed thoroughly in the previous

chapters so only the notable ones common to all chapters will be summarized here. Our

primary covariate of interest in these studies was ethnicity. Since information on

ethnicity, race or culture is not available in administrative health data, immigration

records were used to identify individuals’ country of birth. This reliance on immigration

data results in two limitations in the interpretation of our findings: first, we only have

information on first generation landed immigrants and therefore cannot be certain that

these findings can be applied to non-immigrant members of these ethnic groups;

second, we have to use country of birth as a proxy for ethno-racial group which may

118

introduce some misclassification particularly in immigrants from areas that are more

ethnically heterogeneous. The first limitation is one of generalizability since the process

of immigrating exposes individuals to experiences that temporarily or permanently

change their risk for diabetes and these experiences may differ by race, religion,

culture, age or sex. Despite this limitation, our multivariate analyses in chapters 3 and 5

used an internal comparison group of other immigrants during the same period from

Western Europe and the U.S., and found that the role of ethnicity (measured as country

of birth) in predicting diabetes risk was strong and persisted after controlling for age,

income, education, visa category and time since arrival. Given the strength of the

association, and in the context of the large supporting literature suggesting genetic

origins to ethnic differences in glucose metabolism, our findings likely do reflect (at least

in part) ethnic disparities in risk that are not specific to immigrants. The second limitation

mentioned above relates to the introduction of measurement error in our exposure

definition (i.e. ethnicity). Attempts were made to identify the dominant ethno-racial group

living within each of the 235 countries included in our study and this ethnicity was then

applied to all individuals arriving from that country. However, despite these efforts, some

individuals’ ethnicities will still invariably be misclassified.

Physician-diagnosed diabetes was our main outcome for two of the studies described in

this thesis. In order to identify individuals with diabetes, the Ontario Diabetes Database

(ODD) relies on physician billing codes and hospitalization data – an algorithm that has

been shown to have high sensitivity (86%) and specificity (97%) in the general Ontario

population (Hux et al., 2002). One of the limitations of this definition is that contact with

the health system and the opportunity for diagnosis is a prerequisite for being captured

119

in the ODD. Therefore we may be underestimating diabetes in population sub-groups

that have lower levels of health services utilization. We also will not be capturing a small

number of individuals (estimated at <5%) that receive services in non-OHIP billed

settings (Williams & Young W, 1996). Despite this limitation, when we restricted our

analysis to those individuals with contact with a physician over the study period, we

found that the relative differences by ethnicity persisted. Individuals with no health

insurance were not captured in our data nor were we able to obtain information about

individuals who despite having health insurance, did not access the health care system.

However, the proportion of Canadians who do not have any contact with the health care

system is very low (Statistics Canada, 2006), although this group may represent a

particularly vulnerable population. Chapter 3 of this thesis attempted to address the

question of whether or not immigrants were receiving diabetes screening (and therefore

had the opportunity to be captured by the ODD) and if screening rates differed by

ethnicity. This chapter found that contact with the health system and screening among

those aged 40 and over was high, suggesting that we are capturing the majority of that

population in the administrative data. Finally, our diabetes screening data does not

include lab tests that were conducted in hospitals, which means that a small proportion

of all diabetes tests are missing in our analyses. It is unlikely, however, that ethnicity is

associated with a higher probability of having a diabetes lab test in a hospital as

compared to outside of hospital.

Another limitation of the ODD is that we are unable to differentiate between type 1 and

type 2 diabetes. Although studies suggest that the majority of individuals with diabetes

(90-95%) have type 2, this estimate is likely lower for young adults where type 1

120

diabetes may comprise a higher proportion of cases. We included individuals as young

as 20 years old in our study and it is possible that in our youngest age groups we may

be capturing a higher number of false positives than in our older adult population. There

is evidence from recent studies looking at children and youth (<20 years of age),

however, that among high-risk ethnic groups the proportion of all diabetes that is

comprised of type 2 is high (ranging from 46% in Hispanic to 86% in Aboriginal youth)

and is rising (Ogawa et al., 2007; Dabelea et al., 2007). Therefore, although we are

likely overestimating the incidence of type 2 diabetes in our youngest age groups, it is

likely that the overestimation disproportionately impacts our estimates for our lowest risk

population, thus biasing our findings towards the null.

A final limitation that should be mentioned with respect to defining incident diabetes

cases among recent immigrant populations using the ODD is that we may be

misclassifying a small proportion of prevalent cases as incident cases. This may occur

in cases where an individual has diabetes but their disease was not captured in the

administrative health data within the three-year observation (or ‘wash-in’) period prior to

baseline. A longer ‘wash-in’ period could be used to try to remove those false-positive

incident cases.

In chapter 3 we made the a priori decision to focus our analyses on those aged 40 and

up in order to align with the age range recommended by screening guidelines; however,

as previously discussed, in chapter 4 we found that South Asians, as well as other high

risk ethnic groups have a high risk of developing diabetes younger than age 40.

Therefore, although high risk ethnic groups are largely being screened over age 40, a

121

substantial number of individuals with pre-diabetes and diabetes in younger age groups

may be undiagnosed. Since our analyses did not include these younger age groups, it is

possible that the proportion of undiagnosed diabetes in the overall population may be

higher than what we found in chapter 3. This is an area for future research.

Another limitation of the health administrative data is that we do not have clinical

measures on diabetes risk factors such as BMI, waist circumference or family history

nor do we have data on other relevant health indicators such as hypertension which

may increase the likelihood of early screening for diabetes. These clinical factors may

represent confounders or effect modifiers that we were unable to control for and explore

in our analyses. Although this information would be invaluable in future studies and

could add important information to our results, we feel that due to the population-based

nature of this work, this does not present a serious flaw in our findings. It is possible

(and likely) that the underlying BMI and waist circumference distribution of ethnic groups

differ and therefore controlling for these measures may increase or (less likely) reduce

the observed differences in risk between ethnicities. The purpose of this work was not to

supplant existing individual-level risk algorithms but to ascertain at the population level

what the equivalent risk age cut-offs are for different genders. Currently, age 40-45 (in

Canada, the UK and the U.S.) is being used as a threshold for universal screening and

our intention was to establish similar thresholds by ethnicity for initiating universal

screening irrespective of BMI. Nonetheless, this clinical information would be valuable

for identifying the role of body weight and lifestyle factors in ethnic differences in

diabetes risk which could further be used to design effective clinical prevention and

122

health promotion programs. This is discussed further in the section below entitled ‘future

research’.

5.5 Unanswered Questions and Future Research

Can these results be generalized beyond immigrants to ethnic groups that have been in

Canada for generations?

In order to address this question it will be necessary to identify a cohort of individuals

who are not immigrants but for whom we have information about their ethnic

background. Canadian health surveys have been used for this but have the limitation of

a relatively small non-white ethnic sample (Chiu et al., 2010). Surname algorithms are

another promising method to identify people of South Asian and Chinese ethnic

background (Shah et al., 2012; Khan et al., 2011), however immigrants would have to

be identified and excluded. This work will be important to understand whether, relative

to non-immigrants of the same ethnicity, our findings reflect a higher burden of illness

(perhaps due to a higher risk associated with acculturation, stress, nutrition transition

and competing priorities impacting healthy lifestyle choices) or a lower burden

(reflecting a ‘healthy immigrant effect’ due to self-selection and immigration screening).

123

What is the pattern of screening and burden of undiagnosed diabetes among young

adults (< age 40) belonging to high-risk ethnic groups?

In order to align with screening recommendations we focused our analyses in chapter 3

on those aged 40 and over. However, our results and those of other recent studies

suggest that screening may be appropriate at considerably younger ages in most ethnic

minority groups. Very little research to date has been conducted on diabetes screening

by ethnicity and none to our knowledge has looked at screening among younger adults

in these populations.

What role does BMI play in the relationship between ethnicity and diabetes, as well as

in the relationship between ethnicity, sex and diabetes?

Ethnic differences in risk appear at very young ages, and diabetes occurs at

significantly lower BMI among high risk populations, which both emphasize the genetic

origin to diabetes susceptibility. However, a greater understanding of the impact of BMI

on diabetes risk in different ethnic groups would help researchers disentangle the

genetic versus environmental contributions to diabetes risk. Some preliminary work on

this has been conducted by Chiu and colleagues (Chiu et al., 2011). Understanding sex

differences in the role of BMI is also important to design effective community-based

diabetes prevention programs. These are questions that still need addressing with

respect to diabetes risk.

124

What are the roles of area-level factors and neighbourhood environments? Do the

neighbourhoods in which immigrants settle exert a modifying influence on diabetes risk

through impacting diet and physical activity?

Interesting research in this area is being conducted by Urquia and colleagues in the

area of birth outcomes among immigrant women (Urquia et al., 2009). Their research

using multi-level methods suggest that the effects of neighbourhood deprivation exert

an increasing influence on health outcomes with longer duration of residence. Of course

due to the focus on birth outcomes, this research has been conducted on pregnant

women and it is unclear whether these findings would apply to the general population.

Other research specifically focused on the effects of area-level material deprivation on

health, has found gender differences (Matheson et al., 2008). With respect to diabetes

research, individuals living in lower income neighbourhoods in Ontario have nearly twice

the rate of diabetes as those living in wealthier neighbourhoods (Hux & Tang, 2003). It

may be that some of the reported socioeconomic gradient is related to area-level factors

such as opportunities to exercise, sources of healthy food, safety concerns that may

affect frequency of walking and other outdoor activities and stress in the environment.

Psychological/emotional stress, which can be related to neighbourhood environments

(i.e. crime, noise), has also been suggested to be a risk factor for type 2 diabetes

(Carnethon et al., 2003). Research incorporating multi-level methods would be able to

contribute to furthering our understanding of these relationships.

125

Do immigrants and people of racial-ethnic minorities have poorer diabetes-related

health outcomes in Canada?

Although diabetes-related health outcomes were not the focus of this research, given

our findings that some ethnic groups have a very high risk and that this risk begins at a

young age, a natural question that arises is whether or not these groups have a poorer

prognosis over time. There is now evidence that South Asian populations experience

high rates diabetes-related cardiovascular complications (Tunis et al., 1993; McKeigue

et al., 1993; Anand et al., 2000; Dhawan et al., 1994). There is a paucity of information,

however, about whether these complications are attributable to delays in diagnosis or

inadequate disease management and control. Some work is already being conducted

in this area (Shah et al., 2012), however, more research looking at detailed ethnic

breakdowns and incorporating immigration status and time in Canada, would be useful.

Not only are certain ethnic and socioeconomic groups more likely to develop diabetes,

but the consequences of developing diabetes may be particularly severe for socially

disadvantaged groups (Booth et al., 2012). Our research has shown a very high

incidence of diabetes among ethnic minorities who also experience an earlier age of

onset. Very few studies have looked at the consequences of type 2 diabetes diagnosed

under the age of 40. This is an area to pursue in future work.

126

5.6 Conclusions

We have demonstrated an especially high risk for diabetes in South Asian men and

women that begins at an early age and continues throughout adulthood. The disparity in

risk experienced by South Asians and other high risk ethnic groups as compared with

persons of Western European ancestry was well-established before age 30, suggesting

that diabetes prevention programs in high-risk ethnic groups should begin in childhood

and adolescence. We provide evidence that diabetes screening may be justified 15

years earlier in South Asians than in people of European ancestry and 5 to 10 years

earlier in other ethnic minority groups. High rates of screening observed in high risk,

ethnic-minority populations age 40 and up are promising and suggest that targeted

screening programs aimed at high risk groups beginning at younger ages may be

feasible.

127

Tabl

e 5.

1 S

umm

ary

of th

e el

emen

ts o

f the

The

oret

ical

Mod

el a

nd h

ow th

ey re

late

to th

e cu

rren

t wor

k.

El

emen

t(s)

Exam

ple

and

rele

vanc

e to

stu

dy

Stru

ctur

al/S

ocia

l Det

erm

inan

ts o

f Hea

lth In

equi

ties

(the

soci

al p

roce

sses

that

sha

pe th

e di

strib

utio

n of

the

inte

rmed

iary

det

erm

inan

ts o

f hea

lth

ineq

uitie

s)

Dom

ain

1:

Soci

o-po

litic

al a

nd s

ocia

l co

ntex

t G

over

nanc

e, p

olic

ies,

cul

tura

l and

so

ciet

al n

orm

s Be

yond

the

scop

e of

this

thes

is b

ut re

leva

nt e

xam

ples

incl

ude

redi

strib

utiv

e ta

x la

ws;

the

exte

nt to

whi

ch p

ublic

hea

lth a

nd

univ

ersa

l hea

lth c

are

is a

prio

rity

to th

e go

vern

men

t/soc

iety

and

w

hich

is re

flect

ed in

the

leve

l of r

esou

rces

allo

cate

d

Dom

ain

2:

Soci

al S

tratif

icat

ion/

Soc

ial

Cla

ss

Educ

atio

n/O

ccup

atio

n/In

com

e (S

EP)

Cre

ates

diff

eren

ces

acro

ss p

opul

atio

n gr

oups

in a

cces

s to

re

sour

ces

and

educ

atio

n fo

r hea

lthy

livin

g

Gen

der

Fam

ily-fr

iend

ly la

bour

law

s; c

hild

-car

e po

licie

s; fo

r im

mig

rant

s fro

m

pate

rnal

istic

soc

ietie

s th

is m

ay in

clud

e sy

stem

atic

gen

der

disc

rimin

atio

n th

at h

as d

elet

erio

us e

ffect

s on

hea

lth (a

cces

s to

re

sour

ces,

acc

ess

to e

duca

tion,

food

, saf

e en

viro

nmen

ts, s

exua

l he

alth

, etc

.)

Rac

e/et

hnic

ity

Rel

ated

to d

iscr

imin

atio

n in

peo

ples

' acc

ess

to re

sour

ces

and

oppo

rtuni

ties

base

d on

soc

ially

-con

stru

cted

idea

s of

race

Inte

rmed

iary

/Soc

ial D

eter

min

ants

of

Hea

lth

Dom

ain

1:

Soci

al D

eter

min

ants

of H

ealth

M

ater

ial C

ircum

stan

ces

(i.e.

Hou

sing

, po

tent

ial t

o ac

cess

reso

urce

s,

phys

ical

env

ironm

ent,

occu

patio

nal

envi

ronm

ent,

neig

hbou

rhoo

d)

Abili

ty to

buy

hea

lthy

food

; opp

ortu

nitie

s to

exe

rcis

e; a

bilit

y to

ac

cess

hea

lth c

are

(per

sona

l res

ourc

es to

cov

er tr

ansp

orta

tion,

ch

ildca

re c

osts

and

pos

sibl

y lo

st w

ork

reve

nue)

Biol

ogic

al fa

ctor

s G

enet

ic p

redi

spos

ition

for d

iabe

tes

asso

ciat

ed w

ith e

thni

city

Beha

viou

rs

Die

t, ph

ysic

al a

ctiv

ity; h

ealth

ser

vice

s ut

iliza

tion/

scre

enin

g

128

Psyc

hoso

cial

fact

ors

Stre

ss a

ssoc

iate

d w

ith th

e im

mig

ratio

n ex

perie

nce;

lack

of s

ocia

l su

ppor

ts; r

eset

tlem

ent;

unem

ploy

men

t or i

nfor

mal

or p

reca

rious

em

ploy

men

t - m

ay m

ore

nega

tivel

y in

fluen

ce w

omen

Dom

ain

2:

Hea

lth S

yste

m

Hea

lth S

yste

m (i

.e. A

cces

s to

hea

lth

care

, res

pons

e to

the

heal

th n

eeds

of

diffe

rent

gro

ups,

pro

mot

ing

actio

n to

im

prov

e he

alth

sta

tus)

Avai

labi

lity

of fa

mily

phy

sici

ans

for n

ewco

mer

s; A

cces

s to

sp

ecia

lists

; Tra

nsla

tor s

ervi

ces;

pro

vidi

ng h

ealth

edu

catio

nal

mat

eria

l in

diffe

rent

lang

uage

s; c

omm

unity

-bas

ed p

reve

ntio

n pr

ogra

ms

that

are

cul

tura

lly-s

ensi

tive;

Hea

lth c

over

age

(i.e.

3

mon

th w

aitin

g pe

riod)

; pro

vidi

ng p

rogr

ams

that

add

ress

the

heal

th

need

s of

vul

nera

ble

grou

ps

Dis

trib

utio

n of

Hea

lth O

utco

mes

Diff

eren

ces

in ra

tes

of s

cree

ning

; Diff

eren

ces

in s

tage

of d

isea

se a

t tim

e of

dia

gnos

is; D

iffer

ence

s in

oth

er h

ealth

risk

fact

ors

(i.e.

hy

perte

nsio

n); D

iffer

ence

s in

dis

ease

man

agem

ent;

Diff

eren

ces

in

adve

rse

outc

omes

rela

ted

to d

iabe

tes

(mor

bidi

ty, d

isab

ility,

pr

emat

ure

mor

talit

y)

129

References

Abate, N. & Chandalia, M. (2001). Ethnicity and type 2 diabetes: focus on Asian Indians.

J Diabetes Complications, 15, 320-327.

Abate, N., Chandalia, M., Snell, P. G., & Grundy, S. M. (2004). Adipose tissue

metabolites and insulin resistance in nondiabetic Asian Indian men. J Clin Endocrinol

Metab, 89, 2750-2755.

Abdullah, A., Peeters, A., de Court, & Stoelwinder, J. (2010). The magnitude of

association between overweight and obesity and the risk of diabetes: a meta-analysis of

prospective cohort studies. Diabetes Res Clin Pract, 89, 309-319.

Aguilar-Salinas, C. A., Reyes-Rodriguez, E., Ordonez-Sanchez, M. L., Torres, M. A.,

Ramirez-Jimenez, S., Dominguez-Lopez, A. et al. (2001). Early-onset type 2 diabetes:

metabolic and genetic characterization in the mexican population. J Clin Endocrinol

Metab, 86, 220-226.

Ali, O., Tan, T. T., Sakinah, O., Khalid, B. A., Wu, L. L., & Ng, M. L. (1993). Prevalence

of NIDDM and impaired glucose tolerance in aborigines and Malays in Malaysia and

their relationship to sociodemographic, health, and nutritional factors. Diabetes Care,

16, 68-75.

Allawi, J., Rao, P. V., Gilbert, R., Scott, G., Jarrett, R. J., Keen, H. et al. (1988).

Microalbuminuria in non-insulin-dependent diabetes: its prevalence in Indian compared

with Europid patients. Br Med J (Clin Res Ed), 296, 462-464.

Allender, S., Wickramasinghe, K., Goldacre, M., Matthews, D., & Katulanda, P. (2011).

Quantifying urbanization as a risk factor for noncommunicable disease. J Urban Health,

88, 906-918.

130

Altabe, M. (1998). Ethnicity and body image: Quantitative and qualitative analysis.

International Journal of Eating Disorders, 23, 153-159.

American Diabetes Association (2000). Type 2 diabetes in children and adolescents.

Diabetes Care, 23, 381-389.

American Diabetes Association (2010). Standards of medical care in diabetes - 2010.

Diabetes Care, 33, S11-S61.

Anand, S. S., Yusuf, S., Jacobs, R., Davis, A. D., Yi, Q., Gerstein, H. et al. (2001). Risk

factors, atherosclerosis, and cardiovascular disease among Aboriginal people in

Canada: the Study of Health Assessment and Risk Evaluation in Aboriginal Peoples

(SHARE-AP). Lancet, 358, 1147-1153.

Anand, S. S., Yusuf, S., Vuksan, V., Devanesen, S., Teo, K. K., Montague, P. A. et al.

(2000). Differences in risk factors, atherosclerosis, and cardiovascular disease between

ethnic groups in Canada: the Study of Health Assessment and Risk in Ethnic groups

(SHARE)[see comment]. Lancet.356(9226):279-84.

Benhalima, K., Wilmot, E., Khunti, K., Gray, L. J., Lawrence, I., & Davies, M. (2011).

Type 2 diabetes in younger adults: Clinical characteristics, diabetes-related

complications and management of risk factors. Primary care diabetes, 5, 57-62.

Bhopal, R., Unwin, N., White, M., Yallop, J., Walker, L., Alberti, K. G. et al. (1999).

Heterogeneity of coronary heart disease risk factors in Indian, Pakistani, Bangladeshi,

and European origin populations: cross sectional study. BMJ, 319, 215-220.

Bierman, A. S., Angus, J., Ahmad, F., & et al. (2010). Access to Health Care Services.

In A.S.Bierman (Ed.), Project for an Ontario Women's Health Evidence-Based Report:

Vol. 1 ( Toronto: St.Michael's Hospital and the Institute for Clinical Evaluative Sciences.

131

Block, S. & Galabuzi, G. (2011). Canada's colour coded labour market - the gap for

racialized workers. Ottawa: Canadian Centre for Policy Alternatives and The Wellesley

Institute.

Booth, G. L., Bishara, P., Lipscombe, L. L., Shah, B. R., Feig, D. S., Bhattacharyya, O.

et al. (2012). Universal Drug Coverage and Socioeconomic Disparities in Major

Diabetes Outcomes. Diabetes Care.

Booth, G. L., Creatore, M. I., Gozdyra, P., & Glazier, R. H. (2007). Ethnicity, immigration

and diabetes. In R.H.Glazier, G. L. Booth, P. Gozdyra, M. I. Creatore, & A. M. Tynan

(Eds.), Neighbourhood Environments and Resources for Healthy Living - A Focus on

Diabetes in Toronto (pp. 57-86). Toronto: Institute for Clinical Evaluative Sciences.

Booth, G. L., Rothwell, D. M., Fung, K., & Tu, J. V. (2003). Diabetes and cardiac

disease. In J.E.Hux, G. L. Booth, P. M. Slaughter, & A. Laupacis (Eds.), Diabetes in

Ontario (pp. 95-128). Toronto: Institute for Clinical Evaluative Sciences.

Brancati, F. L., Whelton, P. K., Kuller, L. H., & Klag, M. J. (1996). Diabetes mellitus,

race, and socioeconomic status. A population-based study. Ann Epidemiol, 6, 67-73.

Brancati, F. L., Whittle, J. C., Whelton, P. K., Seidler, A. J., & Klag, M. J. (1992). The

excess incidence of diabetic end-stage renal disease among blacks. A population-

based study of potential explanatory factors. JAMA, 268, 3079-3084.

Bryan, S. N., Tremblay, M. S., Perez, C. E., Ardern, C. I., & Katzmarzyk, P. T. (2006).

Physical activity and ethnicity: evidence from the Canadian Community Health Survey.

Can J Public Health, 97, 271-276.

Burden, A. C., McNally, P. G., Feehally, J., & Walls, J. (1992). Increased incidence of

end-stage renal failure secondary to diabetes mellitus in Asian ethnic groups in the

United Kingdom. Diabet.Med, 9, 641-645.

132

Canadian Diabetes Association Clinical Practice Guidelines Expert Committee &

Diabete Quebec (2011). Diabetes: Canada at the tipping point - charting a new path

Canadian Diabetes Association.

Carnethon, M. R., Kinder, L. S., Fair, J. M., Stafford, R. S., & Fortmann, S. P. (2003).

Symptoms of depression as a risk factor for incident diabetes: findings from the National

Health and Nutrition Examination Epidemiologic Follow-up Study, 1971-1992. Am J

Epidemiol, 158, 416-423.

Chalmers, J. & Cooper, M. E. (2008). UKPDS and the legacy effect. N Engl J Med, 359,

1618-1620.

Chatterji, P., Joo, H., & Lahiri, K. (2012). Racial/ethnic- and education-related disparities

in the control of risk factors for cardiovascular disease among individuals with diabetes.

Diabetes Care, 35, 305-312.

Chen, J., Ng, E., & Wilkins, R. (1996b). The health of Canada's immigrants in 1994-95.

Health Rep, 7, 33-50.

Chen, J., Ng, E., & Wilkins, R. (1996a). The health of Canada's immigrants in 1994-95.

Health Reports.7(4):33-45, 37-50.

Chen, J., Wilkins, R., & Ng, E. (1996c). Health expectancy by immigrant status, 1986

and 1991. Health Rep, 8, 29-38.

Chiu, M., Austin, P. C., Manuel, D. G., Shah, B. R., & Tu, J. V. (2011). Deriving ethnic-

specific BMI cutoff points for assessing diabetes risk. Diabetes Care, 34, 1741-1748.

Chiu, M., Austin, P. C., Manuel, D. G., & Tu, J. V. (2010). Comparison of cardiovascular

risk profiles among ethnic groups using population health surveys between 1996 and

2007. CMAJ, 182, E301-E310.

133

Citizenship and Immigration Canada (2012). OP 15 Medical Procedures. Citizenship

and Immigration Canada [On-line]. Available:

http://www.cic.gc.ca/english/resources/manuals/op/op15-eng.pdf

Citizenship and Immigration Canada, R. a. E. B. (2007). Facts and Figures Immigration

Overview (Rep. No. Catalogue no. C&I-880-06-07E). Ottawa, Ontario: The Minister of

Public Works and Government Services Canada.

Citizenship and Immigration Canada, R. a. E. B. (2009). Facts and Figures -

Immigration Overview 2008 Ottawa: Minister of Public Works and Government Services

Canada.

Coeli, C. M., Faerstein, E., Chor, D., Lopes, C. S., & Werneck, G. L. (2009). Gender

differences in the socioeconomic gradient in self-reported diabetes: does health service

access play a role? Diabetes Res Clin Pract, 86, 134-139.

Colagiuri, S. & Davies, D. (2009). The value of early detection of type 2 diabetes.

[Review] [47 refs]. Current Opinion in Endocrinology, Diabetes & Obesity.16(2):95-9.

Cooper, H. (2002). Investigating socio-economic explanations for gender and ethnic

inequalities in health. Soc Sci Med, 54, 693-706.

Cowie, C. C., Harris, M. I., & Eberhardt, M. S. (1994). Frequency and determinants of

screening for diabetes in the U.S. Diabetes Care, 17, 1158-1163.

Cowie, C. C., Harris, M. I., Silverman, R. E., Johnson, E. W., & Rust, K. F. (1993). Effect

of multiple risk factors on differences between blacks and whites in the prevalence of

non-insulin-dependent diabetes mellitus in the United States. Am J Epidemiol, 137, 719-

732.

Cowie, C. C., Rust, K. F., Byrd-Holt, D. D., Eberhardt, M. S., Flegal, K. M., Engelgau, M.

M. et al. (2006). Prevalence of diabetes and impaired fasting glucose in adults in the

134

U.S. population: National Health And Nutrition Examination Survey 1999-2002.

Diabetes Care, 29, 1263-1268.

Cowie, C. C., Rust, K. F., Ford, E. S., Eberhardt, M. S., Byrd-Holt, D. D., Li, C. et al.

(2009). Full accounting of diabetes and pre-diabetes in the U.S. population in 1988-

1994 and 2005-2006. Diabetes Care, 32, 287-294.

Creatore, M. I., Booth, G. L., Manuel, D. G., Moineddin, R., & Glazier, R. H. (2012).

Diabetes screening among immigrants: a population-based urban cohort study.

Diabetes Care, 35, 754-761.

Creatore, M. I., Moineddin, R., Booth, G., Manuel, D. H., Desmeules, M., McDermott, S.

et al. (2010). Age- and sex-related prevalence of diabetes mellitus among immigrants to

Ontario, Canada. CMAJ, 182, 781-789.

Cruickshank, J. K., Cooper, J., Burnett, M., MacDuff, J., & Drubra, U. (1991). Ethnic

differences in fasting plasma C-peptide and insulin in relation to glucose tolerance and

blood pressure. Lancet, 338, 842-847.

Dabelea, D., Bell, R. A., D'Agostino, R. B., Jr., Imperatore, G., Johansen, J. M., Linder,

B. et al. (2007). Incidence of diabetes in youth in the United States. JAMA, 297, 2716-

2724.

Dalstra, J. A., Kunst, A. E., Borrell, C., Breeze, E., Cambois, E., Costa, G. et al. (2005).

Socioeconomic differences in the prevalence of common chronic diseases: an overview

of eight European countries. Int J Epidemiol, 34, 316-326.

Danaei, G., Finucane, M. M., Lu, Y., Singh, G. M., Cowan, M. J., Paciorek, C. J. et al.

(2011). National, regional, and global trends in fasting plasma glucose and diabetes

prevalence since 1980: systematic analysis of health examination surveys and

epidemiological studies with 370 country-years and 2.7 million participants. Lancet, 378,

31-40.

135

Daniel, M., Rowley, K. G., McDermott, R., Mylvaganam, A., & O'Dea, K. (1999).

Diabetes incidence in an Australian aboriginal population. An 8-year follow-up study.

Diabetes Care, 22, 1993-1998.

De Maio, F. G. & Kemp, E. (2010). The deterioration of health status among immigrants

to Canada. Glob Public Health, 5, 462-478.

Desmeules, M., McDermott, S., Cao, Z., Manuel, D., Kazanjian, A., Vissandjee, B. et al.

(2004). Immigrant health and health care utilization in Canada: Phase II of the National

Immigrant Health Initiative.

Dhawan, J., Bray, C. L., Warburton, R., Ghambhir, D. S., & Morris, J. (1994). Insulin

resistance, high prevalence of diabetes, and cardiovascular risk in immigrant Asians.

Genetic or environmental effect? Br.Heart J, 72, 413-421.

Dowse, G. K., Gareeboo, H., Zimmet, P. Z., Alberti, K. G., Tuomilehto, J., Fareed, D. et

al. (1990). High prevalence of NIDDM and impaired glucose tolerance in Indian, Creole,

and Chinese Mauritians. Mauritius Noncommunicable Disease Study Group. Diabetes,

39, 390-396.

Dunn, J. R. & Dyck, I. (2000). Social determinants of health in Canada's immigrant

population: results from the National Population Health Survey. Social Science &

Medicine.51(11):1573-93.

Dyck, R., Osgood, N., Lin, T. H., Gao, A., & Stang, M. R. (2010). Epidemiology of

diabetes mellitus among First Nations and non-First Nations adults. CMAJ, 182, 249-

256.

Ebrahim, S., Kinra, S., Bowen, L., Andersen, E., Ben-Shlomo, Y., Lyngdoh, T. et al.

(2010). The effect of rural-to-urban migration on obesity and diabetes in India: a cross-

sectional study. PLoS Med, 7, e1000268.

136

Engelgau, M. M., Geiss, L. S., Saaddine, J. B., Boyle, J. P., Benjamin, S. M., Gregg, E.

W. et al. (2004). The evolving diabetes burden in the United States. Ann Intern Med,

140, 945-950.

Feltbower, R. G., McKinney, P. A., Campbell, F. M., Stephenson, C. R., & Bodansky, H.

J. (2003). Type 2 and other forms of diabetes in 0-30 year olds: a hospital based study

in Leeds, UK. Arch Dis Child, 88, 676-679.

Finucane, M. M., Stevens, G. A., Cowan, M. J., Danaei, G., Lin, J. K., Paciorek, C. J. et

al. (2011). National, regional, and global trends in body-mass index since 1980:

systematic analysis of health examination surveys and epidemiological studies with 960

country-years and 9.1 million participants. Lancet, 377, 557-567.

Fischbacher, C. M., Bhopal, R., Rutter, M. K., Unwin, N. C., Marshall, S. M., White, M.

et al. (2003). Microalbuminuria is more frequent in South Asian than in European origin

populations: a comparative study in Newcastle, UK. Diabet.Med, 20, 31-36.

Fischbacher, C. M., Hunt, S., & Alexander, L. (2004). How physically active are South

Asians in the United Kingdom? A literature review. [Review] [41 refs]. Journal of Public

Health.26(3):250-8.

Flegal, K. M., Carroll, M. D., Ogden, C. L., & Johnson, C. L. (2002). Prevalence and

trends in obesity among US adults, 1999-2000. JAMA, 288, 1723-1727.

Franco, L. J. (1996). Diabetes in Japanese-Brazilians--influence of the acculturation

process. Diabetes Res Clin Pract., 34 Suppl, S51-S57.

Fujimoto, W. Y., Bergstrom, R. W., Boyko, E. J., Kinyoun, J. L., Leonetti, D. L., Newell-

Morris, L. L. et al. (1994). Diabetes and diabetes risk factors in second- and third-

generation Japanese Americans in Seattle, Washington. Diabetes Res Clin Pract., 24

Suppl, S43-S52.

137

Gagnon, A. J. (2002). Responsiveness of the Canadian health care system towards

Newcomers (Rep. No. Discussion Paper No.40). Ottawa: Commission on the Future of

Health Care in Canada.

Gelberg, L., Andersen, R. M., & Leake, B. D. (2000). The Behavioral Model for

Vulnerable Populations: application to medical care use and outcomes for homeless

people. Health Serv Res, 34, 1273-1302.

Glazier, R. H., Creatore, M. I., Gozdyra, P., Matheson, F. I., Steele, L. S., Boyle, E. et

al. (2004). Geographic methods for understanding and responding to disparities in

mammography use in Toronto, Canada. J Gen Intern Med, 19, 952-961.

Goel, M. S., McCarthy, E. P., Phillips, R. S., & Wee, C. C. (2004). Obesity among US

immigrant subgroups by duration of residence. JAMA, 292, 2860-2867.

Goel, V. (1994). Factors associated with cervical cancer screening: results from the

Ontario Health Survey. Can J Public Health, 85, 125-127.

Gourdy, P., Ruidavets, J. B., Ferrieres, J., Ducimetiere, P., Amouyel, P., Arveiler, D. et

al. (2001). Prevalence of type 2 diabetes and impaired fasting glucose in the middle-

aged population of three French regions - The MONICA study 1995-97. Diabetes

Metab., 27, 347-358.

Gray, L. J., Tringham, J. R., Davies, M. J., Webb, D. R., Jarvis, J., Skinner, T. C. et al.

(2010). Screening for type 2 diabetes in a multiethnic setting using known risk factors to

identify those at high risk: a cross-sectional study. Vasc Health Risk Manag, 6, 837-842.

Gray, L. J., Yates, T., Davies, M. J., Brady, E., Webb, D. R., Sattar, N. et al. (2011).

Defining obesity cut-off points for migrant South Asians. PLoS One, 6, e26464.

Griffin, S. J., Borch-Johnsen, K., Davies, M. J., Khunti, K., Rutten, G. E., Sandbaek, A.

et al. (2011). Effect of early intensive multifactorial therapy on 5-year cardiovascular

138

outcomes in individuals with type 2 diabetes detected by screening (ADDITION-

Europe): a cluster-randomised trial. Lancet, 378, 156-167.

Gupta, L. S., Wu, C. C., Young, S., & Perlman, S. E. (2011). Prevalence of diabetes in

New York City, 2002-2008: comparing foreign-born South Asians and other Asians with

U.S.-born whites, blacks, and Hispanics. Diabetes Care, 34, 1791-1793.

Guthrie, B., Emslie-Smith, A., & Morris, A. D. (2009). Which people with Type 2

diabetes achieve good control of intermediate outcomes? Population database study in

a UK region. Diabet.Med, 26, 1269-1276.

Guttmann, A., Manuel, D., Stukel, T. A., Desmeules, M., Cernat, G., & Glazier, R. H.

(2008). Immunization coverage among young children of urban immigrant mothers:

findings from a universal health care system. Ambul.Pediatr, 8, 205-209.

Hara, H., Egusa, G., Yamakido, M., & Kawate, R. (1994). The high prevalence of

diabetes mellitus and hyperinsulinemia among the Japanese-Americans living in Hawaii

and Los Angeles. Diabetes Res Clin Pract., 24 Suppl, S37-S42.

Harris, M. I. & Eastman, R. C. (2000). Early detection of undiagnosed diabetes mellitus:

a US perspective. Diabetes Metab.Res Rev, 16, 230-236.

Harris, S. B., Gittelsohn, J., Hanley, A., Barnie, A., Wolever, T. M., Gao, J. et al. (1997).

The prevalence of NIDDM and associated risk factors in native Canadians. Diabetes

Care, 20, 185-187.

Hedemalm, A., Schaufelberger, M., & Ekman, I. (2010). Reported adherence and social

support among immigrants compared to native Swedes. J Immigr.Minor.Health, 12,

124-131.

139

Hennis, A., Wu, S. Y., Nemesure, B., Li, X., & Leske, M. C. (2002). Diabetes in a

Caribbean population: epidemiological profile and implications. Int J Epidemiol, 31, 234-

239.

Hillier, T. A. & Pedula, K. L. (2003). Complications in young adults with early-onset type

2 diabetes: losing the relative protection of youth. Diabetes Care, 26, 2999-3005.

Holman, R. R., Paul, S. K., Bethel, M. A., Matthews, D. R., & Neil, H. A. (2008). 10-year

follow-up of intensive glucose control in type 2 diabetes. N Engl J Med, 359, 1577-1589.

Hosmer, D. W. & Lemeshow, S. (1999). Applied Survival Analysis. New York: John

Wiley & Sons, Inc.

Hux, J. E., Ivis, F., Flintoft, V., & Bica, A. (2002). Diabetes in Ontario: determination of

prevalence and incidence using a validated administrative data algorithm. Diabetes

Care, 25, 512-516.

Hux, J. E., Jacka, R., Fung, K., & Rothwell, D. M. (2003). Diabetes and peripheral

vascular disease. In J.E.Hux, G. L. Booth, P. Slaughter, & A. Laupacis (Eds.), Diabetes

in Ontario: An ICES Practice Atlas (pp. 129-149). Toronto: Institute for Clinical

Evaluative Sciences.

Hux, J. E. & Tang, M. (2003). Patterns of prevalence and incidence of diabetes. In

J.E.Hux, G. L. Booth, P. M. Slaughter, & A. Laupacis (Eds.), Diabetes in Ontario: An

ICES Practice Atlas (pp. 1-18). Toronto: Institute for Clinical Evaluative Sciences.

Iacoviello, L., Arnout, J., Buntinx, F., Cappuccio, F. P., Dagnelie, P. C., de Lorgeril, M.

et al. (2001). Dietary habit profile in European communities with different risk of

myocardial infarction: the impact of migration as a model of gene-environment

interaction. The IMMIDIET Study. Nutr.Metab.Cardiovasc Dis., 11, 122-126.

140

Imkampe, A. K. & Gulliford, M. C. (2011). Increasing socio-economic inequality in type 2

diabetes prevalence--repeated cross-sectional surveys in England 1994-2006. Eur J

Public Health, 21, 484-490.

International Diabetes Federation (2009). website. International Diabetes Federation

[On-line]. Available: http://www.idf.org/node/1052?unode=3B9689B0-C026-2FD3-

879219B2881892E7

International Diabetes Federation (IDF) (2003). Diabetes Atlas, second edition Brussels,

Belgium: International Diabetes Federation.

Jacobson, A. M., Hauser, S. T., Willett, J., Wolfsdorf, J. I., & Herman, L. (1997).

Consequences of irregular versus continuous medical follow-up in children and

adolescents with insulin-dependent diabetes mellitus. J Pediatr, 131, 727-733.

Janssen, P. G., Gorter, K. J., Stolk, R. P., & Rutten, G. E. (2009). Randomised

controlled trial of intensive multifactorial treatment for cardiovascular risk in patients with

screen-detected type 2 diabetes: 1-year data from the ADDITION Netherlands study.

Br.J Gen.Pract, 59, 43-48.

Jeffery, R. W., French, S. A., Forster, J. L., & Spry, V. M. (1991). Socioeconomic status

differences in health behaviors related to obesity: the Healthy Worker Project. Int J

Obes, 15, 689-696.

Jenum, A. K., Holme, I., Graff-Iversen, S., & Birkeland, K. I. (2005). Ethnicity and sex

are strong determinants of diabetes in an urban Western society: implications for

prevention. Diabetologia, 48, 435-439.

Jimenez-Corona, A., Rojas, R., Gomez-Perez, F. J., & guilar-Salinas, C. A. (2010).

Early-onset type 2 diabetes in a Mexican survey: results from the National Health and

Nutrition Survey 2006. Salud Publica Mex, 52 Suppl 1, S27-S35.

141

Joshi, P., Islam, S., Pais, P., Reddy, S., Dorairaj, P., Kazmi, K. et al. (2007). Risk

factors for early myocardial infarction in South Asians compared with individuals in other

countries. JAMA, 297, 286-294.

Kahn, R., Alperin, P., Eddy, D., Borch-Johnsen, K., Buse, J., Feigelman, J. et al. (2010).

Age at initiation and frequency of screening to detect type 2 diabetes: a cost-

effectiveness analysis. Lancet, 375, 1365-1374.

Kapral, M. K., Rothwell, D. M., Fung, K., Tang, M., Booth, G. L., & Laupacis, A. (2003).

Diabetes and stroke. In J.E.Hux, G. L. Booth, P. Slaughter, & A. Laupacis (Eds.),

Diabetes in Ontario. An ICES Practice Atlas. (pp. 151-163). Toronto: Institute for Clinical

Evaluative Sciences.

Katz, D. & Gagnon, A. J. (2002). Evidence of adequacy of postpartum care for

immigrant women. Can J Nurs Res, 34, 71-81.

Katzmarzyk, P. T. (2002). The Canadian obesity epidemic, 1985-1998. CMAJ, 166,

1039-1040.

Khambalia, A., Phongsavan, P., Smith, B. J., Keke, K., Dan, L., Fitzhardinge, A. et al.

(2011). Prevalence and risk factors of diabetes and impaired fasting glucose in Nauru.

BMC Public Health, 11, 719.

Khan, N. A., Wang, H., Anand, S., Jin, Y., Campbell, N. R., Pilote, L. et al. (2011).

Ethnicity and sex affect diabetes incidence and outcomes. Diabetes Care, 34, 96-101.

King, H. & Rewers, M. (1993). Global estimates for prevalence of diabetes mellitus and

impaired glucose tolerance in adults. WHO Ad Hoc Diabetes Reporting Group. Diabetes

Care, 16, 157-177.

Kinnon, D. (1999). Canadian research on immigration and health: An overview Ottawa:

Minister of Public Works and Government Services.

142

Klein, R. & Klein, B. E. K. (1995). Vision disorders in diabetes. In National Diabetes

Data Group (Ed.), Diabetes in America (2nd ed., pp. 293-338). Washington: National

Insitutes of Health.

Kliewer, E.V. & Kazanjian, A. (2000). The health status and medical services utilization

of recent immigrants to Manitoba and British Columbia: A pilot study. Vancouver,B.C.:

Office of Health Technology Assessment, University of British Columbia.

Krieger, N. (1992). Overcoming the absence of socioeconomic data in medical records:

validation and application of a census-based methodology. Am J Public Health, 82, 703-

710.

Kumar, P. J. & Clark, M. L. (1999). Diabetes mellitus and other disorders of metabolism.

In P.J.Kumar & M. L. Clark (Eds.), Clinical Medicine (4th ed., pp. 959-1005). London:

Sounders.

Laroche, M. (2000). Health Status and Health Services Utilization of Canada's

Immigrant and Non-Immigrant Population. Canadian Public Policy, XXVI, 51-73.

Leiter, L. A., Barr, A., Belanger, A., Lubin, S., Ross, S. A., Tildesley, H. D. et al. (2001).

Diabetes Screening in Canada (DIASCAN) Study: prevalence of undiagnosed diabetes

and glucose intolerance in family physician offices. Diabetes Care, 24, 1038-1043.

Lipscombe, L. L. & Hux, J. E. (2007). Trends in diabetes prevalence, incidence, and

mortality in Ontario, Canada 1995-2005: a population-based study. Lancet, 369, 750-

756.

Lofters, A. K., Moineddin, R., Hwang, S. W., & Glazier, R. H. (2010). Low rates of

cervical cancer screening among urban immigrants: a population-based study in

Ontario, Canada. Med Care, 48, 611-618.

143

Lynch, J. W., Smith, G. D., Kaplan, G. A., & House, J. S. (2000). Income inequality and

mortality: importance to health of individual income, psychosocial environment, or

material conditions. BMJ, 320, 1200-1204.

Ma, R. C. & Chan, J. C. (2009). Pregnancy and diabetes scenario around the world:

China. Int J Gynaecol.Obstet., 104 Suppl 1, S42-S45.

Mackenbach, J. P., Looman, C. W., & van der Meer, J. B. (1996). Differences in the

misreporting of chronic conditions, by level of education: the effect on inequalities in

prevalence rates. Am J Public Health, 86, 706-711.

Manuel, D. G. & Schultz, S. E. (2003). Diabetes health status and risk factors. In

J.E.Hux, G. L. Booth, P. M. Slaughter, & A. Laupacis (Eds.), Diabetes in Ontario: An

ICES Practice Atlas. (pp. 77-94). Toronto: Institute for Clinical Evaluative Sciences.

Manuel, D. G. & Schultz, S. E. (2004). Health-related quality of life and health-adjusted

life expectancy of people with diabetes in Ontario, Canada, 1996-1997. Diabetes Care,

27, 407-414.

Marmot, M. G., Smith, G. D., Stansfeld, S., Patel, C., North, F., Head, J. et al. (1991).

Health inequalities among British civil servants: the Whitehall II study. Lancet, 337,

1387-1393.

Mather, H. M., Chaturvedi, N., & Kehely, A. M. (1998). Comparison of prevalence and

risk factors for microalbuminuria in South Asians and Europeans with type 2 diabetes

mellitus. Diabetic Medicine, 15, 672-677.

Mather, H. M. & Keen, H. (1985). The Southall Diabetes Survey: prevalence of known

diabetes in Asians and Europeans. Br.Med J (Clin Res Ed.), 291, 1081-1084.

144

Matheson, F. I., Moineddin, R., & Glazier, R. H. (2008). The weight of place: a multilevel

analysis of gender, neighborhood material deprivation, and body mass index among

Canadian adults. Soc Sci Med, 66, 675-690.

Matuk, L. C. (1996). Health status of newcomers. Canadian Journal of Public

Health.Revue Canadienne de Sante Publique.87(1):52-5, -Feb.

Maty, S. C., James, S. A., & Kaplan, G. A. (2010). Life-course socioeconomic position

and incidence of diabetes mellitus among blacks and whites: the Alameda County

Study, 1965-1999. Am J Public Health, 100, 137-145.

Mbanya, J. C., Cruickshank, J. K., Forrester, T., Balkau, B., Ngogang, J. Y., Riste, L. et

al. (1999). Standardized comparison of glucose intolerance in west African-origin

populations of rural and urban Cameroon, Jamaica, and Caribbean migrants to Britain.

Diabetes Care, 22, 434-440.

McBean, A. M., Li, S., Gilbertson, D. T., & Collins, A. J. (2004). Differences in diabetes

prevalence, incidence, and mortality among the elderly of four racial/ethnic groups:

whites, blacks, hispanics, and asians. Diabetes Care, 27, 2317-2324.

McDermott, R. A., Li, M., & Campbell, S. K. (2010). Incidence of type 2 diabetes in two

Indigenous Australian populations: a 6-year follow-up study. Med J Aust, 192, 562-565.

McDonald, J. T. & Kennedy, S. (2004). Insights into the 'healthy immigrant effect': health

status and health service use of immigrants to Canada. Soc Sci Med, 59, 1613-1627.

McDonald, J. T. & Kennedy, S. (2005). Is migration to Canada associated with

unhealthy weight gain? Overweight and obesity among Canada's immigrants. Soc Sci

Med, 61, 2469-2481.

McKeigue, P. M. (1992). Coronary heart disease in Indians, Pakistanis, and

Bangladeshis: aetiology and possibilities for prevention. Br.Heart J., 67, 341-342.

145

McKeigue, P. M., Ferrie, J. E., Pierpoint, T., & Marmot, M. G. (1993). Association of

early-onset coronary heart disease in South Asian men with glucose intolerance and

hyperinsulinemia. Circulation, 87, 152-161.

McKeigue, P. M., Pierpoint, T., Ferrie, J. E., & Marmot, M. G. (1992). Relationship of

glucose intolerance and hyperinsulinaemia to body fat pattern in south Asians and

Europeans. Diabetologia, 35, 785-791.

McLaren, L. (2007). Socioeconomic status and obesity. Epidemiol Rev, 29, 29-48.

Meisinger, C., Thorand, B., Schneider, A., Stieber, J., Doring, A., & Lowel, H. (2002).

Sex differences in risk factors for incident type 2 diabetes mellitus: the MONICA

Augsburg cohort study. Arch Intern Med, 162, 82-89.

Misra, A. & Ganda, O. P. (2007). Migration and its impact on adiposity and type 2

diabetes. [Review] [119 refs]. Nutrition.23(9):696-708.

Misra, A. & Khurana, L. (2009). The metabolic syndrome in South Asians: epidemiology,

determinants, and prevention. Metab Syndr Relat Disord, 7, 497-514.

Mokdad, A. H., Ford, E. S., Bowman, B. A., Nelson, D. E., Engelgau, M. M., Vinicor, F.

et al. (2000). Diabetes trends in the U.S.: 1990-1998. Diabetes Care, 23, 1278-1283.

Mokdad, A. H., Serdula, M. K., Dietz, W. H., Bowman, B. A., Marks, J. S., & Koplan, J.

P. (2000). The continuing epidemic of obesity in the United States. JAMA, 284, 1650-

1651.

Mukhopadhyay, B., Forouhi, N. G., Fisher, B. M., Kesson, C. M., & Sattar, N. (2006). A

comparison of glycaemic and metabolic control over time among South Asian and

European patients with Type 2 diabetes: results from follow-up in a routine diabetes

clinic. Diabetic Medicine, 23, 94-98.

146

Narayan, K. M., Boyle, J. P., Thompson, T. J., Sorensen, S. W., & Williamson, D. F.

(2003). Lifetime risk for diabetes mellitus in the United States. JAMA, 290, 1884-1890.

National Institute for Health and Clinical Excellence (NICE) (2011). Preventing type 2

diabetes - risk identification and interventions for individuals at high risk: draft guidance.

Newbold, K. B. (2005). Self-rated health within the Canadian immigrant population: risk

and the healthy immigrant effect. Soc Sci Med, 60, 1359-1370.

Newbold, K. B. (2009). Health care use and the Canadian immigrant population. Int J

Health Serv, 39, 545-565.

Newbold, K. B. & Danforth, J. (2003). Health status and Canada's immigrant population.

Social Science & Medicine.57(10):1981-95.

Ng, C., Corey, P. N., & Young, T. K. (2011). Socio-economic patterns of obesity among

aboriginal and non-Aboriginal Canadians. Can J Public Health, 102, 264-268.

Ng, E., Wilkins, R., Gendron, F., & Berthelot, J. (2005). Dynamics of immigrants' health

in Canada: Evidence from a National Population Health Survey Ottawa: Statistics

Canada.

NHS (2012). NHS Health Check and national vascular risk assessment program.

Diabetes UK [On-line]. Available:

http://www.diabetes.org.uk/About_us/Our_Views/Position_statements/Early_identificatio

n_of_people_with_Type_2_diabetes/

O'Dea, K., Patel, M., Kubisch, D., Hopper, J., & Traianedes, K. (1993). Obesity,

diabetes, and hyperlipidemia in a central Australian aboriginal community with a long

history of acculturation. Diabetes Care, 16, 1004-1010.

147

Ogawa, Y., Uchigata, Y., Otani, T., & Iwamoto, Y. (2007). Proportion of diabetes type in

early-onset diabetes in Japan. Diabetes Care, 30, e30.

Oliver, M. J., Lok, C. E., Shi, J., & Rothwell, D. M. (2003). Dialysis therapy for persons

with diabetes. In J.E.Hux, G. L. Booth, P. Slaughter, & A. Laupacis (Eds.), Diabetes in

Ontario: An ICES Practice Atlas (pp. 165-177). Toronto: Institute for Clinical Evaluative

Sciences.

Oza-Frank, R., Ali, M. K., Vaccarino, V., & Narayan, K. M. (2009). Asian Americans:

diabetes prevalence across U.S. and World Health Organization weight classifications.

Diabetes Care.32(9):1644-6.

Patel, J. V., Vyas, A., Cruickshank, J. K., Prabhakaran, D., Hughes, E., Reddy, K. S. et

al. (2006). Impact of migration on coronary heart disease risk factors: comparison of

Gujaratis in Britain and their contemporaries in villages of origin in India.

Atherosclerosis.185(2):297-306.

Pavkov, M. E., Hanson, R. L., Knowler, W. C., Bennett, P. H., Krakoff, J., & Nelson, R.

G. (2007). Changing patterns of type 2 diabetes incidence among Pima Indians.

Diabetes Care, 30, 1758-1763.

Perez, C. E. (2002). Health status and health behaviour among immigrants. Health

Reports, 13, 89-100.

Pincus, G. & White, P. (1933). On the inheritance of diabetes mellitus. Analysis of 675

family histories. Am J Med Sci, 186, 1-14.

Pomerleau, J., McKeigue, P. M., & Chaturvedi, N. (1999). Factors associated with

obesity in South Asian, Afro-Caribbean and European women. Int J

Obes.Relat.Metab.Disord., 23, 25-33.

148

Public Health Agency of Canada (2008). Diabetes in Canada: Highlights from the

National Diabetes Surveillance System 2005-2006 Ottawa: Minister of Health.

Public Health Agency of Canada (2009). Report from the National Diabetes Surveillance

System: Diabetes in Canada, 2009 Ottawa: Centre for Chronic Disease Prevention and

Control.

Public Health Agency of Canada (2011). Diabetes in Canada: Facts and figures from a

public health perspective Ottawa.

Qiao, Q., Hu, G., Tuomilehto, J., Nakagami, T., Balkau, B., Borch-Johnsen, K. et al.

(2003). Age- and sex-specific prevalence of diabetes and impaired glucose regulation in

11 Asian cohorts. Diabetes Care, 26, 1770-1780.

Quan, L., Xu, Y., Luo, S. P., Wang, L., LeBlanc, D., & Wang, T. (2006). Negotiated care

improves fluid status in diabetic peritoneal dialysis patients. Perit.Dial.Int, 26, 95-100.

Raji, A., Seely, E. W., Arky, R. A., & Simonson, D. C. (2001). Body fat distribution and

insulin resistance in healthy Asian Indians and Caucasians. J Clin Endocrinol Metab,

86, 5366-5371.

Ramachandran, A., Mary, S., Yamuna, A., Murugesan, N., & Snehalatha, C. (2008).

High prevalence of diabetes and cardiovascular risk factors associated with

urbanization in India. Diabetes Care, 31, 893-898.

Ramachandran, A., Snehalatha, C., Baskar, A. D., Mary, S., Kumar, C. K., Selvam, S.

et al. (2004). Temporal changes in prevalence of diabetes and impaired glucose

tolerance associated with lifestyle transition occurring in the rural population in India.

Diabetologia, 47, 860-865.

149

Ramachandran, A., Snehalatha, C., Kapur, A., Vijay, V., Mohan, V., Das, A. K. et al.

(2001). High prevalence of diabetes and impaired glucose tolerance in India: National

Urban Diabetes Survey. Diabetologia, 44, 1094-1101.

Rathmann, W., Haastert, B., Icks, A., Lowel, H., Meisinger, C., Holle, R. et al. (2003).

High prevalence of undiagnosed diabetes mellitus in Southern Germany: target

populations for efficient screening. The KORA survey 2000. Diabetologia, 46, 182-189.

Raymond, N. T., Varadhan, L., Reynold, D. R., Bush, K., Sankaranarayanan, S.,

Bellary, S. et al. (2009). Higher prevalence of retinopathy in diabetic patients of South

Asian ethnicity compared with white Europeans in the community: a cross-sectional

study. Diabetes Care, 32, 410-415.

Renzaho, A. M., Mellor, D., Boulton, K., & Swinburn, B. (2010). Effectiveness of

prevention programmes for obesity and chronic diseases among immigrants to

developed countries - a systematic review. Public Health Nutr, 13, 438-450.

Research and Evaluation Branch Citizenship and Immigration Canada (2009). Facts

and Figures - Immigration Overview 2008 Ottawa: Minister of Public Works and

Government Services Canada.

Rimoin, D. L. (1969). Ethnic variability in glucose tolerance and insulin secretion. Arch

Intern Med, 124, 695-700.

Robbins, J. M., Vaccarino, V., Zhang, H., & Kasl, S. V. (2001). Socioeconomic status

and type 2 diabetes in African American and non-Hispanic white women and men:

evidence from the Third National Health and Nutrition Examination Survey. Am J Public

Health, 91, 76-83.

Roderick, P. J., Jones, I., Raleigh, V. S., McGeown, M., & Mallick, N. (1994). Population

need for renal replacement therapy in Thames regions: ethnic dimension. BMJ, 309,

1111-1114.

150

Ross, N. A., Gilmour, H., & Dasgupta, K. (2010). 14-year diabetes incidence: the role of

socio-economic status. Health Rep, 21, 19-28.

Rotimi, C. N., Cooper, R. S., Okosun, I. S., Olatunbosun, S. T., Bella, A. F., Wilks, R. et

al. (1999). Prevalence of diabetes and impaired glucose tolerance in Nigerians,

Jamaicans and US blacks. Ethn.Dis., 9, 190-200.

Rubin, L. R., Fitts, M. L., & Becker, A. E. (2003). "Whatever feels good in my soul":

Body ethnic and aesthetics among African American and Latina women. Culture,

Medicine and Psychiatry, 27, 49-75.

Rull, J. A., guilar-Salinas, C. A., Rojas, R., Rios-Torres, J. M., Gomez-Perez, F. J., &

Olaiz, G. (2005). Epidemiology of type 2 diabetes in Mexico. Arch Med Res, 36, 188-

196.

Salinero-Fort, M. A., del Otero-Sanz, L., Martin-Madrazo, C., de Burgos-Lunar, C.,

Chico-Moraleja, R. M., Rodes-Soldevila, B. et al. (2011). The relationship between

social support and self-reported health status in immigrants: an adjusted analysis in the

Madrid Cross Sectional Study. BMC Fam Pract, 12, 46.

Sandbaek, A., Griffin, S. J., Rutten, G., Davies, M., Stolk, R., Khunti, K. et al. (2008).

Stepwise screening for diabetes identifies people with high but modifiable coronary

heart disease risk. The ADDITION study. Diabetologia, 51, 1127-1134.

Shah, B. R. (2008). Utilization of physician services for diabetic patients from ethnic

minorities. Journal of Public Health.30(3):327-31.

Shah, B. R., Anand, S., Zinman, B., & Duong-Hua, M. (2003). Diabetes and First

Nations People. In J.E.Hux, G. L. Booth, P. Slaughter, & A. Laupacis (Eds.), Diabetes in

Ontario: An ICES Practice Atlas (pp. 231-248). Toronto: Institute for Clinical Evaluative

Sciences.

151

Shah, B. R., Cauch-Dudek, K., Anand, S. S., Austin, P. C., Manuel, D. G., & Hux, J. E.

(2012). Absence of disparities in the quality of primary diabetes care for South Asians

and Chinese in an urban Canadian setting. Diabetes Care, 35, 794-796.

Shields, M., Tremblay, M. S., Laviolette, M., Craig, C. L., Janssen, I., & Gorber, S. C.

(2010). Fitness of Canadian adults: results from the 2007-2009 Canadian Health

Measures Survey. Health Rep, 21, 21-35.

Sicree, R., Shaw, J., & Zimmet, P. (2006). Prevalence and projections. In D.Gan (Ed.),

Diabetes Atlas (3rd ed., pp. 16-104). Brussels: International Diabetes Federation.

Singh-Manoux, A., Clarke, P., & Marmot, M. (2002). Multiple measures of socio-

economic position and psychosocial health: proximal and distal measures. Int J

Epidemiol, 31, 1192-1199.

Solar, O. & Irwin, A. (2010). A conceptual framework for action on the social

determinants of health Geneva: World Health Organization.

Song, S. H. & Hardisty, C. A. (2009). Early onset type 2 diabetes mellitus: a harbinger

for complications in later years--clinical observation from a secondary care cohort. QJM,

102, 799-806.

Statistics Canada. (2006a). Canadian community health survey, cycle 3.1, 2005

[computer file]. Statistics Canada [producer]. Ottawa, Ont., Data liberation initiative

[distributor].

Statistics Canada. (2006b). Profile for Urban Areas, 2006 Census, for Canada,

provinces and territories - 20% sample data (table). Ottawa, Statistics Canada.

Statistics Canada (2007a). Low-income rates among immigrants entering Canada. The

Daily.

152

Statistics Canada. (2007b). Population by immigrant status and period of immigration,

percentage distribution (2006), for Canada, provinces and territories - 20% sample data

(table). Ottawa., Statistics Canada. Immigration and Citizenship Highlight Tables. 2006

Census.

Stephen, E. H., Foote, K., Hendershot, G. E., & Schoenborn, C. A. (1994). Health of the

foreignborn population. Advanced Data from Vital and Health Statistics, 241, 1-10.

Swanson, J. G. & Kaczorowski, J. (2007). Mammography rates for 20 community-based

family practices in Ontario: a full practice audit. Can J Public Health, 98, 374-378.

Tang, M., Chen, Y., & Krewski, D. (2003). Gender-related differences in the association

between socioeconomic status and self-reported diabetes. Int J Epidemiol, 32, 381-385.

Tillin, T., Forouhi, N., Johnston, D. G., McKeigue, P. M., Chaturvedi, N., & Godsland, I.

F. (2005). Metabolic syndrome and coronary heart disease in South Asians, African-

Caribbeans and white Europeans: a UK population-based cross-sectional study.

Diabetologia, 48, 649-656.

Tremblay, M. S., Bryan, S. N., Perez, C. E., Ardern, C. I., & Katzmarzyk, P. T. (2006).

Physical activity and immigrant status: evidence from the Canadian Community Health

Survey. Can J Public Health, 97, 277-282.

Tseng, C. H., Tseng, C. P., Chong, C. K., Huang, T. P., Song, Y. M., Chou, C. W. et al.

(2006). Increasing incidence of diagnosed type 2 diabetes in Taiwan: analysis of data

from a national cohort. Diabetologia, 49, 1755-1760.

Tunis, S. R., Bass, E. B., Klag, M. J., & Steinberg, E. P. (1993). Variation in utilization of

procedures for treatment of peripheral arterial disease. A look at patient characteristics.

Arch Intern Med, 153, 991-998.

153

UK Prospective Diabetes Study Group (1994). UK Prospective Diabetes Study. XII:

Differences between Asian, Afro-Caribbean and white Caucasian type 2 diabetic

patients at diagnosis of diabetes. UK Prospective Diabetes Study Group. Diabet.Med,

11, 670-677.

Ur, E., Chiasson, J., Ransom, T., & Rowe, R. (2008). Screening for Type 1 and Type 2

Diabetes - 2008 Clinical Practice Guidelines Canadian Diabetes Association.

Urquia, M. L., Frank, J. W., Glazier, R. H., & Moineddin, R. (2007). Birth outcomes by

neighbourhood income and recent immigration in Toronto. Health Rep., 18, 21-30.

Urquia, M. L., Frank, J. W., Glazier, R. H., Moineddin, R., Matheson, F. I., & Gagnon, A.

J. (2009). Neighborhood context and infant birthweight among recent immigrant

mothers: a multilevel analysis. Am J Public Health, 99, 285-293.

van den Donk, M., Sandbaek, A., Borch-Johnsen, K., Lauritzen, T., Simmons, R. K.,

Wareham, N. J. et al. (2011). Screening for Type 2 diabetes. Lessons from the

ADDITION-Europe study. Diabet.Med, 28, 1416-1424.

Vimalananda, V. G., Rosenzweig, J. L., Cabral, H. J., David, M. M., & Lasser, K. E.

(2011). Comparison of diabetes control among Haitians, African Americans, and non-

Hispanic whites in an urban safety-net hospital. Diabetes Care, 34, 58-60.

Waugh, N., Scotland, G., McNamee, P., Gillett, M., Brennan, A., Goyder, E. et al.

(2007). Screening for type 2 diabetes: literature review and economic modelling. Health

Technol Assess, 11, iii-xi, 1.

Webb, D. R., Gray, L. J., Khunti, K., Srinivasan, B., Taub, N., Campbell, S. et al. (2011).

Screening for diabetes using an oral glucose tolerance test within a western multi-ethnic

population identifies modifiable cardiovascular risk: the ADDITION-Leicester study.

Diabetologia, 54, 2237-2246.

154

Webb, D. R., Khunti, K., Srinivasan, B., Gray, L. J., Taub, N., Campbell, S. et al. (2010).

Rationale and design of the ADDITION-Leicester study, a systematic screening

programme and randomised controlled trial of multi-factorial cardiovascular risk

intervention in people with type 2 diabetes mellitus detected by screening. Trials, 11, 16.

Whincup, P. H., Nightingale, C. M., Owen, C. G., Rudnicka, A. R., Gibb, I., McKay, C.

M. et al. (2010). Early emergence of ethnic differences in type 2 diabetes precursors in

the UK: the Child Heart and Health Study in England (CHASE Study). PLoS Med, 7,

e1000263.

Wild, S., Roglic, G., Green, A., Sicree, R., & King, H. (2004). Global prevalence of

diabetes: estimates for the year 2000 and projections for 2030. Diabetes Care, 27,

1047-1053.

Wilkins, R. (2008). PCCF+ Version 5D User's Guide. Automated Geographic Coding

Based on the Statistics Canada Postal Code Conversion Files, Including Postal Codes

through March 2008. Ottawa, Health Information and Research Division, Statistics

Canada.

Williams, J. I. & Young W (1996). Appendix: A summary of studies on the quality of

administrative databases in Canada. In V.Goel, J. I. Williams, G. M. Anderson, & et al.

(Eds.), Patterns of health care in Ontario. The ICES Practice Atlas. (2nd ed., pp. 339-

346). Ottawa: Canadian Medical Association.

Wilson, S. E., Lipscombe, L. L., Rosella, L. C., & Manuel, D. G. (2009). Trends in

laboratory testing for diabetes in Ontario, Canada 1995-2005: a population-based study.

BMC Health Serv Res, 9, 41.

Wilson, S. E., Rosella, L. C., Lipscombe, L. L., & Manuel, D. G. (2010). The

effectiveness and efficiency of diabetes screening in Ontario, Canada: a population-

based cohort study. BMC Public Health, 10, 506.

155

Yajnik, C. S. & Deshmukh, U. S. (2008). Maternal nutrition, intrauterine programming

and consequential risks in the offspring. Rev Endocr.Metab Disord, 9, 203-211.

Young, C. (1992). Mortality, the ultimate indicator of survival: the differential experience

between birthplace groups. In J.Donovan, E. M. d'Espaignet, C. Merton, & M. van

Ommerent (Eds.), Immigrants in Australia: A Health Profile (pp. 34-70). Canberra:

Australian Government Publishing Service.

Young, T. K. (2001). Undiagnosed diabetes: burden and significance in the Canadian

population. Advances in Experimental Medicine & Biology, 498, 7-10.

Young, T. K. & Mustard, C. A. (2001). Undiagnosed diabetes: does it matter? CMAJ,

164, 24-28.

Young, T. K., Reading, J., Elias, B., & O'Neil, J. D. (2000). Type 2 diabetes mellitus in

Canada's first nations: status of an epidemic in progress. CMAJ, 163, 561-566.

Zaninotto, P., Mindell, J., & Hirani, V. (2007). Prevalence of cardiovascular risk factors

among ethnic groups: results from the Health Surveys for England. Atherosclerosis,

195, e48-e57.

Zimmet, P., Alberti, K. G., & Shaw, J. (2001). Global and societal implications of the

diabetes epidemic. Nature, 414, 782-787.

Zimmet, P., Taft, P., Guinea, A., Guthrie, W., & Thoma, K. (1977). The high prevalence

of diabetes mellitus on a Central Pacific Island. Diabetologia, 13, 111-115.

156

Appendices

157

Appendix A. Data Sources

The Registered Persons Database (RPDB)

The Ontario Registered Persons Database (RPDB) is an electronic registry of all

individuals who have ever been eligible for health coverage in Ontario. Health care

eligibility is extended to virtually all Canadian citizens, permanent residents or landed

immigrants who have Ontario as their primary place of residence. This database

contains information received from the Ministry of Health and Long-Term Care (date of

birth, sex, date of death, time period(s) for which the individual was eligible for OHIP

coverage, residential postal code) and is then enriched with additional datasets

maintained at the Institute for Clinical Evaluative Sciences (ICES) including

"Date of Last Contact" (DOLC) information. With the scrambled identification number

contained in the RPDB, individuals can be linked to other health administrative files

including physician billing files (OHIP), hospital discharge files (DAD), emergency

department use (NACRS) and disease registries such as the Ontario Diabetes

Database (ODD).

The principal limitations of the RPDB are due to the fact that individuals are not forced

to inform the Ministry of Health when they move, resulting in: 1) the RPDB containing

individuals who are eligible for OHIP but who no longer reside in the province; 2) out-of-

date information about postal code of residence, reducing the reliability of using RPDB

postal codes to identify where an individual lives or assign them area-level indicators

such as neighbourhood socioeconomic status.

Landed Immigrants Data System (LIDS)

The Canadian Landed Immigrant Database (Landed Immigrant Data System or LIDS) is

an electronic database maintained by Citizenship and Immigration Canada (CIC) that

assembles the information from the IMM 1000 forms filled out by immigration officers

158

when immigrants receive landed resident status. These forms serve as a visa for entry

into Canada. The database used for this thesis is a subset of the Canadian LIDS and

contains information on all legal immigrants to Canada between the years 1985 and

2000 who indicated that their destination province was Ontario. The data items included

in the database are sex, date of birth, country of birth, date of landing, education level,

intended occupation, language ability at landing time, and immigration visa category

(refugee in Canada, refugee abroad, economic class, family class, etc.). Only those

individuals granted landed immigrant status (i.e. permanent residents) are included in

this database, therefore refugee claimants who never obtained official refugee status,

temporary residents, such as foreign workers and students, are excluded.

LIDS – RPDB linkage

A probabilistic linkage was carried out linking the Ontario LIDS to the Ontario health

care registry (the RPDB, see below for more information) using surname, given names,

sex, date of birth as well as a manual review of low-quality match, and non-matching

records. Of 1,666,473 records initially in the Ontario LIDS, 1,414,977 (84.4%) were

successfully linked.

An initial validation of the linkage between the Ontario LIDS and RPDB suggested that

there was little difference in the percentage linked between various groups and

differences were therefore unlikely to introduce bias into research results.1 The

differences that were noted include the following:

1) Landing Year – There was some variation by landing year with higher percentages of

successful linkages on or after 1990 (82-88%) attributed to the introduction of new

electronic health cards in 1991. The lowest linkage was in 1985 at 72%;

2) Visa Category – There was a lower linkage for “Business” class immigrants (72.3%)

as compared to other categories. This observation is likely due to a higher percentage

159

in this visa category of people who apply for immigration status for business or

investment reasons, yet who continue to live in their home country;

3) Class – “Investors” (68.4%) and “independent entrepreneurs” (59.3%) had the lowest

linkage and “refugees” had the highest (90.1%), likely due to the same phenomenon as

described for differences by visa category;

4) Country – There was some variation in the linkage by country of birth, however, the

countries with the lowest linkage (< 60%) were: Lichtenstein, Monaco, Taiwan, St.

Pierre Miquelon, Algeria, Morocco, Tunisia, Mauritania, Niger and Brunei. It is possible

that people may have indicated that they were coming to Ontario but then ended up

moving somewhere else. This may particularly be true for individuals coming from the

African countries since there is a large African (particularly French-speaking) community

in Montreal and Quebec may have become an attractive alternative to Ontario.

The principal limitation of this dataset is related to the method used for data linkage.

By restricting the immigration data available for linkage to those who claim Ontario to be

their destination province, and by not including the entire Canadian LIDS for linkage, we

have lost information on immigrants who initially indicated that they were going to settle

in a province other than Ontario and who moved to Ontario after arrival in Canada.

These people would not be identified as immigrants through our LIDS linkage and could

potentially be mis-classified as non-immigrants. In order to circumvent this issue, in all

our analyses we excluded non-LIDS individuals who received first every health

insurance eligibility after 1991.

Ontario Diabetes Database (ODD)

The Ontario Diabetes Database (ODD) is a cumulative disease registry that contains all

Ontario diabetic patients identified since 1991. The database is re-created annually

using hospital discharge abstracts from CIHI (including same day surgery) database,

physician service claims from OHIP database and information regarding the

demographics of persons eligible for health care coverage in Ontario from the

160

Registered Persons Database (RPDB). Persons enter the ODD as incident cases when

they are defined as having diabetes according to the following criteria: individuals with

at least one hospitalization or at least two claims for physicians' services (within two

years) bearing a diagnosis of diabetes. Gestational diabetes is excluded from the

registry by removing any hospital record with a diagnosis of pregnancy care or delivery

close to a diabetic record (i.e. diabetic record date between 120 days before and 180

days after a gestational admission date ). See Appendix C for the ODD inclusion flow

chart.

The ODD has been validated and the algorithm was found to be highly sensitive (86%)

and specific (97%) for identifying patients in whom diabetes was recorded in primary

care charts.2

There are several limitations related to the ODD that affect the results of this research

(and have been discussed in the limitations sections of the previous chapters). Briefly,

the ODD is based on health services utilization patterns and will therefore under-

diagnose diabetes in populations that either have very poor access or who for other

reasons have infrequent contact with the health care system. This, however, likely

represents a very small proportion of the Ontario, urban population. Previous research

has shown that 81% of Canadians see a health care provider annually (Statistics

Canada, CCHS 3.1), and this percentage is likely to be higher for urban-dwelling

populations, which are the focus of this research. Secondly, the ODD cannot distinguish

between type 1 and type 2 diabetes. Type 2 represents the vast majority of diabetes

cases and by limiting our study population to adults (>=20 years of age) we have further

increased the probability that most cases in our study represent type 2 diabetes.

161

References

1. Cernat G, Wall C, Iron K, Manuel D. Initial validation of Landed Immigrant Data System (LIDS) with the Registered Person's Database (RPDB) at ICES. Internal ICES Report to Health Canada. 2002. Toronto, Institute for Clinical Evaluative Sciences (ICES).

2. Hux JE, Ivis F, Flintoft V, Bica A. Diabetes in Ontario: determination of prevalence and incidence using a validated administrative data algorithm. Diabetes Care 2002; 25(3):512-516.

162

Appendix B. ODD algorithm

CIHI records with any diagnosis code 250.x (ICD-9)

OHIP physician service claims with diagnosis code 250.x

Single OHIP claim only

2 OHIP claims or 1 discharge in 2 years

Candidate cases for DM

Presumed gestational DM

Previously in ODD?

No

Incident Cases

Prior prevalent

Cases

Total Cases

Yes

163

Appendix C. Countries included in the Citizenship and Immigration Canada database and the assigned World Region of Origin.

COB code from CIC data Country Name Assigned World Region of Origin620 Anguilla Caribbean621 Antigua And Barbuda Caribbean658 Aruba Caribbean622 Bahama Islands, The Caribbean610 Barbados Caribbean601 Bermuda Caribbean624 Cayman Islands Caribbean650 Cuba Caribbean625 Dominica Caribbean651 Dominican Republic Caribbean626 Grenada Caribbean653 Guadeloupe Caribbean654 Haiti Caribbean602 Jamaica Caribbean655 Martinique Caribbean627 Montserrat Caribbean652 Netherlands Antilles, The Caribbean628 Nevis Caribbean899 Ocean Nes Caribbean656 Puerto Rico Caribbean629 St. Kitts-Nevis Caribbean630 St. Lucia Caribbean631 St. Vincent and the Grenadines Caribbean605 Trinidad & Tobago, Republic of Caribbean632 Turks and Caicos Islands Caribbean633 Virgin Islands, British Caribbean657 Virgin Islands, U.S. Caribbean299 Asia Nes East Asia and the Pacific399 Australia Nes East Asia and the Pacific256 Cambodia East Asia and the Pacific202 China, People's Republic of East Asia and the Pacific840 Cook Islands East Asia and the Pacific801 Fiji East Asia and the Pacific845 French Polynesia East Asia and the Pacific832 Guam East Asia and the Pacific204 Hong Kong East Asia and the Pacific200 Hong Kong Sar East Asia and the Pacific222 Indonesia, Republic of East Asia and the Pacific207 Japan East Asia and the Pacific831 Kiribati East Asia and the Pacific257 Korea, People's Democratic Republic of East Asia and the Pacific258 Korea, Republic of East Asia and the Pacific260 Laos East Asia and the Pacific261 Macao East Asia and the Pacific172 Madagascar East Asia and the Pacific242 Malaysia East Asia and the Pacific262 Mongolia, People's Republic of East Asia and the Pacific241 Myanmar (Burma) East Asia and the Pacific341 Nauru East Asia and the Pacific

164

822 New Caledonia East Asia and the Pacific830 Pacific Islands, US Trust Territory of the East Asia and the Pacific342 Papau New Guinea East Asia and the Pacific227 Philippines East Asia and the Pacific842 Pitcairn Island East Asia and the Pacific903 Reunion East Asia and the Pacific843 Samoa, American East Asia and the Pacific844 Samoa, Western East Asia and the Pacific246 Singapore East Asia and the Pacific824 Solomons, The East Asia and the Pacific203 Taiwan East Asia and the Pacific267 Thailand East Asia and the Pacific268 Tibet East Asia and the Pacific846 Tonga East Asia and the Pacific823 Vanuatu East Asia and the Pacific270 Vietnam, Socialist Republic of East Asia and the Pacific841 Wallis and Futuna East Asia and the Pacific081 Albania Eastern Europe and Central Asia049 Armenia Eastern Europe and Central Asia050 Azerbaijan Eastern Europe and Central Asia051 Belarus Eastern Europe and Central Asia048 Bosnia-Hercegovina Eastern Europe and Central Asia083 Bulgaria Eastern Europe and Central Asia043 Croatia Eastern Europe and Central Asia070 Fyr Macedonia Eastern Europe and Central Asia052 Georgia Eastern Europe and Central Asia025 Greece Eastern Europe and Central Asia026 Hungary Eastern Europe and Central Asia053 Kazakhstan Eastern Europe and Central Asia 054 Kyrgyzstan Eastern Europe and Central Asia019 Latvia Eastern Europe and Central Asia020 Lithuania Eastern Europe and Central Asia055 Moldova Eastern Europe and Central Asia033 Poland Eastern Europe and Central Asia088 Romania Eastern Europe and Central Asia056 Russia Eastern Europe and Central Asia016 Slovak Republic Eastern Europe and Central Asia047 Slovenia Eastern Europe and Central Asia057 Tadjikistan Eastern Europe and Central Asia045 Turkey Eastern Europe and Central Asia058 Turkmenistan Eastern Europe and Central Asia059 Ukraine Eastern Europe and Central Asia042 Union of Soviet Socialist Republics Eastern Europe and Central Asia060 Uzbekistan Eastern Europe and Central Asia044 Yugoslavia Eastern Europe and Central Asia703 Argentina Latin America and Mexico541 Belize Latin America and Mexico751 Bolivia Latin America and Mexico709 Brazil Latin America and Mexico721 Chile Latin America and Mexico722 Colombia Latin America and Mexico542 Costa Rica Latin America and Mexico753 Ecuador Latin America and Mexico543 El Salvador Latin America and Mexico754 French Guiana Latin America and Mexico544 Guatemala Latin America and Mexico

165

711 Guyana Latin America and Mexico545 Honduras Latin America and Mexico501 Mexico Latin America and Mexico546 Nicaragua Latin America and Mexico548 Panama Canal Zone Latin America and Mexico547 Panama, Republic of Latin America and Mexico755 Paraguay Latin America and Mexico723 Peru Latin America and Mexico752 Surinam Latin America and Mexico724 Uruguay Latin America and Mexico725 Venezuela Latin America and Mexico131 Algeria North Africa and the Middle East253 Bahrain North Africa and the Middle East255 Brunei North Africa and the Middle East183 Djibouti, Republic of North Africa and the Middle East101 Egypt North Africa and the Middle East223 Iran North Africa and the Middle East224 Iraq North Africa and the Middle East206 Israel North Africa and the Middle East225 Jordan North Africa and the Middle East226 Kuwait North Africa and the Middle East208 Lebanon North Africa and the Middle East171 Libya North Africa and the Middle East133 Morocco North Africa and the Middle East263 Oman North Africa and the Middle East213 Palestinian Authority (Gaza/West Bank) North Africa and the Middle East265 Qatar North Africa and the Middle East231 Saudi Arabia North Africa and the Middle East210 Syria North Africa and the Middle East135 Tunisia North Africa and the Middle East280 United Arab Emirates North Africa and the Middle East274 Yemen, People's Democratic Republic of North Africa and the Middle East273 Yemen, Republic of North Africa and the Middle East252 Afghanistan South Asia212 Bangladesh South Asia254 Bhutan South Asia205 India South Asia901 Maldives, Republic of South Asia264 Nepal South Asia209 Pakistan South Asia201 Sri Lanka South Asia199 Africa Nes Sub-Saharan Africa151 Angola Sub-Saharan Africa160 Benin, Peoples Republic of Sub-Saharan Africa153 Botswana, Republic of Sub-Saharan Africa188 Burkino-Faso Sub-Saharan Africa154 Burundi Sub-Saharan Africa155 Cameroon, Federal Republic of Sub-Saharan Africa911 Cape Verde Islands Sub-Saharan Africa157 Central Africa Republic Sub-Saharan Africa156 Chad, Republic of Sub-Saharan Africa905 Comoros Sub-Saharan Africa159 Congo, People's Republic of the Sub-Saharan Africa158 Congo,Democratic Republic of Sub-Saharan Africa162 Eritrea Sub-Saharan Africa161 Ethiopia Sub-Saharan Africa

166

521 Greenland Western Europe & U.S.

163 Gabon Republic Sub-Saharan Africa164 Gambia Sub-Saharan Africa165 Ghana Sub-Saharan Africa178 Guinea, Equatorial Sub-Saharan Africa166 Guinea, Republic of Sub-Saharan Africa167 Guinea-Bissau Sub-Saharan Africa169 Ivory Coast, Republic Sub-Saharan Africa132 Kenya Sub-Saharan Africa152 Lesotho Sub-Saharan Africa170 Liberia Sub-Saharan Africa111 Malawi Sub-Saharan Africa173 Mali, Republic of Sub-Saharan Africa174 Mauritania Sub-Saharan Africa902 Mauritius Sub-Saharan Africa175 Mozambique Sub-Saharan Africa122 Namibia Sub-Saharan Africa176 Niger, Republic of the Sub-Saharan Africa177 Nigeria Sub-Saharan Africa179 Rwanda Sub-Saharan Africa914 Sao Tome e Principe Sub-Saharan Africa180 Senegal Sub-Saharan Africa904 Seychelles Sub-Saharan Africa181 Sierra Leone Sub-Saharan Africa182 Somalia, Democratic Republic of Sub-Saharan Africa121 South Africa, Republic of Sub-Saharan Africa185 Sudan, Democratic Republic of Sub-Saharan Africa186 Swaziland Sub-Saharan Africa130 Tanzania, United Republic of Sub-Saharan Africa187 Togo, Republic of Sub-Saharan Africa136 Uganda Sub-Saharan Africa184 Western Sahara Sub-Saharan Africa112 Zambia Sub-Saharan Africa113 Zimbabwe Sub-Saharan Africa000 Unknown Unknown origin/Stateless082 Andorra Western Europe & U.S.305 Australia Western Europe & U.S.011 Austria Western Europe & U.S.035 Azores Western Europe & U.S.012 Belgium Western Europe & U.S.511 Canada Western Europe & U.S.039 Canary Islands Western Europe & U.S.009 Channel Islands Western Europe & U.S.221 Cyprus Western Europe & U.S.015 Czech Republic Western Europe & U.S.014 Czechoslovakia Western Europe & U.S.017 Denmark Western Europe & U.S.002 England Western Europe & U.S.018 Estonia Western Europe & U.S.099 Europe Nes Western Europe & U.S.912 Falkland Islands Western Europe & U.S.021 Finland Western Europe & U.S.022 France Western Europe & U.S.046 German Democratic Republic Western Europe & U.S.024 Germany, Federal Republic of Western Europe & U.S.084 Gibraltar Western Europe & U.S.

167

008 Wales Western Europe & U.S.

090 Holy See Western Europe & U.S.085 Iceland Western Europe & U.S.027 Ireland, Republic of Western Europe & U.S.028 Italy Western Europe & U.S.086 Liechtenstein Western Europe & U.S.013 Luxembourg Western Europe & U.S.036 Madeira Western Europe & U.S.030 Malta Western Europe & U.S.087 Monaco Western Europe & U.S.031 Netherlands, The Western Europe & U.S.339 New Zealand Western Europe & U.S.006 Northern Ireland Western Europe & U.S.032 Norway Western Europe & U.S.034 Portugal Western Europe & U.S.089 San Marino Western Europe & U.S.007 Scotland Western Europe & U.S.037 Spain Western Europe & U.S.915 St. Helena Western Europe & U.S.531 St. Pierre et Miquelon Western Europe & U.S.040 Sweden Western Europe & U.S.041 Switzerland Western Europe & U.S.461 United States of America Western Europe & U.S.

168

Appendix D. Incidence Cohort Creation

Ontario population (RPDB), eligible at any time 1991-2010

N=16,393,637

Immigrant cohort Long-term residents N=15,016,843 N=1,376,793

Exclude rural postal codes

N=1,350,744 N=12,679,645

N=1,295,841 N=12,253,543

Exclude those with no record of contact with health system†

Exclude those with diagnosed diabetes on or prior to March 31st, 1994

N=1,278,140 N=11,985,081

N= 7,312,766

Exclude those with first ever eligibility after April 1, 1991*

N= 818,258

Exclude those who: arrived prior to 1991, had >2 years between arrival and first health care eligibility, or who lost eligibility prior to baseline.

N= 592,376

N= 5,421,654

Exclude those less than 20 year of age at baseline

*In order to remove individuals from Long-term resident cohort who may be recent immigrants (but not in LIDSs). † No date of last contact (DOLC = missing)

169

Appendix E - Age-Period Cohort Effects

Due to the nature of our study cohort where we have multiple waves of immigrants

arriving in different years, born in different time periods, and who arrive at different

ages, we were faced with the issue of possible age-period-cohort effects. Briefly, there

is an intrinsic and mathematical relationship between these three variables whereby the

following can be written:

Period = age + cohort

This relationship makes standard regression modeling techniques insufficient for

controlling for the impact of all three characteristics on the outcome. In order to

investigate the impact of period of arrival and age at arrival, our immigrant study

population was divided into multiple groups based on their age at arrival (30-39, 40-49,

50-59, 60-69) as well as their immigration period. Immigration period was categorized

into three groups, those arriving between 1991 and 1993, 1994 and 1996, and 1997 and

2000. We then used a simple Poisson regression model with person-year offset to look

at whether the probability of being diagnosed with diabetes differed by different

immigrant cohorts (was time-dependent) and we saw that, indeed, more recent waves

of immigrants were at higher risk of being diagnosed with diabetes than earlier waves of

immigrants, both for women (RR with 1991-1993 as baseline, and 95% confidence

intervals: 0.93 (0.88-0.98), p= 0.0039 for 1994-96; 0.76 (0.72-0.80), p<0.0001 for 1997-

2000) and for men (RR, and 95% confidence intervals: 0.88 (0.83-0.93), p<0.0001 for

1994-96; 0.71 (0.68-0.76), p<0.0001). After controlling for characteristics of immigrant

waves including age distribution, world regions of birth, income and visa category (i.e.

composition of immigrant cohort of refugees, family class, etc.) this cohort effect

remained significant. This likely reflects the global rise in obesity rates and

accompanying diabetes burden over the past few decades that have been readily

170

described in the literature.1, 2 Due to this effect, it was necessary to take this into

account in our analyses and we created a variable to represent wave of immigration in

order to be able to control for cohort effects in our Cox model. These cohort effects

should be explored more fully when additional waves of immigration data are available.

For the purposes of this study I only controlled for this effect and did not fully explore it.

References

1. Finucane MM, Stevens GA, Cowan MJ et al. National, regional, and global trends in body-mass index since 1980: systematic analysis of health examination surveys and epidemiological studies with 960 country-years and 9.1 million participants. Lancet 2011; 377(9765):557-567.

2. Danaei G, Finucane MM, Lu Y et al. National, regional, and global trends in fasting plasma glucose and diabetes prevalence since 1980: systematic analysis of health examination surveys and epidemiological studies with 370 country-years and 2.7 million participants. Lancet 2011; 378(9785):31-40.

171

Appendix F – Detailed methodology of the Cox Proportional Hazard Model Model building

Product limit estimator methods, Kaplan-Meier survival curves and log-rank chi-

square scores were used to identify which covariates had an association with

diabetes diagnosis after immigration and should be included in the multivariate

model. Log-rank test results were adjusted using the Tukey method for multiple

comparisons.

The first step was to determine whether censoring was evenly distributed across

risk groups. If the pattern of censoring is not random, this can introduce bias in

our estimates. This data contains only right-censored data with a combination of

Type I (people who are censored at the end of follow-up time) and random

(people who are censored due to loss of eligibility) censoring. Using PROC

FREQ we looked at the proportion of censored individuals across different values

of our covariates of interest (i.e. sex, world region of origin, income quintile,

immigration visa category, education). There were no significant discrepancies of

proportion censored across groups apart from the following: slightly higher %

censored among immigrants from Western Europe and the U.S. at 19% versus a

mean of 11.7% across all world regions; refugees had 6% censored observations

versus a mean of 12% across all other visa immigration categories.

Testing the Cox Proportional Hazard Assumptions

Assumptions of the Cox P-H model are (ref: Survival Analysis using the

proportional hazards model – course notes. SAS Institute Inc., Cary, NC USA.

2001.):

• The relationship between the continuous predictor variables and the log

hazard should be linear

172

• In the absence of interaction terms, the effects of the predictor variables

should be additive

• The effects of the predictor variables are the same at all values of time –

i.e. the hazard ratio is constant over time

Linearity Assumption

The only covariate in our model that is continuous is age at baseline. We used

two methods to test the linearity assumption for our immigrant cohort data. First,

we replaced the continuous variable age with dummy variables representing

increasing age groups and fit a model with these dummy variables. We then

plotted the parameter estimates of the dummy variables and fitted a smoothed

line to the parameter estimate plot. If the linearity assumption is met, the line

should display a linear trend. Since the line was not linear (see figure F.1 below),

was actually an inverse ‘U’ shape, the linearity assumption was not met. As a

second test, we plotted the Martingale residuals of age. The results of this test as

well indicated that age does not meet the linearity assumption (results not

shown).

This lack of linearity was not unexpected, since due to the ‘healthy immigrant’

effect and screening associated with immigration, we would expect that people

that meet the immigrant health assessment criteria at older ages are likely not

the people who carry the highest genetic risk of diabetes. We could expect a

certain ‘survivor effect’ whereby people who come younger have a longer period

of their lives in which them may develop the disease. Whereas those who have

reached their 60’s without yet developing diabetes or another serious chronic

disease detectable in the pre-immigration medical exam are likely to be healthier

individuals. To test this hypothesis we re-ran the tests for linearity for long-term

residents. As predicted by our hypothesis, the linearity assumption for the

relationship between age and risk was violated for immigrants but was

reasonably met for our long-term resident group (see figure F.2 below). To deal

173

with this violation of the linearity assumption for age as a continuous variable we

ran the Cox Proportional Hazard models only stratified by 5-year age groups.

Within each age-grouping, the linearity assumption was met and age was

entered into each stratified model as a continuous variable.

Figure F.1 Immigrants

174

Figure F.2 Long-term Residents

175

Testing the proportional hazards assumption

The proportional hazards assumption was tested in 2 ways. First we plotted the

log-negative-log of the Kaplan-Meier survival curves (generated using PROC

LIFETEST in SAS) versus the log of survival time (see figures F.3-F.7 below). If

the proportionality assumption is upheld (and the hazard ratios are constant) then

the plot should show parallel lines.

The second method we used to test the proportionality assumption was to plot

the Schoenfeld residuals (Schoenfeld 1982). A smoothed line fitted to the

residuals plotted against time should have an intercept and a slope around 0

(Hosmer and Lemeshow 1999). For all the covariates the proportional hazards

assumption was upheld (see figures F.8-F.12 below).

176

Figures F.3-F.7 Log-negative-log tests of proportionality

177

178

179

Figures F.8-F.12 Plots of Schoenfeld Residuals to assess proportionality

180

181

182

In studyWith diabetes at baseline In study

With diabetes at baseline

Population 5,421,654 241,665 592,376 22,903Median Age at baseline 42 64 38 58% aged 65+ 15.5 48.7 7.0 32.7% male 47.5 52.6 47.2 48.4Income quintile§ of neighbourhood of residence (%):

Q1 (lowest income) 18.4 25.3 29.1 32.9Q2 19.6 22.6 23.3 25.5Q3 19.5 19.5 19.8 19.4Q4 20.1 16.9 16.3 13.8Q5 22.2 15.5 11.3 8.4

World Region of BirthEast Asia & the Pacific - - 179,200 (30.3%) 5,848 (25.5%)

South Asia - - 141,538 (23.9%) 9,292 (40.6%)Eastern Europe & Central Asia - - 97,651 (16.5%) 1,551 (6.8%)North Africa & the Middle East - - 46,515 (7.9%) 1,544 (6.7%)

Latin America & Mexico - - 32,922 (5.6%) 1,475 (6.4%)Western Europe & U.S. - - 32,475 (5.5%) 629 (2.7%)

Sub-Saharan Africa - - 30,579 (5.2%) 1,012 (4.4%)The Caribbean - - 30,003 (5.1%) 1,531 (6.7%)

Unknown/Stateless - - 414 (0.1%) 21 (0.09%)Immigration Visa Category

Family - - 257,494 (43.6%) 14,674 (64.1%)Economic - - 242,627 (41.0%) 5,421 (23.7%)

Refugee - - 76,069 (12.9%) 2,356 (10.3%)Other - - 15,107 (2.6%) 452 (2.0%)

Educational Qualifications at Landing0-9 yrs schooling - - 99,707 (19.8%) 9,292 (40.6%)

10 yrs up to secondary - - 145,781 (29.4%) 5,983 (26.1%)

Non-University Qualifications/ Some university - - 122,292 (21.1%) 2,924 (12.8%)University Degree or Higher - - 175,877 (29.7%) 4,454 (19.4%)

Percent with English Language Proficiency at Landing (%) 60.3 53.2Year of arrival

1991-1993 - - 187,766 (31.8%) 7,350 (32.1%)1994-1996 - - 178,438 (30.2%) 7,252 (31.7%)1997-2000 - - 225,093 (38.0%) 8,301 (36.2%)

Appendix G. Characteristics of the Ontario long-term resident and recent immigrant study populations*, as well as those excluded due to prior diabetes diagnosis, 2010.

§ Most recent postal code of residence was used and linked to the most relevant Census income information for that year.

*In order to be included in the study, at baseline individuals had to be: alive, eligible for provincial health care for a minimum of 3 years, urban dwelling, over age 20 and diabetes free.

Recent ImmigrantsLong-term Residents

183

Appendix H - Diabetes incidence rates before and after restriction to those who have received a diabetes test.

Age-standardized average incidence (1994-2010), by sex, and world region of birth.

02468

10121416182022

South

Asia

Caribb

ean

Mexico

& La

tin Ameri

ca

Sub-S

ahara

n Afric

a

N. Afric

a & M

iddle

East

East A

sia &

Pac

ific

E Europe

& C

entra

l Asia

W E

urope

& U.S.

All Immigr

ants

Non-im

migran

tsInci

denc

e ra

te p

er 1

,000

per

son

year

s

MaleFemale

Age-standardized average incidence (1994-2010), by sex, and world region of birth. Study population restricted to those with a DM test between 1991-2010

02468

10121416182022

South

Asia

Caribb

ean

Mexico

& La

tin Ameri

ca

Sub-S

ahara

n Afric

a

N. Afric

a & M

iddle

East

East A

sia &

Pac

ific

E Europe

& C

entra

l Asia

W E

urope

& U.S.

All Immigr

ants

Non-im

migran

ts

Inci

denc

e (p

er 1

,000

per

son-

year

s)

MaleFemale

184

Appendix I – Cox Proportional Hazard model sensitivity analyses

185

Figures I.1- I.2 - Unadjusted diabetes incidence for men and women by ethnicity and age.

Men

0

5

10

15

20

25

30

20-24 25-29 30-34 35-39 40-44 45-49 50-54 55-59 60-64

Age

Inci

denc

e pe

r 1,0

00 p

erso

n-ye

ars

Caribbean East Asia & Pacific E Europe & Central AsiaLatin America & Mexico N Africa & Middle East South AsiaSub-Saharan Africa W Europe & U.S.

Women

0

5

10

15

20

25

30

20-24 25-29 30-34 35-39 40-44 45-49 50-54 55-59 60-64

Age

Inci

denc

e pe

r 1,0

00 p

erso

n-ye

ars

Caribbean E Asia & Pacific E Europe & Central Asia

Latin America & Mexico N Africa & Middle East South Asia

Sub-Saharan Africa W Europe & U.S.

186

Figures I.3-I.4 - Adjusted diabetes incidence for men and women by ethnicity and age (based on risk at 1-year of follow-up)

Men

0

5

10

15

20

25

30

35

40

20-24 25-29 30-34 35-39 40-44 45-49 50-54 55-59 60-64 65-69

Age

Adj

uste

d in

cide

nce

per 1

,000

per

son-

year

s

Caribbean East Asia & Pacific E. Europe & Central AsiaLatin America & Mexico N Africa & Middle East South AsiaSub-Saharan Africa W Europe & U.S.

Women

0

5

10

15

20

25

30

35

20-24 25-29 30-34 35-39 40-44 45-49 50-54 55-59 60-64 65-69

Age

Adj

uste

d in

cide

nce

per 1

,000

per

son-

year

s

Caribbean East Asia & Pacific E. Europe & Central AsiaLatin America & Mexico N Africa & Middle East South AsiaSub-Saharan Africa W Europe & U.S.

187