the nature of wellbeing and its relationship to health inequalities health surveys user meeting 14...
TRANSCRIPT
The nature of wellbeing and its relationship to
Health Inequalities
Health Surveys User meeting14th July 2010
Tom Hennell Senior Public Health Analyst
Department of Health North [email protected]
0161 952 4559
The context: a debate on ‘being well’
• Liberal Economics approach: ‘being well’ is about the accumulation and distribution of economic welfare.
– Proxied by the aggregated monetary value of traded goods and services– Readily quantifiable and modelled by econometric techniques – Gross
Domestic Product, Gross Value Added– Relating to the market economy; hence tending to be a discourse of the
‘right’
• Public Health approach: ‘being well’ is about the accumulation and distribution of good health; WHO definition as ‘complete physical, mental and social wellbeing’
– Proxied by life expectancy, hospitalisation rates, disability rates, self-reported ‘health in general’
– Quantified indicators readily analysable through econometric techniques; modelled in England, Scotland and Wales through successive NHS resource allocation formulae
– Relating to the actions of public agencies ; tending to be a discourse of the ‘left’
• Social Dynamics approaches : ‘being well’ is about establishing and sustaining status and reciprocal obligation within the domains of household, neighbourhood, workplace and nation. Two current flavours in current UK discourse (with much cross-fertilisation) :
– an internal critique of the ‘right’, to do with changing family structures, time preference and consequent generational inequity;
– an internal critique of the ‘left’, focussing on social justice and inequity of economic power
– So far, proposed social dynamics instruments are yet to establish recognition as quantifiable at the individual level; and hence not amenable to econometric techniques : ‘life satisfaction’, ‘happiness’
Wellbeing and ‘being well’: three approaches
1. Being well as “not being ill”; the response of the person in the street,
– if so, not separately quantified at all.
2. Being well as an ideal state of “complete physical, mental and social wellbeing”; analysed in terms of protection against loss, and promotion of recovery,
– if so, a fluid concept whose quantification may be expected to vary according to the balance of domains within which questions may be framed.
– ‘Wellbeing’ metrics typically constructed by aggregation: ‘Adding Up’
3. Being well as an acquired and mutual capacity for being better able to gain from social opportunities, and being able to recover sooner from setbacks; potentially transferable from one social domain to another,
– if so, the extent of being well may be solid and consistently quantified, if a technique can be found to extract the underlying common factor of improved functioning within any population survey (so long as the topics covered are wide-ranging enough).
– ‘Being well’ metric quantified by data reduction: ‘Boiling Down’
• I am here using ‘Wellbeing’ to refer to values calibrated from specific survey instruments; and ‘being well’ to refer to an extracted underlying factor
Wellness and IllnessHealth Survey for England 2006
0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100%
1
2
3
4
5
6
7
8
9
10
Dec
iles
of
incr
easi
ng
wel
lnes
s
Percentage by illness category
no long-term illness Non-limiting long-term illness Limiting long-term illness
CATPCA wellness and long-term illness
Wellness and IllnessHealth Survey for England 2006
0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100%
1
2
3
4
5
6
7
8
9
10
Dec
iles
of
incr
easi
ng
wel
lnes
s
Percentage by illness category
no long-term illness Non-limiting long-term illness Limiting long-term illness
divide illness at ‘limiting’
Wellness and IllnessHealth Survey for England 2006
0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100%
1
2
3
4
5
6
7
8
9
10
Dec
iles
of
incr
easi
ng
wel
lnes
s
Percentage by illness category
no long-term illness Non-limiting long-term illness Limiting long-term illness
divide wellness at lowest quintile
Wellness and IllnessHealth Survey for England 2006
0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100%
1
2
3
4
5
6
7
8
9
10
Dec
iles
of
incr
easi
ng
wel
lnes
s
Percentage by illness category
no long-term illness Non-limiting long-term illness Limiting long-term illness
well notill well ill
unwellnotill
unwell ill
simplify into four categories
Proportion of adults in each categoryHSE 2008
well notill well ill
66% 14%
unwell notill unwell ill
9% 11% >>>> becoming ill >>>>
>>
bei
ng w
ell >
>
Odds of reporting diabetic illness, for those with a doctor diagnosis of diabetes; adjusted for age, gender and general health.
Adults in the Health Survey for England 2006
0
0.5
1
1.5
2
2.5
3
well notill well ill unwell ill unwell notill
wellness and illness category
Od
ds
Rat
io c
om
par
ed t
o 'w
ell
no
till
'
Becoming ill better
Adults reporting chronic musculo-skeletal illness (first) in HSE 2006odds of reporting illness as "limiting" - adjusted for age and self assessed pain
0
0.5
1
1.5
2
2.5
3
3.5
best second third fourth worst
quintiles of wellness
Od
ds
of
rep
ort
ing
illn
ess
as "
limit
ing
“ co
mp
are
wit
h m
ost
wel
l
Recover from, and manage, illness sooner
well notill well ill
unwell notill unwell ill
>>>>> becoming ill >>>>>
>>
bei
ng w
ell >
>Poor wellbeing and inhibitions
against becoming ‘ill’ and ‘not ill’
Dimensions of Being Well• “Being well” is not the same as simply “not becoming ill”.
We propose a common underlying dimension of wellbeing; related to dimensions of: employment status, education, health and household/neighbourhood characteristics.
• These characteristics interact with one another; overall “being well” is both an aggregate of these interactions, and a determinant in each separate dimension or ‘domain’.
• Within each dimension, being “unwell” is strongly associated with inhibition against benefiting from the social opportunities associated with that dimension; with a consequent lower degree of perceived control, and lower levels of social confidence.
– Consequently, those who are “unwell” and “notill” tend to be systematically inhibited against recognising their unwellness as relating to a long-term illness or clinical condition; and hence may be unable to access resources for managing that condition.
– But; those who are “unwell” and “ill” tend to be systematically inhibited against attaining control over the management of their condition, such as to overcome or transcend consequent limitations.
Four domains of Being Well
well workless well inwork
unwell workless unwell inwork
>>> employment status of household >>>
bein
g w
ellwell notill well ill
unwell notill unwell ill
>>>>> becoming ill >>>>>
be
ing
we
ll
well nohouse well housed
unwell nohouse unwell housed
>>>> household and neighbourhood >>>>
bein
g w
ell well noqual well qualified
unwell noqual unwell qualified
>>>> education >>>>
be
ing
we
ll
household & housing education & training
health and illness work & participation
Carousel of Being Well and domains of advantage/disadvantage
Components of ‘not good’ health
Quantified explanation of individual ratings of health as ‘not good’ for adults (16+) in the Health Survey for England, using multi-stage logistic regression:
• Individual factors = 76%– Prior morbidity and individual variation =
65%– Age (10 year intervals) and Sex = 9%
• Systematic factors = 24%– Health deficit risk factors = 8%– Cohorts of birth and residence = 8%– Health and wellbeing asset factors = 8%
A third, a third, a thirdSystematic differences in the health of populations appear to
be perpetuated through three mechanisms (which seem to have roughly equal degrees of effect; although inter-relationships make quantification uncertain)
• Differences in biomedical health risk factors: (e.g. obesity, smoking, excess alcohol, poor diet, low levels of education)– Policy response in prevention strategies– Deficit approach: ‘ how not to do the things that are bad for
you’
• Differences in cohort risk factors: (where and when born, where and how lived since)– Policy response in screening and early diagnosis
• Differences in positive wellbeing; individual, social and reciprocal: (Everyone may expect to become ill at some time; but those with high levels of wellbeing, have the capacity to recognise their illness better, access services easier, recover sooner, and manage their condition fuller.)– Policy response in promotion of ways to wellbeing, healthy
workplaces and social environments, community development– Asset approach: ‘what will enable you to do what you aspire to
do’
Assets and Deficits• Much centrally directed policy has been assessed against deficit factors:
– poverty, illness, health risks
• Wellbeing functions as an asset factor, not as a deficit factor
• Assets and deficits are not opposite extremes of one-dimensional distributions:
– some behaviours can be configured as both assets and deficits (proportion of smokers quitting, versus proportion of population smoking)
• Assets tend to relate more to differences of quality: – less doing different things, than doing the same things better
• Deficits can be assessed individually; assets need to be assessed collectively
– personal, social & reciprocal
• Asset factors are not amenable to central specification– what counts as an asset for a particular population is subject to local
determination
• Overall, asset factors and deficit factors can be quantified as having approximately equal impact on systematic inequality
The value of data reduction as applied to surveys of health and
wellbeingHealth Surveys User meeting
14th July 2010
Tom Hennell Senior Public Health Analyst
Department of Health North [email protected]
0161 952 4559
Four Population Health Surveys• Health Survey for England: 2006 and 2008 (adults and
teenagers)– 14,142 adults (16+) in 2006, 15,102 (16+) in 2008; 1,570
(13-15) in 2008– Approx 1,000 items of information recorded for each
respondent– Focus on social capital (2006), physical activity and fitness
(2008)– Structured samples of household population, weighted for
non-response– children under 16 were surveyed; collecting different
questions, and according to different protocols. 2008 boost sample ages 2-15
• North West Mental Wellbeing Survey 2009– Questions asked of 18,500 adults – Approx 230 items of information recorded for each
respondent– Focus on questions assessing mental wellbeing (WEMWBS)
and quality of life (EQ5D)– Structured samples of household population, weighted for
non-response
Data Reduction on Health Surveys
• Lengen, C; Blasius, J (2007) Constructing a Swiss health space model of self-perceived health.Social Science and Medicine, 65, 1, 80-94.
• Technique of Categorical Principal Component Analysis (CATPCA)
– Over 40 input characteristics, 2 extracted summary dimensions
– About half questions overlap in all three surveys: age, sex, ethnicity, education, marital status, economic activity, household type, alcohol use, smoking, physical activity, general health, Multiple Deprivation quintile, components of EQ5D; the overlap includes most questions with a high statistical communality (variance accounted for)
– In all three surveys, the two extracted dimensions account for slightly less than 20% of overall individual level variance
– Rotated to align with ‘ageing’ in the horizontal dimension; resulting in a counterpart ‘being well’ alignment of the vertical dimension
Select input variables• Establish dependent variable: general health = “not good”
• Basic Model: age, sex, behavioural risk factors, education
• Identify additional significant characteristics – successive logistic regressions
• Remove duplication (summary variables and components)
• Remove components of the EQ5D (so that they can be applied to adjust for degrees of mental or physical distress
• Collapse small number categories, remove variables with high proportion ‘missing’.
• Rescale into positive integer categories (values of 0 are treated as ‘missing’).
input characteristics 2008• BASIC MODEL: Age, sex, fruit & veg, binge drinking, smoking,
highest level education, body mass index
• General health; plus long-term illnesses (mental, cardiac, nervous, respiratory, cancer, gum, musculo-skeletal, digestive, endocrine, eyes), acute sickness in lasts 14 days
• GHQ components (overcoming difficulties, enjoying life, feeling worthless, being useful, able to concentrate, losing sleep)
• Physical activity levels: (sports, walking, activity at work, housework)
• Economic activity, tenure, IMD, household income, cars, household socio-economic classification, industry, rurality,
• Marital status, household type, ethnicity, household smokers, frequency of drinking
CATPCA procedure in SPSS• Declare characteristics as nominal. ordinal, or numeric
• Declare procedure for handling missing variables: (default is ‘exclude’ and for correlations ‘impute after quantification’)
• Restrict number of dimensions ( start with two)
• Save variable quantifications and summary object scores to file, output sub-category centroid co-ordinates into excel.
• Load variable quantifications into SPSS routine ‘factor analysis’ (specify for missing: ‘ replace with mean’)
• Run SPSS factor for two dimensions with varimax rotation.
• Identify the variables with high communality (those most fully conserved in the summary dimensions; and those closely aligned with varimax rotation).
• Select primary characteristic for orientation ( as principal components are fully orthogonal, both orientation and aspect are entirely discretionary)
Scree plot adults 200822% variance accounted for
Scree plot ages 13-15; 200816% variance accounted for
Health Survey for England 2008unrotated factor scores
dimension 1
dim
ensi
on
2
category centroids
age bands
Health Survey for England 2008 rotated factor scores
>>>> ageing >>>>
>>
>>
be
ing
we
ll >
>>
>
category centroids
age bands
Health Survey for England 2008: summary plot of ageing and being well
16-24
25-34
35-4445-5455-64
65-74
75+
>>> weighted ageing >>>
>>
>
we
igh
ted
be
ing
we
ll
>>
>
Health Survey for England 2008: General Health Questionnaire (GHQ12) grouped scores
ghq 1-3
ghq 4+
16-24
25-3435-4445-54
55-6465-74
75+
ghq 0
>>> weighted ageing >>>
>>
>
we
igh
ted
be
ing
we
ll
>>
>
Health Survey for England 2008: General Health Questionnaire (GHQ12) grouped scores
ghq 1-3
ghq 4+
16-24
25-3435-4445-54
55-6465-74
75+
ghq 0
>>> w e igh te d age in g >>>
>>
>
we
igh
ted
be
ing
we
ll
>>
>
Health Survey for England 2008: Components of EQ5D: none, moderate, extreme
>>> w e ig h te d age ing >>>
>>
>
we
igh
ted
be
ing
we
ll
>>
>
anxiety/depression
mobility
pain
selfcare
usual activities
NW mental wellbeing survey 2009: components of EQ5D (excluded)
>>> weighted ageing >>>
>>
>
wei
gh
ted
bei
ng
wel
l >
>>
mobility
self-care
usual activities
pain/discomfort
anxiety/depression
Health Survey for England 2006: test on components of EQ5D
>>>>> weighted ageing >>>>
>>
>>
wei
gh
ted
bei
ng
wel
l >
>>
>mobility
self care
usual activities
pain/discomfort
anxiety/depression
None ofthe time
Rarely Some of the time
Often All ofthe time
I’ve been feeling optimistic about the futureI’ve been feeling useful I’ve been feeling relaxedI’ve been dealing with problems wellI’ve been thinking clearlyI’ve been feeling close to other peopleI’ve been able to make up my own mind about things
Warwick and Edinburgh Mental Wellbeing Score (WEMWBS)
0%
2%
4%
6%
8%
10%
12%
14%
7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35WEMWBS score
Low wellbeing (16.8%)
Moderate wellbeing (62.8%)High
wellbeing(20.4%)
Mean score (27.70)
NW mean score = 27.7Scotland mean = 25.5England mean = health survey 2010
NW mental wellbeing survey 2009: components of the WEMWBS score
>>> weighted ageing >>>
>>>
wei
gh
ted
bei
ng
wel
l >>
>
optimistic
useful
relaxed
clearthinking
close to others
makemind
coping
NW mental wellbeing survey 2009: WEMWBS scores
Below average
Average
Above average
>>> weighted ageing >>>
>>
>
we
igh
ted
be
ing
we
ll
>>
>
NW mental wellbeing survey 2009: life changing events
separation/divorce
grandchild
redundant
retirement
bereavement
noneuniversity
engaged
married
housebuyingchild born
moving
repossession
change job
>>> weighted ageing >>>
>>
>
we
igh
ted
be
ing
we
ll
>>
>NW mental wellbeing survey 2009:
participation in local groups
WI
political party
trades union environmentparents/school
tenantsarts religion
older peoples group
youthw omens group
social club
sporting
none
>>> weighted ageing >>>
>>
>
we
igh
ted
be
ing
we
ll
>>
>
Health Survey for England 2008: Body Mass Index, for persons under 35
underweight
normal weight overweight
obese
>>> weighted ageing >>>
>>
>
we
igh
ted
be
ing
we
ll
>>
>
North West Mental Wellbeing Survey 2009: Warwick Edinburgh Mental Wellbeing Scale; age < 40
>>> weighted ageing >>>
>>
>
we
igh
ted
be
ing
we
ll
>>
>
Health Survey for England 2008: Recreational activity level; age < 35
inactive
lightmoderate
vigorous
>>> weighted ageing >>>
>>
>
we
igh
ted
be
ing
we
ll
>>
>Health Survey for England 2008:
Smoking and Quitting, for persons under 35
current smoker
recent quit
quit 5+ yrs
never smoker
>>> weighted ageing >>>
>>
>
we
igh
ted
be
ing
we
ll
>>
>
Health Survey for England 2008: Frequency of alcohol consumption; age < 35
stopped for health reasons
drinks dailydrinks weekly
occasional drinker
never drinker stopped drinker
>>> weighted ageing >>>
>>
>
we
igh
ted
be
ing
we
ll
>>
>
Health Survey for England 2008: Alcohol consumed on heaviest day; age < 35
did not drink
light
moderate
binge
>>> weighted ageing >>>
>>
>
we
igh
ted
be
ing
we
ll
>>
>
Health Survey for England 2008 - persons aged 13 - 15GHQ scores
0
1-3
4 plus
>>>> weighted material affluence >>>>
>>
>>
wei
gh
ted
bei
ng
wel
l >>
>>
Health Survey for England 2008 - persons aged 13 - 15Gender
male
female
>>>> weighted material affluence >>>>
>>
>>
we
igh
ted
be
ing
we
ll >
>>
>
Health Survey for England 2008 - persons aged 13 - 15Age
13
14
15
>>>> weighted material affluence >>>>
>>
>>
we
igh
ted
be
ing
we
ll >
>>
>
Health Survey for England 2008 - persons aged 13 - 15Household income quintiles
lowestsecond
third
fourthhighest
>>>> weighted material affluence >>>>
>>
>>
we
igh
ted
be
ing
we
ll >
>>
>
Health Survey for England 2008 - persons aged 13 - 15Car ownership
one
nonetwo
three plus
>>>> weighted material affluence >>>>
>>
>>
we
igh
ted
be
ing
we
ll >
>>
>
Health Survey for England 2008 - persons aged 13 - 15Strengths and Difficulties Questionnaire (components)
>>>> weighted material affluence >>>>
>>
>>
we
igh
ted
be
ing
we
ll >
>>
>
prosocial behaviour
hyperactivity
emotional
conduct
peer problems
Health Survey for England 2008 - persons aged 13 - 15Strengths and Difficulties Questionnaire (total scores)
low 0-13
medium 14-16
high 17-40
>>>> weighted material affluence >>>>
>>
>>
we
igh
ted
be
ing
we
ll >
>>
>
Health Survey for England 2008 - persons aged 13 - 15frequency of smoking
don’t smoke
< once a week
more than once a week
>>>> weighted material affluence >>>>
>>
>>
we
igh
ted
be
ing
we
ll >
>>
>
Health Survey for England 2008 - persons aged 13 - 15frequency of drinking
never drink
more than once a week
once a week
once a fortnight
once a monthfew times year
>>>> weighted material affluence >>>>
>>
>>
we
igh
ted
be
ing
we
ll >
>>
>
Health Survey for England 2008 - persons aged 13 - 15Age first smoked a cigarette
9
1113
14
15
never smoked
10
12
>>>> weighted material affluence >>>>
>>
>>
we
igh
ted
be
ing
we
ll >
>>
>
Health Survey for England 2008 - persons aged 13 - 15Age first drank alcohol
9
1314
15
never drank
10
11 12
>>>> weighted material affluence >>>>
>>
>>
we
igh
ted
be
ing
we
ll >
>>
>
Health Survey for England 2008 - persons aged 13 - 15Sedentary on a weekend day
< 2 hr
4 hr plus
2 - 4 hr
>>>> weighted material affluence >>>>
>>
>>
we
igh
ted
be
ing
we
ll >
>>
>
Health Survey for England 2008 - persons aged 13 - 15How do you spend your school breaks?
sitting down
hanging around
walking
running and playing
>>>> weighted material affluence >>>>
>>
>>
we
igh
ted
be
ing
we
ll >
>>
>
Health Survey for England 2008 - persons aged 13 - 15Ethnicity
White
Mixed
British Asian
Black British
Other
>>>> weighted material affluence >>>>
>>
>>
we
igh
ted
be
ing
we
ll >
>>
>
Health Survey for England 2008 - persons aged 13 - 15Recreational activity in last week - football
none
< 1hr1-3 hr
3-5 hr5-7 hr
7hr plus
>>>> weighted material affluence >>>>
>>
>>
we
igh
ted
be
ing
we
ll >
>>
>
Observations on the nature of ‘being well”• ‘Being well’ for adults increases with age up to mid 60s.
– Supports the theory that 'being well' functions as an acquired social capacity, rather than as an ideal state
– Different populations acquire ‘being well’ at different rates
• Indicators of positive mental health and social resilience align more closely with ‘being well’ than do indicators of physical health
– ‘being well’ has a wider field of application than does positive mental wellbeing (e.g. WEMWBS)
• It appears that, through acquiring and maintaining the capacity to manage health behaviours, some health risks, if managed and controlled can also function as health assets (e.g. alcohol, being overweight)
– In terms of 'being well'; why people act can be as significant as how they act.
• For teenagers, however, early ‘coming into age’ (early drinking and smoking, hanging about at break time) appears negative
– ‘being well’ appears to decrease in later school years; 13-15
• ‘Being well’ has a wider field of application than conventional indicators of positive mental wellbeing; and appears to function in three domains:
– Personal: individual feeling and functioning (e.g. how confident can I be in myself?)– Social: functioning of individual in their social environment (e.g how confident can I be in
my social environment? )– Reciprocal: the quality of response within a social environment to the functioning of the
individual (how confident can I expect my social environment to be in me?)
• 'Becoming ill' is a game you can play on your own; 'Being Well' is a game that can only be played in company,
Issues on ‘being well’• Can ‘being well’ be quantified?
• Can relationships of wellbeing be visualised?
• What conclusions may be suggested on the nature of ‘being well’
• How does ‘being well’ relate to ‘becoming ill’
• How much does it matter?