Which influences the selfreporting of health: country of birth or country of residence

1
Which influences the self-reporting of health
country of birth or country of residence? A
British analysis using individual-level data
Paul Norman1, Paul Boyle2 Mark Brown1 1
Centre for Census Survey Research, University
of Manchester 2 School of Geography and
Geosciences, University of St Andrews Thanks
to ESRC RES-163-25-0012 Understanding Population
Trends Processes UK Census data via National
Statistics Offices, ESRC, JISC CCSR
2

• Background
• 1991 2001 UK Censuses have self-reported
  health limiting long-term illness (LLTI)
• Differences in LLTI found between England, Wales
  Scotland in 1991, difference changed by 2001
• Self-assessed health (e.g. LLTI) strongly
  correlates with
• Mortality all cause cause specific
• Morbidity general health, serious conditions,
  physical conditions rather than psychological
• Sickness benefit claims
• True for different countries, populations
  question formulations

3

• People from different cultural backgrounds
• May report their health differently to other
  population sub-groups from each other in
  different places at different times
• There is potential for self-assessment to be
  affected by subjective factors
• Differences in health between areas and
  population sub-groups perhaps due to varying
  interpretations of health
• (Bailis et al. 2003)
• In terms of LLTI, what constitutes a limiting
  and/or a long-term illness
• (Gooberman-Hill et al. 2003 Senior, 1998)
• Self-reported poor health is influenced by
  expectation and comparison. These may be
  culturally determined
• (Mitchell 2005)

4

• UK Census LLTI research
• Aggregate level
• Various geographical scales
• (Boyle et al. 1999 Rees 1993/4 Senior 1998)
• Individual level microdata
• Sample of Anonymised Records (SARs)
• (Boyle et al. 2001 Boyle et al. 2002 Borooah
  1999 Gould Jones 1996)
• ONS Longitudinal Study
• (Bartley Plewis 1997 Boyle et al. 2004
  Harding 2002 Norman et al. 2005 Sloggett
  Joshi 1998 Wiggins et al. 2002)
• Used as input to local health profiles

5

• Where we live affects our health
• (Macintyre et al. 2002 Tunstall et al. 2004)
• Environment most influential on health may not
  be the conditions contemporary with disease
  diagnosis or death
• (Boyle et al. 2002)
• Place of birth childhood conditions affect
  later health
• (Curtis et al. 2004 Osmond et al. 1990 Kaplan
  Salonen, 1991)
• Place of residence in middle age influential
• (Blane et al. 1996 Elford Ben-Shlomo 1997)
• Relative importance of influences at different
  life stages found to vary by cause of death
• (Davey Smith et al. 1997)
• Differences because lag periods between contact
  with potential risk factors subsequent health
  deterioration may vary by condition
• (Polissar 1980 Rogerson Han 2002)

6

• Migration process important
• As we move between residential areas we
  experience different conditions and meet a range
  of people against whom we make comparisons when
  assessing our health
• Health differences between migrants and between
  geographical areas attributed to lifecourse
  disadvantage
• (Brimblecombe et al. 2000)
• Healthy people migrate into less deprived
  locations
• (Norman et al. 2005)
• Ill people may drift towards or becoming stuck
  in deprived places (Boyle et al. 2002)
• International migrants tend to be relatively
  healthy
• (Bentham 1988)
• For non-migrants changing area characteristics
  affect health (Boyle et al. 2004)

7

• Health differences between England, Scotland
  Wales 1991 2001
• Can we compare self-reported measures of illness
  between countries and culture over time?
• (Mitchell 2005)
• Is there something about being born in or living
  in England, Scotland or Wales which affects your
  health?
• For example
• For the population resident in Wales, do those
  people born in Wales report different health to
  those born elsewhere?
• Do those who are Welsh born report their health
  differently when resident in another country?

8

• Data
• UK Census Sample of Anonymised Records (SARs) for
  1991 2001
• Study population household residents in England,
  Scotland Wales, all resident for at least one
  year
• 1991 SAR (2) 1,116,181 individuals
• 2001 SAR (3) 1,712,474 individuals
• Use of SARs enables variable definitions
  modelling not possible with aggregate data
• Method
• Binary logistic regression
• Dependent variable reporting of LLTI
• Models consistent for both 1991 2001

9
Method Explanatory variables Age-group, Sex,
Educational qualification, Ethnicity, Social
Class, Access to car, Unemployment,
Tenure Country of birth (CoB), Country of
residence (CoR) (gt 1 year in the
country) Country of birth by country of
residence All variable definitions
consistently defined 1991 2001
10
1. Country of residence 1991 2001
11
2. Country of birth 1991 2001
12
3. Country of residence by country of
birth 1991 2001
13
4. Country of birth country of residence
1991 2001
14

• Conclusions
• Odds of LLTI consistent with area models for 1991
• (Rees 1993/4 Senior 1998)
• Change by 2001 also consistent with area research
• (Mitchell 2005)
• International migrants are relatively healthy
• (Bentham 1988)
• Is there a Welsh-born effect?
• Yes, those born in Wales tend to have higher
  LLTI odds wherever they live
• Is there a Wales effect?
• Yes, generally, whatever their country of birth,
  residents in Wales tend to have higher LLTI odds

15

• Conclusions
• Is there a Scotland effect for the 1991-2001
  change?
• Yes. All countries of birth have higher odds in
  2001 than in 1991 for residents in Scotland
• Is this due to a change in age structure?
• No. Scotlands population is ageing similarly to
  England and Wales
• Or a loss of healthy Rest of the World-born and
  gain of less healthy Welsh-born?
• No. The population structure by country of birth
  remains constant
• So are people in Scotland more willing to report
  LLTI or have environmental socio-economic
  conditions become worse?
• We cannot tell from the SARs

16
Conclusions How does the influence of Country of
Birth compare with the influence of Country of
Residence? Country of birth has a significant
effect of health, the relative difference between
countries is consistent in 1991 2001 Country of
residence has an additional and potentially
variable effect on LLTI odds Findings consistent
with early life influences persisting and with
people experiencing different effects as their
location personal circumstances change How
else can we compare self-reported measures of
illness between British countries and culture
over time? Transitions using the longitudinal data