nfhk2011 mika kivimaki_turku2011
DESCRIPTION
K2-1 Bidrar dagens arbetsliv till en ökning eller minskning av de socioekonomiska skillnaderna i hälsan / Does current working life increase or decrease socioeconomic inequalities in healthProf. Mika KivimäkiUniversity College London, Helsingfors universitet, ArbetshälsoinstitutetUniversity College London; University of Helsinki; Finnish Institute of Occupational HealthTRANSCRIPT
Work life and social inequalities in
healthhealth
Professor Mika Kivimaki
Department of Epidemiology & Public Health University
College London, UK
Finnish Institute of Occupational Health, Finland
Collaborators:
Prof. Jussi Vahtera, Drs. Marianna Virtanen, Tuula Oksanen,
Paula Salo, and Jaana Halonen, FIOH;
Prof. Sir Michael Marmot, Drs. Archana Singh-Manoux, G. David
Batty, Martin Shipley, Jane E. Ferrie, Eric Brunner, and Mark
Hamer, University College London, UK
Funding:
Academy of Finland, Finnish Work Environment Foundation, EU
New OSH ERA research programme, BUPA Foundation, British
Heart Foundation, Medical Research Council, UK, NIH, US.
Session Outline
� The Social causation assumption (SES health) – is it justifiable?
� Understanding how work is linked with disease � Understanding how work is linked with disease risk?
� Work as an explanation for social inequalities –history and current evidence?
Relative inequalities in the rate of death from any cause
A real public health problem
Mackenbach et al. N Engl J Med 2008
1. The social causation hypothesis
• SES determines the ability to consume goods and services – for
example, high-quality food and health care – which in turn affects health.
• Low SES is associated with higher exposure to occupational health
hazards, potentially contributing to health problems.
• Differences in social values and behavioural preferences between SES
groups may create variations in health.
2. The health-related selection hypothesis
SES Health
SESHealth2. The health-related selection hypothesis
• Childhood health is linked to educational achievement and labour
market prospects and thus to adult SES.
• Severe and limiting health problems during adulthood may increase the
risk of an income shortfall and poor career prospects.
3. The common cause hypothesis
• Common causes, such as genetic influences and personality, determine
both SES and health.
SESHealth
SES Health
Trait
Common cause: genes
Denmark, >20,000 adoptees
Hazard ratio for mortality in adoptees in relation to biological and adoptive father’s social class.
SES Health
Trait
Osler et al. Int J Epidemiol 2006
Note: not replicated; specific SES-related genetic variants not identified.
Common cause: personality
GAZEL cohort, France
SES Health
Trait
BDHI, Buss-Durkee Hostility Inventory
Nabi et al. Int J Epidemiol 20080 . 0
0 . 5
1 . 0
1 . 5
2 . 0
Unadjusted Adjusted
High
Low
29%
RR
Personality in adolescence predicts education in adulthood
Young Finns, Finland
SES Health
Trait
Pulkki et al. Int J Epidemiol 2003
Health predicting social mobility:
The Whitehall II study, the UK
SESHealth
Elovainio et al. Am J Epidemiol 2011
Socioeconomic circumstances influence health
SES Health
Elovainio et al. Am J Epidemiol 2011
The social causation
• Important at least in adulthood
The health-related selection
SES Health
SESHealth
Brief summary
• Important in childhood
The common cause
• Eg. effects of personality are not trivial
SES Health
Trait
Theoretical models on unhealthy work
� To identify key elements within complex and
diverse work environmentsdiverse work environments
� To provide new predictions and explanations of
less well understood health/disease outcomes
� To orient interventions towards healthy work and
well-being of workers
F Kittel 2010
Karasek 1979
Karasek & Theorell 1990
Job Strain Model
Job motivation and job stress
models
rewardWhat I put in my work What I get from my work
J. S. Adams: Equity Theory
on job motivation 1963
effort
J. Siegrist: Effort-Reward
Imbalance model 1996
What I put in my work What I get from my work
Organizational justice theory― 3 forms of justice perceptions
� Distributive justice: fairness of outcomes (equity, equality, and
needs)
� Procedural justice: fairness of the methods or procedures used � Procedural justice: fairness of the methods or procedures used
(decision criteria, voice, control of the process)
� Relational justice: fairness of the interpersonal treatment
received (dignity and respect)
Moorman 2001, Greenberg & Cropanzano 2001, Kivimaki et al Arch Intern Med 2005
Organisational justice questionnaire items
Decisions…• are well-informed,
• are consistently applied (the rules are applied equally for
everyone).everyone).
Management… • listens to the concerns of all those affected by the decision,
• provides opportunities to appeal against or challenge the
decision,
• tries to deal with us in a truthful manner.Kivimäki et al Psychol Med 2003
A quick look at evidence
1. Depression1. Depression
2. CHD
Relative risk of depression or depressive
symptoms according to job strain
Job demands
1.
Bonde Occup Environ Med 2008
Low job control
Relative risk of depression or depressive
symptoms according to job strain
Job strain
Bonde Occup Environ Med 2008
Low social
support
Summary estimates (relative risk) for job
strain components:
1.31 (95% CI 1.08 – 1.59) for high demands
Bonde Occup Environ Med 2008
1.20 (95% CI 1.08 – 1.39) for low job control
1.44 (95% CI 1.24 – 1.68) for low social support
Other work stressors
Bonde Occup Environ Med 2008
Summary
• Both job strain and effort reward imbalance
show associations with mental health
problems, but not unanimously
• Aspects of social relations at work also related
to mental health problems
• Threat to causal inference: Reverse causation
An attempt to exclude the reverse causation
explanation...Ward overcrowding - a person-independent source of work
stress for nurses
Participating hospitals routinely collect monthly figures on
bed occupancy for each ward according to a standard
procedure.procedure.
We examined a subcohort of somatic ward personnel
(n=7340) from the Public Sector Study.
Virtanen et al. Am J Psychiatry 2008
Overcrowding as a time-dependent
exposure (illustration)
00
0
1
0
0
0
Antidepressant treatment
0 = no
1 = yes
0
Person A
Person B
Virtanen et al. Am J Psychiatry 2008
00
0
1
0
0
0
00
0
01
6 months 8 months 10 months
Person C
Virtanen et al. Am J Psychiatry 2008
PART 1:CAUSAL MODELS
Risk factors Preclinical disease CardiovascularManifest disease
Psychosocialfactors
indirect effect prognostic factoretiological factor trigger
Risk factors Preclinical disease Manifest disease Cardiovascular
Psychosocialfactors
2. Work stress and cardiovascular disease
Risk factorsx
Preclinical diseaseprocesses
Cardiovasculardeath
Manifest diseasex
e.g., obesity, smoking,
physical inactivity,
high LDL cholesterol
e.g., atherosclerosis,
endothelic dysfunction
e.g., angina, myocardial
infarction
PART 2:ALTERNATIVEEXPLANATIONS Psychosocial
factors stress
confounding, bias, reversed causality
Risk factorsx
Preclinical diseaseprocesses
Manifest diseasex
Cardiovasculardeath
Psychosocial
factors
Kivimäki et al. Scand J Work Environ Health 2006
Underlying mechanisms
16%
16%
32%
Physical inactivity, poor diet and the metabolic syndrome the most important
explanatory factors in this cohortChandola et al. Eur Heart J 2008
Meta-analysis:
Job strain and
CVD 4/9
Kivimäki et al. Scand J
Work Environ Health
2006Decreases risk Increases risk
2/10
3-year risk of cerebrovascular disease among 48,361 women aged 18–65
years (the Finnish Public Sector Study)
Kivimäki et al. Int J Epidemiol. 2009Kivimäki et al. Int J Epidemiol. 2009
10-year risk of cerebrovascular disease among 49 259 women aged
30 to 50 years (The Women’s Lifestyle and Health Cohort Study, Sweden)
Kuper et al. Stroke. 2007
In sum, reasonable evidence to assume
social causation and to link work and
disease risk.
But are work characteristics linked with
social inequalities in health?
Marmot et al. Lancet 1997
”CONCLUSION: Much of the inverse social gradient in incident CHD can be attributed to
differences in psychosocial work environment…” P. 235.
Marmot et al. Lancet 1997
Thompson ISI web of science: 519 citations in 17/08/2011
“…psychological distress explained only 2% of the association between SES and all-cause
mortality when assessed at baseline (hazard ratio for mortality changed from 1.60; 95% CI
1.26-2.04, to 1.58; 95% CI, 1.24-2.02) and 5% when assessed longitudinally (adjusted hazard
ratio, 1.56; 95% CI, 1.23-1.99).”
Stringhini et al. JAMA 2010
Marmot et al. Diabetologia 2008
A contemporary cohort of 48,000 employed women, 3.5-y follow-up
The Finnish Public Sector Study
Kivimaki et al. Int J Epidemiol 2009
Relative risk (95% CI) for SES and sickness absence
MEN WOMEN
Adjusted for age and family status
1.37 (1.21 to 1.55) 1.30 (1.14 to 1.47)
The Danish Work Environment Cohort Study
(DWECS)
1.37 (1.21 to 1.55) 1.30 (1.14 to 1.47)
+ health behaviours
1.33 (1.17 to 1.51) (3%) 1.24 (1.09 to 1.41) (5%)
+ physical work environment
1.10 (0.95 to 1.28) (20%) 1.14 (0.99 to 1.32) (12%)
+ psychosocial work environment
1.09 (0.93 to 1.29) (20%) 1.09 (0.93 to 1.28) (16%)
Christensen et al. J Epidemiol Community Health 2008