social inequality, psychosocial factors and chd: what have we learned

53
Social Inequality, Psychosocial Factors and CHD: What Have We Learned from the Research in Rich Countries? John Lynch Department of Epidemiology, Biostatistics and Occupational Health McGill University, Montreal, Canada PURE - Dubai, January 2006

Upload: cyma

Post on 21-Jan-2016

28 views

Category:

Documents


0 download

DESCRIPTION

Social Inequality, Psychosocial Factors and CHD: What Have We Learned from the Research in Rich Countries? John Lynch Department of Epidemiology, Biostatistics and Occupational Health McGill University, Montreal, Canada PURE - Dubai, January 2006. - PowerPoint PPT Presentation

TRANSCRIPT

Page 1: Social Inequality, Psychosocial Factors and CHD: What Have We Learned

Social Inequality, Psychosocial Factors and CHD:

What Have We Learned

from the Research in Rich Countries?

John Lynch

Department of Epidemiology, Biostatistics and Occupational Health

McGill University, Montreal, Canada

PURE - Dubai, January 2006

Page 2: Social Inequality, Psychosocial Factors and CHD: What Have We Learned

CH

D R

ate

/ 1

00K

SocialAdvantage

The Social Gradient in CHD

SocialDisadvantage

What have we learned from social epidemiology?

Page 3: Social Inequality, Psychosocial Factors and CHD: What Have We Learned

The First Whitehall Study

0

1

2

3

4

5

Administrative Professional Clerical Other

RR

Rose and Marmot. Br Heart J (1981)

2.1

3.2

4.0*Adjust for smoking, blood pressure, cholesterol, overweight etc

1.8

2.3

2.6

35%

Page 4: Social Inequality, Psychosocial Factors and CHD: What Have We Learned

“The social class difference was partly explained by known

coronary risk factors: men in the lower grades smoked more

and exercised less, they were shorter and more overweight,

and they had higher blood pressures and lower levels of

glucose tolerance. Most of the difference, however, remains

unexplained. It seems that there are major risk factors yet to

be identified, …”

Rose and Marmot (1981, p.13)

Page 5: Social Inequality, Psychosocial Factors and CHD: What Have We Learned

“… if these aspects of lifestyle account for less than a third of the social gradient in

mortality, what accounts for the other two thirds? The second question occupies the

rest of this book.” (p. 45 of 320) Marmot (2004)

Page 6: Social Inequality, Psychosocial Factors and CHD: What Have We Learned

1. There must be other mechanisms at the individual and contextual level that

generate social gradients in CHD such as stress, job control, autonomy,

social participation, neighborhood effects, social capital, etc.

i.e., a range of psychosocial factors

Implications (1)

Page 7: Social Inequality, Psychosocial Factors and CHD: What Have We Learned

Type A Behavior Pattern

Cynical Distrust

Anger/Anger-in/Anger-out/

Hostility

Social Isolation

Social Support

Control

Sense of Control

Mastery

Hopelessness

Depression

Stress

Sense of Coherence

Hardiness

Optimism

John Henryism

Job Strain

Ways of Coping

Anxiety

Self esteem

Resilience

Social cohesion

Social capital

Psychosocial factors in social epidemiology

CHD

Page 8: Social Inequality, Psychosocial Factors and CHD: What Have We Learned

2. Interventions focused on health behaviors and conventional risk factors are unlikely

to appreciably reduce the social gradient in health.

For instance, a 1998 JAMA study that has been cited more than 200 times (Web of

Science), stated that, “Thus, public health policies and interventions that exclusively

focus on individual risk behaviors have limited potential for reducing socioeconomic

disparities in mortality.” (p. 1707)

Implications (2)

Behaviours do not matter

Behaviours matter but they are just not the “big story”

The downplaying of behaviours and their physiological correlates

in understanding social inequalities in CHD

influences research, funding, interventions and policy.

Page 9: Social Inequality, Psychosocial Factors and CHD: What Have We Learned

Social Inequality

CHD

Traditional RiskFactors

P-N-E Immune Function

ControlDepression

Stress

behaviours

INTERHEART: PAR ~ 75%

Page 10: Social Inequality, Psychosocial Factors and CHD: What Have We Learned

• Conventional risk factors – smoking, hypertension, dyslipidemia and diabetes do explain

most CHD in populations

• INTERHEART study (2004) show PAR for 4 conventional risk factors is 76%

• Emberson, Whincup, et al (2005) in BRHS show smoking, blood pressure and cholesterol

account for 81% PAR (adjusted for regression dilution bias)

• Stamler (1992,1999) and Greenland (2003) show that 75-100% of all CHD cases occur

among those exposed to at least one conventional risk factor

What have we learned from CHD epidemiology?

Page 11: Social Inequality, Psychosocial Factors and CHD: What Have We Learned

So how can it be that the factors accounting for most CHD in a population

do not seem to account for most of the social gradient in CHD, when the

social gradient simply emerges from sub-grouping the population

according to some indicator of social position?

A Paradox?

Page 12: Social Inequality, Psychosocial Factors and CHD: What Have We Learned

1000 cases of CHD

Least educated

600

Middle educated

300

Most educated

100

0

100

200

300

400

500

600

700

LeastEducated

MostEducated

Lipids, hypertension, smoking, diabetes

A Paradox?

Page 13: Social Inequality, Psychosocial Factors and CHD: What Have We Learned

• Population sample of 2,682 Finnish men – Kuopio IHD Study

• Stratified the population into lower and higher risk groups based on standard clinical

definitions - current smoking, hypertension, dyslipidemia and prevalent diabetes

• 34.7% were current smokers, 58.7% had hypertension, 42.7% had dyslipidemia, 6.5%

had diabetes, and 84.9% had at least one of these risk factors

• Fatal (ICD9 codes 410-414) and nonfatal CHD, classified according to MONICA criteriia

• There were 425 CHD events (108 fatal, 317 non-fatal) during an average follow-up of

10.5 years.

An Illustration

Lynch, Davey Smith, Harper, Bainbridge. J Epidemiol Community Health (in press)

Page 14: Social Inequality, Psychosocial Factors and CHD: What Have We Learned

• Of the 425 total cases - 402 (94.6%) occurred among men exposed to at least one of

the four conventional risk factors

• 70%+ of cases occurred in men with at least 2 risk factors

• PAR ~ 70%

• Similar to INTERHEART (PAR ~ 75%) and with US cohorts where 90%+ of cases occur

among those exposed to at least one conventional risk factor

Thus these results are consistent with the idea that

conventional risk factors DO EXPLAIN the vast majority of CHD cases

Accounting for Cases of CHD in this Population

Page 15: Social Inequality, Psychosocial Factors and CHD: What Have We Learned

• Social gradient in CHD comparing high to low education is RR = 1.90

• After adjustment for conventional risk factors RR = 1.68

• This corresponds to a 24% reduction in the excess RR

Thus these results are consistent with the idea that

conventional risk factors DO NOT EXPLAIN the social gradient in CHD

Accounting for the Social Gradient in CHD in this Population

Page 16: Social Inequality, Psychosocial Factors and CHD: What Have We Learned

What would population levels and social gradients in CHD look like

if there were no conventional risk factors in this population?

Page 17: Social Inequality, Psychosocial Factors and CHD: What Have We Learned

N(%)

No. of Cases (%)

Risk(per K)

RR Risk

ER (per K)

Education

< Primary school

1,121(41.8)

218 (51.3)

194 1.83 88

Some high school

1,128(42.1)

161(37.9)

143 1.35 37

High school

graduate

433(16.1)

46(10.8)

106 1.0 0

Total 2,682 425 158 - -

N(%)

No. of Cases (%)

Risk(per K)

RR Risk

ER (per K)

Education

< Primary school

122 (30.2)

8 (34.8)

66 1.63 25

Some high school

184 (45.5)

11(47.8)

60 1.46 19

High school

graduate

98 (24.3)

4(17.4)

41 1.0 0

Total 404 23 57 - -

Whole KIHD PopulationN = 2,682

Low Risk Segment of KIHD PopulationN = 404 (15.1%)

Page 18: Social Inequality, Psychosocial Factors and CHD: What Have We Learned

0

50

100

150

200

250

Least Educated Most Educated

CH

D R

isk

/ 10

00

66 6041

Excess Risk (whole pop) = 88 per 1000Excess Risk (“low risk” pop) = 25 per 1000

Make it a low risk pop = 72% reduction in the excess risk of social inequality in CHD194

143

106

Low risk population

Low risk population

Low risk population

RR = 1.8

RR = 1.6

Page 19: Social Inequality, Psychosocial Factors and CHD: What Have We Learned

In a low risk population with no smoking, hypertension, dyslipidemia and

diabetes

• Relative social gradient in CHD remains

RR = 1.8whole vs 1.6low risk

• Absolute social gradient is reduced by 72%

ER = 88whole vs 25low risk per 1,000

Page 20: Social Inequality, Psychosocial Factors and CHD: What Have We Learned

CH

D R

ate

/ 1

00K

Social Disadvantage

Original social inequality in CHD

Give the disadvantaged the same risk factor levels as the most advantaged

If we intervened to substantially reduce risk factors in all social groups

Which social inequality is “better”?

Social Advantage

Page 21: Social Inequality, Psychosocial Factors and CHD: What Have We Learned

• Within an absolute risk framework, there is no paradox between the observations from

CHD and social epidemiology. Conventional risk factors do account for most CHD and for

most of the absolute social gradient in CHD (72% of the excess risk).

• However, when explaining relative social gradients in CHD, the apparent paradox may

arise that the factors which explain most cases of CHD do not explain the relative social

gradient.

• Adjustment for conventional risk factors only reduced the relative social gradient in CHD by

24%. This is normally interpreted to mean that most of the effects of social inequality on

CHD do not work through mechanisms linked to conventional risk factors.

Page 22: Social Inequality, Psychosocial Factors and CHD: What Have We Learned

• The extent to which a 3rd variable reduces the RR in an exposure-outcome association

(an indication of confounding of the association) depends on the relative distributions of

the exposure over strata of the 3rd variable and the strength of its association with the

outcome.

• In this case there is a more extreme relative distribution of education over strata of

hopelessness than over strata of the conventional risk factors (partly because the

prevalence of conventional risk factors is high), and so hopelessness/depression appears

to be a stronger confounder (in this case interpreted as a mechanism) of the association

between education and CHD.

What does it take to be a confounder?

Page 23: Social Inequality, Psychosocial Factors and CHD: What Have We Learned

Education

CHD

Depression

Conventional Risk Factors

28%

24%

SocialGradient RR=1.9

on the basis of the adjusted RR analysis we would normally interpret hopelessness/depression

as an important mechanism in the causation of the social gradient in CHD

Page 24: Social Inequality, Psychosocial Factors and CHD: What Have We Learned

• intervene to eliminate hopelessness / depression

reduces the relative social gradient 28%

eliminates 14% of CHD cases (confounded estimate)

• intervene to eliminate smoking, hypertension, dyslipidemia and diabetes

smaller reduction in the relative social gradient – 24%

eliminate 90%+ of CHD cases and 72% of the absolute social gradient

How important are these 2 mechanisms?

Page 25: Social Inequality, Psychosocial Factors and CHD: What Have We Learned

• 72% reduction in ER is an underestimate because “low risk” population

is not really low risk

• No differences in blood pressure and have higher BMI

• but low risk population has ex-smokers, higher cholesterol and LDL,

and are 5 cm shorter

Page 26: Social Inequality, Psychosocial Factors and CHD: What Have We Learned

• Deciding on the “importance” of various risk factors for elucidating the mechanisms

behind social gradients in CHD cannot be done on the basis of relative comparisons

alone.

• As Geoffrey Rose commented:

“Relative risk is not what decision-taking requires … relative risk is only for researchers;

decisions call for absolute measures.”

p. 19, Strategy of

Preventive Medicine.

Page 27: Social Inequality, Psychosocial Factors and CHD: What Have We Learned

0

50

100

150

200

250

300

350

400

1900 1910 1920 1930 1940 1950 1960 1970 1980 1990

Infant M

ortalit

y per

1,000

0

0.5

1

1.5

2

2.5

3

Relati

ve Dis

parity

Black – White Inequality in Infant Mortality

over the 20th Century, USA

Relative Inequality

White

Black

0

50

100

150

200

250

300

350

400

1900 1910 1920 1930 1940 1950 1960 1970 1980 1990

Infant M

ortality

per 1,0

00

0.0

20.0

40.0

60.0

80.0

100.0

120.0

140.0

160.0

180.0

200.0

Absolut

e Dispa

rity

Absolute Inequality

White

Black

Page 28: Social Inequality, Psychosocial Factors and CHD: What Have We Learned

0

20

40

60

80

100

1

3

5

7

9

RR

Age-specific Mortality Differences between the

Richest and Poorest 20% of the World’s Population

Gwatkin (2000)

Poorest 20%

Richest 20%

Mo

rtal

ity

Rat

e p

er 1

000

RR

Page 29: Social Inequality, Psychosocial Factors and CHD: What Have We Learned

• Explaining and reducing relative social gradients in CHD is a legitimate and important

focus of research and intervention because it can help reveal novel CHD risk factors and

mechanisms that are unevenly distributed across social groups.

• However, explanations for relative social gradients need to be understood within the

context of what causes most cases of CHD. Otherwise they may deflect attention from

the most important population-level causes.

• Rose – if everyone in the population smoked we would be finding that radon or

asbestos were the most important causes of lung cancer

Page 30: Social Inequality, Psychosocial Factors and CHD: What Have We Learned

• In populations where the prevalence of conventional risk factors is high, it is possible that

there are no or small social gradients in those risk factors such that they cannot account

for relative CHD differences across social groups but contribute substantially to the

absolute risk of CHD in all social groups.

• In the Whitehall Study there are small social differences in blood pressure and no social

differences in total cholesterol

• So reducing major, high prevalence risk factors from the population may have little effect

on the relative social gradient but a large effect on the absolute social gradient as

indicated by a reduction in the excess risk.

Page 31: Social Inequality, Psychosocial Factors and CHD: What Have We Learned

• Importantly, this also means that whatever proximal (CRP) and distal factors (social

capital) are identified as causes of the relative social gradient in CHD, if their behavioral

and biological mechanisms do not involve conventional risk factors then they probably

account for a small proportion of CHD cases.

Page 32: Social Inequality, Psychosocial Factors and CHD: What Have We Learned

Social Inequality

CHD

Traditional RiskFactors

P-N-E Immune Function

ControlDepression

Stressbehaviours

INTERHEART: PAR ~ 80%

?

Page 33: Social Inequality, Psychosocial Factors and CHD: What Have We Learned

200

300

400

500

600

700

1900 1910 1920 1930 1940 1950 1960 1970 1980 1990 2000

rate

pe

r 1

00

,00

0

10

20

30

40

50

60

70

80

gin

i co

eff

icie

nt/

po

vert

y

heart disease

income inequality

Income Inequality, Poverty and Heart Disease: 1900-2000, USA

poverty

Page 34: Social Inequality, Psychosocial Factors and CHD: What Have We Learned

700.00

800.00

900.00

1,000.00

1,100.00

1,200.00

1,300.00

1,400.00

1,500.00

1,600.00

1,700.00

1968 1970 1972 1974 1976 1978 1980 1982 1984 1986 1988 1990 1992 1994 1996 1998

40.0

45.0

50.0

55.0

60.0

65.0

70.0

Race-specific Voting Participation in Presidential Elections and age-adjusted, all-cause mortality, USA, 1968 - 1998

Year

All-

cau

se m

orta

lity

per

100

,000

% V

oting in Pre

sidential elections

Black Mortality

White Mortality

Black Voting

White Voting

Lynch and Davey Smith (2003)

Page 35: Social Inequality, Psychosocial Factors and CHD: What Have We Learned

0

100

200

300

400

500

600

700

800

1900 1910 1920 1930 1940 1950 1960 1970 1980 1990 2000

0

500

1000

1500

2000

2500

3000

AA

DR

Per C

apita Consum

ptionSex-Specific Smoking Trends and Heart Disease, 1900-1998, USA

Male Heart Disease

Female Heart DiseaseMale Smoking

Female Smoking

Page 36: Social Inequality, Psychosocial Factors and CHD: What Have We Learned

0

10

20

30

40

50

60

70

80

90

100

1900 1910 1920 1930 1940 1950 1960 1970 1980 1990 2000

0

500

1000

1500

2000

2500

3000

Gin

iP

er Capita C

onsumption

Sex-Specific Smoking Trends and Income Inequality, 1900-1998, USA

Male Smoking

Female Smoking

Income Inequality

Page 37: Social Inequality, Psychosocial Factors and CHD: What Have We Learned

37

38

39

40

41

42

43

44

45

46

47

1980 1985 1990 1995 2000

4.9

5

5.1

5.2

5.3

5.4

5.5

5.6

Sex-Specific Cholesterol Trends and Income Inequality, 1980-2002, USA

Total Cholesterol Trends from Minnesota Heart Survey: Arnett et al. Circulation (2005)

Gini

Male Cholesterol

Female Cholesterol

Page 38: Social Inequality, Psychosocial Factors and CHD: What Have We Learned

• We should re-affirm that smoking, hypertension, dyslipidemia and diabetes are the most

important causes of CHD in populations and of social gradients in CHD.

• If our concern is to reduce the overall population health burden of CHD and the

population health burden of CHD inequalities, then reducing conventional risk factors will

do the job.

• Eliminating the social gradient in CHD attributable to non-traditional risk factor

mechanisms will make a modest contribution to improving overall population health

because the magnitude of the between social group CHD differences is small relative to

the total risk variation in the population and to that due to traditional risk factors.

Conclusion

Page 39: Social Inequality, Psychosocial Factors and CHD: What Have We Learned

We should increase our efforts to find ways to influence the multiple pathways

from international, national and local policy through to individual behaviour that will

reduce conventional risk factors among current and future generations in richer

and poorer countries.

Conclusion

Page 40: Social Inequality, Psychosocial Factors and CHD: What Have We Learned

Individual and Population Level Causation

Issues for PURE

Page 41: Social Inequality, Psychosocial Factors and CHD: What Have We Learned

0

100

200

300

400

500

600

700

“Heart Disease”

All Stroke

RiskFactors

SEP

INTERHEART

Psychosocial – CHD

PAR ~ 30%Psychosocial

CHD

Stress

SEP

?

CHD

Black Africans Coloured White

SEP Steyn, et al Circulation (2005)

Page 42: Social Inequality, Psychosocial Factors and CHD: What Have We Learned

Social Inequality

Major RiskFactors

PsychosocialFactors

CHD

PURE

Social, economic, demographic, behavioural transition

ReportingTendency

Page 43: Social Inequality, Psychosocial Factors and CHD: What Have We Learned

Thank you

Page 44: Social Inequality, Psychosocial Factors and CHD: What Have We Learned

Triangulating Epidemiology

Populations

Sub-groups

Individuals

Biology

Genetics

Disease

Disease Trends

Populations

Sub-groups

Individuals

Biology

Genetics

Page 45: Social Inequality, Psychosocial Factors and CHD: What Have We Learned

CH

D R

ate

/ 1

00K

1) Original population levels and social inequality in CHD

2) Social inequality and levels of CHD after reduction in social inequality and/or psychosocial causes?

3) Social inequality and levels of CHD after population-wide reductions in risk factors?

Which social inequality is “better”?

SocialAdvantage

SocialDisadvantage

Page 46: Social Inequality, Psychosocial Factors and CHD: What Have We Learned

• in terms of CHD and probably many other diseases as well, we should focus

on the developmental influences (political, economic, social, psychological,

biological, genetic) on population levels and social inequalities in the major

health behaviors in rich and poor countries

Implications for Research on Human Growth and Development

Page 47: Social Inequality, Psychosocial Factors and CHD: What Have We Learned

Important to reiterate Geoffrey Rose

• The causes of individual cases of CHD.

The positive predictive value for individual risk prediction is very low because lots of

people with conventional risk factors don’t get CHD

– what makes people susceptible to the risk factors?

Vs

Page 48: Social Inequality, Psychosocial Factors and CHD: What Have We Learned

Current or Former Smoker

CHD (fatal or non-fatal) Lung Cancer Death

172 / 1000 15 / 1000

Never Smoker 12 / 1000 0

Risk of CHD and Lung Cancer by Smoking Status

Page 49: Social Inequality, Psychosocial Factors and CHD: What Have We Learned

Vs

• The causes of population levels of CHD

Almost all cases of CHD do have the conventional risk factors and so population levels almost completely accounted for 3-4 conventional risk factors

Thus eliminating the risk factors eliminates the disease

Page 50: Social Inequality, Psychosocial Factors and CHD: What Have We Learned

“If causes can be removed, then susceptibility ceases to matter”

Geoffrey Rose

Page 51: Social Inequality, Psychosocial Factors and CHD: What Have We Learned

Extensions

1. Understand why a relative social gradient (RR=1.6) remains even in a low risk population

2. Given the differential distribution of exposure to at least 1 risk factor across educational

groups, is the exposure distribution enough to account for the greater disease burden

generated among the low educated?

Greater exposure or greater susceptibility?

% exposed to 1+ CHD Risk per 1,000

High ed. 77 65

Mid ed. 84 83

Low ed. 89 128

15% 97%

Page 52: Social Inequality, Psychosocial Factors and CHD: What Have We Learned

(1997)

Page 53: Social Inequality, Psychosocial Factors and CHD: What Have We Learned