Social Conditions as Fundamental Causes of Health Disparities

Bruce LinkGlasgow Scotland

December 11, 2007

New York City All Cause Age-adjusted Death Rates In Lowest Versus Highest Income Neighborhoods 2001

New York City All Cause Age-adjusted Death Rates Per 100,000

by Race – 2004

New York City Percent Fair or Poor Self-Reported Health by Income Level (% Poverty Level) and

Race/Ethnicity

New York Centric View

All Cause Age-adjusted Death Rates Per 100,000 People Ages 25-64 by

Education -- 2004839

619

251

486

345

167

0100200300400500600700800900

Males Females

< 12 years12 years13+ years

US All Cause Age-adjusted Death Rates Per 100,000 by Race – 2002

993

1341

701

902

0

200

400

600

800

1000

1200

1400

males females

whiteblack

US Percent Fair or Poor Self-Reported Health by Poverty Level and Race/Ethnicity (age adjusted)

21

14

6

21

14

6

26

17

10

20

15

9

0

5

10

15

20

25

30

Overall White Black Hispanic

Below 100%100%-less than 200%200% or more

Age Adjusted Death Rates 100,000 (Men) 25-64 England and Wales

Scottish National Survey 2003 -- Percent Reporting Fair/Bad/Very Bad Health (Age Adjusted)

SES,Race

MechanismsSmoking, Diet, Exercise, Stress,

Etc, etc. ?

MortalityMorbidity

?

What Are the Mechanisms that Accountfor the Association?

Deaths Per 1000 Among Taxpayers andNon-Taxpayers in Rhode Island 1865

Age Categories Taxpayers Non-Taxpayers(examples)

Under 1 93.4 189.8

30-39 4.5 15.5

60-69 15.1 39.5

Chapin AJPH 1924

Deaths per 1000 (age adjusted) by SESof Census Tract -- Chicago 1930

SES Males Females

1--Lowest 15.1 12.3

2 11.6 10.2

3 10.2 9.0

4 9.2 7.9

5--Highest 8.7 6.8

Coombs, Medical Care, 1941

What is the point?

• Imagine yourself back in Rhode island in 1865 and doing what we just did for the current data –we might have asked what were the mechanisms involved?

• Contaminated water, poor sanitation, crowded substandard housing – the diseases were cholera, TB, small pox…

• We did something about the risk factors, we developed vaccines, and people don’t die of TB, small pox and cholera in Rhode Island any more.

• But the SES association is resilient.

The Concept of Fundamental Social Causes

Fundamental social causes involve resources such asknowledge, money, power, prestige and beneficial social connections that determine the extent to which people areable to avoid risks and adopt protective strategies so as toreduce morbidity and mortality.

Because such resources can be used in different ways in different situations, fundamental causes have effects ondisease even when the profile of risk and protective factors and diseases changes radically.

It is their persistent effect on health in the face ofdramatic changes in mechanisms that leads us to call them“fundamental.”

How Social and Economic Resources Affect Health – The Importance of Contexts

• Resources operate at the individual level – people use their knowledge, money, power, prestige and beneficial social connections to obtain healthy outcomes.

• But resources also provide access to generally salutary contexts – neighborhoods, occupational conditions, marriages – access to health consequential circumstances comes with access to contexts in a sort of “package deal.”

US Life Expectancy at Birth 1900-2000

47

55

61

7074

77

40

45

50

55

60

65

70

75

80

1900 1920 1940 1960 1980 2000

Scottish Life Expectancy at Birth 1910-2000

50.1

56

64.467.3

72.1

53.2

59.5

68.7

73.777.6

40

45

50

55

60

65

70

75

80

1910 1930 1950 1970 1995

Women

Men

US: Heart Disease -- Age-adjusted Death Rates Per 100,000 People

587559

493

412

321 293258

40

140

240

340

440

540

640

1950 1960 1970 1980 1990 1995 2000

Cancer (green) and Stroke (Yellow) -- Age-adjusted Death Rates Per 100,000 People

181 178

148

96

65 63 6150

194 194 199208

216 210200

186

406080

100120140160180200220240

1950 1960 1970 1980 1990 1995 2000 2004

National Center for Health Statistics – Health United States 2006

US : Flu (blue) and HIV (green) -- Age-adjusted Death Rates Per 100,000 People

1016

5

4854

42

3137

33

24

0

10

20

30

40

50

60

1950 1960 1970 1980 1990 1995 2000

Scotland Ischamemic Heart Disease -- Age-adjusted Death Rates Per

100,000 People 15-74

350395 400 390

300

185195160 150 150

125

6540

90

140

190

240

290

340

390

440

1950 1960 1970 1980 1990 2000

Men

Women

Leon et al: Understanding the Health of Scotland in and International Context 2003

Scotland Lung Cancer -- Women (green) and Men (Yellow) -- Age-adjusted Death Rates Per 100,000

People Ages15-74

68

100

140

120

98

65

12 15 20

45 45 43

10

30

50

70

90

110

130

150

1950 1960 1970 1980 1990 2000

Leon et al: Understanding the Health of Scotland in and International Context 2003

Percentage Self Reporting Health as Excellent or Good by Age Group (40-49 yellow and 60-69 blue)

and Decade of Birth using 1972 to 2004 General Social Surveys

74%76%

82% 82%84%

52%

57.00%

63%

74%

82%

50.00%

60.00%

70.00%

80.00%

90.00%

1900's 1910's 1920's 1930's 1940's 1950's 1960's

Age40-49

Age60-69

Adapted from: Robert Warren and Elaine Hernandez (In Press) Journal of Health and Social Behavior,Table 2

Something is Driving these Dramatic Improvements in Health

X ?

Shouldn’t whatever “x” is be an important part of our explanations of health disparities?

Do Key Explanatory Variables in Theories of Disparities Account for Trends Toward

Improvement in Health Over Time?

• How about genetic factors?

• Stress?

• Social involvement and participation?

• How about income inequality?

Of course, X is not any one thing but many things

• The discovery of the germ theory is a strong candidate for declines in rates of infectious diseases in the first half of the 20th century.

• Recent declines in age adjusted rates of death from lung cancer are probably influenced by the lagged effects of declines in smoking rates in earlier decades.

• The rapid decline in HIV/AIDS mortality is probably related to the new anti-retroviral drugs that were developed and disseminated in the late 1990’s

• And then screening for disease, public health efforts to increase the consumption of fruits and vegetables, promote exercise, eradicate smoking, and smog control, flu shots, seat belts, angioplasty, screening for early detection of cancer, etc. etc.

• So X is clearly not just one thing and is likely different things for different diseases…and probably different things at different times….But the confluence of all of these things has clearly had an enormously positive impact on population health.

• Clearly human beings have dramatically increased their capacity to control disease and death.

Fundamental Cause Reasoning Concerning the Sources of Disparities: The Core Proposition

• Our enormous capacity to control disease and death combined withsocial and economic inequality creates health disparities.

• It does so because of a very basic principle – When we develop the ability to control disease and death, the benefits of this new found capacity are not distributed equally throughout the population, but are instead harnessed more securely by individuals and groups who are less likely to be exposed to discrimination and who have more knowledge, money, power, prestige and beneficial social connections.

• People who are more advantaged with respect to resources such asthese and who are less likely to be held back by discrimination benefit more and have lower death rates as a consequence. Disparities are the result.

How Does Happen -- Examples

• When Bad things happen --– Titanic, WTC, Katrina– All the quotidian tragedies of life

• When Good things happen --– Discovery that smoking is harmful to health– New technology -- cancer screens

1954 Percent Answering that they had Recently Read or Heard Anything Saying that Cigarette Smoking May be a Cause of Cancer

81

90

95.5 95.1

70

75

80

85

90

95

100

1954

< 12 years12 years13-1516+

Nationwide Gallup Surveys Jan and June 1954

Percent Responding “Yes” to a Question asking whether Smoking is a Cause of Lung

Cancer by Education 1954

41 41.947.5 46.3

0102030405060708090

100

1954

< 12 years12 years13-1516+

Nationwide Gallup Surveys Jan and June 1954

Percent Responding “Yes” to a Question asking whether Smoking is a Cause of

Lung Cancer

41 42.8

70.1 71.1 71.181.6 84 82.8

94.4 92.9

0102030405060708090

1001954 Jan1954 June19691971197219771981198519901999

Multiple National Public Opinion Surveys

Percent Responding “Yes” to a Question asking whether Smoking is a Cause of Lung

Cancer by Education 1954 thru 1999

4146.3

62.8

84.7

72.1

92.486.6

96.8

30

40

50

60

70

80

90

100

1954 1969-1972 1977-1985 1990-1999

< 12 years12 years13-1516+

Nationwide Surveys 1954 -1999

Percent Current Smokers1954-1999

47.243.4

40.9 42 42.138.3

34.629.5

24.822.3

05

101520253035404550

1954 Jan1954 June19691971197219771981198519901999

Multiple National Public Opinion Surveys

Percent Current Smokers by Education 1954 thru 1999

43.245.5

40.5

34.438

22.5

32.2

14.6

05

101520253035404550

1954 1969-1972 1977-1985 1990-1999

< 12 years12 years13-1516+

Nationwide Surveys 1954 -1999

Age Adjusted Lung Cancer Mortality (Men 25-64) 1950-1998 by SES of County of

Residence

0

10

20

30

40

50

60

70

80

90

50 56 62 68 74 80 86 92 98

Low SES

High SES

Middle SES

From Singh et al. 2002 Journal of the National Cancer Institute

Points to be Made From the Very Familiar Smoking Story

• It exemplifies the process specified by the fundamental causes theory –there is a social shaping of new knowledge and this social shaping creates health disparities by socioeconomic status.

• The story sensitizes us to the fact that many other forms of health knowledge and technology are shaped in this way – e.g. the same states in the US that were quick to adopt the use of beta blockers had been rapid in their deployment of hybrid corn 30 to 50 years earlier.

• The length of time it takes for knowledge and behavior change tospread is instructive. It is not simply a matter of being smart enough to know the evidence and act. It is again likely deeply social. And this tells me an important story about disparities.

Percent Responding “Yes” to a Question asking whether Smoking is a Cause of Lung

Cancer by Education 1954 thru 1999

4146.3

62.8

84.7

72.1

92.486.6

96.8

30

40

50

60

70

80

90

100

1954 1969-1972 1977-1985 1990-1999

< 12 years12 years13-1516+

Nationwide Surveys 1954 -1999

Percent Current Smokers by Education 1954 thru 1999

43.245.5

40.5

34.438

22.5

32.2

14.6

05

101520253035404550

1954 1969-1972 1977-1985 1990-1999

< 12 years12 years13-1516+

Nationwide Surveys 1954 -1999

Percent Believing that Sun Goes Around Earth and that Sound Travels Faster than

Light – US Population 1999

24.522.6

0

5

10

15

20

25

30

% Sun Goes Around Earth % Sound Faster than Light

NSF Survey of Public Understanding of Science and Technology 1999

Among People Who Believe that Sun Goes Around Earth and that Sound Travels Faster than Light -- %

Who Think Smoking Causes Cancer

91.1 92

0102030405060708090

100

Sun - Earth Sound - Light

NSF Survey of Public Understanding of Science and Technology 1999

Test #1 --SES Associations with More and Less Preventable Causes of

Death

• We say that SES differences arise because people of higher SES use flexible resources to avoid risks and adopt protective strategies

• it follows that the SES gradient should be more pronounced for diseases that we can do something about… for which there are known and modifiable risk and protective factors…

• Our first test involves ratings of the preventability of death from specific causes

US National Longitudinal Mortality Survey

• Very large study of a nationwide sample of over 350,000 people.

• Interviewed as part of the US Current Population Survey (assesses unemployment etc.) and followed for 9 years with National Death Index for mortality and cause of death

Relative Risks of Death by Income -- NationalLongitudinal Mortality Study

Income (1980 $) Men 45-64Women 45-64

< 5000 2.32 3.13

5000-9999 1.79 2.63

10000-14999 1.56 2.03

15000-19999 1.35 1.69

20000-24999 1.21 1.47

25000-49999 1.09 1.28

50000+ (reference category) 1.00 1.00

Sorlie et al. AJPH 1995

The Rating Task

• Thinking of both our ability to prevent a disease from occurring and to treat it once it occurs, to what degree was it possible, in the early 1990’s to prevent death from this disease?

• Rated on a 5 point scale from “virtually impossible to prevent death” to “virtually all deaths preventable”

• Inter-rater reliability .85. Correlation with Rutstein independent ratings .57.

Examples of Hi and Lo Preventability Diseases

• Low Preventability:brain cancer, ovarian cancer, gallbladder

cancer, multiple sclerosis, pancreatic cancer,

• High Preventability:lung cancer, ischemic heart disease,

colon cancer, pneumonia

National Longitudinal Mortality Study Percent Dying During 9 Year Follow-Up:

Men and Women 25-44

0.5

1

1.8

0.20.3

0.4

00.20.40.60.8

11.21.41.61.8

Hi Preventability Lo Preventability

16+ Years12 -15 years< 12 Years

Phelan, Link, Diez-Roux, Kawachi and Levin. 2004 JHSB

National Longitudinal Mortality Study Percent Dying During 9 Year Follow-Up:

Men and Women 45-64

4.15

8.2

1.8 1.8 2.1

0123456789

Hi Preventability Lo Preventability

16+ Years12 -15 years< 12 Years

Phelan, Link, Diez-Roux, Kawachi and Levin. 2004 JHSB

Test # 2 Evidence Bearing on the Hypothesis Trends Over Time

• If the core proposition is true we should find that disparities by SES and race emerge when new health enhancing information or technology is obtained: – E.g. Heart disease, Hodgkins Disease, Colon Cancer

• If death from a disease remains unpreventable – disparities will not change dramatically with time– E.G. Brian cancer, Ovarian Cancer, Pancreatic Cancer

Trends by Race

Brain Cancer -- Age-adjusted Death Rates Per 100,000 1950-1999 (Males)

2.843.21

2.91 3.1 3.08

3.864.18

4.54.81 4.97 5.17 5.04 5.26 5.47

0

1

2

3

4

5

6

50-54 55-59 60-64 65-69 70-74 75-79 80-84 85-89 90-94 95-99

White Males

Black Males

Ovarian Cancer -- Age-adjusted Death Rates Per 100,000 1950-1999 (Females)

7.216.66 6.56 6.43 6.72

8.358.91 8.9 9.03 8.88 8.58

8.11 8.08 8.13

0123456789

10

50-54 55-59 60-64 65-69 70-74 75-79 80-84 85-89 90-94 95-99

White Females

Black Females

Pancreatic Cancer -- Age-adjusted Death Rates Per 100,000 1981-2002

1112.3

11.712.3 12.1

11.5 11.3 11

8.4 8.3 8.2 8.2 8.2 8 8.2 8.2

0

2

4

6

8

10

12

14

1981 1984 1987 1990 1993 1996 1999 2002

White

Black

Heart Disease -- Age-adjusted Death Rates Per 100,000 1950-2000

586.7548.3

512

455.3

391.5

324.8

584.8559

492.2

409.4

317

253.4

200

250

300

350

400

450

500

550

600

1950 1960 1970 1980 1990 2000

White

Black

All Cancers Combined-- Age-adjusted Death Rates Per 100,000 1950-2000

176.4199.1

225.3

256.4279.5

248.5

194.6 193.1 196.7 204.2 211.6197.2

0

50

100

150

200

250

300

1950 1960 1970 1980 1990 2000

White

Black

Breast Cancer-- Age-adjusted Death Rates Per 100,000 1950-2000

25.327.9 28.9

31.7

38.134.5

32.232.4 32 32.5 32.1 33.2

26.323.9

0

5

10

15

20

25

30

35

40

45

1950 1960 1970 1980 1990 2000 2004

White

Black

Colon, Rectum and Anus -- Age-adjusted Death Rates Per 100,000 1960-2000

22.8

26.1

28.3

30.6

28.2

30.9

29.2

27.4

24.1

20.320

22

24

26

28

30

32

34

1960 1970 1980 1990 2000

White

Black

Age Adjusted Cancer Mortality (All Types) 1950-1998 by SES of County of Residence

100

120

140

160

180

200

220

240

260

50 54 58 62 66 70 74 78 82 86 90 94 98

Low SES

High SES

Middle SES

From Singh et al. 2002 Journal of the National Cancer Institute

Age Adjusted Colon Cancer Mortality (Men 25-64) 1950-1998 by SES of County of

Residence

0

5

10

15

20

25

50 56 62 68 74 80 86 92 98

Low SES

High SES

Middle SES

From Singh et al. 2002 Journal of the National Cancer Institute

Percent with Total Cholesterol Above 240 mg/dL -- Age-adjusted

24.423.5

19.3 18.9

28.9

26.5

19.4

17.5

28.9 29

19.6

1615

17

19

21

23

25

27

29

1971-1974 1976-1980 1988-1994 2001-2004

< 100% Pov Level

> 200% Pov Level >100% Pov <200%

Conclusions

• When we examine Race and SES disparities in mortality by particular diseases we find dramatic evidence that such disparities are created over time.

• Groups with more resources and who face less discrimination benefit more greatly from our new found capacity and disparities emerge

• This means that explanations that propose relatively unchanging causes of disparities like genes, relative deprivation, social participation etc. cannot be the main reasons for health disparities.

Colon, Rectum and Anus -- Age-adjusted Death Rates Per 100,000 1950-2000

22.8

26.1

28.3

30.629.3

28.2

30.929.2

27.4

24.1

2220.320

22

24

26

28

30

32

34

1950 1960 1970 1980 1990 1995 2000

White

Black

Age Adjusted Lung Cancer Mortality (Men 25-64) 1950-1998 by SES of County of

Residence

0

10

20

30

40

50

60

70

80

90

50 56 62 68 74 80 86 92 98

Low SES

High SES

Middle SES

From Singh et al. 2002 Journal of the National Cancer Institute

Policy Issues

• Social policies that provide resources to the resource poor

• Population health policy that focuses on changing contextual factors that affect everyone in a context regardless of resources

• Population health policies that “contextualize” individual risk

Social Policies that Provide Resources to the Resource Poor

Probabilityof

DiseaseAnd Death

Hi

Lo

SESLo Hi

Lop off low end of SES distribution

Policies That Affect How Steep the SES Gradient Becomes

Current Situation

More of a Gradient

Less of a Gradient

Probabilityof

DiseaseAnd Death

Hi

Lo

SESLo Hi

• Create interventions that affect entire contexts rather than interventions that require individual action– Fluoridation versus flossing– Air bags versus seat belts– Cleaning up lead paint versus instructing parents how

to keep their children from ingesting paint chips– Meat inspection versus information to carefully wash

cutting boards

• If interventions require individual action “contextualize” risk factors – that is find out what puts people at risk of risk – then intervene broadly and comprehensively to remove existing barriers to individual action directed towards health.

• If interventions require individual action “contextualize” risk factors – that is find out what puts people at risk of risk – then intervene broadly and comprehensively to remove existing barriers to individual action directed towards health.

Thank you for your Attention!