The Social Shaping of Health
Disparities: The Fundamental Cause
Hypothesis
Bruce Link
Heron
April 7, 2011
U.S. All Cause Age-adjusted Death
Rates Per 100,000 by Race – 2005
1200
1000
1016.5
785.3
800
600
white
black
400
200
0
National Center for Health Statistics – Health United States 2008
U.S. All Cause Age-adjusted Death
Rates Per 100,000 People Ages 2564 by Education -- 2005
900
821
800
700
600
606
472
500
400
< 12 years
12 years
13+ years
352
300
249
165
200
100
0
Males
Females
National Center for Health Statistics – Health United States 2008
U.S. Percent Fair or Poor SelfReported Health by Poverty Level
and Race/Ethnicity
30
26
25
20
21
21
20
17
15
14
15
14
10
10
5
6
6
Overall
White
9
Below 100%
100%-less than 200%
200% or more
0
Black
Hispanic
ional Center for Health Statistics – Health United States 2006 –
From NHANES Surveys
Age Standardized Mortality Rates by SES
Classification (NS-SEC) in the North East and
South West, Men 25-64, 2001-2003
Hi Manager Prof.
700
700
600
Lo Manager Prof.
500
Intermediate
400
400
Small Employers
300
200
210
195
100
0
North East
South West
Lower supervisory and
technical
Semi-Routine
Routine
NS-SEC= National Statistics Socioeconomic Classification
What Are the Mechanisms that Account
for the Association?
?
SES,
Race
Mechanisms
Smoking, Diet,
Exercise, Stress,
Etc, etc. ?
Mortality
Morbidity
Deaths Per 1000 Among Taxpayers and
Non-Taxpayers in Rhode Island 1865
Age Categories
(examples)
Taxpayers
Non-Taxpayers
93.4
189.8
30-39
4.5
15.5
60-69
15.1
39.5
Under 1
Chapin AJPH 1924
Deaths per 1000 (age adjusted) by SES
of 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 as
knowledge, money, power, prestige and beneficial social
connections that determine the extent to which people are
able to avoid risks and adopt protective strategies so as to
reduce morbidity and mortality.
Because such resources can be used in different ways
in different situations, fundamental causes have effects on
disease even when the profile of risk and protective
factors and diseases changes radically.
It is their persistent effect on health in the face of
dramatic 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 19002000
80
77
75
74
70
70
65
61
60
55
55
50
45
47
40
1900
1920
1940
1960
1980
2000
US: Heart Disease -- Age-adjusted
Death Rates Per 100,000 People
640
587
540
559
493
440
412
340
321
293
240
258
140
40
1950
1960
1970
1980
1990
1995
2000
Cancer (green) and Stroke (Yellow) -- Age-adjusted
Death Rates Per 100,000 People
240
220
200
180
194
181
194
178
199
208
216
210
200
186
160
148
140
120
100
96
80
65
60
63
61
40
1950
1960
1970
1980
1990
1995
2000
50
2004
National Center for Health Statistics – Health United States 2006
US : Flu (blue) and HIV (green) -- Ageadjusted Death Rates Per 100,000 People
60
50
54
48
42
40
37
33
31
30
24
20
16
10
10
5
0
1950
1960
1970
1980
1990
1995
2000
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
90.00%
Age
40-49
80.00%
74%
82%
82%
84%
76%
74%
70.00%
63%
60.00%
50.00%
Age
60-69
57.00%
52%
1900's
1910's
1920's
1930's
1940's
1950's
1960's
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?
• Social involvement and participation?
• How about income inequality?
• Relative position on hierarchies?
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 with
social 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 as
these 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.
Explanations for Race and SES Disparities
that Have Different Emphases than a Social
Shaping Fundamental Cause Approach
•
•
•
•
•
Genetic Differences
Health Selection
Relative Deprivation
Job Control
Stress of Lower Position
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 -- National
Longitudinal Mortality Study
Income (1980 $)
Women 45-64
Men 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 45-64
9
8.2
8
7
6
5
4
5
16+ Years
12 -15 years
< 12 Years
4.1
3
1.8
2
1.8
2.1
1
0
Hi Preventability
Lo Preventability
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 County-Level SES and
Race in the US
Brain Cancer -- Age-adjusted Death Rates Per
100,000 1950-1999 (Males) US
6
White Males4.97
5
4
3
2
3.86
4.18
4.5
5.17
4.81
3.21
2.84
5.04
2.91
5.26
3.1
5.47
3.08
Black Males
1
0
50-54 55-59 60-64 65-69 70-74 75-79 80-84 85-89 90-94 95-99
Ovarian Cancer -- Age-adjusted Death Rates
Per 100,000 1950-1999 (Females) US
10
9
8
7
6
5
White Females 9.03
8.35
8.91
8.9
8.88
7.21
8.58
6.66
8.11
8.08
8.13
6.56
6.43
6.72
Black Females
4
3
2
1
0
50-54 55-59 60-64 65-69 70-74 75-79 80-84 85-89 90-94 95-99
Pancreatic Cancer -- Age-adjusted Death
Rates Per 100,000 1981-2002 US
14
12
Black 12.3
11
11.7
12.3
12.1
11.5
11.3
11
8
8.2
8.2
10
8
8.4
8.3
White
8.2
8.2
8.2
6
4
2
0
1981
1984
1987
1990
1993
1996
1999
2002
Heart Disease -- Age-adjusted Death Rates
Per 100,000 1950-2000 US
600
586.7
584.8
559
548.3
550
512
492.2
500
450
455.3
400
409.4
Black
391.5
350
White
300
317
324.8
253.4
250
200
1950
1960
1970
1980
1990
2000
Breast Cancer-- Age-adjusted Death Rates Per
100,000 1950-2000
45
40
Black38.1
35
30
25
32.4
32
27.9
32.5
28.9
32.1
31.7
33.2
34.5
White 26.3
25.3
32.2
23.9
20
15
10
5
0
1950
1960
1970
1980
1990
2000
2004
Colon, Rectum and Anus -- Age-adjusted
Death Rates Per 100,000 1960-2000 US
34
32
30.9
30.6
30
29.2
28.3
27.4
28
28.2
26.1
26
White
24
22
Black
24.1
22.8
20.3
20
1960
1970
1980
1990
2000
Age-, sex-, race-adjusted pancreatic cancer mortality
per 10,000 persons 45 or more years, 1968-2005
Age-, sex-, race-adjusted pancreatic cancer mortality
per 10,000 persons 45 or more years, 1968-2005
Age-, sex-, race-adjusted lung cancer mortality per 10,000
persons 45 or more years, 1968-2005
Age-, sex-, race-adjusted lung cancer mortality per 10,000
persons 45 or more years, 1968-2005
Age-, sex-, race-adjusted lung cancer mortality per 10,000
persons 45 years and over by county SES percentile, 19682005
What do These Tests Tell Us?
• This is consistent with the social shaping perspective and it
says that the scope of problem is large… BUT
• The link to new knowledge and technology is not as direct
we would like. Let me turn now to three stories where the
linking is somewhat better….
Income Disparities in Cholesterol
• Chang and Lauderdale use data on
cholesterol levels from NHANES before
(1976-1980) and after the introduction of
highly effective statins (1999 -2004)
• Income is assessed as the poverty income
ratio
Income Gradients for Total Cholesterol 1976-80 and
1999-2004: Predicted Lipid Levels from NHANES
for Women
219
216
214
213
1976-1980
209
204
200
1999-2004
199
195
194
0
1
2
3
4
5
Chang, Virginia and Diane Lauderdale. 2009. Journal of Health and
Social Behavior 50:245-260
Income Gradients for Total Cholesterol 1976-80 and
1999-2004: Predicted Lipid Levels from NHANES
for Men
218
219
214
1976-1980
212
209
205
204
200
1999-2004
199
194
0
1
2
3
4
5
Chang, Virginia and Diane Lauderdale. 2009. Journal of Health and
Social Behavior 50:245-260
Medical Advances and Race/ Ethnic Disparities in
Cancer Survival
• Tehranifar, Neugut, Phelan, Link, Liao, Desai and Terry.
2009. Cancer Epidemiology Biomarkers Prevention.
• Cancer cases (N=580,225) in SEER ages 20+ diagnosed
with one invasive cancer in 1995-1999.
• Used 5-year relative survival rates to measure degree to
which mortality from each cancer is amenable to medical
interventions (early detection and treatment) – ranged from
5% for pancreatic cancer to 99% for prostate cancer.
Do Racial/Ethnic Differences in Survival Increase as Cancers
become more amenable to medical interventions?
1.80
Hazard Ratio
1.60
1.40
American
Indian/Alaska Native
1.20
Asian or Pacific
Islander
1.00
African American
0.80
Hispanic
0
10
20
30
<<<<<Less Amenable
40
50
60
70
80
90
100
More Amenable >>>>>
HIV Mortality
• One potentially dramatic example might be
HIV-AIDS mortality.
• In particular Highly Active Anti-Retroviral
Therapy (HARRT) as a new life saving
technology.
HIV Mortality 1987-2005
• Rubin, Colen and Link. 2010. American Journal of Public
Health
• HIV mortality in every county in the United States from
the National Center for Health Statistics by Age, Race and
Gender.
• Constructed rates for every year using mortality data for
the numerator and census data for the denominator.
• Constructed SES measures for each county using
indicators of education, income, occupation and poverty.
• We identified a pre (1987-1994), a peri (1995-1998) and a
post (1999-2005) HARRT period.
• We expect an interaction between SES and period and
between race and period such that the benefit HAART is
more pronounced in high SES counties and among Whites
as opposed to low SES counties and among Blacks
HIV Deaths among Whites per 100,000 by Age
40
30
15-24 years
25
25-34 years
20
35-44 years
15
45-54 years
55-64 years
10
5
Year
2005
2004
2003
2002
2001
2000
1999
1998
1997
1996
1995
1994
1993
1992
1991
1990
1989
1988
0
1987
Deaths per 100,000
35
HIV Deaths among Blacks per 100,000 by Age
160
120
15-24 years
100
25-34 years
80
35-44 years
45-54 years
60
55-64 years
40
20
Year
20
05
20
03
20
01
19
99
19
97
19
95
19
93
19
91
19
89
0
19
87
Deaths per 100,000
140
Incidence Rate Ratios – Blacks Versus Whites Before (Pre),
During (Peri) and After (Post) the Introduction of Highly
Active Anti-Retroviral Therapy in the United States
10
9
8
7
6
5
4
3
2
1
0
7.84
5.64
White
Black
3.66
1
Pre HAART
1
Peri HAART
1
Post HAART
IRRs adjusted for age, sex, and SES and urbanicity of county of residence.
Incidence Rate Ratios – Comparing a County at the 95th
Percentile of SES to a County at the 5th Percentile of SES Pre,
Peri and Post the Introduction of Highly Active AntiRetroviral Therapy (HAART) in the United States
4
3.5
3
2.72
2.5
1.91
2
1.41
1.5
1
1
1
1
High SES County (5th
percentile)
Low SES County (95th
percentile)
0.5
0
Pre HAART Peri HAART Post HAART
IRRs adjusted for age, sex, race, and urbanicity of county of residence
30
20
10
0
1985
1990
1995
Year
95% CI
Fitted Values for Blacks
2000
Fitted Values for Whites
2005
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, health
selection due to health induced disability, relative
deprivation, job control etc. cannot be the main reasons for
health disparities.
Colon, Rectum and Anus -- Age-adjusted
Death Rates Per 100,000 1950-2000
34
32
30.9
30
30.6
29.2
28
Black 29.3
28.3
27.4
28.2
26.1
26
White
24
24.1
22.8
22
22
20.3
20
1950
1960
1970
1980
1990
1995
2000
Age-, sex-, race-adjusted lung cancer mortality per 10,000
persons 45 years and over by county SES percentile, 19682005
Puzzles
• Combinations of SES and gender, race, ethnicity,
immigration status and sexual preference
sometimes produces “paradoxes.” Understanding
these paradoxes can be a door to newer and deeper
knowledge.
– Women disadvantaged compared to men but live
longer.
– African Americans disadvantaged with respect to SES
but have lower rates of major depression.
– Some immigrants groups --though disadvantaged with
respect to SES -- enjoy longer life.
Conclusion
• Many prominent disparities are created
when the benefits of new knowledge about
disease causation and new approaches to
prevention and cure are distributed
unequally in populations.
• The HERON’s Gaze needs to be fixed on
these unequal distributions and it needs tobe
ready to strike when the processes that
enable their emergence are evident.