This preview shows page 1. Sign up to view the full content.
Unformatted text preview: Original Contributions Socioeconomic Factors,
Health Behaviors, and Mortality
Results From a Nationally Representative
Prospective Study of US Adults
Paula M. Lantz, PhD; James S. House, PhD; James M. Lepkowski, PhD;
David R. Williams, PhD; Richard P. Mero, MS; Jieming Chen, PhD Context.—A prominent hypothesis regarding social inequalities in mortality is
that the elevated risk among the socioeconomically disadvantaged is largely due
to the higher prevalence of health risk behaviors among those with lower levels of
education and income.
Objective.—To investigate the degree to which 4 behavioral risk factors (cigarette smoking, alcohol drinking, sedentary lifestyle, and relative body weight)
explain the observed association between socioeconomic characteristics and allcause mortality.
Design.—Longitudinal survey study investigating the impact of education,
income, and health behaviors on the risk of dying within the next 7.5 years.
Participants.—A nationally representative sample of 3617 adult women and
men participating in the Americans’ Changing Lives survey.
Main Outcome Measure.—All-cause mortality veriﬁed through the National
Death Index and death certiﬁcate reviews.
Results.—Educational differences in mortality were explained in full by the
strong association between education and income. Controlling for age, sex, race,
urbanicity, and education, the hazard rate ratio of mortality was 3.22 (95% conﬁdence interval [CI], 2.01-5.16) for those in the lowest-income group and 2.34 (95%
CI, 1.49-3.67) for those in the middle-income group. When health risk behaviors
were considered, the risk of dying was still signiﬁcantly elevated for the lowestincome group (hazard rate ratio, 2.77; 95% CI, 1.74-4.42) and the middle-income
group (hazard rate ratio, 2.14; 95% CI, 1.38-3.25).
Conclusion.—Although reducing the prevalence of health risk behaviors in lowincome populations is an important public health goal, socioeconomic differences
in mortality are due to a wider array of factors and, therefore, would persist even
with improved health behaviors among the disadvantaged.
JAMA. 1998;279:1703-1708 OVER THE PAST several decades,
health behavior or lifestyle factors—
smoking cigarettes, being overweight,
drinking alcoholic beverages, and being
physically inactive or leading a sedentary lifestyle—have often been cited as the
major determinants of premature and pre- From the Survey Research Center (Drs Lantz, House,
Lepkowski, and Williams and Mr Mero), the School of
Public Health (Drs Lantz, House, and Lepkowski), and
the Department of Sociology (Drs House, Williams, and
Chen), University of Michigan, Ann Arbor. Dr Chen is
now with the Department of Psychology and Sociology,
Texas A&M University at Kingsville.
Reprints: Paula M. Lantz, PhD, Department of Health
Management and Policy, School of Public Health, University of Michigan, 109 Observatory, Ann Arbor, MI
48109-2029 (e-mail: email@example.com). JAMA, June 3, 1998—Vol 279, No. 21 ventable morbidity and mortality.1-7 More
recently, differences in health outcomes
by socioeconomic position have been recognized as a persisting and perhaps even
increasing public health problem.8-12 Less
well understood, however, is the relationship between health risk behaviors and socioeconomic differentials in health, especially in nationally representative samples.
In a number of longitudinal studies, important socioeconomic indicators—such as
income and education—have been shown
to be inversely associated with various
mortality outcomes, including premature mortality, cardiovascular mortality,
and death from all causes.13-18 In addition,
it is well documented that people of lower
socioeconomic position are significantly more likely to lead a sedentary lifestyle,
to be overweight, and to smoke cigarettes.19-22 Thus, a prominent hypothesis
is that the elevated mortality risk associated with low levels of income and education is primarily due to the higher prevalence of health risk behaviors among
people who are poor and/or have low educational attainment.3,23-25 However, previous efforts to explain socioeconomic
differences in mortality in a variety of subpopulations have found that strong differences remain after controlling for major lifestyle risk factors.16,18,26-29
For editorial comment see p 1745.
There are some serious limitations in
the samples of most prior prospective
studies on the contribution of health
risk behaviors to socioeconomic differences in mortality. Although populationbased samples were used, the populations were generally confined to a limited geographic area, such as a single
city, county, or small region of a country,
and, in many cases, samples were
urther restricted by including only
males.16,18,20,26-29 In addition, much previous work has not provided a careful
analysis of 2 primary socioeconomic indicators—education and income—even
though it is quite possible that the
mechanisms by which income and education are related to health behaviors
and/or mortality differ significantly.
The degree to which health behaviors
explain or mediate the influence of socioeconomic factors on mortality has important ramifications for health policy. The
research presented here attempts to bring
greater clarity to this issue by addressing the following questions: (1) what is the
relationship between the socioeconomic
factors of education and income and health
behaviors, such as cigarette smoking, body
weight, consumption of alcoholic beverages, and physical activity; (2) what are Socioeconomic Factors, Health Behaviors, and Mortality—Lantz et al ©1998 American Medical Association. All rights reserved.
Downloaded from www.jama.com at Oregon State University, on September 1, 2006 1703 the relative magnitudes of the effects of
education, income, and health behaviors
on all-cause mortality; and (3) to what extent do health behaviors explain education and income differences in mortality,
and does this vary by age, race, or sex?
Our approach uses a nationally representative, longitudinal sample that includes
both men and women, considers the effects of income and education separately,
and investigates demographic subgroup
variation in the relationship between education, income, health behaviors, and
Study Design and Sample
The data analyzed for this study are
from the Americans’ Changing Lives
(ACL) longitudinal survey conducted by
the University of Michigan Survey Research Center. A stratified, multistage
area sample of noninstitutionalized persons 25 years of age or older living in the
coterminous United States was selected
for study over time. Persons aged 60
years and older and blacks were oversampled. Initial face-to-face interviews
were conducted with 3617 persons in
1986, representing 70% of all sampled
households and 68% of sampled individuals. Information on the independent
variables being studied (as described below) was taken from the 1986 ACL wave
1 survey. Two subsequent waves were
conducted in 1989 and 1994. Additional
details on the ACL survey design and
methods are provided elsewhere.12,30
Information on deaths among sample
respondents from mid-1986 through
March 1994 was obtained from informants and through the National Death
Index. The main outcome variable is allcause mortality. In addition, underlying
causes of death (obtained from death certificates) were grouped into 4 categories
based on the International Statistical
Classification of Diseases, 10th Revision
(ICD-10): (1) tumors, (2) cardiovascular
diseases, (3) all other diseases, and (4) external causes, such as unintentional injury, suicide, homicide, or legal intervention. To date, 90.3% of all deaths have
been verified with death certificates. Reports of the 9.7% of deaths (n = 53) not yet
verified with death certificates were reviewed carefully, and actual death appears to be certain in each case. For these
cases, the month and year of death were
ascertained from information about the
deaths obtained from informants.
Socioeconomic Factors and
Other Sociodemographic Measures
The socioeconomic factors being studied are education and income, based on
self-reported information from the ACL
1704 JAMA, June 3, 1998—Vol 279, No. 21 wave 1 survey. Education is measured
as respondents’ total years of completed
education and is grouped as a 3-category
classification: 0 through 11 years; 12
through 15 years; and 16 or more years.
Income is measured as the combined income from all sources of the respondent
and his or her spouse in the preceding
year, and also is grouped into 3 categories: $0 through $9999; $10 000 through
$29 999; and $30 000 or more. More refined categories of education and income
produced similar results for the analyses presented, as did adding controls for
household size and assets.
Age is grouped into 6 categories: 25
through 34 years; 35 through 44 years;
45 through 54 years; 55 through 64 years;
65 through 74 years; and 75 years or
older. Other sociodemographic variables
being studied include sex (male vs female), race (nonblack vs black), and urbanicity of residence (central city, suburban, or rural). Previous research has
found these demographic variables to be
related to socioeconomic factors, health
risk behaviors, and mortality. Thus, they
are included in the analysis primarily as
controls for potential confounders.
Behavioral Risk Factor Measures
Health behavior indicators are based
on self-reported information from respondents at ACL wave 1. Cigarette
smoking is coded as never smoked,
former smokers, and current smokers.
Alcohol drinking is coded using 3 categories based on the number of drinks consumed in the past month: nondrinkers
(0 alcoholic drinks in past month), moderate drinkers (1-89 drinks), and heavy
drinkers ( 90 drinks). Body weight was
measured using the body mass index
(BMI), weight in kilograms divided by
the square of height in meters , based on
self-reported weight and height. The
body weight variable was coded as normal body weight, overweight, and underweight. Following the methods of
Berkman and Breslow,1 those in the
highest 15% of the weighted sex-specific
BMI distributions were coded as overweight and those in the lowest 5% of the
weighted sex-specific BMI distributions
were coded as underweight.
A physical activity index was computed based on answers to questions regarding how often the respondent engaged in active sports or exercise, gardening or yard work, and taking walks.
Physical activity index scores were divided into quintiles to create 5 groupings of near-equal sample size. The group
in the top quintile represents the 21% of
the weighted sample that is the most
Health Status.—Three variables were
available to measure baseline health sta- tus: (1) self-rated health measured with a
single 5-category scale classified as excellent, very good, good, fair, and poor; (2)
the number of major chronic conditions experienced in the last year from a list of 10
conditions; and (3) an index of functional
status, with the lowest score of 1 representing confinement to a chair or bed and
the highest score of 4 representing the
ability to do heavy work inside or outside
Statistical Analysis.—In all analyses,
the data were weighted to adjust for differential response rates and variation in
probabilities of selection into the sample.
Poststratification weights adjust ACL
wave 1 sample results to the July 1, 1986,
Bureau of the Census population estimates by sex, age, and region of the country. Descriptive statistics were obtained
through the Statistical Analysis System,
SAS Institute, Inc, Cary, NC, including
frequency distributions of all variables
being studied, cross tabulations of the
socioeconomic variables and health risk
behaviors, and cross tabulations of socioeconomic variables and mortality. In creating contingency tables regarding the
relationship between socioeconomic factors and health risk behaviors, direct
standardization to the age distribution of
the weighted ACL wave 1 population was
used to account for the strong association
between age and socioeconomic factors.31
The Cox proportional hazards model
was used to estimate the relative risk of
mortality in terms of various background, socioeconomic, and health behavior variables. Taylor series linearization procedures using SUDAAN, Research Triangle Institute, Research
Triangle Park, NC, were used to make
adjustments to standard errors for the
complex sample design. The effects of
each independent variable being studied on mortality were analyzed separately. A series of multiple predictor
models were then estimated. First, the
relative hazard rate of mortality was estimated for income and education groups
both separately and together, controlling for age, sex, race, and urbanicity.
Second, the behavioral risk factors being studied were added to the base model
to investigate how much of the socioeconomic differentials in mortality could be
attributed to these factors. Models were
also run in which controls for baseline
health status were added and in which
possible interactive effects between
health behaviors and variables such as
education, income, sex, and race were
A significant portion of sample respondents (representing the national
population) were socioeconomically dis- Socioeconomic Factors, Health Behaviors, and Mortality—Lantz et al ©1998 American Medical Association. All rights reserved.
Downloaded from www.jama.com at Oregon State University, on September 1, 2006 advantaged (Table 1). A total of 25.6% of
the weighted sample reported 0 to 11
years of education, and 19.2% reported
annual incomes of less than $10 000 at
ACL wave 1. A total of 546 respondents
(15.1% of the overall sample and 9.9% of
the weighted sample) died during the
7.5-year follow-up period. The deaths included 255 males and 291 females, 338
nonblacks and 208 blacks, and 147 persons younger than 65 years and 399 persons aged 65 years and older.
The distribution of the 4 behavioral
risk factors being studied significantly
varied by educational attainment and
annual household income, adjusting for
age (Table 2). For example, persons with
the least amount of education and with
the lowest incomes were significantly
more likely to be current smokers, overweight, and in the lowest quintile for
physical activity. Additional analyses
suggest that there was a high degree of
stability in individuals’ health behaviors
across ACL study waves. For example,
of those who were overweight at wave 1,
84% were overweight at wave 2, and of
those who were current smokers at wave
1, 79% were still smoking at wave 2.
Table 3 presents the hazard rate ratios of mortality by education and income
for males and females separately. Those
with low educational attainment were
significantly more likely to die than
those with 16 or more years of education.
The relationship between education and
mortality and between income and mortality was stronger for females. Both
men and women in the lowest-income
category were more than 3 times as
likely to die during the follow-up period
of the study than those in the highest
group, controlling for age and other sociodemographic variables (Table 3).
While education was strongly related to
health behaviors, income was more predictive of mortality than education.
The relationship between socioeconomic factors, health behaviors, and
mortality was explored by conducting a
sequence of Cox proportional hazards
models. The results of a model including
statistical controls for age, race, urbanicity, sex, education, and income are presented as model 1 in Table 4. The results
show that the effect of income on mortality was strong and significant when
controlling for educational attainment
and background demographic variables.
However, when these sociodemographic
variables were considered simultaneously, the bivariate effect of education
on mortality attenuated to a statistically
insignificant level. Additional model
testing (results not shown) demonstrated that the mechanism by which
education was related to mortality was
through its association with income.
JAMA, June 3, 1998—Vol 279, No. 21 Table 1.—Distribution of Study Variables in ACL Population*
(Total N = 3617) Unweighted, % Weighted, % Age, y
25-34 740 20.5 29.0 35-44 591 16.3 23.2 45-54 390 10.8 14.6 55-64 685 18.9 13.8 65-74 765 21.2 12.5 75 446 12.3 7.0 Variable Sex
Male 1358 37.5 47.1 Female 2259 62.5 52.9 Race
Nonblack 2243 67.5 89.0 Black 1174 32.5 11.0 Residence
City 1204 33.3 24.4 Suburban 1346 37.2 47.0 Other 1067 29.5 28.6 Education, y
0-11 1349 37.3 25.6 12-15 1768 48.9 54.7 16
10 000 500 13.8 14.7
19.2 1176 Normal
Quintile 1 (low) 40.3 29.3 30.4 941 26.0 27.5 1616 44.7 42.1 1837 50.8 41.2 1650 45.6 54.5 130 Never
Alcohol drinks in past month
Body mass index†
Overweight 40.5 26.7 1060 Past Moderate 40.8 966 30 000
Current 32.5 1475 10 000-29 999 3.6 4.3 679 18.8 15.3 2752 76.1 79.6 186 5.1 5.1 1037 28.7 21.3 Quintile 2 540 14.9 14.9 Quintile 3 952 26.3 27.4 Quintile 4 439 12.1 15.2 649 17.9 21.3 3071 84.9 90.1 546 15.1 9.9 Quintile 5 (high)
Dead *ACL indicates Americans’ Changing Lives.
†Body mass index is a measure of weight in kilograms divided by the square of height in meters. When the 4 health behaviors being
studied were added individually to
model 1 (results not shown), the effect of
income on mortality attenuated slightly
yet remained significant for both the
lowest-income and the middle-income
groups. For example, when physical activity was added to the model, the coefficient for the effect of income attenuated a small amount, suggesting that
physical activity explains only a small
proportion of the relationship between
income and mortality. The results of the
full model when all health behaviors
were considered simultaneously (model
2, Table 4) show that there was still a
strong and significant income effect on mortality for both the middle-income
(odds ratio [OR] = 2.14; CI, 1.38-3.25)
and the low-income groups (OR = 2.77;
CI, 1.74-4.42). The 4 health behaviors together accounted for 12% to 13% of the
predictive effect of income on mortality.
In terms of the health behaviors, the results suggest that being severely underweight or having lower levels of physical
activity were significant risk factors for
subsequent mortality, controlling for demographic and socioeconomic characteristics (Table 4). The relationship between physical activity and mortality
appeared to be monotonic, suggesting that
there are gains not only from being physically active but also from increasing Socioeconomic Factors, Health Behaviors, and Mortality—Lantz et al ©1998 American Medical Association. All rights reserved.
Downloaded from www.jama.com at Oregon State University, on September 1, 2006 1705 Table 2.—Age-Adjusted Prevalence of Health Risk Behaviors by Socioeconomic Factors in ACL Population*
Factors Income, $ 0-11 12-15 16 10 000 10 000-29 999 30 000 Smoking, %
Current 42.0 33.1 19.6 37.7 34.2 27.4 Former 22.4 25.0 26.5 20.4 25.3 28.3 Never 35.6 41.9 53.9 41.9 40.5 100
Alcohol drinks in past month, %
4 = 134.6 (P .001) 100 44.3 100
4 = 30.86 (P .001) 100 58.0 42.0 33.0 59.3 46.0 31.3 1-89 37.6 54.0 63.3 37.2 50.3 64.2 90 4.4 4.0 3.7 3.5 3.7 100
Body mass index, %†
4 = 139.2 (P .001) 100 4.5 100
4 = 159.1 (P .001) 100 5.4 4.2 6.4 5.9 3.7 78.9 84.7 69.2 76.1 82.3 27.5 Overweight 5.7 67.1 Normal 15.4 11.1 100
Physical activity, %
Quintile 1 (low) 100
4 = 103.8 (P .001) 24.4 18.0 100 14.0 100
4 = 48.2 (P .001) 100 37.3 22.1 13.6 33.7 25.5 14.7 Quintile 2 14.3 15.0 16.6 14.0 15.7 15.3 Quintile 3 26.0 27.1 27.1 30.3 26.6 26.1 Quintile 4 9.1 14.6 16.7 9.3 13.2 17.3 Quintile 5 13.3 21.2 26.0 12.7 19.0 26.6 100
2 4 100
= 301.63 (P .001) 100
2 4 100
= 160.7 (P .001) 100 *ACL indicates Americans’ Changing Lives.
†Body mass index is a measure of weight in kilograms divided by the square of height in meters.
Table 3.—Sex-Speciﬁc Hazard Rate Ratios of Mortality by Socioeconomic Factors*
Male (n = 1358) Female (n = 2259) Age-Adjusted
(95% CI) Multivariate
(95% CI) Age-Adjusted
(95% CI) Multivariate
(95% CI) 1.60 (1.08-2.36) 1.51 (0.99-2.29) 2.54 (1.25-5.16) 2.46 (1.14-5.0) 12-15 1.20 (0.81-1.23) 1.19 (0.97-1.75) 1.73 (0.79-3.78) 1.75 (0.80-3.82) 16
10 000 1.0 1.0 1.0 1.0 3.32 (2.16-5.10) 3.13 (1.97-4.95) 3.90 (1.92-7.92) 3.82 (1.86-7.85) 2.27 (1.39-3.71) 2.34 (1.43-3.82) 2.64 (1.27-5.47) 2.64 (1.28-5.42) 1.0 1.0 1.0 1.0 Factors
0-11 10 000-29 999
30 000 *Multivariate odds ratios were adjusted for age, race, and urbanicity. CI indicates conﬁdence interval. amounts of activity. In regard to being underweight, descriptive information on the
severely underweight individuals who
died shows that the majority (78%) were
age 75 years or older. Notably, the effects of smoking and drinking were no
longer significant once they were adjusted for the demographic, socioeconomic, and other health behavior variables, and being overweight was not
significant in any of the models.
It is plausible that baseline differences
in both income and health behaviors reflect differences in health status to some
degree. The 3 ACL wave 1 health status
variables (self-reported health, number
of chronic conditions, and functional status) were added separately and simultaneously to a model controlling for background characteristics, income, educa1706 JAMA, June 3, 1998—Vol 279, No. 21 tion, and health behaviors. The results
(not shown) do not suggest any different
patterns or effects from those shown in
Table 4. The relationship between income and mortality remained strong and
significant (P .001) controlling for baseline health status and health behaviors
Additional analyses, including an examination of interaction tests, were conducted to see if the patterns and results
observed for the full sample were the
same across subpopulations of interest.
Six subgroups were examined: males, females, nonblacks, blacks, persons ages
25 through 64 years, and persons ages 65
years and older. The results (not shown)
did not reveal findings that were substantially different from those for the total sample. Overall, health behaviors ex- plained only a small proportion of income
differences in mortality across sex, race,
and age groups.
For those descendents with death certificate information (n = 493), the
weighted underlying cause of death was
tumors, 30%, cardiovascular disease, 28%,
other diseases, 37%, and external causes,
5%. Controlling for income and other sociodemographic variables, education was
not significantly related to any cause-ofdeath category. Those in the lowestincome group had significantly higher
rates of tumor deaths and cardiovascular disease deaths, and those in the
middle-income group had a significantly
higher rate of tumor deaths. Several
health behaviors were associated with a
significantly higher risk of death in specific categories (ie, both current and
former smoking was associated with an
increased risk of tumor deaths, heavy
drinking was associated with increased
risk of death from external causes, and
low physical activity was associated with
increased risk of tumor and cardiovascular deaths). However, for both tumor
and cardiovascular disease deaths separately, controlling for health behaviors attenuated the association between low and
moderate income with mortality to the
same degree observed for death from all
causes. The income effects decreased by
12% to 17% when health risk behaviors
were added to the models, similar to what
was observed in analyses where all causes
of death were grouped together.
The ACL survey findings show that
lower levels of education and income are
associated with a significantly higher
prevalence of health risk behaviors, including smoking, being overweight, and
physical inactivity. The results also show
that lower income (net of demographic
characteristics) leads to a significant increase in mortality risk, yet the influence of major health risk behaviors explains only a modest proportion of this
Our findings of strong socioeconomic differences in mortality (including larger socioeconomic differentials for women than
men, and a stronger mortality effect for
income than for education for both women and men) are consistent with previous
longitudinal research.13-18 In addition, our
findings regarding the association between socioeconomic factors, health behaviors, and mortality are similar to previous studies conducted using limited
samples. For example, in a 20-year study
of Ontario males, Hirdes and Forbes6 concluded that smoking and other health practices are not the primary mechanisms linking socioeconomic status and mortality.
Similarly, the Alameda County Study28 Socioeconomic Factors, Health Behaviors, and Mortality—Lantz et al ©1998 American Medical Association. All rights reserved.
Downloaded from www.jama.com at Oregon State University, on September 1, 2006 showed that the risk of mortality associated with living in high-poverty areas of
Oakland, Calif, changed little after adjusting for smoking, alcohol consumption,
physical activity, BMI, and sleep patterns. Our results contribute to previous
studies by providing evidence regarding
the association between education, income, health behaviors, and mortality from
a nationally representative sample that includes both men and women.
While there appears to be little debate
regarding the need to improve the health
of populations with low levels of income
and education, the appropriate focus of
policy and program responses is less
clear. An important area on which both
policy rhetoric and action have focused is
that of health education and health promotion at the individual level. A tacit assumption among some policymakers and
health authorities is that an important
way to reduce socioeconomic gaps in
health status is to improve the health behaviors among those with low levels of
income and education. This position is obvious in the Department of Health and
Human Services’ Healthy People 2000:
National Health Promotion and Disease
Prevention Objectives and other reports
on the state of health among poor and minority persons in the United States.2,3,23-25
This position has also been articulated in
the lay press. For example, an opinion
piece in the Wall Street Journal32 criticized public health researchers’ growing
focus on social systems and institutions,
arguing that poor people tend to have
worse health and shorter life expectancies, primarily “because unhealthy habits
are more prevalent on the lower rungs of
the socioeconomic ladder.”
Our results suggest that despite the
presence of significant socioeconomic differentials in health behaviors, these differences account for only a modest proportion of social inequalities in overall
mortality. Thus, public health policies and
interventions that exclusively focus on individual risk behaviors have limited potential for reducing socioeconomic disparities in mortality. While reducing the
prevalence of behavioral risk factors is an
important and critical public health goal,
socioeconomic differentials in mortality are
due to a wider array of factors and, therefore, would persist even with improved
health behaviors. Increasing health promotion and disease prevention efforts
among the disadvantaged is not a “magic
policy bullet” for reducing persistent socioeconomic disparities in mortality.
If health risk behaviors do not explain
much of the relationship between socioeconomic factors and mortality, what
else can account for this strong association? First, differences in exposure to
occupational and environmental health
JAMA, June 3, 1998—Vol 279, No. 21 Table 4.—Mortality Hazard Rate Ratios From Explanatory Models*
Model 1 Hazard Rate Ratio
(95% CI) Variable
25-34 Model 2 Hazard Rate Ratio
(95% CI) 1.0 1.0 35-44 2.72 (1.15-6.42) 2.66 (1.11-6.37) 45-54 3.71 (1.28-10.70) 3.46 (1.20-9.95) 55-64 9.87 (4.76-20.49) 9.30 (4.53-19.10) 65-74 17.64 (8.50-36.60) 16.78 (8.17-34.47) 75 47.47 (22.70-99.50) 40.00 (19.1-83.93)
Male 1.0 1.0 0.44 (0.33-0.57) Female 0.41 (0.30-0.54) Race
Nonblack 1.0 1.0 1.21 (0.94-1.55) Black 1.19 (0.92-1.48) Residence
Rural 1.0 1.0 Suburban 1.19 (0.92-1.52) 1.16 (0.91-1.48) City 1.63 (1.17-2.27) 1.52 (1.10-2.10) Education, y
16 1.0 1.0 12-15 1.06 (0.73-1.54) 0.95 (0.61-1.32) 0-11
30 000 1.08 (0.76-1.54) 0.90 (0.62-1.46) 1.0 1.0 2.34 (1.49-3.67) 2.14 (1.38-3.25) 10 000
Never 3.22 (2.01-5.16) 2.77 (1.74-4.42) ... 1.0 Current ... 1.26 (0.93-1.69) ... 1.28 (0.95-1.74) 10 000-29 999 Former
Alcohol drinks in past month
Body mass index†
Normal ... 1.0 ... None 1.13 (0.88-1.44)
0.85 (0.46-1.59) ... 1.0 Underweight ... 2.03 (1.32-3.12) Overweight
Quintile 5 (high) ... 0.94 (0.72-1.23) ... 1.0 Quintile 4 ... 1.46 (0.87-2.45) Quintile 3 ... 1.60 (1.04-2.47) Quintile 2 ... 2.25 (1.41-3.58) Quintile 1 (low) ... 2.91 (1.94-4.56) *CI indicates conﬁdence interval; ellipses, data not applicable.
†Body mass index is a measure of weight in kilograms divided by the square of height in meters. hazards across social strata do exist and,
thus, may be playing a role in mortality
inequalities.33-35 Second, although not a
panacea for eliminating socioeconomic
differences in health status, improved
equity regarding access to and use of
preventive and appropriate therapeutic
medical care is viewed as having some
potential for preventing the further deterioration of health in disadvantaged
Third, socioeconomic stratification itself may be a social force that has deleterious health effects for those in the lower
strata. As Blane41 explains, socioeconomic inequalities in societies “structure
the life experiences of their members so
that advantages and disadvantages tend
to cluster cross-sectionally and accumulate longitudinally.” Persons in lower so- cioeconomic strata have increased exposure to a broad range of psychosocial
variables predictive of morbidity and mortality. This includes (1) a lack of social relationships and social supports; (2) personality dispositions, such as a lost sense
of mastery, optimism, sense of control, and
self-esteem or heightened levels of anger and hostility; and (3) chronic and acute
stress in life and work, including the stress
of racism, classism, and other phenomena related to the social distribution of
power and resources.25,30,34,42-45 Furthermore, Lynch et al46 report that both the
psychosocial orientations and health risk
behaviors of adults are more common
among those whose parents were poor
when they were children. Thus, many individual characteristics, such as personality factors, psychosocial attitudes and Socioeconomic Factors, Health Behaviors, and Mortality—Lantz et al ©1998 American Medical Association. All rights reserved.
Downloaded from www.jama.com at Oregon State University, on September 1, 2006 1707 orientations, and health risk behaviors,
should be viewed as products of or responses to social environments (eg, family, school, neighborhood, cultural context, etc) rather than strictly as individual
There are a number of limitations in our
study methods. First, the health behaviors being investigated were selfreported and were not assessed retrospectively. Literature on the accuracy of
self-reported health behaviors suggests
that, although most people report honestly for behaviors that are not illegal, the
biases that do exist are in the direction
of underreporting negative health behaviors.48-50 Thus, the result of any problems in the reporting of health behaviors would likely be an underestimation
of their effects. Second, the length of the
follow-up period in this prospective study
limits our ability to investigate the longerterm effects of income, education, and
health behaviors on mortality. Third, the
small number of deaths for some of the
demographic groups puts limits on the
multivariate subgroup analysis that could
be performed. Fourth, it is possible that
additional health behaviors and risk
factors not studied explain more of the
relationship between income and mortality. Lynch et al26 report that, in a longitudinal study of Finnish men, the association between socioeconomic status
and mortality from all causes and from
cardiovascular disease was eliminated by
simultaneous adjustment for biologic factors, psychosocial factors, and health risk
behaviors. A full explanation of social inequalities in mortality, however, needs
to address why all of these risk factors
tend to be patterned by socioeconomic
Our results suggest that both health
behaviors and socioeconomic factors are
important determinants of mortality.
While health behaviors are related to
both income and education, they account
for a small proportion of observed socioeconomic differences in mortality. Thus,
the problem of lifestyle and mortality is
not just one of inadequate education or
income, and the problem of socioeconomic differentials in mortality is not
just a problem of lifestyle choices. We
must look to a broader range of explanatory risk factors, including structural elements of inequality in our society.
This study was supported by grants P01AG05561
and R01AG09978-01 from the National Institute on
Aging, National Institutes of Health, Bethesda, Md,
and by a Health Investigator Award (Dr House) from
the Robert Wood Johnson Foundation, Princeton, NJ.
The authors would like to thank the Technical
Sections of the Survey Research Center for conducting the sampling, interviewing, and coding for
the Americans’ Changing Lives survey, additional
colleagues for their assistance in various phases of
the work, and the respondents. John Lynch, PhD, 1708 JAMA, June 3, 1998—Vol 279, No. 21 Marc Muscik, PhD, and several anonymous reviewers offered instructive comments on earlier drafts of
1. Berkman LF, Breslow L. Health and Ways of
Living. New York, NY: Oxford University Press;
2. US Department of Health and Human Services.
Healthy People 2000: National Health Promotion
and Disease Prevention Objectives. Washington,
DC: US Dept of Health and Human Services; 1990.
DHHS publication 91-50212.
3. McGinnis MJ, Foege WH. Actual causes of death
in the United States. JAMA. 1993;270:2207-2212.
4. Wiley JA, Camacho TC. Life-style and future
health: evidence from the Alameda County Study.
Prev Med. 1980;9:1-21.
5. Wilson PS. Established risk factors and coronary
artery disease: the Framingham Study. Am J Hypertens. 1994;7(pt 2):75-125.
6. Hirdes JP, Forbes WF. The importance of social
relationships, socioeconomic status and health practices with respect to mortality among Ontario males.
J Clin Epidemiol. 1992;554:175-182.
7. Patterson RE, Haines PS, Pokin BM. Health lifestyle patterns of US adults. Prev Med. 1994;23:453460.
8. Adler NE, Boyce WT, Chesney MA, Folkman S,
Syme LS. Socioeconomic inequalities in health: no
easy solution. JAMA. 1993;269:3140-3145.
9. Marmot MG, Kogevinas M, Elston MA. Social/
economic status and disease. Annu Rev Public
10. Pappas G, Queen S, Hadden W, Fisher G. The
increasing disparity in mortality between socioeconomic groups in the United States, 1960 and 1986.
N Engl J Med. 1993;329:103-109.
11. Elo IT, Preston SH. Educational differences in
mortality: United States, 1979-85. Soc Sci Med. 1996;
12. House JS, Kessler RC, Herzog AR, et al. Age,
socioeconomic status and health. Milbank Q. 1990;
13. Stevenson THC. The social distribution of mortality from different causes in England and Wales,
1910-12. Biometrika. 1923;15:382-400.
14. Marmot MG, Shipley MJ, Rose G. Inequalities
in death—specific explanations of a general patterns? Lancet. 1984;1:1003-1006.
15. Sorlie PD, Backlund E, Keller JB. US mortality
by economic, demographic and social characteristics: the National Longitudinal Mortality Study. Am
J Public Health. 1995;85:949-956.
16. Salonen JT. Socioeconomic status and risk of
cancer, cerebral stroke and death due to coronary
heart disease and any disease: a longitudinal study
in eastern Finland. J Epidemiol Community
17. Kaplan GA, Keil JE. Socioeconomic factors and
cardiovascular disease: a review of the literature.
18. Davey Smith G, Shipley MJ, Rose G. Magnitude
and causes of socioeconomic differentials in mortality: further evidence from the Whitehall Study.
J Epidemiol Community Health. 1990;44:265-270.
19. Winkleby MA, Fortmann SP, Barrett DC. Social class disparities in risk factors for disease: eightyear prevalence patterns by level of education. Prev
20. Osler M. Social class and health behavior in Danish adults: a longitudinal study. Public Health. 1993;
21. Wagenknecht LE, Perkins LL, Cutler GR, et al.
Cigarette smoking is strongly related to educational
status: the CARDIA Study. Prev Med. 1990;19:158169.
22. Liu K, Cedres LB, Stamler J, et al. Relationship of education to major risk factors and death from
coronary heart disease, cardiovascular diseases, and
all causes. Circulation. 1982;66:1308-1314.
23. Krieger N, Rowley DL, Herman AA, Avery B,
Phillips MT. Racism, sexism and social class: implications for studies of health, disease and well-being.
Am J Prev Med. 1993;9(suppl):82-122.
24. Mechanic D. Socioeconomic status and health: an explanation of underlying processes. In: Bunker JP,
Gomby DS, Kehrer BH, eds. Pathways to Health:
The Role of Social Factors. Menlo Park, Calif: Henry
J Kaiser Family Foundation; 1989:9-26.
25. Williams DR. Socioeconomic differentials in
health: a review and redirection. Soc Psychol Q.
26. Lynch JW, Kaplan GA, Cohen RD, et al. Do cardiovascular risk factors explain the relation between
socioeconomic status, risk of all-cause mortality, cardiovascular mortality, and acute myocardial infarction? Am J Epidemiol. 1996;144:934-942.
27. Rose G, Marmot MG. Social class and coronary
heart disease. Br Heart J. 1981;45:13-19.
28. Haan M, Kaplan G, Camacho T. Poverty and
health: prospective evidence from the Alameda
County Study. Am J Epidemiol. 1987;125:989-998.
29. Duijkers TJ, Kromhout D, Spruit IP, Doornbos
G. Intermediary risk factors in the relation between
socioeconomic status and 25-year mortality (the
Zutphen Study). Int J Epidemiol. 1989;18:658-662.
30. House JS, Lepkowski JM, Kinney AN, Mero RP,
et al. The social stratification of aging and health.
J Health Soc Behav. 1994;35:213-234.
31. Shyrock HS. The Methods and Materials of Demography. Orlando, Fla: Academic Press Inc; 1976.
32. Satel S. The politicization of public health. Wall
Street Journal. December 12, 1996;editorial section:
33. Antonosky A. Social class, life expectancy and
overall mortality. Milbank Q. 1967;45:31-73.
34. Williams DR, Collins C. US socioeconomic and
racial differences in health: patterns and explanations. Annu Rev Sociol. 1995;21:349-386.
35. Moore ME, Hayward MD. Occupational careers
and mortality of elderly men. Demography. 1990;27:
36. Riessman CK. The use of health services by the
poor. Soc Policy. 1974;5:41-49.
37. Dutton DB. Explaining the low use of health
services by the poor: costs, attitudes or delivery
systems? Am Soc Rev. 1978;43:348-368.
38. Lantz PM, Weigers ME, House JS. Education
and income differentials in breast and cervical cancer screening: policy implications for rural women.
Med Care. 1997;35:219-236.
39. Blendon R, Aiken L, Freeman H, Corey C. Access to medical care for black and white Americans.
40. Woolhandler S, Himmelstein DU, Silber R,
Bader M, et al. Medical care and mortality: racial
differences in preventable deaths. Int J Health Serv.
41. Blane D. Social determinants of health—socioeconomic status, social class and ethnicity. Am J
Public Health. 1995;85:903-905.
42. House JS, Williams DR. NHLBI Report of the
Conference on Socioeconomic Status and Cardiovascular Health and Disease: Psychosocial Pathways Linking SES and CVD. Bethesda, Md: National Institutes of Health; 1995:119-124.
43. Rodin J. Aging and health: effects of the sense of
control. Science. 1986;233:1271-1276.
44. Kessler RC, Neighbors HW. A new perspective
on the relationship between race, social class and
psychological distress. J Health Soc Behav. 1986;27:
45. Kaplan G, Pamuk E, Lynch JW, Cohen RD,
Balfour JL. Inequality in income and mortality in
the United States: analysis of mortality and potential pathways. BMJ. 1996;312:999-1003.
46. Lynch JW, Kaplan GA, Salonen JT. Why do poor
people behave poorly? variation in adult health behaviors and psychosocial characteristics by stages
of the socioeconomic lifecourse. Soc Sci Med. 1997;
47. Mirowsky J, Ross CE. Social Causes of Distress. New York, NY: Aldine de Gruyter; 1989.
48. Baranowski T. Methodologic issues in self-report
of health behavior. J Sch Health. 1985;55:179-182.
49. Cohen BB, Vinson DC. Retrospective self-report of alcohol consumption: test-retest reliability by
telephone. Alcohol Clin Exp Res. 1995;19:1156-1161.
50. Durante R, Ainsworth BE. The recall of physical activity: using a cognitive model of the questionanswering process. Med Sci Sports Exerc. 1996;28:
1282-1291. Socioeconomic Factors, Health Behaviors, and Mortality—Lantz et al ©1998 American Medical Association. All rights reserved.
Downloaded from www.jama.com at Oregon State University, on September 1, 2006 ...
View Full Document
This note was uploaded on 09/08/2010 for the course SOCI 166 at San Jose State University .
- The Land