346
EPIDEMIOLOGY
Black-white
mortality differences by family income
Death rates among US black men and women under 75 years of age are higher than for their white counterparts. The explanation for this excess risk, though attributed to socioeconomic factors, remains unclear. We calculated mortality rates by family income for blacks and whites in a respresentative sample of the US population (National Longitudinal Mortality Study). For persons aged less than 65 years of age, mortality rates are lower in those with higher family income for both blacks and whites, and both men and women. However, at each level of income, blacks have higher mortality than whites. Higher levels of family income are also associated with lower death rates from cardiovascular disease, cancer, and deaths from causes other than cardiovascular disease or cancer. After adjustment for income, blacks have higher death rates from each of these three general causes. For subjects below 65 years, the mortality gradient by income is larger than the gradient by race. The differences in mortality rates by race not accounted for by income may be due to other differences such as access to health care, type or quality of medical care, or behavioural risk factors that disadvantage black populations.
In a large representative sample of the US noninstitutionalised population, annual family income and other socioeconomic characteristics were measured by interviewers.8 Follow-up of the sample was done with the national death index for the years 1979-85. Mortality rates by income were obtained and compared using over 2 million person-years of experience for whites and 200 000 personyears for blacks aged 25 years or more. This study, the National Longitudinal Mortality Study (NLMS), provides an opportunity to study in detail the relation between income and mortality in blacks and whites, and provides an estimate of the magnitude of the excess black mortality that can be accounted for by differences in income. The validity and precision of the findings from the NLMS are enhanced by its large population size and its representation of the non-institutionalised US population. Recent attention has focused on the importance of delineating the differences between race and socioeconomic class in relation to health outcomes.9,10 In the USA, apart from the older Kitagawa and Hauser report that used 1960 census data," only small studies have investigated this issue. The NLMS provides a large national database capable of showing the profound effect that socioeconomic status has on
mortality.
Subjects and
sample population for the NLMS was selected from the current population survey (CPS).12 The CPS is a complex sample of households that are surveyed monthly to obtain demographic, economic, and social information about the US population with an emphasis on employment, unemployment, and The
census
Introduction for blacks is higher than for whites for both men and women less than 75 years of age.1 In older age groups, death rates for blacks are the same or less than for whites. The reasons for the excess in those less than age 75 have not been identified, though disparities in socioeconomic status and opportunity are likely to be important. Persons residing in low-income areas, or with low family incomes, have higher rates of disease prevalence and severity,2,3 are more likely to be admitted to hospital for medical reasons,4and have higher mortality rates3,5-7 than persons with higher socioeconomic status. The explanation for this excess risk generally involves patterns in health-risk behaviours and opportunities for primary and secondary medical care. In the
USA, the death
rate
methods
bureau’s
other labour force characteristics. The survey is used by the Bureau of Labor Statistics to prepare monthly estimates of unemployment rate and is conducted by personal and telephone interview, with a response rate close to 96%. For the current length of follow-up, the NLMS has identified 10 CPS surveys for mortality assessment using the national death index. The surveys chosen were done in ADDRESSES National Institutes of Health, National Heart, Lung, and Blood Institute, Bethesda, Maryland (P. Sorlie, PhD, E. Rogot, MA); Bowman Gray School of Medicine, Department of Public Health Sciences, Section on Epidemiology, Winston-Salem, North Carolina (R. Anderson, PhD); Bureau of the Census, Suitland, Maryland (N. J. Johnson, PhD, E. Backlund, MS) Correspondence to Dr Paul D Sorlie, National Heart, Lung, and Blood Institute, Federal Building, Room 3A10, Bethesda, Maryland 20892, USA
347
TABLE I-DEATHS IN THE NATIONAL LONGITUDINAL MORTALITY STUDY*
TABLE II-% POPULATION WITH ANNUAL FAMILY INCOMES LESS THAN$10 000* Men
*10 surveys, 1979-85
a
total combined income of all members of the
respondent’s family. The income from the various samples was adjusted to 1980 levels by use of the consumer-price index. Income was divided into categories based on the groupings stated in the CPS interview. Data are not presented for the highest income in blacks because there were fewer than 5 deaths in this category for each age-sex group. While there are many measures of social and economic status, we have chosen family income as the primary measure for this analysis. It represents current status and reflects the financially determined opportunities for adequate housing, nutrition, and medical care. Each survey in the NLMS was matched to the national death index of the National Center for Health Statistics. 13 The national death index is a computer file of all deaths in the USA since 1979.
Annual Income Black Men
($1000 in 1980)
Annual Income
-+-- White Men
($1000 in 1980)
Black Women
Fig 1-All-cause death Left-hand
follow-up
Data records with up to 14 personal identifying items are submitted to the national death index and possible matches are produced based on 12 criteria established by the National Center for Health Statistics. Identification of correct matches and false positives is made by a careful review of the matching items. The national death index is an effective and accurate means of ascertaining deaths according to personal identifiers.14,15 Except for the younger age group (< 15 years), mortality rates for the NLMS are consistent with estimates from the non-institutional population of the USA, were indicating that only a small proportion of deaths (
EPIDEMIOLOGY
Black-white
mortality differences by family income
Death rates among US black men and women under 75 years of age are higher than for their white counterparts. The explanation for this excess risk, though attributed to socioeconomic factors, remains unclear. We calculated mortality rates by family income for blacks and whites in a respresentative sample of the US population (National Longitudinal Mortality Study). For persons aged less than 65 years of age, mortality rates are lower in those with higher family income for both blacks and whites, and both men and women. However, at each level of income, blacks have higher mortality than whites. Higher levels of family income are also associated with lower death rates from cardiovascular disease, cancer, and deaths from causes other than cardiovascular disease or cancer. After adjustment for income, blacks have higher death rates from each of these three general causes. For subjects below 65 years, the mortality gradient by income is larger than the gradient by race. The differences in mortality rates by race not accounted for by income may be due to other differences such as access to health care, type or quality of medical care, or behavioural risk factors that disadvantage black populations.
In a large representative sample of the US noninstitutionalised population, annual family income and other socioeconomic characteristics were measured by interviewers.8 Follow-up of the sample was done with the national death index for the years 1979-85. Mortality rates by income were obtained and compared using over 2 million person-years of experience for whites and 200 000 personyears for blacks aged 25 years or more. This study, the National Longitudinal Mortality Study (NLMS), provides an opportunity to study in detail the relation between income and mortality in blacks and whites, and provides an estimate of the magnitude of the excess black mortality that can be accounted for by differences in income. The validity and precision of the findings from the NLMS are enhanced by its large population size and its representation of the non-institutionalised US population. Recent attention has focused on the importance of delineating the differences between race and socioeconomic class in relation to health outcomes.9,10 In the USA, apart from the older Kitagawa and Hauser report that used 1960 census data," only small studies have investigated this issue. The NLMS provides a large national database capable of showing the profound effect that socioeconomic status has on
mortality.
Subjects and
sample population for the NLMS was selected from the current population survey (CPS).12 The CPS is a complex sample of households that are surveyed monthly to obtain demographic, economic, and social information about the US population with an emphasis on employment, unemployment, and The
census
Introduction for blacks is higher than for whites for both men and women less than 75 years of age.1 In older age groups, death rates for blacks are the same or less than for whites. The reasons for the excess in those less than age 75 have not been identified, though disparities in socioeconomic status and opportunity are likely to be important. Persons residing in low-income areas, or with low family incomes, have higher rates of disease prevalence and severity,2,3 are more likely to be admitted to hospital for medical reasons,4and have higher mortality rates3,5-7 than persons with higher socioeconomic status. The explanation for this excess risk generally involves patterns in health-risk behaviours and opportunities for primary and secondary medical care. In the
USA, the death
rate
methods
bureau’s
other labour force characteristics. The survey is used by the Bureau of Labor Statistics to prepare monthly estimates of unemployment rate and is conducted by personal and telephone interview, with a response rate close to 96%. For the current length of follow-up, the NLMS has identified 10 CPS surveys for mortality assessment using the national death index. The surveys chosen were done in ADDRESSES National Institutes of Health, National Heart, Lung, and Blood Institute, Bethesda, Maryland (P. Sorlie, PhD, E. Rogot, MA); Bowman Gray School of Medicine, Department of Public Health Sciences, Section on Epidemiology, Winston-Salem, North Carolina (R. Anderson, PhD); Bureau of the Census, Suitland, Maryland (N. J. Johnson, PhD, E. Backlund, MS) Correspondence to Dr Paul D Sorlie, National Heart, Lung, and Blood Institute, Federal Building, Room 3A10, Bethesda, Maryland 20892, USA
347
TABLE I-DEATHS IN THE NATIONAL LONGITUDINAL MORTALITY STUDY*
TABLE II-% POPULATION WITH ANNUAL FAMILY INCOMES LESS THAN$10 000* Men
*10 surveys, 1979-85
a
total combined income of all members of the
respondent’s family. The income from the various samples was adjusted to 1980 levels by use of the consumer-price index. Income was divided into categories based on the groupings stated in the CPS interview. Data are not presented for the highest income in blacks because there were fewer than 5 deaths in this category for each age-sex group. While there are many measures of social and economic status, we have chosen family income as the primary measure for this analysis. It represents current status and reflects the financially determined opportunities for adequate housing, nutrition, and medical care. Each survey in the NLMS was matched to the national death index of the National Center for Health Statistics. 13 The national death index is a computer file of all deaths in the USA since 1979.
Annual Income Black Men
($1000 in 1980)
Annual Income
-+-- White Men
($1000 in 1980)
Black Women
Fig 1-All-cause death Left-hand
follow-up
Data records with up to 14 personal identifying items are submitted to the national death index and possible matches are produced based on 12 criteria established by the National Center for Health Statistics. Identification of correct matches and false positives is made by a careful review of the matching items. The national death index is an effective and accurate means of ascertaining deaths according to personal identifiers.14,15 Except for the younger age group (< 15 years), mortality rates for the NLMS are consistent with estimates from the non-institutional population of the USA, were indicating that only a small proportion of deaths (
