| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
RESEARCH AND PRACTICE |
The authors are with the Johns Hopkins Bloomberg School of Public Health, Baltimore, Md. Nan Marie Astone is with the Department of Population and Family Health Sciences. Margaret Ensminger and Hee Soon Juon are with the Department of Health Policy and Management.
Correspondence: Requests for reprints should be sent to Nan Marie Astone, PhD, Department of Population and Family Health Sciences, Johns Hopkins Bloomberg School of Public Health, 615 N Wolfe St, Baltimore, MD 21205
(e-mail: nastone{at}jhsph.edu).
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMETHODSRESULTSDISCUSSIONReferences |
|---|
Objectives. This study examined predictors of longevity in a cohort of inner-city African American women.
Methods. Data were derived from a cohort study of inner-city African American mothers whose median age in 1966 was 31 years. Analyses involved single-decrement life tables and pooled logistic regression.
Results. Giving birth for the first time before age 25 and having at least a high school education predicted longevity in this sample. Effects of later age at first delivery in terms of mortality risk were stronger after 55 years and, especially, after 70 years.
Conclusions. The findings offer support for Geronimus's weathering hypothesis. Predictors of longevity among African Americans may be distinct from predictors for the population as a whole.
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONReferences |
|---|
Epidemiologists hypothesize that differences in mortality rates across populations are due to differences in the prevalence of individual-level risk factors across populations. Recent studies have identified a number of individual-level risk and protective factors associated with longevity. The longer lives of people from higher social class backgrounds have been well documented.1,6–8 In addition, other social background characteristics, such as family and household composition,1,9–11 have been shown to affect mortality. A theme in many recent studies of mortality is the protective effect of social networks and religious involvement.1,12–15
In general, introduction of controls for these individual-level risk factors in multivariate models of longevity has not eliminated the negative effect of being African American.1,16 One possible reason is that there are differences between African Americans and the groups with which they have been compared in terms of risk and protective factors for longevity. For example, Schoenbach and his colleagues14 found that social network ties promote longevity among European Americans but not African Americans.
We used data from a longitudinal study17 (the Woodlawn Project) to examine whether predictors of longevity that have demonstrable effects among European Americans have a significant impact among inner-city African American women.
|
METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONReferences |
|---|
Since the 1960s, the population of the community has declined dramatically (from 81 000 in 1960 to approximately 27 000 in 1990).17 Woodlawn may be typical of the inner-city areas described by Wilson.18 That is, housing segregation resulted in the community's being heterogeneous with regard to social class; with increasing residential integration, those who could afford to leave the inner city did (often moving to other neighborhoods in Chicago), leaving behind those without the resources to relocate.
The Woodlawn Project. In 1966, all children who attended first grade in Woodlawn's 12 schools (9 public and 3 parochial) were asked to participate in a research and intervention program. Of the 1255 children eligible for inclusion in the study, 1242 (99%) agreed to participate. In 1967, a mother or mother surrogate (hereafter "mothers") was interviewed for each child (1158 mothers and 84 mother surrogates completed interviews). Twice since the initial 1967 interview, the mothers have been followed up, once in 1976 and once in 1997–1998. The mothers eligible for inclusion in these 2 follow-ups were those who were neither known to be dead nor known to have moved away from the Chicago area in 1976. There were 1136 mothers in this group, 91% of those eligible to participate in the initial round of data collection. All of these 1136 mothers were eligible for inclusion in the present study sample; their median age in 1967 was 32 years.
Study sample. In each of the follow-up studies of the Woodlawn Project, the investigators obtained information on whether or not mothers were alive at the time of the follow-up. This included the 1992 follow-up of the Woodlawn children, during which the mothers were not interviewed (i.e., each Woodlawn child who participated in the 1992 survey was asked whether his or her mother was alive).
Table 1
shows that 884 (78%) of the 1136 mothers who were eligible for the 1976 follow-up were alive in 1997. Of the 252 reported to be dead, a date of death was available for 162, or almost two thirds. The 90 cases of deceased mothers for whom we had no date of death could be broken down into 3 groups. The first contained only one case of a mother who was not interviewed in 1976 and whose child was not interviewed in 1992. The second group contained 19 women who were reported as deceased during the locating phase of the 1997 interview, who were known to be alive in 1976, and whose child was not interviewed in 1992. Finally, the members of the third and largest group of 70 women were reported by their children to be alive in 1992 but were reported as deceased during the locating phase of the 1997 follow-up study.
|
In addition to the low attrition rate in our study sample, a compelling advantage of the Woodlawn data in regard to the present study was their prospective, longitudinal nature. All of our predictor variables were measured in 1967, the beginning of the study; most of the deaths observed occurred at least 15 years later. Thus, the problem of reverse causality was greatly minimized in our study.
To illustrate this point, consider the case of religious involvement and social participation and their putative effects on longevity. In studies in which the time lag between measurement of predictors and mortality is short, there is always the possibility that poor health or another factor endogenous to mortality is responsible for a person's disinclination to participate in social activities or to attend church, resulting in a spurious relationship between the predictor and the outcome. Although controlling for health status, which some studies are able to do, minimizes this problem, it does not completely eliminate it.
Predictors of Mortality
Demographic factors.
We examined a number of demographic factors to assess their impact on longevity. Some have argued that urban African Americans whose families were part of the large migration north in the 1950s and 1960s were disadvantaged (particularly regarding the quality of their education) relative to their counterparts who had been born in northern cities.19 Thus, we included birthplace (Chicago, the South, or elsewhere) as a variable, hypothesizing that Chicago-born mothers would live longer than others. Nontraditional familial arrangements are associated with higher risks of death,6–8 so we included household headship in 1966 (couple, mother only, or "other"), hypothesizing that mothers residing in married-couple households would live longer than others. We also included an indicator of number of moves during the previous 5 years (0 to 5 or more). Past research has shown high mobility to be associated with negative outcomes.20–22
Finally, we included age at which women first gave birth (younger than 25 years vs 25 years or older). There are competing hypotheses as to the effect of age at first delivery on mortality. For example, there is an extensive literature on the association between maternal age at first delivery and subsequent socioeconomic status. Contributors to this literature regard delaying first deliveries as positive and as reflective of an "orderly" life course.23 This suggests the hypothesis that a relatively early age at first delivery will be a risk factor for mortality.
Geronimus, however, suggested that African American women in their 20s are subject to higher levels of stress than both European Americans in the same age range and African Americans younger than 20.24 Noting that African American women who delay giving birth until their 20s have higher neonatal mortality rates, Geronimus proposed the "weathering hypothesis": in the case of African American women, owing to their uniquely disadvantaged social and economic position, the age range commonly regarded as normative for the transition to parenthood is not ideal.
Evidence for this hypothesis is the higher risk of ill health among the offspring of African American women who delay giving birth than among the offspring of African American women who become mothers for the first time at earlier ages. The assumption is that African American women giving birth for the first time at later ages are in poorer health than those doing so at earlier ages. On the basis of the weathering hypothesis, one would expect a later age at first delivery to be negatively associated with longevity among African American women.
Social class. We hypothesized that mothers from higher social class backgrounds would live longer. Our social class measures included education, income, and receipt of welfare. Indicators were as follows: educational attainment in 1966 (less than high school, high school or more), poverty status (poor, not poor), and welfare receipt in 1966 (yes, no).
Health status. It is ideal that studies of mortality include indicators of health status, because policymakers need to understand the mechanisms through which structural correlates of mortality operate proximately. Recent research indicates that, in some populations, health status early in life affects later outcomes, including longevity.25,26 We used 4 indicators of health status: 2 measures of physical health and 2 measures of mental health. The first physical health measure was a dichotomous dummy variable indicating whether the mother suffered from a chronic condition in 1966. The second was a dummy variable indicating whether the mother had health problems during the pregnancy with the focal child. The mental health indicators were dummy variables indicating whether the mother was tense and sad at the time of the baseline interview.
Religiosity. As mentioned earlier, one of the most robust findings in the literature on US mortality rate differentials is the protective effect of religious involvement on longevity. Many studies of religion and mortality are limited by the data available to measuring religiosity in terms of church attendance. It is difficult in these studies to know the factors for which church attendance is acting as a proxy (e.g., a temperate lifestyle, serenity stemming from religious commitment, the social support of an intentional community). Our indicator of religiosity was more direct, indicating the importance of religion to the mother (extremely, very, not important).
Social participation. Social participation and social network ties have been found to predict longevity.1,14,15 We used 4 indicators of social integration and participation. First, in separate items, mothers were asked how often they visited friends and relatives. These 2 items were combined, and the mothers were divided into 4 groups according to their score on this sum. Second, mothers were asked about PTA membership and involvement. Third, mothers were asked about their participation in up to 4 civil rights and political organizations (the 1960s were a time when there was much political activity in Chicago and in Woodlawn specifically). Finally, the mothers were asked whether they had voted in the most recent election.
Analytic Techniques
We used 2 analytic techniques. The first, a single-decrement life table, allowed us to examine whether levels and patterns of mortality in our study sample were similar to those observed in other, comparable samples.
The second technique we used was pooled logistic regression. In this analysis, log odds of dying at a given age were regressed linearly on a set of covariates. All relevant years were included in each analysis (n = 57 644). This approach allowed us to use all of the information available to us, to incorporate age as a time-varying covariate, and to examine whether predictors of longevity have different effects across the life course.27–29
In our presentation of the results, all of the effects we report are net of the effects of age (linear and quadratic), even in the case of effects labeled "unadjusted."
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONReferences |
|---|
, we present our estimates of the life table quantity q (one's probability of dying during the interval from age x to age x + 1, given that one was alive at exact age x), bounded by the confidence intervals of the estimates; also, we present similar values from the 1990 life table for female African American residents of Illinois. For ease of presentation, we have limited the graph to the ages of 55 through 77 years (death rates at earlier ages were very low).
|
Predictors of Longevity
Table 2 presents (1) the distribution of the predictor variables; (2) "unadjusted" odds ratios (i.e., adjusted for age only), along with their associated confidence intervals, for probability of dying in a given year from the pooled logistic regression models; and (3) odds ratios for each predictor after adjustment for the effects of all of the other predictors. The only social or background factor significantly associated with longevity was age at which a woman first gave birth. Women who delayed childbearing until they were 25 years or older had higher death rates than women who gave birth at younger ages. This effect persisted after control for the other predictors, as can be seen in the third column of Table 2.
|
Before adjustment for other predictors, mothers who reported a chronic condition in 1966 had higher odds of dying than those who did not, and this association narrowly failed a test of statistical significance (P = .08). None of the other indicators of health status were significantly associated with mortality. Also, none of the indicators of religiosity or social participation exhibited associations with mortality that reached statistical significance, before or after adjustment for other predictors.
None of the findings in Table 2
were substantially different when we conducted the regressions with the sample of 1049 women who were alive or had valid dates of death. On the basis of the findings shown in Table 2, we estimated a parsimonious model of mortality. This model included all variables that had a significant unadjusted effect on mortality, as well as the indicator of chronic conditions, which we included because of the importance of controlling for health status in models of mortality. These results are presented in Table 3 alongside linear and quadratic effects of age, which were highly significant. Neither poverty status nor the presence of chronic conditions had a significant effect on mortality in the parsimonious model.
|
, we report the results of the only significant interaction we found: that between current age and the age at which women first gave birth. As can be seen in Table 4, giving birth for the first time at 25 years or older was associated with a lower risk of death while women were younger than 55 years. Between the ages of 55 and 70 years, however, mothers who had given birth for the first time at 25 years or older had more than twice the odds of dying, and after the age of 70 years their risk was much higher than that among women who gave birth for the first time at younger ages.
|
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONReferences |
|---|
With the exception of our findings regarding social class, our results in general are not in accord with those observed in the recent literature on mortality rate differentials, which identifies social participation and religiosity as important factors promoting longevity among Americans. In our population, these predictors were not associated with mortality, even when no adjustment was made for other predictors. Many of the studies assessing religiosity and social participation as risk factors for mortality have involved European American populations or large, nationally representative samples in which European Americans represent the majority of the subjects. The one study that looked for interactions between African American ethnicity and religiosity or social participation revealed no effects.
Our findings underscore the possibility that there may be interactions such as those just described, particularly in the case of predictors that reflect the influence of institutional context and involvement. Residential and social segregation result in de facto segregation in American institutions such as churches, neighborhood organizations, and political groups. Thus, the meanings and prevalence of involvement in these institutions vary greatly among population subgroups. There is every reason, therefore, to believe that involvement in such groups may have different effects on health and well-being among different groups.
|
Acknowledgments |
|---|
We are grateful to Richard Cain for computational assistance and to Robert Schoen, Kenneth Hill, and Young Kim for helpful comments.
|
Footnotes |
|---|
Accepted for publication December 19, 2001.
|
References |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONReferences |
|---|
2.
Geronimus AT, Bound J, Waidmann TA, Hillemeier MM, Burns PB. Excess mortality among blacks and whites in the United States. N Engl J Med.1996;335:1552–1558.
3. Laveist TA. Segregation, poverty, and empowerment: health consequences for African Americans. Milbank Q.1993;71:41–64.[Medline]
4. Guest AM, Almgren G, Hussey JM. The ecology of race and socioeconomic distress: infant and working-age mortality in Chicago. Demography.1998;35:23–34.[Medline]
5. Elo IT, Preston SH. Estimating African-American mortality from inaccurate data. Demography.1994;31:427–458.[Medline]
6. Hemstrom O. Is marriage dissolution linked to differences in mortality risks for men and women? J Marriage Fam.1996;58:366–378.
7. Lillard LA, Panis CW. Marital status and mortality: the role of health. Demography.1996;33:313–327.[Medline]
8. Tucker JS, Friedman HS, Wingard DL, Schwartz JE. Marital history at midlife as a predictor of longevity: alternative explanations to the protective effect of marriage. Health Psychol.1996;15:94–101.[Medline]
9. Marmot M. Social inequalities in mortality: the social environment. In: Wilkinson R, ed. Class and Health: Research and Longitudinal Data. London, England: Tavistock; 1986:21–33.
10. Kitagawa E, Hauser PM. Differential Mortality in the United States: A Study in Socioeconomic Epidemiology. Cambridge, Mass: Harvard University Press; 1973.
11.
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.
12. McCullough ME, Hoyt WT, Larson DB, Koenig HG, Thoresen C. Religious involvement and mortality: a meta-analytic review. Health Psychol.2000;19:211–222.[Medline]
13.
Oman D, Reed D. Religion and mortality among the community-dwelling elderly. Am J Public Health.1998;88:1469–1475.
14.
Schoenbach VJ, Kaplan BH, Fredman L, Kleinbaum DG. Social ties and mortality in Evans County, Georgia. Am J Epidemiol.1986;123:577–591.
15.
Berkman LF, Syme SL. Social networks, host resistance, and mortality: a nine-year follow-up study of Alameda County residents. Am J Epidemiol.1979;109:186–204.
16. 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;9:82–122.[Medline]
17. Ensminger ME, Juon HS. The influence of patterns of welfare receipt during the child-rearing years on later physical and psychological health. Women Health.2001;32:25–46.[Medline]
18. Wilson WJ. When Work Disappears: The World of the New Urban Poor. New York, NY: Random House; 1996.
19. Lieberson S. A Piece of the Pie: Blacks and White Immigrants Since 1880. Berkeley, Calif: University of California Press; 1980.
20. Juon HS, Ensminger ME. Childhood, adolescent, and young adult predictors of suicidal behaviors: a prospective study of African Americans. J Child Psychol Psychiatry.1997;38:553–563.[Medline]
21. Myers SM. Childhood migration and social integration in adulthood. J Marriage Fam.1999;61:774–789.
22. Pribesh S, Downey DB. Why are residential and school moves associated with poor school performance? Demography.1999;36:521–534.[Medline]
23. Jones AS, Astone NM, Keyl P, Kim YJ, Alexander CS. Teen childbearing and educational attainment: a comparison of methods. J Fam Econ Issues.1999;20:387–418.
24. Geronimus AT. The weathering hypothesis and the health of African American women and infants: implications for reproductive strategies and policy analysis. In: Sen G, Snow RC, eds. Power and Decision: The Social Control of Reproduction. Cambridge, Mass: Harvard University Press; 1994:75–100.
25.
Schwartz JE, Friedman HS, Tucker JS, Tomlinson-Keasey C, Wingard DL, Criqui MH. Sociodemographic and psychosocial factors in childhood as predictors of adult mortality. Am J Public Health.1995;85:1237–1245.
26. Lamont DW, Parker L, Cohen MA, et al. Early life and later determinants of adult disease: a 50 year follow-up study of the Newcastle Thousand Families cohort. Public Health.1998;112:85–93.[Medline]
27. Yamaguchi K. Event History Analysis. Newbury Park, Calif: Sage Publications; 1991.
28. Abbott R. Logistic regression in survival analysis. Am J Epidemiol.1985;212:465–471.
29. D'Agostino RB, Lee ML, Belanger AJ, Cupples LA, Anderson K, Kannel WB. Relation of pooled logistic regression to Cox regression analysis: the Framingham Heart Study. Stat Med.1990;9:1501–1515.[Medline]
30. Armstrong R. US Decennial Life Tables for 1989–91: Volume II, State Life Tables Number 14, Illinois. Washington, DC: US Dept of Health and Human Services; 1998.
This article has been cited by other articles:
|
|
 |
|
  C. G. Colen, A. T. Geronimus, J. Bound, and S. A. James Maternal Upward Socioeconomic Mobility and Black-White Disparities in Infant Birthweight Am J Public Health, November 1, 2006; 96(11): 2032 - 2039. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
A. T. Geronimus, M. Hicken, D. Keene, and J. Bound "Weathering" and Age Patterns of Allostatic Load Scores Among Blacks and Whites in the United States Am J Public Health, May 1, 2006; 96(5): 826 - 833. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 J. Warren-Findlow Weathering: Stress and Heart Disease in African American Women Living in Chicago Qual Health Res, February 1, 2006; 16(2): 221 - 237. [Abstract] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
