Phantom of the Area: Poverty-Area Residence and Mortality in the United States Norman J. Waitzman, PhD, and Ken R. Smith, PhD Introduction The inverse gradient between individ- ual and household socioeconomic status and mortality is well established.'3 Recent work has suggested that this gradient may be partly attributable to areal characteristics.4i6 A study of 1811 Oakland, Calif, residents aged 35 years and older who were followed up from 1965 to 1974 found that residence in a federally designated poverty area had a significant association with all-cause mortal- ity, even after selective adjustments for age, race, baseline measures of individual health behaviors, certain physiological measures, and individual and household socioeco- 4 nomic and demographic characteristics. Contradictory evidence, however, surfaced from a study in 1981 of residents aged 16 through 65 years in poorer wards in England who were followed up through 1989.7 To date, no national-level analysis has been performed of poverty-area residence and mortality in the United States, nor have national-level analyses addressed a period of growing socioeconomic disparities in mor- tality3 or in specific causes of death. Such extensions would determine whether the association uncovered in Oakland applies nationally in similarly designated locales; these refinements might also provide insight as to the differences in findings between the English study and the US studies. We per- formed such an analysis, using a nationally representative sample of civilian, noninstitu- tionalized adults in the United States exam- ined in the early 1970s and followed through 1987. Methods The data for our analysis were taken from the first National Health and Nutrition Examination Survey (NHANES I), con- ducted between 1971 and 1975, and the 1987 NHANES I Epidemiologic Follow-Up Survey. The NHANES I was the second in a series of surveys that constitute the largest nationally representative surveys to include both a medical exam by a physician and an extensive interview. The follow-up survey is the second and latest available follow-up survey of adults aged 25 through 74 years who completed the NHANES I medical examination (n= 14407). The first-stage sur- vey design of the NHANES 1 (1971-1974) included an oversampling of respondents in poverty areas (n= 11 348). Our study sample included only adult respondents from this sample who were either White or Black and for whom there were complete covariate data and information on vital status after the baseline interview (n= 10161). We performed Cox proportional hazards analyses to measure the effect of poverty-area residence separately on all-cause mortality and mortality due to cardiovascular disease, cancers, extemal causes, and all other causes. The proportional hazards assumption was found to be consistent with the data. Given the well-documented diminution of the effect of socioeconomic status on mortality with age,8 we included an interaction term between age and poverty-area indicators. The results shown in Table 1 are from the interac- tion models, based on the significance of the age-poverty-area interaction. The age split (25-54 years or 55+ years at baseline) was based on an earlier analysis of the data that resulted in a division of the sample into 3 age cohorts9 and on our determination that the middle age interval could be broken evenly between the younger and older cohorts with- out substantive changes to the results. The regression results (Tables 1 [last column] and 2) and crude rate ratios (Table 3) reflect both the incorporation of sample weights and the adjustment for the complex, multistage design effect of the NHANES survey, which was performed with SUDAAN, Version 6.4 (Research Triangle Institute, Research Trian- gle Park, NC). The fully adjusted model (Table 1) included as covariates baseline measures of household income as a percentage of the Norman J. Waitzman is with the Department of Economics and Ken R. Smith is with the Depart- ment of Family and Consumer Studies, University of Utah, Salt Lake City. Requests for reprints should be sent to Nor- man J. Waitzman, PhD, Department of Economics, University of Utah, 1645 E Central Campus Dr-Front, Salt Lake City, UT 84112-9300. This paper was accepted July 17, 1997. American Journal of Public Health 973