Abstract
STUDY OBJECTIVE: The aim was to compare four different measures of women's social class (interview and census based) as predictors of well known social gradients in reproductive outcomes. The intent was to address two obstacles that confront research in the United States regarding social gradients in women's health: how women's social class should be measured, and the absence of socioeconomic data in most health records. STUDY DESIGN: The study was a retrospective cohort analysis, using a community based random sample. Setting--Alameda County, California, 1987. SUBJECTS: 51 black and 50 white women, ages 20 to 80 years, identified by random digit dialling, were interviewed by telephone. MEASUREMENTS AND MAIN RESULTS: Census data were linked to individual records via the respondents' addresses. Using number of full term pregnancies as an example, multiple linear regression analyses demonstrated that individual class was not significantly associated with this outcome, whereas household class was: women from non-working-class households had 0.8 fewer such pregnancies than women from working class households (95% confidence interval [CI] = -1.4, -0.1). The block group measure functioned most like the household class measure (beta = -0.7, 95% CI = -1.4, 0.1), while the census tract measure was non-significant (beta = -0.4, 95% CI = -1.2, 0.4). Similar results were obtained for the outcomes: age at first full term pregnancy, percent of early terminated pregnancies, and yearly income. CONCLUSIONS: These results suggest block group data may offer a uniform source of social class information that can be appended to individual health records, and that this strategy is not invalidated by concerns regarding ecological fallacy.