Analyses of black and white differentials in the age trajectory of mortality in two closed cohort studies

We examine the relationship of age to mortality in blacks and whites in two cohort studies, the 20-year follow-up of the Evans County, Georgia. Study population and the 25-year follow-up of the Charleston Heart Study population. We conducted analyses with two parametric forms of hazard models (Gompertz and Weibull) for total mortality experience and considered the fit of the two hazard models in each study both separately and with the data pooled. We evaluated the robustness of conclusions about differences in the age pattern of mortality for blacks and whites by comparing results from the two hazard models. Where the tests were non-nested, we used an information-based statistic (AIC) to compare the fit of the models to the data. Results were robust to the selection of the hazard function, that is both models provided evidence that mortality rates at younger ages were lower for white than black females but that mortality rates increased more rapidly with age for white females. The absolute differences and the differences in the rates of increase were in the same direction, but smaller, for males. Though both models represented the general features of mortality patterns, the information statistic suggested better performance of the Gompertz function. The Gompertz function was also less sensitive to the initial age of determination of mortality exposure.