The use of mortality time series data to produce hypothetical morbidity distributions and projects mortality trends

It is difficult to obtain direct empirical estimates of chronic disease prevalence in the U.S. population. The available estimates are usually derived from epidemiological studies of selected populations. In this paper we present strategies for estimating morbidity distributions in the national population using auxiliary biomedical evidence and theory to estimate transitions to morbidity states from a cohort mortality time series. We present computational methods which employ these estimates of morbid state transitions to produce life table functions for both primary (morbidity) and secondary (mortality) decrements. These methods are illustrated using data on stomach cancer mortality for nine white male cohorts, aged 30 to 70 in 1950, observed for a 28-year period (1950 to 1977).