Socioeconomic mortality differentials at ages 95 and higher in the United States. An application of the extinct-cohort method

Roland Rau, University of Rostock
Magdalena Muszynska, Duke University

Our paper deals with the question whether socioeconomic differentials in mortality exist also at ages 95 and higher in the United States. To answer this question, we extracted death counts by age, sex, year of death and level of education from NCHS's multiple cause of death data. The corresponding population data to compute age-specific mortality and life-tables have been estimated with the same data using the extinct-cohort method. As expected, women could expect to live slightly longer than men at age 95 (~0.5 years). The differences between socioeconomic groups as measured by the variable ``education'' were about 2.5 months in remaining life expectancy at age 95. Surprisingly, we could not detect a social gradient. Also plotting age-specific mortality resulted in the same outcome: higher socioeconomic status does not translate to lower mortality at ages 95 and higher in the United States. Our results support the hypothesis of converging mortality differentials with increasing age.