Search

Objectives: Sociodemographic trends in the United States may influence future dementia-associated mortality, yet there is little evidence about their potential impact. Our study objective was to estimate the effect of dementia on survival in adults stratified by sex, education, and marital status. Methods: Using survey data from the Health and Retirement Study (HRS) linked to Medicare claims from 1991 to 2012, we identified a retrospective cohort of adults with at least one International Classification of Diseases—ninth revision—Clinical Modification (ICD-9-CM) dementia diagnosis code ( n = 3,714). For each case, we randomly selected up to five comparators, matching on sex, birth year, education, and HRS entry year ( n = 9,531), and assigned comparators the diagnosis date of their matched case. Participants were followed for up to 60 months following diagnosis. We estimated a survival function for the entire study population and then within successive strata defined by sex, education, and marital status. Results: On average, dementia cases were 80.5 years old at diagnosis. Most were female, had less than college-level education, and approximately 40% were married at diagnosis. In multivariate analyses, dementia diagnosis was associated with earlier mortality for women (predicted median survival of 54.5 months vs. 62.5 months; dementia coefficient = −0.13; 95% confidence interval [CI] = [−0.22, −0.04]; p = .003), but even more so among men (predicted median survival of 35.5 months vs. 54.5 months; dementia coefficient = −0.42; 95% CI = [−0.52, −0.31]; p < .001). We found substantial heterogeneity in the relationship between dementia and survival, associated with both education and marital status. Conclusion: Both sex and level of education moderate the relationship between dementia diagnosis and length of survival.