Diabetes is a risk factor for dementia, but relatively little is known about the epidemiology of the association. A retrospective population study using Western Australian hospital inpatient, mental health outpatient, and death records was used to compare the age at index dementia record (proxy for onset age) and survival outcomes in dementia patients with and without preexisting diabetes (n = 25,006; diabetes, 17.3%). Inpatient records from 1970 determined diabetes history in this study population with incident dementia in years 1990-2005. Dementia onset and death occurred an average 2.2 years and 2.6 years earlier, respectively, in diabetic compared with nondiabetic patients. Age-specific mortality rates were increased in patients with diabetes. In an adjusted proportional hazard model, the death rate was increased with long-duration diabetes, particularly with early age onset dementia. In dementia diagnosed before age 65 years, those with a ≥15-year history of diabetes died almost twice as fast as those without diabetes (hazard ratio = 1.9, 95% confidence interval: 1.3, 2.9). These results suggest that, in patients with diabetes, dementia onset occurs on average 2 years early and survival outcomes are generally poorer. The effect of diabetes on onset, survival, and mortality is greatest when diabetes develops before middle age and after 15 years' diabetes duration. The impact of diabetes on dementia becomes progressively attenuated in older age groups.
Keywords: Alzheimer disease; dementia; diabetes mellitus; mortality; proportional hazards models; retrospective studies; survival.