Timing and duration of diabetes and incident dementia in older adults: a prospective cohort study

Abstract

Diabetes is associated with dementia, but how this association varies with attained age and cumulative exposure remains unclear. We integrated time-updated exposure histories with repeated age-specific landmark cohorts, enabling complementary evaluation of overall and age-conditional dementia risk. We analysed 18,028 dementia-free participants aged 50 years or older from the Health and Retirement Study. Incident dementia was classified using the Langa–Weir algorithm. Time-updated Cox models evaluated risk according to current diabetes status. Age-anchored landmark models used delayed entry at six prespecified ages from 50 to 75 years. Formal age interactions and competing mortality were also assessed. Diabetes was associated with higher dementia risk in the time-updated analysis (HR, 1.29; 95% CI, 1.19–1.41). The association attenuated with older attained age but remained evident across the landmark cohorts. An estimated interval exceeding 10 years since first diabetes ascertainment identified groups at higher risk. The findings remained consistent after accounting for competing death. Diabetes was associated with dementia throughout later life, with stronger associations at younger attained ages and among participants with a longer estimated interval since first ascertainment. These observational findings support considering the timing of diabetes ascertainment in cognitive risk assessment.