Abstract
Alzheimer's disease and related dementias (ADRD) are a growing public health crisis. Estimates on the prevalence and incidence of ADRD across and within population-based studies have varied in part due to competing measures to assess dementia status. Disentangling these inconsistencies is crucial for understanding dementia disparities among racial/ethnic, and nativity groups among older adults. Based on the Health and Retirement Study we examined across (Whites, Blacks) and within-group differences (US- and non-US-born Latinos) in estimates of dementia life expectancy, using four competing algorithmic techniques (i.e., the Langa-Weir, Expert, Hurd, and Lasso) for the classification of dementia ascertainment. Estimates of dementia life expectancy across algorithms largely point to dementia disparities in the prevalence of the disease across racial/ethnic, and nativity groups, regardless of the algorithmic technique utilized. Elucidating algorithms that can be utilized with different racial/ethnic groups may reduce bias in dementia assessment in the future.