Assessing the 10/66 Dementia Classification Algorithm for International Comparative Analyses with the U.S.

Abstract

Cross-national comparisons of dementia prevalence are essential for identifying unique determinants and cultural-specific risk factors, but methodological differences in dementia classification across countries hinder global comparisons. This study maps the 10/66 algorithm for dementia classification, widely used and validated in low- and middle-income countries (LMICs), to the U.S. Aging, Demographics, and Memory Study (ADAMS), the dementia sub-study of the Health and Retirement Study, and assesses its performance in ADAMS. We identified the subset of 10/66 algorithm items comparably measured in ADAMS, then used these items to re-train the 10/66 algorithm against the ADAMS clinical dementia diagnosis, employing k-fold cross-validation to assess performance. We compared the modified 10/66 algorithm to four other dementia classification algorithms previously validated in ADAMS, both for overall dementia estimation as well as for estimating education gradients. The modified 10/66 algorithm had higher sensitivity (87%) and specificity (93%) than the comparison algorithms. All of the algorithms over-estimated the education gradient in dementia, although the modest ADAMS sample size precludes precise comparisons of education gradient accuracy. Overall, we found that the modified 10/66 algorithm performs well in classifying dementia status in the U.S. Our results support the validity of risk factor comparisons between U.S. and 10/66 LMIC dementia datasets.