Assessment of cognition in American Indians poses challenges, including barriers to healthcare, unvalidated clinical standards, and confounding social determinants of health. Alternative strategies for case identification include algorithmic methods, which can outperform clinical judgment in some circumstances. Algorithmic methods can be maximized using single-domain tests with multiple-serial trial tasks, such as the California Verbal Learning Test II-Short Form (CVLT-SF). We collected CVLT-SF and detailed clinical data, including dementia gold standard by consensus adjudication, in 818 American Indians aged 65-95 in 2010-2013 and repeated in 403 returning participants in 2017-2019 (mean follow-up 6.7 years, range: 4-9). Our algorithm categorized CVLT-SF scores into four memory deficit categories: none, encoding, storage, and retrieval. At Visit 1, 75.4% had no memory deficit, 15.6% encoding deficit, 3.5% storage deficit, and 5.5% retrieval deficit. At Visit 2, comparable percentages were 68.7%, 10.6%, 6.5%, and 14.2% (respectively). The majority with any deficit at Visit 1-especially encoding-were lost to follow-up by Visit 2. Most with deficits at Visit 2 were newly categorized from those previously intact. The performance of our memory algorithm, compared with adjudication for dementia, was moderately good: correct classification 69%, sensitivity 51%, and specificity 91%. These descriptive findings encompass a novel contribution in defining memory impairment among American Indians from a single cognitive test. However, more work is needed to improve the sensitivity of this algorithm and maximize its utility for case identification over conventional methods. Altogether, these data provide an important step toward better cognitive characterization and dementia care for an understudied, underserved population. (PsycInfo Database Record (c) 2024 APA, all rights reserved).
Read full abstract