Alzheimer's Disease Risk In Asian American Elders: A Classification and Regression Tree Approach

Background and Purpose: Alzheimer’s disease (AD) represents a growing public health concern, particularly among aging populations in the United States. Despite increasing recognition of racial and ethnic disparities in AD prevalence and diagnosis, Asian American older adults remain understudied, with limited culturally specific evidence on risk factors and disease patterns. Existing research has largely aggregated diverse populations, obscuring subgroup-specific mechanisms. This study addresses this gap by examining the complex interplay of socio-demographic characteristics, cardiovascular risk factors, sleep conditions, and antidepressant use in relation to AD among Asian American elders. The primary objective was to identify key predictors and their interactions using a Classification and Regression Tree (CART) approach. The study sought to answer: Which factors and combinations of factors are most strongly associated with AD risk in this population? It was hypothesized that functional status, age, and health-related conditions would emerge as primary determinants of AD risk.

Methods: This study utilized a cross-sectional design based on secondary data from the 2017 Uniform Data Set of the National Alzheimer’s Coordinating Center (NACC), including 4,343 Asian American older adults. Participants were drawn from multiple Alzheimer’s Disease Research Centers across the United States. The dependent variable was AD diagnosis, determined through standardized clinical assessments. Independent variables included socio-demographic factors (age, sex, education, independence in daily living), cardiovascular risk indicators (smoking history, body mass index [BMI], stroke history), and health conditions (apnea and antidepressant use). Bivariate analyses were conducted using chi-square tests and independent samples t-tests to identify significant associations. Multivariate analysis employed the CART method using Minitab 23, with the Gini index for node splitting and 10-fold cross-validation to optimize model performance. Model evaluation included metrics such as area under the curve (AUC), sensitivity, and specificity.

Results: Bivariate analyses revealed significant associations between AD and several variables, including age, education, independence level, antidepressant use, smoking history, and BMI. The CART model demonstrated strong predictive performance (AUC ≈ 0.79–0.80) and identified level of independence as the most influential factor. Individuals living dependently exhibited substantially higher AD prevalence (72.7%) compared to those living independently (18%). Age was the second most important factor, with thresholds at 73.5 and 84.5 years differentiating risk levels. Apnea emerged as a critical factor in multivariate analysis, despite nonsignificant bivariate findings, with markedly higher AD prevalence among those affected. Additional important predictors included antidepressant use and BMI. The model further identified distinct high-risk subgroups based on combinations of these factors.

Conclusions and Implications: This study highlights the importance of functional independence, age, and comorbid health conditions in understanding AD risk among Asian American older adults. Clinically, these results support early screening efforts targeting individuals with reduced independence and specific health profiles. From a policy perspective, the study emphasizes the need for culturally tailored interventions and improved access to diagnostic and preventive services for Asian American communities. Future research should incorporate longitudinal designs and additional biological and environmental factors to further refine risk prediction and support personalized approaches to AD prevention and care.