Personal Assistant Service Patterns and Health Behaviors Among Adults with Disabilities: A Latent Class Analysis

Purpose: This study aimed to classify patterns of Personal Assistant Service (PAS) among adults with disabilities in Korea using latent class analysis (LCA) based on the Behavioral Model for Vulnerable Populations (BMVP). Methods: A cross-sectional descriptive study was conducted using data from the Disability and Life Dynamics Panel (2018-2023). The most recent observation per individual was selected to ensure analytic independence, yielding 482 adults. LCA was performed using six service use domains. Multinomial logistic regression examined factors associated with class membership, and modified Poisson regression assessed associations between latent classes and five health behaviors. Results: Three classes were identified: Comprehensive Support (36.3%), Physical and Social Support (36.1%), and Household Support (27.6%). Compared with non-users, service users showed greater socioeconomic and health vulnerability, including more severe disability, lower income, living alone, lower social support, and higher depression. Women and individuals with greater disability severity and daily-living limitations showed higher odds of belonging to the Comprehensive Support class. After adjusting for covariates, the Comprehensive Support class showed a significantly lower prevalence of physical activity and alcohol consumption compared to the Household Support class, while previously observed differences in fast food intake were attenuated.Conclusion: PAS is associated with health behaviors among adults with disabilities, although current service structures may not fully address health enhancing behaviors. Expanding PAS to include tailored physical activity support, health management, and psychosocial care may help reduce health disparities among adults with disabilities.