Multidimensional Frailty-Related Vulnerability Profiles Among Community-Dwelling Older Adults: A Data-Driven Cluster Analysis of the 2023 National Survey of Older Koreans

Background/Objectives: Frailty-related vulnerability spans health, physical function, health behaviors, cognition, and psychological well-being. A single summary score may not fully convey how these features co-occur within individuals. This study identified data-driven frailty-related profiles among community-dwelling older adults and compared the magnitude of between-profile differences. Methods: We conducted a cross-sectional secondary analysis of the 2023 National Survey of Older Koreans. The analytical sample included 9950 participants. Ten indicators—self-rated health, chronic disease count, 1-year weight loss, sleep quality, weekly exercise frequency, fatigue, stair-climbing ability, 300-m walking ability, cognitive function, and depressive symptoms—were standardized and entered into K-means clustering. Candidate solutions for K = 2–10 were evaluated using within-cluster sums of squares and mean silhouette coefficients. Profile characteristics and effect sizes were compared using independent-samples t-tests, Pearson’s chi-squared tests, Cohen’s d, and Cramér’s V. Results: The K = 2 solution had the highest mean silhouette coefficient (0.273) among the candidate solutions, indicating modest separation between clusters. Within the analytical sample, 6962 participants (70.0%) formed a lower-vulnerability profile and 2988 (30.0%) a multidimensional higher-vulnerability profile. The latter profile showed less favorable health, functional, psychological, and cognitive characteristics. Among the cluster-defining indicators, the largest descriptive differences were observed in 300-m walking (d = 2.013) and stair-climbing ability (d = 2.001). Conclusions: The two clusters represented a relative multidimensional vulnerability gradient rather than clearly distinct clinical phenotypes. Although the findings describe co-occurring vulnerability characteristics, their potential clinical utility requires independent and longitudinal validation.