Heterogeneous trajectories of intrinsic capacity and their predictors: a Bayesian network analysis of a longitudinal cohort in Chinese nursing homes

Objectives To identify distinct developmental trajectories of intrinsic capacity (IC) among nursing home residents in China and to elucidate the complex network relationships among their predictors using a Bayesian network model, guided by the Socio-Ecological Theory. Methods An 18-month longitudinal study was conducted with 592 older adults from three nursing homes in Northeast China from March 2022 to February 2024. IC was assessed at baseline (T0), 6 (T1), 12 (T2), and 18 (T3) months using the ICOPE screening tool. Latent growth mixture modeling (GMM) identified trajectory subgroups. A BN was constructed to model the interrelationships among significant predictors identified from multinomial logistic regression. Results Three distinct IC trajectories were revealed: ‘High Functioning-Stable’ (16.0%), ‘Medium Functioning-Slow Decline’ (40.5%), and ‘Low Functioning-Rapid Decline’ (43.4%). The Bayesian network model identified four direct predictors of trajectory membership: family function, number of chronic diseases, grip strength, and perceived stress. Health-promoting behaviors and loneliness were key indirect predictors, operating through mediators such as grip strength and perceived stress, respectively. Conclusion Intrinsic capacity development among nursing home residents is heterogeneous and driven by a complex network of interlinked factors. Our findings advocate for a paradigm shift from single-factor interventions to integrated, multi-target strategies. Clinicians should simultaneously address family support, multimorbidity, physical capacity, and psychological well-being to effectively delay IC decline.