Longitudinal co-development of social interaction, physical frailty, and subjective cognitive decline among community-dwelling older adults in Japan: a six-year parallel growth study

Abstract

Background Population aging requires community-based approaches to identify multidomain risk patterns before severe disability or clinical cognitive impairment develops. Although social interaction, physical frailty, and subjective cognitive decline are relevant to healthy aging, their parallel longitudinal trajectories remain insufficiently characterized in community-dwelling older adults. Objective This study examined the six-year co-development of social interaction, screening-defined physical frailty, and screening-defined subjective cognitive decline among community-dwelling older adults in Japan. Methods This longitudinal cohort study analyzed 307 older adults (34.6% of 887 baseline-eligible participants) with valid measurements of all three focal variables in 2014, 2017, and 2020. Social interaction was measured using the Index of Social Interaction. Physical frailty (PF) was operationalized using the five-item motor domain of the Kihon Checklist (KCL), and screening-defined subjective cognitive decline (SCD) was operationalized using the three-item KCL Cognitive Function domain (KCL-CF). A Bayesian parallel-process latent-basis growth model was fitted, adjusting for baseline sociodemographic, lifestyle, and chronic disease factors. Results The model showed acceptable posterior predictive fit. Social interaction scores decreased from 16.82 to 16.09, PF prevalence increased from 8.8 to 19.5%, and screening-defined SCD prevalence increased from 31.3 to 62.9%. A higher social-interaction intercept factor was associated with lower PF intercept liability (standardized estimate −0.282, 95% credible interval (CrI) −0.503 to −0.044) and lower SCD intercept liability (−0.433, −0.651 to −0.179). The social-interaction and PF slope factors were inversely associated (−0.490, −0.809 to −0.026), whereas slope associations involving SCD were imprecisely estimated. Conclusion Social, physical, and subjective cognitive indicators showed distinct but interrelated longitudinal patterns. Repeated multidomain assessment may support community-based risk characterization and follow-up for healthy aging.