The associations between multimorbidity, healthy lifestyles and health status among older adults: a network analysis

With the prevalence of multimorbidity among older adults has shown a marked upward trend, there is an urgent need to explore the patterns and intervention strategies of multimorbidity. Based on network analysis, this study investigates the associations among lifestyle, health status, and chronic diseases in the older adults, reveals the underlying links among the three, identifies key nodes in the network, and provides critical evidence to support the prevention and control of multimorbidity. Data were obtained from the China Health and Retirement Longitudinal Study (CHARLS). We constructed an Ising conditional-dependence network for chronic diseases, and a mixed graphical model (MGM) incorporating multimorbidity, lifestyle behaviors, and health-status indicators. Ten multiply imputed datasets were analysed separately and network estimates were pooled. Edge accuracy and centrality stability were evaluated using nonparametric and case-dropping bootstrap procedures. The analysis included 9,602 participants. The chronic-disease network contained 52 nonzero edges (density = 0.57); its strongest positive conditional associations were chronic lung disease–asthma, psychiatric problems–memory-related diseases, dyslipidemia–diabetes, stroke–memory-related diseases, and liver disease–kidney disease. In the mixed network (21 nodes; 133 nonzero edges; density = 0.63), ADL impairment had the highest strength and betweenness. ADL impairment, BMI category, smoking, depressive symptoms, and physical activity had the highest bridge strength. In the mixed network, CS coefficients were 0.75 for strength and bridge strength and 0.52 for betweenness. The findings characterize conditional association patterns among chronic diseases, lifestyle behaviours, and health status in older adults with established multimorbidity. We generated hypotheses about potentially important nodes, but longitudinal studies are required to confirm the observed patterns.