Longitudinal Network Dynamics of Cardiometabolic, Mobility, and Psychological Indicators Among Older Adults in Rural South Africa

Abstract Background and Objectives Evidence used to define later-life health remains dominated by high-income settings, despite rapid population ageing in sub-Saharan Africa. This study examined how cardiometabolic indicators, symptom burden, mobility limitation, and psychological functioning co-evolve over time among older adults in rural South Africa. Research Design and Methods We analysed three waves of the Health and Ageing in Africa: A Longitudinal Study of an INDEPTH Community in South Africa (HAALSI; 2014-2022) among adults 60+ years at baseline (N = 2,807). Seven indicators were modelled: body mass index (BMI), systolic blood pressure (SBP), pain burden, sleep duration, mobility limitation, depressive symptoms, and life satisfaction. We estimated within-wave Gaussian graphical models and a cross-lagged panel network (CLPN) using full information maximum likelihood. Results Cardiometabolic indicators were more temporally stable than psychological indicators (autoregressive coefficients: BMI = 0.77; SBP = 0.49; depressive symptoms = 0.06; life satisfaction = 0.06). Temporal out-strength was highest for BMI (0.843), followed by SBP (0.705) and mobility limitation (0.668), whereas pain (0.165) and sleep duration (0.112) showed lower connectedness. The CLPN showed acceptable model fit (CFI=.96, RMSEA=.03, 90%CI[.03,.04], SRMR=.03). Mobility limitation predicted higher later life satisfaction (β = 0.16, p<.05) and showed a small inverse association with later depressive symptoms (β=-0.09, p<.10). Discussion and Implications Results indicate cardiometabolic indicators and mobility limitation were more temporally connected components of later-life health. These findings support integrated primary-care approaches that combine cardiometabolic monitoring with attention to locomotor function, while the mobility-well-being association warrants cautious interpretation and replication. These associations are observational and should not be interpreted as causal or as purely within-person effects.