Abstract
Background: Social prescribing is expanding across the Western Pacific Region premised in part on the assumption that social disconnection drives poor mental health and wellbeing. Yet whether depression instead generates social withdrawal — reversing the assumed directional order — has not been longitudinally tested in East Asian populations. Methods: We analyzed three waves (2015, 2018, 2020) of the China Health and Retirement Longitudinal Study (n=4813 adults aged ≥45 years), spanning pre-pandemic and pandemic-transition periods. Cross-lagged panel models with coarsened exact matching estimated directional associations between social disconnection (social isolation, loneliness) and subjective health (self-rated health, depressive symptoms). Multi-group analyses tested moderation by functional limitation. Subgroup analyses examined cross-wave variations within each group. Findings: Depressive symptoms predicted subsequent loneliness two to three times more strongly than the reverse across both periods (β₂₀₁₅₋₂₀₁₈ D→L=0·237 vs L→D=0·080; β₂₀₁₈₋₂₀₂₀ D→L=0·179 vs L→D=0·078; all p<0·001). During pandemic transition, depressive symptoms negatively predicted subsequent social isolation (β=-0·060, p=0·001), and social isolation positively predicted subsequent self-rated health (β=0·040, p=0·007). Functional limitations amplified depression’s impact on later loneliness pre-pandemic, but during pandemic transition loneliness more strongly predicted subsequent depressive symptoms among individuals without functional limitations. Interpretation: Depression more strongly precedes loneliness than the reverse — challenging a core directional assumption underpinning social prescribing for mental health in ageing populations. Depression-first intervention strategies may be more effective than social prescribing alone for loneliness prevention in the Western Pacific region, and crisis protocols require differential targeting by functional status.