Cohort-specific childhood adversity measures and incident cardiovascular disease across four ageing cohorts: analyses of childhood hunger, later-life psychosocial indicators, and life-course burden

Childhood adversity has been linked to later cardiovascular disease (CVD), but retrospective exposure measures differ across ageing cohorts, limiting direct comparison. We analyzed the China Health and Retirement Longitudinal Study (CHARLS), Health and Retirement Study (HRS), English Longitudinal Study of Ageing (ELSA), and Survey of Health, Ageing and Retirement in Europe (SHARE). Primary analyses related incident CVD to a cohort-specific observed-domain childhood adversity index (CAI; 0–4), with measured domains varying by participant. Secondary analyses examined direct childhood hunger (CHARLS and SHARE) and a three-domain life-course burden; an exploratory analysis assessed joint hunger with cohort-specific later-life loneliness/isolation. Adapted cardiovascular-kidney-metabolic (CKM) staging was assessed separately in CHARLS. Model 2 adjusted for age, sex, birth cohort, smoking, alcohol, and body mass index (BMI). Education, income, and depressive symptoms were added sequentially to assess attenuation, not mediation; cohort estimates were pooled by restricted maximum likelihood. The primary complete-case samples were 9,336 (CHARLS), 2,142 (HRS), 6,224 (ELSA), and 13,912 (SHARE). The pooled hazard ratio (HR) was 1.07 (95% confidence interval [CI] 1.03–1.10) per additional observed adverse domain, a descriptive average across non-equivalent cohort-specific measures rather than a common-scale or population-average effect; I²=0% indicates statistical homogeneity, not measurement equivalence. The conventional HRS Model 2 estimate was not statistically significant (HR 1.04, 95% CI 0.93–1.16; p = 0.546), and time-varying estimates indicated attenuation after five years. Multiple imputation of selected covariates gave HR 1.06 (Hartung–Knapp CI 1.02–1.09; p = 0.016). Direct-hunger estimates were positive in CHARLS and SHARE, but the two-cohort Hartung–Knapp CI included 1 (0.95–1.41). The exploratory 5-year additive interaction was supported only in CHARLS; SHARE was imprecise. The three-domain burden showed a positive average association, with non-uniform categorical patterns. Adapted CKM results were separate, cross-sectional, and CHARLS-only. Among the analyzed complete-case risk sets, higher observed-domain CAI was associated with incident CVD under cohort-specific measures. These findings support further life-course risk research, not causal attribution, intervention efficacy, or preventive targeting. No absolute-risk difference was estimated for the primary CAI contrast; the results are not population-average effects or a validated common-scale measure. Adapted CKM findings remain a separate cross-sectional CHARLS analysis.