**Background:** Multimorbidity, the co-occurrence of at least two chronic conditions unrelated to an index diagnosis, is a major 21st-century health challenge with high prevalence and escalating rates. Adverse childhood experiences (ACEs) are known to increase the risk of multimorbidity, but the mechanisms remain unclear. Two proposed mediation models—psychosocial chain (e.g., social engagement) and biological embedding (e.g., allostatic load)—have not been empirically validated together. This study aimed to test whether social engagement and allostatic load mediate the ACE–multimorbidity relationship in a large, national longitudinal sample.
**Methods:** Data came from the Canadian Longitudinal Study on Aging (CLSA) Comprehensive Cohort, including 27,170 community-dwelling adults aged 45–85 at baseline (recruitment 2011–2015). ACEs (14 items adapted from the Childhood Experiences of Violence Questionnaire and National Longitudinal Study of Adolescent to Adult Health) were assessed retrospectively at first follow-up (2015–2018). Allostatic load was constructed from 18 biomarkers (e.g., HbA1c, CRP, blood pressure, BMI) using high-risk quartile cutoffs. Social engagement was a latent variable combining social support (Medical Outcomes Study Social Support Survey) and social participation frequency. Multimorbidity was a count of 21 chronic conditions (e.g., heart disease, diabetes, cancer, mood disorders) diagnosed by a health professional. Covariates included age, sex, income, smoking, alcohol, nutrition, physical activity, and baseline multimorbidity. Structural equation modeling with full information maximum likelihood was used, with model fit assessed by CFI (>0.95), SRMR (<0.05), and RMSEA (<0.05).
**Key Results:** The overall model showed good fit (RMSEA=0.06, SRMR=0.02, CFI=0.97). ACEs were directly associated with multimorbidity (β=0.12, 95% CI 0.11–0.13). Indirect pathways: ACEs were negatively associated with social engagement (β=−0.14, −0.16 to −0.12), and social engagement was negatively associated with multimorbidity (β=−0.10, −0.12 to −0.08). ACEs were positively associated with allostatic load (β=0.04, 0.03–0.05), and allostatic load was positively associated with multimorbidity (β=0.16, 0.15–0.17). The total effect of ACEs on multimorbidity was β=0.14 (0.13–0.15). The model explained 26.5% of the variance in multimorbidity. Sex-stratified analyses showed stronger effects in females (direct β=0.13 vs. 0.10; indirect β=0.03 vs. 0.01). Age-stratified analyses revealed diminishing effects with age: in the 75–85 group, the indirect pathway was non-significant (β=0.006, n.s.), though direct ACE–multimorbidity and social engagement–multimorbidity paths remained significant. When baseline multimorbidity was added as a covariate, all paths remained significant, and the model explained 73% of variance.
**Clinical Implications:** This is the first study to demonstrate that social engagement and allostatic load mediate the ACE–multimorbidity link, supporting a lifespan developmental model. The findings suggest that early-life adversity programs both social and biological pathways that increase multimorbidity risk. Clinically, ACE screening may identify at-risk individuals for early intervention. Enhancing social engagement at any age could reduce multimorbidity burden, especially given its modifiable nature. The results also underscore the need for early childhood interventions to prevent the cascade from ACEs to chronic disease. The model provides a platform for future research incorporating additional mediators (e.g., self-regulation) and moderators (e.g., genetic factors) to further elucidate multimorbidity mechanisms.