Multilevel predictors of interpersonal social health: findings from the social health survey in urban Iran

Social health is increasingly recognised as an integral component of holistic health frameworks. However, empirical research examining social health as a distinct outcome remains limited, with relatively little research distinguishing between its interpersonal and institutional aspects. Using multilevel modelling, this study examined interpersonal social health and its sociodemographic, psychological, and structural associated factors among urban residents in Iran. This cross-sectional study used a proportional, stratified, multistage cluster sampling design to recruit adults from urban areas in all 31 provinces of Iran in 2025. The analysis included 18,754 participants aged ≥ 18 years. Interpersonal social health was assessed using the validated 39-item Interpersonal Social Health Questionnaire (ISHQ). Multilevel mixed-effects models examined sociodemographic, psychological distress, subjective well-being, and structural factors associated with interpersonal social health, accounting for the hierarchical clustering of individuals within counties and provinces. Model fit was assessed using the Akaike information criterion (AIC), Bayesian information criterion (BIC), and likelihood-ratio tests. Interpersonal social health had a mean score of 142.83 (SD = 22.57; range: 39–195), and showed significant clustering at both the province (ICC = 5.5%) and county (ICC = 11.4%) levels in the null multilevel model. In the final model, older age, being female, being married, higher educational attainment, employment, and higher socioeconomic status were associated with higher interpersonal social health scores. In contrast, divorced or separated status and lower educational attainment were associated with lower interpersonal social health scores. A non-linear socioeconomic status (SES) gradient was observed, with no further increase in the outcome at the highest SES quintile. Greater life satisfaction and health satisfaction were positively associated with interpersonal social health, whereas psychological distress was negatively associated. Social hope was positively associated with the outcome (B = 0.23, β = 0.04, p < 0.001), whereas provincial-level income inequality was negatively associated (B = −50.81, β = −0.10, p = 0.03). The interaction showed that the positive association between social hope and interpersonal social health was weaker in provinces with greater income inequality. Among urban adults in this study, interpersonal social health was associated with sociodemographic, psychological, and structural factors, with significant variation across both provincial and county contexts. The positive association between social hope and interpersonal social health was weaker in provinces with higher levels of income inequality. Given the cross-sectional design, causal inferences cannot be made. Future longitudinal and mixed-method research could further clarify how changes in psychosocial and structural contexts relate to interpersonal social health over time.