When widowhood meets hospitalisation risk: the dynamic, time-dependent association in older adults

In China’s rapidly ageing society, widowhood is a prevalent stressor, yet the temporal dynamics of its impact on healthcare utilisation remain poorly understood. While “widowhood shock” is documented in Western settings, how the association evolves with duration in China’s distinct sociocultural context is unknown. This study aimed to investigate the heterogeneous and dynamic association between widowhood and hospitalization risk among Chinese older adults, and to verify the robustness of the findings through rigorous causal identification methods. Using five waves of longitudinal data (2011–2020) from the China Health and Retirement Longitudinal Study (CHARLS), including a total of 9,428 participants aged ≥ 60 years (942 widowed and 8,486 non-widowed). We adopted the heterogeneous difference-in-differences (CSDID) approach combined with doubly robust inverse probability weighted regression adjustment (DRIPW) to estimate cohort-specific average treatment effects on the treated (ATT). Propensity score matching was used to balance 16 covariates across sociodemographic, health, social support, economic, and insurance domains. Gender-stratified analyses were conducted as a robustness check to explore potential disparities. After achieving covariate balance (standardized bias reduction > 70% for all key variables), widowhood significantly increased hospitalization risk among Chinese older adults aged ≥ 60 years, with an overall ATT of 0.051 (SE = 0.019, P = 0.007). Heterogeneity was observed across cohorts and calendar time: Cohort G3 (first widowed in 2015) had the strongest effect (ATT = 0.061, P = 0.024), while Cohorts G2 (2013) and G5 (2020) showed no significant positive effects; calendar-time effects were significant in T3 (2015, ATT = 0.073, P = 0.009) and T5 (2020, ATT = 0.054, P = 0.039) but not in T2 (2013) and T4 (2018), reflecting differences between cohort-specific treatment effects and overall temporal trends. Dynamic analysis showed a significant immediate effect in the year of widowhood (period 0: ATT = 0.060, SE = 0.021, P = 0.004). This effect attenuated and became statistically non-significant in the subsequent 1–3 waves, indicating an immediate but non-persistent “shock” pattern. Robustness checks confirmed the reliability of the core findings. Widowhood exerts a significant positive impact on hospitalization risk among Chinese older adults, with heterogeneous effects across cohorts and calendar time and an immediate but non-persistent dynamic pattern. These findings highlight the need for targeted interventions, particularly during the immediate year following spousal loss and for specific high-risk cohorts. Practical steps include enhancing family-based health management support and optimizing healthcare accessibility for rural widowed older adults.