Postwar Decline in Healthcare Services Utilization With Rising Disparities: A Repeated Cross-Sectional Study Using...

IntroductionTigray experienced a devastating armed conflict in November 2020 that systematically dismantled one of the country's best-performing health systems, with only 30% of hospitals and 17% of health centers remaining functional six months into the war. This study aimed to compare several count data models in analyzing postwar healthcare services utilization disparity and associated factors in Eastern Tigray, Ethiopia.MethodsRepeated cross-sectional analytical study design employed secondary data from 20 facilities across 18 districts of Eastern Tigray comprising 2,040 observations. We compared many models and ZINB-GLMM was selected to handle the healthcare utilization with substantially excess zeros, unmeasured heterogeneity of facility, and potential period-specific interaction. The model considered demographic, service-related, geographic, and conflict-period factors. Model performance was compared using AIC and BIC.ResultsMean utilization declined from 335 recipients prewar to 41 postwar, an 87.8% reduction. The prewar period was significantly associated with higher rate of utilization compared with the postwar period ($p<\hspace{0.17em}0.001$). The ZINB-GLMM outperformed (AIC=15,772.66, BIC=16,109.90), with the moderate variability of random intercept (${\sigma }_{ri}^2=0.16,{\sigma }_{ri}=0.4$) and $\left(ICC=0.970,95\%\hspace{0.17em}CI:0.908-0.986\right)$ indicates significant between-facility proportion of variability (97.0%) in healthcare utilization attributable to differences between facilities after accounting for fixed effects. The zero-inflation component of the model was crucial and the prewar period was significantly associated with lower odds of excess zero counts relative to the postwar period ($p<\hspace{0.17em}0.001$).ConclusionsThe armed war produced terrible declines in healthcare utilization with noticeable geospatial disparities. ZINB-GLMM substantially outperformed, confirming its appropriateness for overdispersion, excess zeros, unmeasured heterogeneity for clustering of facilities, and period-specific interaction effects. The findings underline the significance of strengthening resilient reproductive health systems and ensuring continuity of essential maternal health services during humanitarian crises and post-conflict recovery periods. Recovery efforts must prioritize rural and peripheral areas while rebuilding reproductive health services.