Geographical inequalities and factors associated with emergency obstetric and newborn care readiness in Ethiopia: a national Bayesian spatial analysis

Emergency obstetric and newborn care readiness in Ethiopia shows substantial geographical variation, with higher readiness in general hospitals compared to medium clinics. Oxytocin availability is positively associated with facility readiness.

Summary and findings

The study assessed emergency obstetric and newborn care readiness in Ethiopia using a Comprehensive Emergency Readiness Index (CERI) based on data from 776 health facilities. The nationally weighted mean CERI was 57.4, with significant geographical variation observed. Factors associated with CERI included provider availability and oxytocin availability.

Abstract

Abstract Background Maternal and neonatal survival depends not only on access to health facilities but also on the capacity of those facilities to provide timely, high-quality emergency obstetric and newborn care (EmONC). Despite substantial investments in maternal and newborn health and expansion of emergency obstetric services in Ethiopia, persistent gaps in facility readiness continue to limit the delivery of life-saving care. Previous national studies have largely focused on service availability or individual readiness components, with limited assessment of multidimensional emergency readiness and its geographical variation. We developed a Comprehensive Emergency Readiness Index (CERI) and assessed its national distribution, geographical variation, inequality, and facility-level factors associated with emergency obstetric and newborn care readiness using nationally representative health facility data from Ethiopia. Methods We conducted a nationally representative cross-sectional study using data from 776 health facilities included in the 2021 Ethiopia Service Provision Assessment. A latent Comprehensive Emergency Readiness Index (CERI) was constructed from essential emergency obstetric and newborn care medicines, equipment, and clinical guidelines using structural equation modelling. National estimates accounted for facility sampling weights. Geographic variation was examined using spatial autocorrelation analyses, and Bayesian hierarchical spatial regression models based on the stochastic partial differential equation approach with integrated nested Laplace approximation were fitted to evaluate associations between facility characteristics and emergency readiness. Model performance was compared using the deviance information criterion, widely applicable information criterion, conditional predictive ordinates, and predictive accuracy measures. Results The nationally weighted mean CERI was 57.4 (SE 2.6), compared with an unweighted mean of 70.6 (SD 36.0). CERI varied substantially across regions, ranging from 27.8 (95% CI 15.9–39.6) in Addis Ababa to 62.5 (50.9–74.1) in Amhara, with moderate national inequality (Gini coefficient 0.374; coefficient of variation 0.510). General hospitals demonstrated the highest readiness (89.1, 95% CI 86.0–92.2), whereas medium clinics had the lowest (5.6, 95% CI 1.3–9.9). Government-managed facilities had higher readiness (78.5, 95% CI 75.0–82.0) than private-for-profit facilities (10.5, 95% CI 5.2–15.9). The Bayesian hierarchical spatial model incorporating both regional and spatial random effects provided the best overall model fit (DIC 2516.49). After accounting for spatial dependence and regional heterogeneity, greater provider availability (posterior mean 4.64, 95% CrI 1.02–8.27) and oxytocin availability (posterior mean 9.96, 95% CrI 7.52–12.40) were positively associated with CERI, whereas medium clinics had substantially lower CERI than the reference facility category (posterior mean − 55.37, 95% CrI − 70.09 to − 40.61). Residual spatial autocorrelation was no longer evident after model adjustment (Moran's I − 0.013; P  = 0.770). Conclusion Emergency obstetric and newborn care readiness remains unevenly distributed across Ethiopian health facilities, with substantial geographical and facility-level variation and moderate national inequality. Provider availability, oxytocin availability, and medium clinic status were independently associated with the Comprehensive Emergency Readiness Index after accounting for spatial dependence and regional heterogeneity. Integrating multidimensional readiness assessment with spatially informed health-system monitoring could support more equitable assessment and prioritisation of emergency obstetric and newborn care readiness within Ethiopia's health system.

Limitations

Not reported in abstract.

Elsewhere in the Oxytocin corpus