A methodological framework to assess temporal trends and sub-national disparities in healthcare quality metrics using facility surveys, with applications to sick-child care in Kenya, Senegal, and Tanzania

Monitoring healthcare quality at a subnational resolution is key to identify and resolve geographic inequities and ensure that no sub-population is left behind. Yet, health facility surveys are typically not powered to report reliable estimates at a subnational scale. In this study, we present a framework to fill this gap and jointly analyse publicly available facility survey data, allowing exploration of temporal trends and subnational disparities in healthcare quality metrics. Specifically, our Bayesian hierarchical model includes random effects to account for differences between survey instruments; space-time processes to leverage correlations in space and time; and covariates to incorporate auxiliary information. We apply this framework to Kenya, Senegal, and Tanzania - three countries with at least four rounds of standardized facility surveys each - and estimate the readiness and process quality of sick-child care over time and across subnational areas. These estimates of readiness and process quality of care over time and at a fine spatial resolution show uneven progress in improving facility-based service provision in Kenya, Senegal, and Tanzania. For instance, while national gains in overall readiness of care improved in Tanzania, geographic inequities persisted; in contrast, Senegal, and Kenya experienced stagnation in overall readiness at the national level, but disparities grew across subnational areas. Overall, providers adhered to about one-third of the clinical guidelines for managing sick-child illnesses at the national level. Yet across subnational units, such adherence greatly varied (e.g., 25% to 85% between counties of Kenya in 2020). Our new approach enables identifies precise estimation of changes in the spatial distribution of healthcare quality metrics over time, at a a programmatic spatial resolution, and with accompanying uncertainty estimates. Use of our framework will provide new insights at a policy-relevant spatial resolution for national and regional decision-makers, and international funders.