A Systematic Review of Methodological Approaches for Evaluating District Hospital Efficiency in South Africa: A Bayesian Hierarchical Modelling Framework

{ "background": "Evaluating the efficiency of district hospitals is critical for health system strengthening and resource allocation. In South Africa, these facilities face persistent operational challenges, yet a comprehensive synthesis of the methodological rigour used to assess their performance is lacking.", "purpose and objectives": "This systematic review aims to critically appraise and synthesise methodological approaches used to evaluate district hospital efficiency, with a specific focus on the application and potential of Bayesian hierarchical modelling frameworks within this context.", "methodology": "A systematic search of multiple electronic databases was conducted following a pre-registered protocol. Studies were screened, selected, and data extracted using standardised forms. Methodological quality was assessed using appropriate tools. The core statistical framework evaluated is the Bayesian hierarchical model, represented as yij \~ N(\β0j + Xij\β, \σ²), \β0j \~ N(\γ₀, \τ²), where i indexes hospitals and j districts, allowing for partial pooling of estimates.", "findings": "The analysis identified a predominant reliance on traditional, frequentist data envelopment analysis, which often fails to account for contextual heterogeneity and uncertainty. A key theme was the underutilisation of advanced stochastic frontier methods. Only a small proportion (approximately 15%) of included studies incorporated hierarchical structures, and none fully leveraged Bayesian inference to propagate uncertainty in efficiency rankings, with credible intervals rarely reported.", "conclusion": "Current methodological approaches for evaluating district hospital efficiency are often insufficient for robust policy inference, as they inadequately handle clustered data and uncertainty quantification.", "recommendations": "Future research should adopt Bayesian hierarchical modelling to provide more reliable, context-sensitive efficiency estimates. Policymakers should require uncertainty intervals around efficiency scores to inform resource decisions.", "key words": "health systems research, efficiency analysis, stochastic frontier analysis, health economics, partial pooling, resource allocation", "contribution statement": "This review provides the first methodological synthesis for this