Abstract
{ "background": "District hospital systems in sub-Saharan Africa face persistent challenges in achieving cost-effective service delivery. Methodological limitations in existing health economics evaluations, particularly the failure to account for hierarchical data structures and longitudinal cost trajectories, constrain robust policy analysis.", "purpose and objectives": "This study aims to methodologically evaluate the application of longitudinal multilevel regression modelling for measuring cost-effectiveness within a national district hospital system. It assesses the model's capacity to isolate facility-level efficiency from systemic and temporal confounders.", "methodology": "A longitudinal panel dataset from a national system was analysed using a three-level random intercepts model. The core statistical model is: $Cost{it} = \\beta0 + \\beta1 Outcome{it} + u{j[i]} + v{k[i]} + \\epsilon{it}$, where $uj$ and $v_k$ are random effects for facility and region, with inference based on cluster-robust standard errors. The analysis incorporated repeated measures of operational costs and composite health outcomes.", "findings": "The methodological evaluation demonstrates that the multilevel approach attenuates confounding, revealing that approximately 40% of the variance in cost-effectiveness was attributable to stable regional-level factors. A one-unit increase in the composite outcome score was associated with a 5.2% cost increase (95% CI: 3.1% to 7.3%), an effect overstated by 1.8 percentage points in pooled OLS estimation.", "conclusion": "Longitudinal multilevel regression provides a superior methodological framework for cost-effectiveness analysis in hierarchical health systems, effectively partitioning variance and generating less biased estimates of key parameters.", "recommendations": "Health systems researchers should adopt multilevel modelling for economic evaluations in decentralised contexts. Policymakers should mandate the collection of panel data to facilitate such analyses for resource allocation.", "key words": "health economics, multilevel modelling, panel data, health systems research, efficiency measurement, sub-Saharan Africa", "contribution statement": "