Abstract
{ "background": "District hospitals are critical nodes in Kenya's healthcare system, yet systematic evaluations of their cost-effectiveness are scarce. Existing analyses often fail to account for the hierarchical structure of health system data and long-term trends.", "purpose and objectives": "This study aimed to develop and apply a novel multilevel modelling framework to assess the cost-effectiveness of district hospital systems, quantifying the influence of facility-level and regional-level determinants on performance metrics over an extended period.", "methodology": "We conducted a longitudinal analysis of administrative and financial data from a nationally representative panel of district hospitals. Cost-effectiveness was measured using a composite outcome of disability-adjusted life years averted per unit of expenditure. A three-level linear mixed model was fitted: $y{ijt} = \\beta0 + \\beta X{ijt} + uj + vk + \\epsilon{ijt}$, where $i$, $j$, and $k$ index hospitals, districts, and provinces, respectively. Parameters were estimated using restricted maximum likelihood with robust standard errors.", "findings": "The model explained 68% of the variance in cost-effectiveness. A one-standard-deviation increase in outpatient utilisation was associated with a 15.2% improvement in cost-effectiveness (95% CI: 12.1 to 18.3). Provincial-level random effects accounted for 22% of the residual variance, indicating significant regional disparities. Projections suggest that without targeted intervention, the efficiency gap between the top and bottom performing regions will widen by over 30%.", "conclusion": "The cost-effectiveness of district hospitals is significantly influenced by operational factors like outpatient throughput and exhibits substantial geographic clustering. The multilevel approach provides a superior framework for attributing performance variation to the correct administrative tier.", "recommendations": "Policy should prioritise interventions that increase appropriate outpatient utilisation. Resource allocation formulae should incorporate the provincial-level variance component identified here to ensure equitable and efficient funding. Future evaluations must adopt hierarchical models to avoid biased inference.", "key words": "health economics