OBJECTIVES: Making health policy decisions based on multiple criteria, including health and non-health components, is crucial in the context of universal health coverage program financing, especially in the priority settings of low- and middle-income countries. This research aims to identify the optimal interventions to enhance health benefits and reduce poverty outcomes for each province in South Africa, constrained by the limited health budget and medical workforce.

METHODS: Using publicly available data related to South Africa, we formulate a bi-objective mixed-integer linear programming model to determine the optimal set of health interventions and funding for each intervention that can maximize both the health benefits (measured using the averted disability-adjusted life years, DALYs) and the averted health-induced poverty cases (measured using financial risk protection, FRP). We also compare our approach with the heuristic baseline of choosing interventions based on the descending order of health benefits/cost ratio.

RESULTS: By iteratively solving the bi-objective model, a Pareto curve was obtained to represent all possible combinations of health benefits and FRP: one end of the curve gets maximum health benefits attainable and 98.3% maximum FRP attainable, and the other end gets maximum FRP attainable and 99.1% maximum health benefits attainable. In contrast, the baseline can only identify one solution (one extreme point of the Pareto curve) and fails to identify the others. Aggregating the amount of funding for each intervention across all possible solutions shows that 17 out of the 26 interventions are, on average, fully funded. The remaining interventions are only partially funded.

CONCLUSIONS: Under multi-criteria decision-making, multi-objective optimization provides policymakers with the whole set of optimal solutions, which allows them to prioritize different criteria under constraints. Analyzing different solutions on the Pareto curve also highlights the most and least effective interventions.