Background: Over $20 billion is spent annually on HIV programs in low- and middle-income countries. The allocation of resources by country and intervention depends on a mixture of need, funding sources, capacity, effectiveness, policy, politics and social factors. We investigated how well the current allocation is optimized for cost-effectiveness.
Methods: We applied an HIV simulation model, Goals, to 55 countries accounting for about 90% of all new infections to determine the cost per infection, death and DALY averted for each of 13 interventions. Units costs were based on Investment Cases for each country. The models were fit to surveillance and survey data and UNAIDS official estimates. Cost-effectiveness ratios were calculated by country and intervention by scaling up each intervention, one-at-a-time, over five years and recording the incremental costs, infections, deaths and DALYs compared to a counterfactual of no scale up of any interventions. The results comprise a database of total cost and cost-effectiveness measures for 716 country/intervention pairs.
Results: Cost-effectiveness varies widely across countries and interventions as shown in Figure 1 for cost per infection averted. Six percent of these combinations are cost savings because the total cost of the intervention is less than the savings generated due to treatment costs averted. ART dominates cost per death and DALY averted and also ranks high in cost-effectiveness for infections averted. The most cost-effective prevention interventions are generally VMMC, PMTCT, outreach to sex workers and condom promotion. These programs currently receive about 14% of direct intervention funding, which is about two-thirds of the need. The most cost-effective programs are in East and Southern Africa where incidence is high and costs are generally low. Currently almost 60% of all resources and 70% of donor resources are focused on ESA. About 15% of donor resources go to West and Central Africa where cost-effectiveness can be considerably worse.
Conclusions: Resources for HIV prevention and treatment are generally targeted appropriately but more focused allocation of resources could improve cost-effectiveness by about a quarter. Resource allocations should be continually assessed because cost-effectiveness can change significantly as incidence patterns change.

Figure 1
[Figure 1]