Costs and cost-effectiveness of HIV counselling and testing modalities in Southern Mozambique

Objective Despite the high HIV associated burden, Mozambique lacks data on HIV counselling and testing (HCT) costs. To help guide national HIV/AIDS programs, we estimated the cost per test for voluntary counselling and testing (VCT) from the patient’s perspective and the costs per person tested and per HIV-positive individual linked to care to the healthcare provider for VCT, provider-initiated counselling and testing (PICT) and home-based testing (HBT). We also assessed the cost-effectiveness of these strategies for linking patients to care. Methods Data from a cohort study conducted in the Manhiça District were used to derive costs and linkage-to-care outcomes of the three HCT strategies. A decision tree was used to model HCT costs according to the likelihood of HCT linking individuals to care and to obtain the incremental cost-effectiveness ratios (ICERs) of PICT and HBT with VCT as the comparator. Sensitivity analyses were performed to assess robustness of base-case findings. Findings Based on costs and valuations in 2015, average and median VCT costs to the patient per individual tested were US$1.34 and US$1.08, respectively. Costs per individual tested were greatest for HBT (US$11.07), followed by VCT (US$7.79), and PICT (US$7.14). The costs per HIV-positive individual linked to care followed a similar trend. PICT was not cost-effective in comparison with VCT at a willingness-to-accept threshold of US$4.53, but only marginally given a corresponding base-case ICER of US$4.15, while HBT was dominated, with higher costs and lower impact than VCT. Base-case results for the comparison between PICT and VCT presented great uncertainty, whereas findings for HBT were robust. Conclusion PICT and VCT are likely equally cost-effective in Manhiça. We recommend that VCT be offered as the predominant HCT strategy in Mozambique, but expansion of PICT could be considered in limited-resource areas. HBT without facilitated linkage or reduced costs is unlikely to be cost-effective.