Abstract
BACKGROUND: HIV incidence remains high among transgender women in Lima, Peru, most of whom report sex work. On the basis of a stakeholder analysis and health system capacity assessment, we designed a mathematical model to guide HIV programmatic planning among transgender women sex workers (TWSW) in Lima. METHODS: Using a deterministic compartmental model, we modelled HIV transmission among TWSW, their stable partners, and their clients to estimate the impact and cost-effectiveness of combinations of interventions compared with the standard of care on reducing HIV incidence over a 10-year period. We simulated HIV transmission accounting for differences in sexual positioning in anal intercourse and condom use by partner type and fitted the model to HIV surveillance data using Latin hypercube sampling. The interventions we considered were 15% relative increase in condom use with clients and 10% relative increase with stable partners; increase in antiretroviral treatment (ART) coverage at CD4 count lower than 500 cells per mm(3) and greater than or equal to 500 cells per mm(3); and 15% pre-exposure prophylaxis (PrEP) coverage using generic and branded formulations. We considered a basic scenario accounting for current limitations in the Peruvian HIV services and an enhanced scenario assuming achievement of the UNAIDS 90-90-90 targets and general improvements in HIV services. The 50 best fits according to log-likelihood were used to give the minimum and maximum values of intervention effect for each combination. We used disability-adjusted life-years (DALYs) to measure the negative health outcomes associated with HIV infection that could be averted through the interventions investigated and calculated incremental cost-effectiveness ratios to compare their cost-effectiveness. FINDINGS: Under the basic scenario, combining the four interventions of increasing condom use with clients and stable partners, extending ART to people with CD4 count greater than or equal to 500 cells per mm(3), and 15% PrEP coverage with generic drugs would avert 47% (range 27-51) of new infections in TWSW, their clients, and their stable partners over 10 years, with an incremental cost-effectiveness ratio of US$509 per DALY averted. Under the enhanced scenario, this strategy would avert 61% (44-64) of new infections among this population with an incremental cost-effectiveness ratio of $1003 per DALY averted. Under both scenarios, implementation of this strategy approaches or surpasses the 50% incidence reduction goal and would represent a cost-effective use of country resources if generic PrEP drugs are used. The total cost of implementing this strategy under the enhanced scenario would be approximately $1·2 million per year over 10 years, corresponding to 10% of the current Global Fund's yearly contribution in Peru. INTERPRETATION: Investments in HIV services among TWSW in Lima would be cost-effective, even under stringent cost-effectiveness criteria when accounting for setting-specific resource constraints. Notable improvements in HIV testing rates, innovative interventions to increase condom use, and reduced PrEP costs will be key to achieving the 50% incidence reduction goal. Modelling studies incorporating stakeholders' perspectives and health system assessments can bring added value to HIV policy making. FUNDING: National Institutes of Health.