Background HIV incidence remains high among transgender women (TW) in Lima, of whom the majority report sex-work. A stakeholder analysis and a health-system capacity assessment informed a mathematical modelling study to devise a tailored combination prevention programme. We modelled impact and cost-effectiveness of realistic combinations of interventions among TW sex-workers (TW-SW) in Lima.

Methods In an HIV policy dialogue, a stakeholder analysis provided data on acceptability, feasibility, appropriate coverage targets and scale-up times of both novel and (improved) existing interventions. The health system study assessed capacity, costs and needs. Using a published model we simulated HIV transmission among TW-SW, their clients and stable partners, with implementation of combinations of the following interventions: 15% and 10% relative increase in condom use with clients and stable partners respectively, 15% pre-exposure prophylaxis (PrEP) coverage, treatment following new WHO guidelines and “test and offer”, both including testing promotion and leading to 65 and 75% coverage respectively. A social/structural component was part of all scenarios. The interventions’ individual and combined impact and cost-effectiveness were assessed.

Results Combining increased condom use with clients and treatment under new WHO guidelines resulted in around 50% of new infections averted over 10 years; this was highly cost-effective ($90/DALY averted), under the World Bank threshold though feasibility of condom use increases remains problematic. Treatment in isolation was over the highly cost-effective threshold. A15% coverage of PrEP might be feasible and adds to impact, but is not cost-effective at $1440/year.

Conclusions Implementing WHO treatment guidelines, combined with increased condom use among TW-SW would be highly effective and cost-effective. Inclusion of PrEP adds to impact but requires drastic cost reductions to become cost-effective. Success of all interventions is contingent on effective adherence support. Involving stakeholders in the elaboration of mathematical modelling studies is feasible, and should result in more relevant cost-effectiveness analyses to support programmatic decision-making.

Declaration of conflicts of interest Dr. Guanira was the principal investigator for the Peruvian IPrEX sites. All authors declare having no conflicts of interest.