A novel method to estimate the indirect community benefit of HIV interventions using a microsimulation model of HIV disease

Background: Microsimulation models of human immunodeficiency virus (HIV) disease that simulate individual patients one at a time and assess clinical and economic outcomes of HIV interventions often provide key details regarding direct individual clinical benefits ("individual benefit"), but they may lack detail on transmissions, and thus may underestimate an intervention's indirect benefits ("community benefit"). Dynamic transmission models can be used to simulate HIV transmissions, but they may do so at the expense of the clinical detail of microsimulations. We sought to develop, validate, and demonstrate a practical, novel method that can be integrated into existing HIV microsimulation models to capture this community benefit, integrating the effects of reduced transmission while keeping the clinical detail of microsimulations.

Methods: We developed a new method to capture the community benefit of HIV interventions by estimating HIV transmissions from the primary cohort of interest. The method captures the benefit of averting infections within the cohort of interest by estimating a corresponding gradual decline in incidence within the cohort. For infections averted outside the cohort of interest, our method estimates transmissions averted based on reductions in HIV viral load within the cohort, and the benefit (life-years gained and cost savings) of averting those infections based on the time they were averted. To assess the validity of our method, we paired it with the Cost-effectiveness of Preventing AIDS Complications (CEPAC) Model - a validated and widely-published microsimulation model of HIV disease. We then compared the consistency of model-estimated outcomes against outcomes of a widely-validated dynamic compartmental transmission model of HIV disease, the HIV Optimization and Prevention Economics (HOPE) model, using the intraclass correlation coefficient (ICC) with a two-way mixed effects model. Replicating an analysis done with HOPE, validation endpoints were number of HIV transmissions averted by offering pre-exposure prophylaxis (PrEP) to men who have sex with men (MSM) and people who inject drugs (PWID) in the US at various uptake and efficacy levels. Finally, we demonstrated an application of our method in a different setting by evaluating the clinical and economic outcomes of a PrEP program for MSM in India, a country currently considering PrEP rollout for this high-risk group.

Results: The new method paired with CEPAC demonstrated excellent consistency with the HOPE model (ICC = 0.98 for MSM and 0.99 for PWID). With only the individual benefit of the intervention incorporated, a PrEP program for MSM in India averted 43,000 transmissions over a 5-year period and resulted in a lifetime incremental cost-effectiveness ratio (ICER) of US$2,300/year-of-life saved (YLS) compared to the status quo. After applying both the direct (individual) and indirect (community) benefits, PrEP averted 86,000 transmissions over the same period and resulted in an ICER of US$600/YLS.

Conclusions: Our method enables HIV microsimulation models that evaluate clinical and economic outcomes of HIV interventions to estimate the community benefit of these interventions (in terms of survival gains and cost savings) efficiently and without sacrificing clinical detail. This method addresses an important methodological gap in health economics microsimulation modeling and allows decision scientists to make more accurate policy recommendations.