Background HIV transmission in many settings is concentrated amongst people who inject drugs (PWID), and can reach high HIV prevalence. Early anti-retroviral treatment (ART) for PWID could be important for reducing HIV transmission in these settings. This modelling analysis projects the impact of such a strategy in a high HIV prevalence (30%) setting, and explores how behavioural and epidemiological factors affect the projections.

Methods A deterministic,model of HIV transmission amongst PWID is calibrated to HIV prevalence data from Manipur, India. The model is used to assess the 20 year impact (on HIV incidence/prevalence) of scaling up early ART to PWID, compared to a baseline of current ART access (20% of HIV+ individuals with CD4 < 350 cells/µL are recruited onto ART annually). An uncertainty analysis was undertaken using posterior model fits to consider which PWID behavioural/epidemiological factors affect impact.

Results Only small reductions in PWID HIV incidence/prevalence (< 10% over 20 years) will occur with current ART provision in Manipur. If current ART recruitment rates are maintained, but all HIV+ PWID become eligible for ART, then a median 12%/11% relative decrease in HIV incidence/prevalence occurs over 20 years compared to baseline. This increases to 22%/20% or 33%/28% if 40% or 80% of HIV+ PWID are recruited per year, respectively. If the ART LTFU amongst PWID is halved to 7.5% per year then impact increases by half. Uncertainty analyses suggest the impact achieved through scaling-up ART is highly dependent on baseline HIV prevalence and the cofactor increase in HIV infectivity during early infection, with less impact occurring for higher HIV prevalences and cofactors. The injecting cessation rate and level of mixing between risk groups are also important.

Discussion HIV treatment could result in large reductions in PWID HIV incidence, but is unlikely to result in local elimination except in low HIV prevalence settings.