Background
People who inject drugs (PWID) continue to experience high HIV incidence in a number of settings, including India; however, HIV is not homogeneously distributed in PWID and varies across social (injection partners) and spatial (injection venues) networks. To maximize impact of evidence-based interventions for PWID, it is critical to identify novel delivery points, including injection venues, to deliver HIV (testing & treatment) and harm-reduction services (medication for opioid use disorder (MOUD) and syringe service programs (SSP).
Methods
2512 PWID (10 index and 2502 recruits) were enrolled between 2017-19 in Delhi, India using a sociometric design (named injection partners). Based on network and behavior data, we constructed an individual-based transmission model of the dynamic injection network and injection venue attendance of the cohort with an exponential random graph model (ERGM) from 2017 to 2023. HIV incidence (baseline = 7.3 per 100 person-years) was simulated for each spatial targeting strategy and with increased service coverage corresponding to 50%, 75% or 100% of their respective UNAIDS/UNODC goals to quantify the impact of spatially targeted strategies on HIV epidemic control.
Results
110 injection venues with >10 visitors were identified at baseline, forming 3 spatial clusters of highly aggregated venues visited by PWID roughly based on their residence. In models targeting the same number of venues (between 1 to 12), prioritizing the most visited venues stratified by spatial clusters covered more individuals vs. prioritizing the overall most visited venues until the total targeted venues surpassed 9 (Figure). Simulation results showed that HIV incidence decreased with increasing service and venue coverage. With all services reaching 50% of their respective coverage goals, HIV incidence decreased to between 4.8 per 100PY [4.7, 4.9, 95%CI] (1 venue) and 3.7 per 100PY [3.6, 3.9] (when targeting 1 and 12 venues, respectively). Impact was greater (reduced to 3.0 per 100PY for 1 venue [2.9, 3.1] and 1.8 per 100PY for 12 venues [1.7, 1.9]) when all service goals were completely reached.
Conclusions
In settings with limited resources, these findings suggest that targeting venues that are spatially distinct with minimal overlap of PWID has a larger impact on reducing HIV incidence than targeting the most highly populated venues which may cluster together limiting reach of PWID in a community. However, as coverage across venues expands the two approaches tend to overlap.