Leveraging data synthesis to identify optimal and robust strategies for HIV elimination among substance-using MSM

Alcohol and methamphetamine use increases risk of HIV among men who have sex with men (MSM) and numerous interventions have been developed to decrease HIV acquisition and transmission among substance using MSM. Yet, despite a considerable body of research documenting these associations, substantial uncertainty remains regarding the specific behavioral pathways between substance use and HIV that are most responsible for this elevated risk (e.g., condom use, sexual partner selection, or HIV medication adherence). Without this knowledge, it is difficult to identify the extent to which substance use drives HIV among MSM or estimate the population level impact of interventions among substance using MSM. In addition, substance use, adherence, risk reduction, and combined interventions have all shown excellent promise to reduce HIV incidence, but large-scale comparative effectiveness trials are extremely challenging and costly and can seldom comprehensively examine the unique value of these interventions to specific subgroups (e.g., by race/ethnicity or age). Accordingly, this project seeks to 1) synthesize data on the relationship between alcohol, methamphetamine, and HIV among MSM, including the impact of substance use on HIV risk behavior and the prevention-care continuum, 2) estimate the plausible range and sources of HIV infections attributable to alcohol and methamphetamine use among MSM using a principled and widely-used approach to network epidemic models (i.e., EpiModel), and 3) determine optimal and robust strategies for reducing HIV incidence among substance using MSM. For each aspect of this work, we will leverage advanced statistical and computational tools to rigorously calibrate our models, validate them against independent data sources, and perform extensive sensitivity analysis. To increase the usefulness of these models for real-world decision making, we will utilize uncertainty quantification to ensure the identified strategies are most likely to succeed after accounting for potential inaccuracy in our model parameters and assumptions. All model development will be conducted using open-source software enabling easy replication, modification, and extensions by other researchers. The project’s team is exceptionally well positioned to achieve these goals with expertise spanning network analysis, drug use epidemiology, epidemic modeling, and high-performance computing. Finally, dissemination activities are designed to directly inform key stakeholders in order to reduce HIV incidence and maximize the impact of this project on HIV elimination efforts.