Public health studies aimed at behavior modification, earlier HIV treatment, and linkage to HIV care among people who use drugs (PWUDs) have made progress towards reducing HIV incidence not only for those directly-treated but also their HIV risk networks. Only some members of the networks assigned to the intervention are actually exposed. In such studies, estimation of individual and disseminated effects is of interest. The individual effect is the effect on the participants who directly received the study intervention beyond being in an intervention network and the disseminated effect is the effect on the participants who shared a network with the directly-treated participant. This project aims to advance HIV treatment and prevention research by contributing new methodology to study causal mechanisms in networks of people who use drugs at high risk for HIV infection. We will develop causal inference statistical methods combined in novel ways with network science to solve some of the remaining challenges for analysis of data from network-based studies of PWUDs. Development of sample size and power formulas for this setting will facilitate conduct of adequately- powered studies better equipped to determine effective HIV treatment and prevention measures among PWUD networks. Results of this work will be motivated and applied to the following studies: Transmission Reduction Intervention Project, HIV, HCV, and STI Risk Associated with Nonmedical Use of Prescription Opioids, STEP into Action, and HIV Prevention Trials Network 037, where there will be individual effects on participants directly receiving interventions or exposures and disseminated effects on their network members in modifying HIV risk behavior, lowering HIV incidence, and improving HIV care. These methodologies will improve the quality of information from network-based studies, expanding the knowledge base of best HIV preventative and treatment practices among this subpopulation, and leveraging network-based effects to reduce risk and improve HIV treatment and prevention. I am well suited to perform this research based on 1) my past research experience in causal inference methods for HIV/AIDS research, 2) the exceptional collaborative team I have assembled to ensure that this research is of the highest quality, including two collaborators with expertise in drug abuse research, and 3) the unique existing study data sources included in this project. Resources at the University of Rhode Island will provide a centrally-located environment to facilitate research, including collaborative opportunities with leading drug abuse and addiction researchers and world-class facilities with excellent computing and library systems. This project will ultimately lead to the development of interventions strengthened by network features to maximize achievable benefits to reduce onward HIV transmission and improve the continuum of care among PWUDs.
Public health studies aimed at behavior modification and earlier HIV treatment have made progress towards reducing HIV incidence not only for those directly-treated drug users but also their social or risk networks; however, research-to-date has been impeded due to a lack of methods to estimate how interventions permeate a risk network and evaluation of interventions has been based mostly on overall effects. I propose to develop statistical inference methods combined in novel ways with network science applicable in settings with social influence and when statistical dependence may be present. These methodologies will improve the quality of information from network-based studies of high-risk drug users, expanding the knowledge base of best practices among this vulnerable subpopulation, and leveraging network-based interventions to reduce risk and improve HIV treatment and prevention.
