We develop new statistical methods for investigation and scale-up of HIV prevention interventions. Challenges arise from the complex dependencies that characterize data from HIV prevention research studies, reflecting the spread of HIV along the sexual contact networks we consider. The data generated from both randomized and observational studies include HIV incidence in different subpopulations, self-reported behavior regarding partner selection, and viral genetic sequences. Such data are likely to be incomplete because of failure to locate intended participants, denial of consent, and dropout. Furthermore even complete data would not allow networks to be fully characterized, or viral genetic linkage analyses to be conducted with certainty. Our methods are intended to make maximal use of incomplete information to estimate quantities that will be useful in guiding scale-up of successful interventions. These include not only the randomized effects of interventions, but also expected effects under policies of delivering them in ways that are likely to be used practice. Optimal scale-up requires knowledge of where and in which populations the interventions succeeded (fully or partially) and to identify factors, such as network features, that predict success. To investigate these questions, we extend methods developed in the first grant period for investigation of viral genetic linkage of incident and prevalent HIV infections. Such linkage provides information about the extent to which new infections arise from strains circulating within or across subpopulations defined by demographic factors, such as age, gender, and residence. It also provides insight into HIV transmission dynamics, such as identifying bridge populations that facilitate entry of HIV into new populations and the impact of interventions on such spread. During the first grant period, we focused on cross-sectional data collected in single villages; here we consider data collected across multiple villages over time that permit investigation of spatio-temporal HIV dynamics. Building on existing network theory and our own prior work, we also propose new methods to investigate how best to use prevention interventions to functionally fragment sexual contact networks and thereby impede epidemic spread.

Public Health Relevance

Development of analytical methods is targeted to estimate quantities that will help guide policies regarding how best to target HIV prevention interventions, choose among them, and prioritize resources for their scale-up. We also describe application of new methods to data from HIV prevention research studies that may be incomplete because of denial of consent, dropout, or other mechanisms.

National Institute of Health (NIH)
Method to Extend Research in Time (MERIT) Award (R37)
Project #
Application #
Study Section
Special Emphasis Panel (NSS)
Program Officer
Gezmu, Misrak
Project Start
Project End
Budget Start
Budget End
Support Year
Fiscal Year
Total Cost
Indirect Cost
Harvard University
Biostatistics & Other Math Sci
Schools of Public Health
United States
Zip Code
Stephens, Alisa; Tchetgen Tchetgen, Eric; De Gruttola, Victor (2014) Locally efficient estimation of marginal treatment effects when outcomes are correlated: is the prize worth the chase? Int J Biostat 10:59-75
Tchetgen Tchetgen, Eric (2014) The control outcome calibration approach for causal inference with unobserved confounding. Am J Epidemiol 179:633-40
Goyal, Ravi; De Gruttola, Victor; Blitzstein, Joseph (2014) Sampling Networks from Their Posterior Predictive Distribution. Netw Sci (Camb Univ Press) 2:107-131
Carnegie, Nicole Bohme; Wang, Rui; Novitsky, Vladimir et al. (2014) Linkage of viral sequences among HIV-infected village residents in Botswana: estimation of linkage rates in the presence of missing data. PLoS Comput Biol 10:e1003430
Wirth, Kathleen E; Tchetgen Tchetgen, Eric J (2014) Accounting for selection bias in association studies with complex survey data. Epidemiology 25:444-53
Goodreau, Steven M; Carnegie, Nicole B; Vittinghoff, Eric et al. (2014) Can male circumcision have an impact on the HIV epidemic in men who have sex with men? PLoS One 9:e102960
Wang, Rui; Goyal, Ravi; Lei, Quanhong et al. (2014) Sample size considerations in the design of cluster randomized trials of combination HIV prevention. Clin Trials 11:309-318
Gleeson, James P; Cellai, Davide; Onnela, Jukka-Pekka et al. (2014) A simple generative model of collective online behavior. Proc Natl Acad Sci U S A 111:10411-5