Observational cohorts are, and will continue to be, indispensable to evaluate clinical strategies for HIV-infected patients. The most clinically relevant strategies are dynamic, that is, strategies that incorporate the patients' time-varying clinical history in the clinical decision. Unfortunately, conventional statistical methods cannot appropriately compare dynamic strategies. Rather methods specifically designed to deal with dynamic strategies and time-varying confounders are needed. We propose to continue to develop analytical methods to enhance the validity of estimates from observational HIV cohorts. We will build upon our previous collaborative efforts to compare the effectiveness of combined antiretroviral therapy (cART) of both static and dynamic strategies. We propose to answer key clinical questions about the management of HIV-infected patients by applying innovative analytic methods to complex observational data. These questions include 'when to switch cART?', 'what cART to start?' (with an eye towards 'when to switch?' and 'what to switch to?'), and 'when to monitor immunologic and virologic parameters?' for the management of HIV patients. Our primary data source will be the HIV-CAUSAL Collaboration, and we will establish collaborations with other HIV consortia in Europe, the U.S., and Africa. We will generate and maintain user-friendly, publicly-available software and detailed documentation to make these methods available to the HIV research community. Our work will implement and extend cutting-edge approaches to emulate randomized trials using observational data, including methods based on inverse probability of weighting of dynamic marginal structural models and the parametric g-formula. One reason for the lack of widespread use of these methods is their perceived technical complexity, which has made them appear inaccessible to many practitioners. After two decades of methods and software development, our group is now applying innovative methods to large prospective HIV cohorts, and showing that these methods can be added to the standard arsenal of HIV researchers. This proposal will strengthen the international role of the HIV-CAUSAL Collaboration as an incubator for new methodologies in observational HIV research.

Public Health Relevance

We will answer key clinical questions about the management of HIV-infected patients by applying innovative analytic methods to complex observational data. We will develop the methods, implement them, show that they work, and make them available to HIV researchers.

National Institute of Health (NIH)
National Institute of Allergy and Infectious Diseases (NIAID)
High Priority, Short Term Project Award (R56)
Project #
Application #
Study Section
AIDS Clinical Studies and Epidemiology Study Section (ACE)
Program Officer
Mckaig, Rosemary G
Project Start
Project End
Budget Start
Budget End
Support Year
Fiscal Year
Total Cost
Indirect Cost
Harvard University
Public Health & Prev Medicine
Schools of Public Health
United States
Zip Code
Hernán, Miguel A (2016) Does water kill? A call for less casual causal inferences. Ann Epidemiol 26:674-680
O'Hagan, Justin J; Lipsitch, Marc; Hernán, Miguel A (2014) Estimating the per-exposure effect of infectious disease interventions. Epidemiology 25:134-8
Supervie, Virginie; Ndawinz, Jacques D A; Lodi, Sara et al. (2014) The undiagnosed HIV epidemic in France and its implications for HIV screening strategies. AIDS 28:1797-804
Hernán, Miguel A; Hernández-Díaz, Sonia; Robins, James M (2013) Randomized trials analyzed as observational studies. Ann Intern Med 159:560-2