Grant: Fellowship: Bisby (Prize)

Significantly reducing HIV incidence will require the combination of behavioral, biomedical, and structural interventions. Conducting rigorous impact evaluation of such complex HIV interventions can be extremely challenging, however. By comprehensively assessing the effects of a health program, impact evaluations have been shown to generate new-evidence, promote accountability, and prioritize public health actions. Community-randomized controlled trials, often considered the gold-standard in medical research, are not always possible for logistical or ethical reasons. In addition, HIV transmission is inherently dynamic and this will further complicate the design, implementation, and analysis of impact evaluations. Mathematical models of disease transmission are able to take disease dynamics into account by explicitly and quantitatively describing relations between variables and a disease's natural history. Recent advances have enabled researchers to combine, within a Bayesian framework, mathematical models of disease transmission with statistical techniques. Although increasingly used, this method has not yet been thoroughly validated. The goal of this research is thus to strengthen impact evaluations of HIV prevention programs by building on these recent accomplishments. My first aim is to validate this approach by applying it to a simulated population (agent-based model). The second research aim of this proposal is to use mathematical models to improve design and implementation of impact evaluations of HIV intervention by providing guidance on critical data needed and design issues to be addressed. The outcomes of this fellowship are expected to improve impact evaluation of HIV prevention programs. Further, the developed framework could be applied to a wide range of infectious diseases such as HPV, tuberculosis, and malaria.