*Harvard PhD Program in Health Policy Alumnus & Faculty Member
Dissertation Title: "Mathematical Modeling to Evaluate Disease Control Policy"
In this dissertation I assessed three distinct policy questions: the implications of introducing a new tuberculosis diagnostic in southern Africa, the potential value of research related to HIV treatment policy in South Africa, and the causal effect of state cigarette taxes imposed between 1996 and 2013 on health outcomes in the United States.
In Chapter 1 I investigated use of a new TB diagnostic, Xpert MTB/RIF, to replace sputum smear-based diagnostic algorithms in southern Africa. Analyses were undertaken using a dynamic mathematical model, taking account of TB transmission and natural history, HIV epidemiology, TB drug resistance, and disease control interventions. Results suggest Xpert may substantially reduce TB morbidity and mortality, and modestly reduce transmission, while significantly increasing demands on healthcare resources. Xpert adoption was estimated to cost $711 to $1,083 per DALY averted, representing good value for money according to conventional benchmarks.
In Chapter 2 I estimated the value of new research on various targets relevant to HIV treatment policy in South Africa. I implemented the analysis with a mathematical model of HIV epidemiology, simulating HIV transmission, disease progression, and receipt of treatment. I used Value of Information (VOI) methods to identify priority research areas, based on the welfare gains possible by obtaining better information prior to decision-making. High priority research targets included issues of cost and implementation, relative infectiousness during late HIV, the reduction in infectiousness following treatment initiation, and the therapeutic benefits of early treatment initiation.
In Chapter 3 I estimated how changes in state cigarette taxes over the period 1996-2013 affected smoking behavior and health outcomes in the United States. The causal effect of tax changes were estimated in the context of a demographic model directly linking smoking behavior with subsequent mortality risks, using data from the Behavioral Risk Factor Surveillance System. Results suggest national cigarette consumption by 2013 was 4.0% lower (3.3, 4.6) compared to a counterfactual with no new state cigarette taxes after 1996, averting 27 thousand (22, 34) deaths and producing 119 thousand (92, 151) extra life-years lived. The health impact of these taxes was projected to be far greater in future years.