Dean for Academic Affairs, Harvard T.H. Chan School of Public Health
*Harvard PhD Program in Health Policy Alumna
Dissertation Title: "Using Mathematical Modeling to Evaluate the Public Health Impact and Cost-Effectiveness of Cervical Cancer Screening Strategies in Different World Regions"
Cervical cancer is the second most common cancer in women worldwide, with the majority of cases and deaths occurring in developing countries. Over the past few decades, our understanding of the natural history of cervical cancer, and specifically the role of human papillomavirus (HPV) as the causal agent of cervical cancer, has increased dramatically. This increase in knowledge has been accompanied by the rapid development of new technologies for cervical cancer screening, which has had important implications for different settings around the world. This dissertation comprises of three separate analyses that utilize distinct mathematical models and methods to evaluate policies relevant to cervical cancer screening.
Chapter I describes the development of a biological model that is used to identify the optimal management of an equivocal cytology result, atypical squamous cells of undetermined significance (ASCUS), in the United States. We conducted a comprehensive cost-effectiveness analysis comparing different strategies for the management of ASCUS, as well as a short-term analysis using data from a clinical trial over a two-year period. Despite independent input data, model structures, and time horizons, both analyses support similar conclusions about the attractiveness of HPV DNA testing as a management strategy for ASCUS.
In Chapter II, we sought to identify interventions that could be included in a package of health services targeted to women undergoing a one-time cervical cancer screen in developing countries. We developed an integer programming (IP) model to maximize health benefits, subject to budget and human resource constraints in four resource-poor regions. Inputs to the IP model were calculated using epidemiological models for seven different diseases. Our results suggest that the addition of other health interventions during a one-time cervical cancer screen can lead to substantial health gains. We conclude that future analyses used to inform resource allocation decisions would be enhanced by the inclusion of non-monetary constraints.
Chapter III presents a two-step calibration approach in the parameterization of a stochastic model of the natural history of cervical cancer. This model builds upon our previous models, but is modified to accommodate more recent data on HPV infection, as well as more complex screening protocols. In the first step of calibration, we capitalized on the availability of primary data from a longitudinal study, to identify a plausible range for each input parameter. In the second step, we conducted a comprehensive search over multiple inputs simultaneously to identify parameter sets that produced output that was consistent with independent data. We then used a sample of the best fitting sets to project a range of benefits and cost-effectiveness of different screening policies, and explored the policy implications associated with different model fit criteria.