Dissertation Title:  "Cervical Cancer Prevention: Using Primary Data to Inform Decision-Making in Developed and Developing Country Contexts"

Globally, cervical cancer is the third most common cancer in women, with more than 85% of cases occurring in developing countries. In the developed world, screening programs based on frequent cytologic tests have reduced cervical cancer incidence and mortality through detection and treatment of precancerous lesions. Cytologic screening programs have not been feasible in low-income countries. Recent advances — including two human papillomavirus (HPV) vaccines that protect against oncogenic genotypes HPV16/18 and a rapid HPV DNA test that allows for same-day screening and treatment of precancerous lesions in resource-poor settings — are promising developments in the field of prevention. This dissertation seeks to evaluate the cost-effectiveness of cervical cancer prevention strategies in developed and developing country contexts, and to examine aspects of the natural history of HPV that may inform clinical decision-making.

Chapter 1 assesses the health and economic impact of HPV16/18 vaccination for pre-adolescent girls and cervical cancer screening options for adult women in Eastern Africa. A simulation model of cervical carcinogenesis was calibrated to epidemiologic data from five countries and used to comparatively assess primary and secondary prevention strategies within each setting. Pre-adolescent vaccination, followed by a single screening with HPV DNA testing between ages 35 and 40, reduced cancer incidence by 50%. Findings indicate that if the cost of vaccination is low, and coverage of pre-adolescent girls is feasible, vaccination provides good value for public health dollars in Eastern Africa.

Chapter 2 examines predictors of oncogenic HPV persistence, which is a known risk factor for progression to precancer and cancer. Applied to primary, longitudinal data from a population-based cohort in Costa Rica, discrete-time survival analysis and Generalized Estimating Equations demonstrated that the presence of additional HPV genotypes does not change the likelihood that an infection will persist.

Chapter 3 evaluates the cost-effectiveness of age-based management protocols for women in the United States with low-grade cervical abnormalities. A decision analytic model informed by data from the ASCUS/LSIL Triage Study suggested HPV DNA triage was the most effective strategy to detect precancer, but adjustment to control for biases in trial design suggested that immediate colposcopy may be a more effective and economically attractive alternative.