Modélisation de l'efficacité populationnelle du vaccin contre le virus du papillome humain au Canada
|Authors:||Van de Velde, Nicolas|
|Advisor:||Boily, Marie-Claude; Brisson, Marc|
|Abstract:||Objective: The two main objectives of this thesis were to develop 1) mathematical models to predict the population-level impact of HPV vaccination in Canada, and 2) methods to quantify uncertainty around model predictions. Methods: We developed three mathematical models: 1) a static compartmental model of cervical cancer natural history (Model 1), 2) an individual-based dynamic model of HPV infection (Model 2), and 3) the first individual-based transmission-dynamic model of partnership formation and dissolution, and natural history of multi-type HPV infection and disease (anogenital warts, and cervical, anogenital and oropharyngeal cancers) (Model 3). For each model, an extensive fitting procedure was conducted, which identified multiple posterior parameter combinations (out of hundreds of thousands of prior parameter sets) that fit simultaneously highly stratified behavioral and epidemiologic data, taken from the literature, population-based datasets, and original studies. Parameter uncertainty was illustrated by presenting the median [10th; 90th percentiles] of predictions, using the posterior parameter combinations. Sensitivity analysis was conducted varying vaccine efficacy, duration of protection, coverage and vaccination strategies. Results: We provided the following evidence for HPV vaccination recommendations. Models 1-3 predicted that girls-only HPV vaccination can substantially reduce HPV-related burden of disease. Predictions were most sensitive to duration of vaccine protection. Model 3 predicted that the bivalent vaccine will be slightly more effective at preventing cervical cancer in the longer term. However, the quadrivalent vaccine will substantially reduce anogenital warts. Finally, the candidate nonavalent vaccine has the potential to produce substantial incremental benefits if its efficacy and duration of protection are at least 85% and 30 years, respectively. From a methodological point of view, we illustrated that parameter uncertainty surrounding HPV natural history parameters is important and must be presented when providing predictions to decision makers. Finally, we identified key structural assumptions that influence predictions: herd immunity, natural immunity, partnership duration, individual genotypes and vaccine waning function. Conclusion: We developed increasingly sophisticated HPV models and calibration techniques to keep track with the increasingly complex policy questions being asked. Our final model is being used to examine the impact of HPV vaccination on health inequalities, evaluate the cost-effectiveness of HPV vaccination, and optimize screening.|
|Document Type:||Thèse de doctorat|
|Open Access Date:||19 April 2018|
|Collection:||Thèses et mémoires|
All documents in CorpusUL are protected by Copyright Act of Canada.