Bayesian modelling assessing the effectiveness of a vaccination strategy to prevent HPV-related diseases: the BEST study

Andrea Marcellusi, Università di Roma "La Sapienza"
Patrizia Giannantoni, Università di Roma "La Sapienza"
Gianluca Baio, University College London
Alessandro Capone, Kingston University
Francesco Saverio Mennini, Università degli Studi di Roma "Tor Vergata"

Background: Cervical cancer is the second most frequent cancer among women worldwide. Each year, 500,000 new cases of cervical cancer are estimated, resulting in about 260,000 deaths in 2005. The risk of been diagnosed of cervical cancer in the life course is 6,2‰, whereas the risk of die is 0,8‰. Human Papillomavirus (HPV) is associated with cervical cancer. An excess of uncertainty associated with the main parameters of commonly utilized models can be observed. The aim of this study was to assess the impact of a quadrivalent-based HPV vaccination strategy on the life expectancy of Italian women using a statistical Bayesian approach. Methods: A full Bayesian Markov model was used, where all unknown quantities were associated with suitable probability distributions reflecting the state of science currently available. These distributions were updated by the observation of Italian available data, and uncertainty was propagated through the entire model with a Markov Chain Monte Carlo procedure. The model was calibrated using age-specific incidence of invasive cervical cancer data derived from Italian female population. It was considered as follow-up population one cohort of girls vaccinated at age 12. The model extrapolated the estimated number of deaths for each year of simulation and each prevention strategy (None, Sreening, Screening+Vaccination) and, through multistate life tables, we compared their life expectancies. Results: Vaccination strategy seems to improve women’s life expectancy of 0.78 year with respect of Screening program. The largest increase in life expectancy (+0.81) was observed at the age 26. Comparisons between prevention and non-prevention policies are still under investigation but we expect major improvements in life expectancy at all ages. Conclusions: The developed Bayesian model reduced the uncertainty associated with HPV vaccination. The quadrivalent-based HPV vaccination programme provided excellent effectiveness results and the life expectancy of the Italian vaccinated women is expected to improve.