Meeting Brochure and registration form      SMDM Homepage

Sunday, 23 October 2005 - 11:00 AM

QUANTIFYING UNCERTAINTY OF PREDICTED HPV VACCINE EFFECTIVENESS

Marc Brisson, PhD, Merck Frosst Canada ltd, Montreal, QC, Canada, Nicolas Van de Velde, MSc, Université Laval, Montreal, QC, Canada, and Marie-Claude Boily, PhD, Imperial College, London, United Kingdom.

Purpose: Clinical trials have shown prophylactic HPV vaccines to be effective against persistent HPV infection, cervical intraepithelial neoplasia (CIN) and genital warts. To aid policy decisions concerning HPV vaccination, model based estimates of population-level effectiveness and cost-effectiveness are needed. The process of model development requires choices and assumptions, which introduce uncertainty. Uncertainty should be quantified to provide policy makers with the necessary information to make appropriate decisions. The aim of this study is to examine the impact of model and parameter assumptions on the effectiveness of HPV vaccination.

Methods: A cohort model of the natural history of HPV infection (types 16, 18, and other Low Risk and High Risk types) was developed to investigate the effectiveness of HPV vaccination at preventing infection, CIN, cervical cancer and genital warts in Canada. Natural history parameter sets, which fit simultaneously North-American longitudinal and cross-sectional data (incidence and prevalence of HPV infection, CIN and cervical cancer), were identified as follows: 1) Uniform prior distributions were defined for each parameter using minimum and maximum values found in the literature; 2) >20 000 sets of different parameter combinations were selected from prior distributions using Latin Hypercube Sampling, and 3) GOF techniques were used to identify “best fit” parameter sets. In combination with multivariate analyses, “Best Fit” parameter sets were used to assess the sensitivity of HPV vaccine effectiveness to plausible model and parameter assumptions.

Results: Assuming the proportion of individuals protected following immunization (take) is 100%, reduction in susceptibility to HPV-16/18/6/11 infection (degree of protection) is 95% and duration of protection is life-long, the model estimates of the type-specific vaccine effectiveness against infection, CIN and cervical cancer ranges between 80-95%. The effectiveness of vaccination is most sensitive to vaccine efficacy (take, degree and duration), level of immunity following natural infection and the age-specific progression and regression rates.

Conclusion: Preliminary model results suggest that vaccination against HPV 16/18/6/11 has the potential to reduce HPV infection and disease. Our modeling framework has the advantage of integrating multivariate sensitivity analysis on all the natural history and vaccine parameters when assessing the effectiveness and cost-effectiveness of vaccination. Hence, the model can provide a non-subjective, flexible and robust framework for decisions regarding implementation of new screening and HPV vaccination programs.


See more of Oral Concurrent Session F - Simulation
See more of The 27th Annual Meeting of the Society for Medical Decision Making (October 21-24, 2005)