Evaluating Human Papillomavirus (HPV) Vaccination Strategies in Canada using the Cancer Risk Management HPV Microsimulation Model (CRM-HPVMM)

Transcription

1 Evaluating Human Papillomavirus (HPV) Vaccination Strategies in Canada using the Cancer Risk Management HPV Microsimulation Model (CRM-HPVMM) ARCC Conference Monday, May 27, 2013 Presented by William Flanagan Statistics Canada The CRM-HPVM model has been made possible through a financial contribution from Health Canada, through the Canadian Partnership Against Cancer. The assumptions and calculations underlying the simulation results were prepared by "The Organization" and the responsibility for the use and interpretation of these data is entirely that of the authors. 1

2 Acknowledgements Keiko Asakawa Marc Brisson Ann Burchell Andy Coldman Laurie Elit William K Evans Natalie Fitzgerald Steve Gribble Fei Fei Liu Meg McLaughlin Anthony Miller Claude Nadeau Cathy Popadiuk Michael Wolfson Statistics Canada Université Laval University of Toronto BC Cancer Agency Juravinski Cancer Centre Juravinski Cancer Centre Canadian Partnership Against Cancer Independent Consultant Canadian Partnership Against Cancer London Health Sciences Centre University of Toronto Statistics Canada Memorial University University of Ottawa 2

3 Province Sex Age Year Cancer Risk Management Model (CRMM) overview Risk Factors Lifestyle Environmental Socio-Economic Status Presence of virus Screening Target populations Participation rates Various modalities New Treatment Cancer Incidence Treatment Progression Outcomes Cancer incidence Cancer deaths Resource needs Direct Health Care costs Cost per life-year gained Life expectancy Health-adjusted LE Economic impacts Δ Cost Δ Survival Δ Health utility 3

4 Cancer Risk Management Model (CRMM) overview CRMM HPV incidence and other inputs Demographics Economics Lung Cancer Colorectal Cancer HPVMM Cervical Cancer (Breast Cancer) (Prostate Cancer) 4

5 Sample key questions HPV vaccination Q: What are the impacts of Vaccinating girls of different ages Vaccinating boys Catch-up programs Vaccine efficacy and waning assumptions Cervical cancer screening Q: What would be the optimal cervical cancer screening program in terms of its health and economics impacts by Screening test modality (e.g., cytology or HPV DNA test) Screening frequency Age to begin and end screening HPV vaccination + screening Q: What balance between vaccination and screening will have the greatest impact? Q: How should screening change over time as more women who have been vaccinated come of screening age and the population has mixed vaccination status? 5

6 HPVMM: overview Continuous-time, interacting-agent, Monte-Carlo microsimulation model Model started with Van de Velde et al (2010) published model description; made enhancements HPV transmission through sexual contact network (age 10+) Six HPV groups explicitly modeled (16, 18, Other-HR, 6, 11, Other-LR) Bivalent and Quadrivalent vaccines can be assessed Little or no empirical data on model parameter values, limited empirical aggregate data 61 parameter values estimated through Latin Hypercube Sampling Find feasible parameter sets (solutions) that meet fit data targets 6

7 HPVMM: Parameter estimation Estimation is staged, each phase sampling selected parameters, as well as all the solutions of the previous phases In each phase, 50,000 simulations run (25,000 actors simulated per run) Assessed model fit to a total of 130 data points Over 6500 solutions found Phase Model Parameters [#of parameters] Empirical Constraints [# data points to fit] 1 Behaviour [41] Various (CCHS, Van de Velde et al) [105] 2 HPV-16 transmission & clearance [4], Sex act frequency [2],Lifetime natural immunity [2], Persistent infection [2] HPV-16 prevalence by age (FOCAL trial) [8] 3 HPV-18 [2] HPV-18 [1], HPV 16/18 co-infection [1] (Moore, adj.) 4 HPV-Other HR [2] HPV-HR prevalence by age (FOCAL trial) [8] HPV-Other HR [1], HPV-HR (Moore) [1] 5 HPV-6 [2] HPV-6 (Moore) [1] 6 HPV-11 [2] HPV-11 [1], HPV 6/11 co-infection [1] (Moore) 7 HPV-Other LR [2] HPV-Other LR [1], HPV-LR [1] (Moore) 7

8 Creating vaccination scenarios using the HPVMM web interface Activate / deactivate specific vaccine program element(s) Specify target vaccination age and sex Specify program implementation target year(s) Specify vaccine type Can target only those never been vaccinated Specify vaccine coverage (%) 3 parameters used to create comprehensive vaccination scenarios 8

9 Vaccinating boys adds protection to females Number of females infected with HPV 16 or 18 At year 50, reduction in prevalence by No vaccinating vaccination girls = 24% Vaccinate 12-yr old Reduction girls (50%) in prevalence by Vaccinate adding boys 12-yr = old 43% girls (50%) and boys (50%) HPVMM ; 250,000 actors simulated; 50% vaccination coverage; 100% of vaccinated population protected lifetime, 95% per-act transmission protection 9

10 Would the vaccine waning assumption influence the choice of the optimal age(s) at vaccination? 1,200,000 1,100,000 1,000,000 No vaccine waning: 100 % of population protected by vaccination for lifetime (# females infected with HPV 16/18) Greater prevalence reduction at year 50, if vaccinated at young age 1,200,000 1,100,000 1,000,000 Vaccine waning: Declining % of population protected by vaccination* (# females infected with HPV 16/18) Greater prevalence reduction at year 50, if vaccinated at ages , , , , ,000 Girls age 10 Girls age 12 Girls age 14 Girls age 16 Girls age , ,000 Girls age 10 Girls age 12 Girls age 14 Girls age 16 Girls age ,000 Benefit of vaccination depends on -2the 3 trade-off between % Years since vaccination Years since vaccination sexually active in the target population and vaccine waning assumption * Normally distributed with mean 20 and SD of 5 years (Distributional assumption taken from Van de Velde et al.) HPVMM ; 250,000 actors simulated; 70% vaccination coverage; Assumed 95% per-act transmission protection 10

11 Summary Web-based, user-friendly, accessible and transparent Only changes to a few parameters allows wide variety of vaccination scenario comparisons Model, documentation and source code (Modgen) available to registered users Incorporates parameter uncertainty in model outcomes Graphical representation of parameter uncertainty possible with simple steps Platform will be used to assess cost-effectiveness of various vaccination and screening programs 11

12 Interested in more information or workshops... Please contact CPAC!

13 Supplemental slides (to be shown if needed) 13

14 HPV - Cervical Cancer Logic Diagram HPV incidence Natural History Incidence & Staging Progression of Cancer Case-fatality Vaccination Screening Treatment of Cancer Treatment of non-cancerous lesions Direct health care costs and Impact on Health tracked throughout Persons are at risk of other causes of death as well 14

15 HPVMM: Fit result (example) Fit data target range (95% CI) from FOCAL data (pre-specified) Each line represents age groupspecific distribution of prevalence generated from a certain combination of parameters Source: Take2, Phase 2 run 15

16 HPVMM web model Vaccine proportion protected: % of a vaccinated population which remains protected as a function of the number of years since vaccination Value = 1 means that all (100%) of the vaccinated population is protected (with a specified degree of per-act protection described in a next slide) When 10 years are passed since vaccination, 98% of the vaccinated population remains protected Users can specify values up to 40 years since vaccination (the value at 40 years continues for 40+ years) 16

17 HPVMM web model Vaccine degree of protection: per-act probability of virus protection : : Values represent per-act probability of protection, i.e., the probability of protection decreases exponentially as the number of sexual act increases Users can specify values up to 40 years since vaccination (the value at 40 years continues for 40+ years) 17

18 Number of females infected with HPV 6/11 Vaccinating 12-year girls only vs. Adding 12-year boys At year 50, reduction in mean prevalence by vaccinating girls Reduction = 16% in mean prevalence by adding boys = 25% 18 HPVMM ; 250,000 actors simulated; 100% of vaccinated population protected lifetime, 95% per-act transmission protection

19 Number of females infected with HPV 16/18 Vaccinating 12-year girls only vs. Adding catch-up program Immediate benefit of catch-up program Catch-up program adds vaccination to year old girls, only at the first year of the vaccination program. 19 HPVMM ; 250,000 actors simulated; 100% of vaccinated population protected lifetime, 95% per-act transmission protection

20 Number of females infected with HPV 6/11 Vaccinating 12-year girls only vs. Adding catch-up program Immediate benefit of catch-up program Catch-up program adds vaccinating year old girls only at the first year of the vaccination program 20 HPVMM ; 250,000 actors simulated; 100% of vaccinated population protected lifetime, 95% per-act transmission protection

21 (For Bill s information) could be deleted completely Useful feature: In CRMM, parameter values can be also displayed graphically to visualize their distributional assumptions. Normally distributed with mean = 20 and standard deviation of 5 : : 21

22 HPVMM web model 3 parameters used to create comprehensive vaccination scenarios Specify vaccination programs - Vaccination age(s) - Sex - Program implementation target year(s) - Vaccination coverage (%) - Vaccine type (bivalent or quadrivalent) 22