IN3 Translating Outcomes from a Dynamic Transmission Model for Varicella Vaccination to Cost-Effectiveness Estimates: The Impact of Different Analytic Approaches on the Results

Compare three different analytic approaches to estimate the cost-effectiveness of a varicella vaccination programme using clinical outcomes from a dynamic transmission model. An age-structured SIR (susceptible, infectious, recovered) dynamic transmission model was developed to predict the impact of routine infant vaccination on varicella incidence. Individuals transitioned between S and I compartments based on UK force-of-infection data. Each compartment was stratified into 8 age groups to track individuals as they aged over time. Input parameters including force of infection, who-acquires-infection-from-whom (WAIFW) matrix structure, vaccine efficacy, vaccine coverage, costs, QALYs, and demographic data were based on published UK data. The model estimated the incremental cost-effectiveness ratio (ICER) for the vaccination programme using three analytic approaches: summing outcomes 1) for the entire population cumulatively over time (CumPop), 2) for the entire population for the steady-state year (SSPop), and 3) for the lifetime of the first vaccinated birth cohort (Cohort). Costs and QALYs were discounted at 3.5% per year. ICERs were compared for the three analyses for different time horizons and vaccine coverage rates. The vaccination programme reached a steady state after 75 years. For this time horizon, the incremental costs per QALY gained for the CumPop (£1,407) and SSPop (£868) analyses were 47% and 68% lower than for the Cohort (£2,678) analysis. In the CumPop and Cohort analyses, the ICER decreased as the model time horizon increased and as vaccine coverage increased. The ICER for the CumPop analysis was always lower than for the Cohort analysis; in the steady-state year, the ICER for the SSPop analysis was the lowest. Cost-effectiveness estimates using data from dynamic transmission models differ depending on the analytic approach, time horizon, and coverage rate used, with the two population approaches yielding lower ICERs because they better capture the full population benefit of herd protection.