Methods. A dynamic population-based influenza transmission model was developed to predict the number of expected influenza cases in an average influenza season in Toronto (population 2.5 million). The results of this model were used in a traditional decision-analytic model to estimate effectiveness and costs of the three intervention options: 1) UIIP, 2) targeted immunization strategy aimed at risk populations and those who may transmit influenza to those at risk, and 3) no immunization. The population of Ontario was stratified into children, adults and elderly. Influenza transmission parameters and probabilities of events, effectiveness and utilities for final health outcomes were derived from the literature. Unit costs were obtained from Ontario cost and fee schedules. Deterministic and probabilistic sensitivity analyses were used to assess uncertainty.
Results. Preliminary analyses suggest that an average influenza season results in approximately 0.8 to 1.8 million infected individuals 30% of which are expected to develop symptomatic illness. The deterministic cost-utility analysis suggests UIIP dominates the other two strategies and results in 3,000 QALYs gained and C$471,500 saved compared to a targeted vaccination strategy. One-way sensitivity analyses revealed that results are sensitive to basic reproductive number, proportion developing symptoms, vaccine efficacy and coverage, costs of vaccine and vaccine administration, hospitalization costs for the elderly, and the probability of developing symptoms in the elderly. However, probabilistic analysis shows that compared to a selective vaccination strategy the probability of UIIP being cost-effective is > 80% at a willingness to pay threshold of C$50,000.
Conclusions. This first analysis indicates that Ontario's UIIP may dominate the other two strategies and may also have the highest probability of being cost-effective. However, the results of this model are subject to several limitations (data availability and quality for influenza transmission parameters and program efficacy, exclusion of program implementation costs). As new information becomes available (e.g. an evaluation of UIIP's program efficacy is currently underway) these limitations will be addressed and the cost-utility of the program will be re-evaluated. Improving the model may shift the balance between the vaccination programs.