The economic evaluation of influenza vaccination in the elderly population: a model based on Bayesian Networks

Influenza infection is a major cause of illness, morbidity and mortality throughout the world, affecting 5-15% of global population each year. The high-risk groups of influenza complication include mainly elderly and patients with cardiovascular or pulmonary disorder, and metabolic disease (diabetes). The influenza vaccination has proved to be effective in reducing acute complications among high-risk patients, particularly in influenza-like illness, hospitalisation and mortality from all causes. The burden of influenza on health care systems becomes highly relevant, since the increasing proportion of population aged over 65. Therefore, substantial public health implications arise from the decision maker perspective, when programming an annual vaccination strategy. In particular, the public decision maker faces a strategic problem when deciding: a) whether to implement a vaccination campaign, and b) which vaccine(s) to prescribe. On the one hand, a vaccination program can result in an effective strategy, both from the financial and the clinical point of view, in light of the reduction of hospitalisations and deaths among the elderly. On the other hand, the cost effectiveness of this choice is highly dependent on some exogenous parameters, such as influenza attack rate (which is likely to vary from year to year), and coverage rate (i.e. the proportion of patients that are actually vaccinated). Given the presence on the pharmaceutical market of at least two types of influenza vaccines (the standard and the more recent subunit vaccine adjuvated with MF59), in order to take into account pros and cons, over the entire population, the decision should be taken integrating pre-constituted opinions and empirical data. The aim of this paper is then to build a decision model which allows the decision makers to evaluate the possible results under different scenarios, and to choose the decision associated to the highest expected utility, expressed in terms of incremental cost effectiveness ratio (ICER). The paper is structured as follows: section 2 presents the statistical approach used in the definition of the model and section 3 describes the main results and conclusions obtained.