D4.4 Report describing the tool and some retrospective applications for validation

The present deliverable of the FLASH project aims to provide a tool to improve payers' and manufacturers' ability to predict the budget impact of pharmaceutical innovations over time in situations that present complexity that more standard approaches that more standard approaches do not fully address. This is done by relying on the properties of Markov models, which are widely employed in economic evaluations of new technologies but not in budget-impact analyses.Unlike previous instances of the use of Markov models in budget-impact analysis, the proposed approach is exible in that it is not it is not condition or product speci c. The main conceptual di erence between the proposed approach and the way BIA is typically carried out is the ability of the former to let the treatment period contribute to the de nition of the Markov state. By doing this, the model can account for some circumstances that are typically ignored in standard BIAs, such as the role of Payment-by-Result agreements. Moreover, in this setting, changes in the size of the treated population that occur over time, e.g. due to improved e ectiveness, are naturally accounted for. We start by illustrating a general version of the model, which allowsto account for several details of the speci c situation, but is potentially demanding in terms of number of input parameters. Then we show that, by introducing some additional assumptions, a data-parsimonious version of the model can be implemented, requiring only information that is typically available to decision-makers at the time when a new product reaches the market.The model is validated through a retrospective application to the case of adoption of a new cancer drug from the perspective of a regional payer in Italy. This exercise shows good ability of the model to predict actual expenditure, based on the information that was available at the time of adoption. An example involving several treatment sequences and several treatments in each sequence is used to illustrate how the model can be used in practice and the the information that can be drawn from the model output.The proposed tool is developed with a focus on cancer medicines, where the challenges tend to be even more serious than in other therapeutic areas. However, there is nothing to prevent the model from being used in other clinical areas, as long as the information available allows the necessary input parameters to be estimated.The code that allows to run the model to obtain a prediction of the budget impact over time will be made publicly available as a separate deliverable of the FLASH project and it will be made publicly available on the project web-page as soon as possible.