PRM7 Using Probabalistic Sensitivity Analysis in Budget Impact Models to Reduce Uncertainty and Improve Decision-Making

Abstract

Budget impact analysis (BIA) is formally required by many national HTA regulatory agencies including NICE and the PBAC, in the UK and Australia, respectively. Current practice only involves the use of point estimates to serve as “best guess” for decision-makers. However, using probabilistic sensitivity analysis (PSA) can serve to reduce parameter uncertainty in order to generate discussion and ultimately improve decision-making. Using the same techniques applied to cost-effectiveness analysis, a PSA was incorporated into a budget impact model used for a client's medical device. This involved creating and running a Monte-Carlo simulation (MCS) over 10,000 iterations to generate a 95% confidence interval (CI) around the overall budget impact in addition to a probability curve. Point-estimate budget impact was found to be a saving of £4,736,893 based on a number of pre-defined input parameters in the model. Running a MCS generated a 95% CI: a saving of £10,367,403 and an incremental cost of £861,166 either side of the point-estimate. In addition, a probability curve was generated with overall budget impact on the x-axis and probability on the y-axis. 25 data points were generated running from a maximum potential saving of approximately £12m (1% probability) to an incremental cost of approximately £3m (100% probability). Using PSA in this budget impact model demonstrates that there is a significant likelihood this medical device could actually generate an incremental cost rather than saving (which the point-estimate shows). This serves as an example of how using this technique could serve to generate discussion among decision-makers in order to make more informed and improved budget impact decisions in the future.