Benefits of probabilistic sensitivity analysis – a review of NICE decisions

Abstract

Objective: Since 2004, the National Institute of Health and Clinical Excellence (NICE) has required manufacturers to conduct a probabilistic sensitivity analysis (PSA) in their technology appraisals. The objective of this review is to assess the cost-effectiveness of different technology appraisals and compare them with the actual decision made by the NICE based on PSA.Methods: The search term ‘probabilistic sensitivity analysis’ was used on the NICE home page (25 January 2012). The appraisals identified in the search were assessed and subjected to further review, if a probability of being cost-effective was provided, regardless of the threshold indicated. If several probabilities were provided, the number provided by the evidence review group was used. If several scenarios were presented, the base case scenario was chosen. Finally, the probabilities of being cost-effective were compared with the actual decision made, which could result in two outcomes: recommended or not recommended.Results: A total of 31 assessments were included for the final review. The results were plotted on a graph to illustrate whether there was a relationship between the PSA outcomes and the final recommendation. The assessments were ranked according to their probability of being cost-effective.Conclusion: A higher probability of a technology being cost-effective was correlated with more positive decision-making. There appeared to be a clear threshold at which technologies with a 40% certainty of being cost-effective tended to be recommended, whereas those below the threshold were not recommended. The reports suggested that the incremental cost-effectiveness ratios (ICER) estimate was not a robust driver of decision-making. A NICE applicant should pay increased attention to the PSA in addition to the ICER estimate.