OBJECTIVES:To assess which technology appraisal characteristics of oncology treatments are most strongly associated with a positive recommendation from NICE for regular commissioning or via the Cancer Drugs Fund (CDF).

METHODS: 60 tumour-related NICE oncology appraisals published since June 2016 were reviewed. Data extracted included characteristics of the decision problem (such as population and number of comparators) and the company submission (including source of clinical evidence, use of indirect treatment comparisons, cost-effectiveness model structure and results). A binomial generalised linear model was fit with a probit link function using R. Stepwise forwards selection identified model parameters in descending order of significance of correlation with a positive recommendation. The Akaike Information Criterion (AIC) was reviewed for each model; if the introduction of a parameter raised the AIC compared to the previous model, the previous model was selected.

RESULTS: Of the 69 decisions issued by NICE, 40 (53.6%) were positive, with an additional 16 (27.5%) recommended via the CDF only. 96% of submissions with a probability of more than 50% of being cost-effective were recommended for regular commissioning or the CDF. The key drivers of a positive recommendation as identified by the regression analysis were: the probability of cost-effectiveness and the number of meetings held by the NICE Committee. These coefficients correctly predicted 91% of the technology appraisals, where the probability of cost-effectiveness was reported, and were jointly significant (p<0.01). The higher the probability of being cost-effective, the higher the probability of being approved, whilst more than one committee meeting was negatively associated with approval.

CONCLUSIONS: The majority of oncology appraisals since June 2016 have received a positive NICE recommendation for regular commissioning or via the CDF and there is evidence to suggest that the key drivers were not associated with the decision problem or methodology but the probability of being cost-effective.