Advertisement

HT1 EVIDENCE OF METHODOLOGICAL HETEROGENEITY IN THE NICE APPRAISAL PROCESS DEPENDING ON ERG: A REVIEW OF SURVIVAL EXTRAPOLATION APPROACHES IN ONCOLOGY

      Background

      The National Institute of for Health and Care Excellence (NICE) makes reimbursement recommendations estimating survival benefits of new oncology therapies based on extrapolations of immature survival data from clinical trials. Two main approaches to survival extrapolation are commonly used. These were proposed by Nicholas Latimer and Adrian Bagust, affiliated with the Liverpool Reviews and Implementation Group (LRiG) and Sheffield School of Health and Related Research (ScHARR) Evidence Review Groups (ERG) respectively. This research evaluates whether LRiG and ScHARR ERGs, assessing companies’ submission to NICE, have preferences for either of the two extrapolation approaches.

      Methodology

      All publicly-available NICE Single Technology Appraisals (STAs) in oncology critiqued by LRiG and SCHARR were reviewed (until 31/12/2019). Statistical comparisons were performed using Chi-square tests.

      Results

      60 oncology STAs were identified (LRiG reviewed 40, ScHARR 20). In 68% (41/60), Latimer’s approach was preferred by companies (LRiG: 65% [26/40]; ScHARR: 75% [15/20]). The ERG preferred approach was Latimer’s in 37% (22/60) of cases (ScHARR: 80% [16/20]; LRiG: 15% [6/40], p<0.001) and Bagust’s in 63% (38/60) (ScHARR: 20% [4/20]; LRiG: 85% [34/40]). LRiG were significantly more likely to prefer Bagust’s approach when the company’s base case adopted Latimer’s (77%, 20/26) than ScHARR (7%, 1/20). ScHARR agreed with Latimer’s approach more often (93%, 14/15) than Bagust’s one (50%, 3/6) while LRiG never preferred Latimer’s approach to Bagust’s when the latter was used in the companies’ base case (0%, 0/14). This propensity for different approaches was statistically significant (p<0.001).

      Conclusions

      Latimer’s fully parametric approach was preferred over Bagust’s piecewise in most companies’ submissions, independently of the ERG reviewing the analyses. A significant preference for Bagust’s approach was observed by LRiG, while ScHARR appeared neutral to submitted methodologies. This heterogeneity might translate into different potential survival estimates and therefore appraisal outcomes depending on the ERG assigned to critique companies’ submissions.