Proportional Hazards in Survival Analysis: A Review of the Variation in Determining and Presenting PH in NICE STAs

OBJECTIVES: Technology appraisals (TAs) submitted to the National Institute for Health and Care Excellence (NICE) describe cost-effectiveness analyses conducted on medical therapies, including survival analysis carried out on time-to-event (TTE) data. This research aims to explore: (1) variation in the reporting of methods of testing for proportional hazards (PH) in company submissions in TAs; (2) the statistical methods used to deal with the results from the PH testing; and (3) the acceptability to the NICE external assessment/evidence research group (EAG/ERG) committee.

METHODS: All publicly available NICE single technology assessments (STAs) in oncology were identified up to April 13, 2023. From these information on the PH tests conducted (and their reporting) and EAG/ERG opinion on the related methods were extracted and analysed descriptively.

RESULTS: The 92 most recent NICE STAs were extracted, reflecting 97 distinct treatment comparisons, of which 90 included survival analysis. In 63 appraisals, the survival analysis focused on within trial comparisons, with 79.4% of these mentioning the PH assumption. Log cumulative hazard plots were reported in 44 appraisals (88%), while the Grambsch Therneau test was reported by just 17 (34%) appraisals. The results of the PH tests were redacted in 19 (38%) of the studies that reported them. 86% presented PH test results in the original company submissions, with 14% providing PH testing at the clarification question stage. PH testing methods were questioned by the EAG/ERG in only 4 appraisals.

CONCLUSIONS: Out of all studies that perform PH testing, the prevalence of each test varies. The most common test presented is the log cumulative hazard plot. Despite the variation observed in NICE TAs, PH testing was questioned by the EAG/ERG in only 4 appraisals. Thus, NICE TA submissions should adhere more firmly to PH testing guidelines outlined in NICE documentation, and this should be enforced by the EAG/ERG.