Accuracy of Overall Survival (OS) Extrapolations in French Cost-Effectiveness Reports Relative to Updated Data from Randomized Controlled Trials of Immune Checkpoint Inhibitors (ICIS)
Author(s)
Gueye A1, Juban A2, Branchoux S3, Gaudin AF4, Colrat F4
1Université de Paris-Dauphine, Paris, France, 2Université de Paris-Dauphine, Paris, 75, France, 3Bristol-Myers Squibb SARL, RUEIL MALMAISON, France, 4Bristol-Myers Squibb SARL, Rueil Malmaison, 92, France
Presentation Documents
OBJECTIVES: OS extrapolations of ICIs are required to inform cost-effectiveness analyses. This is often challenging given the short follow-up of randomized controlled trials (RCTs) available for model fitting. The objective of the study was to assess the accuracy of estimates of long-term OS predicted in French cost-effectiveness assessment reports (CEARs) using longer-term RCTs data.
METHODS: A systematic review of French ICIs CEARs until May 31st, 2022 was performed. A targeted literature review was conducted to identify RCTs with extended follow-up data after these CEARs had been published. Relative differences between projected and observed OS was calculated and analyzed based on length of additional follow-up, type of model and data maturity (Pearson’s correlation).
RESULTS: Among the 174 CEARs identified, 17 met the inclusion criteria. Mean additional follow-up was 32 months. Indications were non-small-cell lung cancer (47%; n=8), renal cell carcinoma (29%; n=5), melanoma (18%; n=3) and urothelial carcinoma (6%; n=1). Extrapolations were based on piecewise (41%; n=7), parametric (35%; n=6), multidimensional treatment effect (18%; n=3) and spline (6%; n=1) models. OS was underestimated in 59% of the CEARs (n=10) and 41% (n=7) did not fall within the confidence interval. Mean relative difference at landmark RCT was -21%. When an additional follow-up of at least 2 years was available (n=10), OS was underestimated in 80% of the CEARs with a relative difference of -33% at landmark RCT. Among piecewise models, OS was underestimated in 71% of CEARs (n=5). No correlation was found between data maturity and relative difference (Pearson r: -0.32).
CONCLUSIONS: OS extrapolations in French CEARs underestimated observed OS and did not fully capture long-term benefit of ICIs. This underestimation was even greater when an additional follow-up of more than 2 years was available.
Conference/Value in Health Info
Value in Health, Volume 25, Issue 12S (December 2022)
Acceptance Code
P21
Topic
Health Technology Assessment, Methodological & Statistical Research
Topic Subcategory
Decision & Deliberative Processes
Disease
sta-drugs