Potential impact on cost-effectiveness estimates of using immature survival data: a case study based on transcatheter edge-to-edge repair (TEER) used for patients with severe mitral regurgitation at high surgical risk

ObjectiveTo review the survival modelling used in cost-effectiveness studies evaluating an interventional procedure and to discuss implications for decision-makers.DesignA case study of three economic evaluations that each used immature data from the EVEREST II High Surgical Risk (HSR) Study of transcatheter edge-to-edge repair (TEER) for patients with severe mitral regurgitation (MR) who were at high risk of surgery.SettingEstimation of patient survival in cost-effectiveness studies.ParticipantsThe EVEREST II HSR Study included 78 patients who had TEER of the mitral valve using the MitraClip device and a retrospectively identified control group of 36 patients who received medical management and were followed up for 12 months. Observed survival (TEER arm only) was updated at 5 years.ResultsTwo studies used 12-month observed mortality from EVEREST II HSR to model survival over lifetime horizons. Observed and modelled survival were associated with considerable uncertainty due to short follow-up and small numbers of participants. Modelling control patients' survival required an approximate 10-fold extrapolation based on 12-month observation of only 38 patients. Observed 5-year survival in the TEER group differed from that less mature follow-up suggesting that survival modelling based on shorter follow-up was unsatisfactory. No public domain data for the control group are available beyond 12-month follow-up so meaningful estimates using mature data for both arms are currently not possible. A third study developed survival models using incompletely reported transitions between MR grades in EVEREST II HSR and mortality rates observed for different MR grades derived from a study in an unrelated population.ConclusionsModelling survival in such small samples followed up for only 12 months is associated with great uncertainty, and cost-effectiveness results based on these analyses should be viewed as premature and used cautiously in reimbursement decisions.