METHODOLOGICAL APPROACHES FOR INCORPORATING REAL-WORLD DATA (RWD) FOR OVERALL SURVIVAL (OS) INTO LONG-TERM SURVIVAL ESTIMATES- A CASE-STUDY NICE TECHNOLOGY APPRAISAL IN EXTENSIVE-STAGE SMALL-CELL LUNG CANCER (ES-SCLC).
Connor S1, Orfanos P2, Flahavan E3, Patterson K4, Lee D5
1Roche Products Ltd, Herts, UK, 2F. Hoffmann La Roche, Basel, Switzerland, 3F. Hoffmann La Roche, Welwyn Garden City, HRT, UK, 4BresMed, Sheffield, UK, 5BresMed Health Solutions Ltd., Sheffield, DBY, UK
OBJECTIVES Increase plausibility of long-term survival estimates for ES-SCLC patients via incorporation of RWD. METHODS The IMpower133 (NCT02763579) trial follow-up was insufficient to describe the proportion of ES-SCLC patients expected to survive long-term, when treated with platinum-etoposide. To improve estimates of long-term survival for ES-SCLC patients, we identified a cohort from the US Flatiron Health electronic health record-derived database. In line with the trial, patients with 0-1 ECOG performance status, and platinum-etoposide treatment in first-line ES-SCLC were selected (N=860) and survival curves generated. Different methodologies were applied to incorporate RWD into the CEM projections. Firstly, RWD could be incorporated from different time points, e.g. at time zero or when 90% of trial observed events occurred. Secondly, either the RWD hazard of death was applied alone, or in combination with the clinical trial hazard of death, assuming a fixed shape over time. These combined into 4 different approaches for estimating long-term survival, with standard parametric extrapolations applied. The intervention arm assumed a constant effect beyond the trial, with alternative assumptions tested in scenarios. RESULTS With the log-logistic extrapolation in the base case, incorporating RWD into the survival model along with the trial data lowered the total life years gained, compared to applying trial data alone: 1.21 versus 1.00-1.11 incorporating the RWD. These lower, life years estimates with RWD were considered by clinical experts to more accurately reflect long-term survival in the control arm and gave improved cost-effectiveness estimates. CONCLUSIONS Incorporating real-world OS data can improve the clinical plausibility of long-term survival estimates, impacting the incremental difference estimated between therapies, and therefore ICER calculations. Further research is needed into the most appropriate methods to use in different circumstances. The methodology described here is one of the first approaches to formally incorporate RWD into a NICE appraisal.
Conference/Value in Health Info
2019-11, ISPOR Europe 2019, Copenhagen, Denmark
Economic Evaluation, Methodological & Statistical Research, Real World Data & Information Systems
Distributed Data & Research Networks, Health & Insurance Records Systems, Modeling and simulation