Authors:
Swaminathan, A, Xu, C, Zhang, S, Du, K, Siu, E, Kalesinskas, L, Lite, S, Song, Y, Snider, JW, Schwemm, A, Bargo, D, Adamson, BJS
OBJECTIVES
Real-world evidence (RWE) generated from electronic health records (EHR) has been shown to be more relevant, timely, and representative for health technology assessment decision-making compared to evidence from clinical trials. We assessed how using EHR-derived RWE instead of published clinical trial data can impact point estimates and uncertainty ranges in cost-effectiveness estimates for advanced non-small cell lung cancer (NSCLC) therapies.
METHODS
We replicated a cost-effectiveness analysis of NSCLC therapies developed by the Institute for Clinical and Economic Review in 2016 (“traditional”), replacing meta-analysis-derived hazard ratios from clinical trials with RWE-derived hazard ratios for progression-free and overall survival (“RWE-enhanced”). Hazard ratios were calculated using Cox proportional hazards models adjusted for age, sex, race, practice type, performance status, and smoking history. The study used the Flatiron Health database, a nationwide (US-based) longitudinal, de-identified EHR-derived database. We compared the cost-effectiveness of immunotherapy (atezolizumab, pembrolizumab, nivolumab) vs. chemotherapy (single-agent docetaxel) in the second line of therapy. The analytic cohort included 3,492 adults with EGFR- NSCLC that progressed after first-line treatment with a platinum-based chemotherapy doublet. We report traditional and RWE-enhanced incremental cost-effectiveness ratios (ICERs) and differences in uncertainty as percent change in the size of 95% credible intervals (CIs).
RESULTS
The traditional vs RWE-enhanced ICERs were as follows: atezolizumab — $84,000/quality-adjusted life year (QALY) vs. $138,000/QALY; nivolumab — $136,000/QALY vs. $123,000/QALY; pembrolizumab — $181,000/QALY vs $111,000/QALY. Compared to uncertainty in traditional ICERs, 95% CIs for RWE-enhanced ICERs were reduced by 37%, 69%, and 83% for atezolizumab, nivolumab, and pembrolizumab respectively.
CONCLUSIONS
This proof-of-concept demonstrated how clinical depth, longer follow-up time, and larger sample sizes in EHR-derived data may reduce uncertainty in cost-effectiveness analysis. The approach has potential to inform dynamic value-based pricing and highlights the importance of reassessments once RWE is available.
Sources:
ISPOR Annual Meeting