Abstract
BACKGROUND AND OBJECTIVE: The Centers for Medicare and Medicaid Services increasingly rely on real-world evidence to inform drug price negotiations under the Inflation Reduction Act. This study aims to evaluate methodological decisions that impact real-world comparative effectiveness outcomes using a case example of first-line pembrolizumab versus therapeutic alternatives in advanced non-small cell lung cancer among the Medicare population. METHODS: This study used a deidentified, electronic health record-derived, advanced non-small cell lung cancer dataset (2011-23) to analyze Medicare-eligible stage IV patients in three indications: (1) non-squamous, epidermal growth factor receptor, and anaplastic lymphoma kinase negative; (2) squamous; and (3) epidermal growth factor receptor and anaplastic lymphoma kinase negative with programmed death ligand-1 expression ≥1%. Indications (1)-(2) involved pembrolizumab combinations, while (3) referred to pembrolizumab monotherapy. Comparators included common non-platinum-based chemotherapy regimens. Propensity score-based inverse probability weighting was applied. The primary outcomes were real-world progression-free survival and overall survival. Scenario analyses examined the influence of time period selection, programmed death ligand-1 inclusion, therapeutic alternatives, and treatment switching on comparative effectiveness estimates. RESULTS: In the non-squamous cohort (1), overall survival benefits of pembrolizumab therapies compared to alternatives varied from a non-significant difference to an improvement of 2.7 months (95% confidence interval 1.2, 4.8), depending on analytical choices. In the squamous cohort (2), pembrolizumab combinations consistently demonstrated overall survival benefits, which ranged from 1.4 months (95% confidence interval 0.1, 3.0) to up to 3.6 months (95% confidence interval 0.1, 5.9). However, for pembrolizumab monotherapy (3), overall survival differences were statistically non-significant. Scenario analyses indicated substantial variability in outcomes based on methodological choices. CONCLUSIONS: This study underscores the importance of transparent reporting and scenario analyses in real-world evidence to support Centers for Medicare & Medicaid Services decision making during drug price negotiations. Findings highlight the need for rigorous methodological standards to ensure the external validity of real-world evidence and its alignment with clinical practice.