Presentation (CTI)

OBJECTIVES: Partitioned survival analysis (PartSA) is a common model structure for evaluating the cost-effectiveness of cancer treatments. Unlike Markov models, PartSA uses progression-free survival (PFS) and overall survival (OS) outcomes to estimate health state membership without explicitly modeling individual transitions, making it challenging to accurately calculate incident post-progression events informing subsequent treatment costs. This study aimed to examine the impact of different incident progression calculation methods on incremental cost-effectiveness ratios (ICERs) in terms of cost per QALY gained.
METHODS: A hypothetical PartSA model was developed to evaluate the impact of alternative incident progression estimation methods, simulating an “aggressive” cancer archetype, with current SoC leading to a median PFS of 2.7 months and median OS of 4.5 months. A constant PFS HR of 0.7 and OS HR of 0.8 versus SoC was assumed for “novel” treatment, while PFS and OS HRs of 0.5 and 0.9, respectively, were also explored to evaluate an alternative survival benefit profile. Methods considered included equal mortality rates for PFS and PD states (approach 1), Euler’s method (approach 2), using general population mortality for the PFS state (approach 3), and assuming all patients progress before dying (approach 4).
RESULTS: ICERs with approaches 3 and 4 were 23.3% and 23.7% lower, respectively, while approach 2 was 8.7% higher, compared to approach 1. ICER variations using the alternative survival benefit profile were directionally consistent with the main analysis, albeit with different magnitudes (14.1% and 14.2% lower for approaches 3 and 4, respectively, 11.1% higher for approach 2). Scenario analyses for other cancer archetypes are ongoing.
CONCLUSIONS: Estimating incident progression in Part SA models requires clinical assumptions, with choice of incident progression estimation method potentially significantly impacting ICER results. Carefully alignment with the specific cancer type under investigation and transparent reporting of assumptions made is crucial for interpretation of cost-effectiveness results for cancer treatments.