Comparison of partitioned survival modeling with state transition modeling approaches with or without consideration of brain metastasis: a case study of Osimertinib versus pemetrexed-platinum.

Background: The partitioned survival model (PSM) and the state transition model (STM) are widely used in cost-effectiveness analyses of anticancer drugs. Using different modeling approaches with or without consideration of brain metastasis, we compared the quality-adjusted life-year (QALY) estimates of Osimertinib and pemetrexed-platinum in advanced non-small cell lung cancer with epidermal growth factor receptor mutations.
Methods: We constructed three economic models using parametric curves fitted to patient-level data from the National Health Insurance Review and Assessment claims database from 2009 to 2020. PSM and 3-health state transition model (3-STM) consist of three health states: progression-free, post-progression, and death. The 5-health state transition model (5-STM) has two additional health states (brain metastasis with continuing initial therapy, and with subsequent therapy). Time-dependent transition probabilities were calculated in the state transition models. The incremental life-year (LY) and QALY between the Osimertinib and pemetrexed-platinum cohorts for each modeling approach were estimated over seven years.
Results: The PSM and 3-STM produced similar incremental LY (0.889 and 0.899, respectively) and QALY (0.827 and 0.840, respectively). However, 5-STM, which considered brain metastasis as separate health states, yielded a slightly higher incremental LY (0.910) but lower incremental QALY (0.695) than PSM and 3-STM.
Conclusions: Our findings indicate that incorporating additional health states such as brain metastases into economic models can have a considerable impact on incremental QALY estimates. To ensure appropriate health technology assessment decisions, comparison and justification of different modeling approaches are recommended in the economic evaluation of anticancer drugs.
(© 2024. The Author(s).)