Methods to estimate the survival benefit from individual patient data meta-analysis. Impact on the health economic outcome

B Lueza, Mauguen, Pignon Jp, Le Péchoux
In economic evaluation, different parametric or non-parametric methods can be used to assess the survival benefit from individual patient data. A challenge, when using individual patient data from a meta-analysis to estimate the survival benefit, is to take into account stratification on clinical trials. Our objective was to compare five different methods to estimate survival benefit and to investigate the impact on the health economic outcome. We chose to compare four non parametric methods
more » ... ssically used in meta-analyses for survival endpoints (the Kaplan-Meier, the Stewart and the Peto methods) and a parametric method (Weibull) to estimate survival functions. We used individual patient data from the Meta-Analysis of Radiotherapy in Lung Cancer (MAR-LC) in locally advanced non-small cell lung cancer patients. Our data set consisted of 10 phase III trials comparing modified RT to conventional RT in 2,000 patients. Depending on the chosen method, mean ICERs ranged from 10,320€ to 34,937€ per life year gained. With a willingness to pay for one life year lower than 50,000€, the five acceptability curves led to different conclusions. The choice of a survival method to estimate health benefit strongly impacts on the health economic results.