To explore the almost ubiquitous problem of performing survival extrapolation with heavily censored data from clinical trials, discuss the implications of current practice and introduce possible novel methods to alleviate the problem.

DESCRIPTION

Economic evaluations as part of health technology assessments (HTA) typically require estimates of lifetime survival benefit for new oncologic therapies. Interim analyses of trials with limited follow up are increasingly used to inform FDA and EMA regulatory approval, but the high degrees of administrative censoring in these trials create significant challenges when it comes time to extrapolate survival outcomes over a lifetime time horizon.

Current approaches of extrapolation often assume that the treatment effect observed in the trial can continue indefinitely, while in reality it is likely to vary over time, particularly in the long term. In this workshop, we will firstly present a brief review of current methods to inform long term survival benefit in the presence of heavily censored data and their main limitations and implications for the wider economic analysis. Secondly, we will present a brief introduction to the advantages of Bayesian modelling, specifically in survival analysis in HTA. Finally, we will present an innovative methodology based on “blending” survival curves as a possible solution. The basic idea is to mix a flexible model (e.g. Cox semi-parametric) to fit as well as possible the observed data and a parametric model encoding assumptions on the expected behaviour of underlying long-term survival. The two are “blended” into a single survival curve that is identical with the flexible model over the range of observed times and increasingly similar to the parametric model over the extrapolation period. This approach allows a wide range of plausible scenarios to be considered as well as the inclusion of genuine information, based e.g. on hard data or expert opinion.