PURPOSE:

Network meta-analysis (NMA) of time-to-event outcomes based on constant hazard ratios can result in biased findings when the proportional hazards (PH) assumption does not hold in a subset of the trials. Several NMA methods that do not rely on the PH assumption have been proposed. However, their application has been limited, possibly due to lack of familiarity among researchers performing cost-effectiveness analysis of oncology drugs. We aim to summarize published non-PH NMA methods for time-to-event outcomes, demonstrate their application, and compare results, as well as outline recommendations regarding basic model selection process.

DESCRIPTION:

Workshop attendees will develop an understanding regarding the alternative non-PH NMA methods and their implementation, including considerations for model selection. Initially, we will review the PH assumption, discuss violations, and consider implications from a cost-effectiveness analysis perspective. After highlighting the need for a more flexible approach, we will summarize each of the published non-PH NMA methods, specifying each model and stepping through the corresponding JAGS code. We will then compare across the methods and models in terms of their key features. We will illustrate their application, describing model selection, outputs, and interpretation. Finally, we will consider the strengths and limitations of the alternative methods and will propose a model selection process to help guide decision-making. The audience will be asked to consider when non-PH NMA methods should be considered, which models should be preferred, and which criteria should factor into decision-making.

Dr. Kurt will chair the session and introduce the importance of considering non-PH NMA methods (15 min). Dr. Campbell will summarize the alternative methods and models (15 min) and Dr. Jansen will illustrate their application to a case study in renal cell carcinoma (15 min). Ms. Cope will propose a transparent and explicit stepwise model selection process considering model fit, external constraints, and clinical validity (15 min).