A Guide to Selecting Flexible Survival Models to Inform Economic Evaluations of Cancer Immunotherapies

•Standard parametric survival models can lack the flexibility to capture the shape of the underlying hazard functions of patients treated with cancer immunotherapies. This can lead to inaccurate long-term survival projections and biased cost-effectiveness estimates. More flexible extrapolation models are being adopted, but their acceptability for healthcare decision making can be an area of contention. Guidance is limited on when flexible models should be used and which, of the many available, should be considered.•An algorithm was developed to supplement existing guidance on extrapolation model selection for economic evaluation. The algorithm recommends an initial review of relevant external evidence identified in a systematic, reproducible way; highlights the need to engage with clinical experts and consider the observed hazard function and how it might change in the future; cautions against overinterpreting observed survival, particularly when data are immature; and recommends that the results of all plausible models are presented.•If followed, this algorithm will provide a systematic and evidence-based approach for flexible survival model selection. This will improve transparency and consistency, reduce the risk of inappropriate model selection, and increase confidence in the results of the cost-effectiveness analysis of cancer immunotherapies. It may also prove useful for other treatments and diseases where more flexible extrapolation models may be warranted.