OBJECTIVES: A framework will be presented and discussed providing an overview of the ways to include all available information in the estimation process of extrapolating survival while ensuring clinical plausibility.

METHODS: A SLR identifying 180 NICE HTA’s was conducted plus a review on the NICE DSU TSDs. Based on this SLR the DIVE framework was developed introducing four categories on how to use information to ensure clinical plausible survival extrapolations; 1) Direct use of clinical trial information, 2) Indirect use of external information, 3) use of external information to Validate assumptions and outputs 4) use of External information to improve the estimation process.

RESULTS: For category Direct, the following methods will be considered: standard survival distributions, mixture cure models (MCM), mixture models, fractional polynomials, splines and landmark models. For category Indirect, network meta analyses (NMAs) on hazard ratios (HRs), parametric NMAs, fractional polynomial NMAs, matching adjusted indirect treatment comparisons (MAIC) and simulated trial comparisons (STC) will be discussed. For category Validation, ways to use external information like expert opinion or historical data for validation purposes will be provided. For category External, ways to improve the estimation process by expressing information from external data in a priori distributions in Bayesian extrapolation methods will be described, together with ways to use external evidence around subsequent treatment prescription in clinical practice for treatment switching adjustments.

CONCLUSIONS: The DIVE framework provides a structured way to evaluate whether all or most relevant information is used to inform extrapolation of survival outcomes.