An exploration of methods to incorporate external evidence into survival modelling in oncology

This thesis explores the role of external evidence in survival analysis for health technology assessment (HTA), specifically when estimating overall survival (OS) over a lifetime horizon, for populations with cancer.

External evidence encompasses data and/or information from sources such as disease registries, observational studies, expert opinion, and general population statistics; which has the potential to enhance decision-making critical survival extrapolations. However, external evidence introduces challenges related to evidence compatibility, methodological rigour, and transparency.

A systematic review was undertaken to identify and categorise methods used in the literature to leverage external evidence. This review was complemented by an analysis of methods used in contemporary HTAs of cancer treatments. Together, these reviews provided a foundation for evaluating the feasibility and relevance of different approaches in real-world contexts.

To test the methods three case studies (squamous cell carcinoma of the head and neck, melanoma, and non-small-cell lung cancer) were used to test selected methods, which illustrated practical considerations and issues with implementation. The method testing resulted in a set of seven comparable methods which could then be explored further.

A neutral comparison study was conducted, evaluating the performance of the seven methods under scenarios with varying data-generating mechanisms. Characteristics that varied across scenarios included: sample size, study follow-up, hazard profiles, survival magnitude, and external evidence similarity (versus the pivotal trial). The findings revealed trade-offs across methods under different scenarios, demonstrating that no single approach was dominant. Instead, the choice of method must account for the decision context including the nature of the external evidence and the characteristics of the target population.

This thesis concludes with practical recommendations for selecting and applying methods to incorporate external evidence into survival analysis for HTA, as well as recommendations for future research.