Economic Evaluation seminar - Back to the Future: Revisiting Conventional Methods to Accommodate Complex Hazard Functions in Survival Modelling for Cost-Effectiveness Analysis
Event details
Abstract: Survival analysis is a key method in cost-effectiveness evaluation as it enables the assessment of health economic outcomes over extended periods, often far exceeding the observation duration of trials or other data collection periods. Traditionally, a standard set of parametric survival models that ignore patient-specific factors are evaluated for goodness of fit and clinical plausibility before selecting one as the base case for extrapolation. However, there is increasing evidence that these models, with their relatively simple hazard functions, are inadequate for capturing more complex hazard dynamics over time. Recent developments include decision algorithms that quickly favour complex methods such as cubic splines. While these methods handle intricate hazards in the short term, they introduce a new set of challenges and issues. This paper argues that a simpler approach incorporating observable heterogeneity, recognised by statisticians in the 1980s and 1990s and capable of addressing a primary driver of complexity, has been overlooked. We expand on the simple underlying theory, draw a distinction between individual and population-level survival and hazard functions, and demonstrate a known and straightforward technique for constructing population-level hazards for use in cohort models. We illustrate the benefits with a flexible user-definable simulation model and real-world data. We conclude by making recommendations to the current NICE/DSU guidance on survival modelling, which in many cases, may negate the need for adopting considerably more complex methods.
If you are not a member of University of York staff and are interested in attending a seminar, please contact alfredo.palacios
Venue details
Wheelchair accessible