## Managing Structural Uncertainty in Health Economic Decision Models: A Discrepancy Approach > [!Abstract]- > Healthcare resource allocation decisions are commonly informed by computer model predictions of population mean costs and health effects. It is common to quantify the uncertainty in the prediction due to uncertain model inputs, but methods for quantifying uncertainty due to inadequacies in model structure are less well developed. We introduce an example of a model that aims to predict the costs and health effects of a physical activity promoting intervention. Our goal is to develop a framework in which we can manage our uncertainty about the costs and health effects due to deficiencies in the model structure. We describe the concept of ‘model discrepancy’: the difference between the model evaluated at its true inputs, and the true costs and health effects. We then propose a method for quantifying discrepancy based on decomposing the cost-effectiveness model into a series of subfunctions, and considering potential error at each subfunction. We use a variance-based sensitivity analysis to locate important sources of discrepancy within the model to guide model refinement. The resulting improved model is judged to contain less structural error, and the distribution on the model output better reflects our true uncertainty about the costs and effects of the intervention. > [!Cite]- > Strong, Mark, Jeremy E. Oakley, and Jim Chilcott. “Managing Structural Uncertainty in Health Economic Decision Models: A Discrepancy Approach.” _Journal of the Royal Statistical Society Series C: Applied Statistics_ 61, no. 1 (January 1, 2012): 25–45. [https://doi.org/10.1111/j.1467-9876.2011.01014.x](https://doi.org/10.1111/j.1467-9876.2011.01014.x). ## Notes %% begin notes %% From the Centre for Bayesian Statistics in Health Economics (CHEBS) at the University of Sheffield (now defunct): Within a health economic decision model there are multiple sources of uncertainty. These uncertainties must be properly accounted for if a decision maker is to have confidence in the model output. Structural uncertainty is present when we are uncertain about the model output because we are uncertain about the functional form of the model. We are not certain that our model adequately reflects reality, and so we are not certain that our result would be correct, even if the true values of all input parameters were known. Standard probabilistic sensitivity analysis will not account for structural uncertainty, which may result in spuriously precise estimate of the model outputs. Moreover, structural uncertainty may well have a greater impact on the model outputs than parameter and methodological uncertainty, yet methods for dealing with structural uncertainty are relatively underdeveloped. To resolve the problem of structural uncertainty two broad approaches have been proposed; model averaging and discrepancy modelling. In the model averaging approach, we calculate the sum of the outputs of a set of plausible models, weighted by some measure of their adequacy. If we believe our weighted average result then we are implicitly assuming that our set of models contains the "true" model, although we do not know which the true model is. In the discrepancy modelling approach, we do not have to assume that any of our plausible models are necessarily "true", instead we try to make judgements about the discrepancy between the model output and the "true" target value. In the context of health economic decision modelling there are difficulties with both these approaches (Strong et al, 2009). %% end notes %% %% begin annotations %% %% end annotations %% %% Import Date: 2024-08-15T07:42:54.759-06:00 %%