Using outcomes data to identify best medical practice: The role of policy models

Increasingly, physicians are attempting to incorporate best evidence into t heir clinical decision making, However, best evidence takes a variety of fo rms, including clinical trials, cohort studies, administrative data, and pa tient preference data. Incorporating multiple data sources in a way that in forms complex clinical decisions is a substantial analytical challenge. One approach to this challenge is to develop a simulation/decision model that explicitly represents the natural history of disease and the impact of trea tments on that natural history. The model should be requisite-that is, suff icient in form to address the decision problem-but not overly complex. Such a model can be of value because it (1) allows a variety of viewpoints to b e considered, (2) incorporates the best scientific evidence, and (3) permit s sensitivity analyses to evaluate the impact of alternative clinical scena rios and uncertainty in model inputs. The Stroke Prevention Policy Model (S PPM) illustrates this approach. The SPPM is a simulation model designed to predict the best among various treatment alternatives for preventing stroke s. Similar models can be applied to treatment outcomes for liver disease.