| Summary: | “Evidence Based Medicine” (EBM) urges that medical recommendations be based on the best research evidence, rather than on clinical judgement. While I strongly endorse attention to relevant research evidence, I argue that the related downplaying of clinical judgement is a step backwards. This is because actual models of EBM encourage physicians to focus exclusively on research probabilities and so to neglect relevant information about patients. I call this feature of EBM the “Problem of Extra Information” (PEI), and contend that it leads to predictions and prescriptions based on the wrong probabilities. The PEI has been largely neglected by EBM, which has construed the challenge of clinical care as a matter of developing better research evidence, and of reminding physicians to attend to patients’ preferences and values. And although meritorious attempts have been made to connect research with individuals through sophisticated methodological improvements, these only address the PEI partially, and do not eliminate the need for clinical discretion. In this dissertation I contend that, in response to the PEI, clinical medicine requires a more Discretionary Approach (DA). This approach recognizes that the objective probabilities that matter for clinical recommendations are those in the reference class defined by everything the physician knows about the patient, and argues that the central role for judgment in clinical practice is to estimate these probabilities. So understood, the DA has two main advantages over the EBM approach: prudential adequacy and evidential flexibility. My defence of the DA consists of addressing criticisms of the role ascribed to judgment and clinical experience within this approach. The final two chapters of this doctoral dissertation complement my arguments with two meta-analytical empirical studies: one which compares “therapeutic guidelines based on evidence” with “usual care” with respect to patients’ outcomes, and another which examines the relative predictive performance of statistical models and physicians’ judgment in the context of diagnosis and prognosis. These studies refute previous evidence cited against judgment and vindicate the plausibility of the Discretionary Approach to clinical care.
|
|---|
