[Knowledge must be combined with values for decision making in uncertainty]

On the basis of a clinical presentation of a woman with atrial fibrillation and diabetes mellitus, this article examines how medical evidence can be combined with patient preferences in the analysis of clinical choice. Such choices tend to be analysed and discussed without due consideration for patient or public values. It is argued that such an analysis is incomplete until the patient's preferences are incorporated into the model. This is also a way of implementing patient-centred medicine. We discuss how decision theory can be used to clarify clinical choices, sometimes with surprising results for the physician. How to handle situations when the patient's preferences are in conflict with the physician's judgement, is also discussed. It is concluded that good clinical practice must be based on evidence, but it should also take into account the patient's values in choices with uncertain outcomes.