Communicating risk to patients.

Abstract

Introduction Tests that screen for disease must maximize sensitivity (true positives) and specificity (true negatives), so as not to misdiagnose, potentially miss life-threatening disorders or increase patient distress, as well as health care costs. Sensitivity and specificity alone may, however, be misunderstood by patients, and indeed clinicians themselves, especially when disease prevalence is low in the population. Patients testing positive are usually more interested in the probability that they actually have the disease1; known as the positive predictive value. This probability is obtainable using a ‘simple’ Bayesian probability formula, yet it is rarely presented in brochures and other material provided to and/or readily accessible to patients. Professor Gerd Gigerenzer1 and others have conducted many studies, including randomized trials, on how best to present screening information to facilitate understanding and more informed choices. Aims To present simple and understandable methods of illustrating screening test probabilities, including positive predictive values. Methodology Tables and probability trees, that include frequencies, appear to be very informative. These and similar methods could usefully and readily be employed in patient brochures1 and consultations. Such techniques could also be implemented in relevant new and existing patient databases and registries, to provide local, individual and updated probabilities3. Results Illustrative published medical data and custom graphs will be presented.