Clinical Chemistry’s NIH Book Club Corner, NIH Biomedical Computing Interest Group Book Club Review of How Doctors Think by Jerome Groopman. Jim DeLeo, Alan T. Remaley, G. William Moore, Gerald L. McLaughlin, Ellen J. Bicknell, and Melanie Swan.

Abstract

This book explores how physicians make difficult diagnoses. Cognitive processes, not merely learning correlations between symptoms and diseases, must be an integral part of clinical medicine. We are introduced to a woman with abdominal complaints, undiagnosed for more than 15 years. Finally, a physician carefully listens to her entire story, and makes the correct diagnosis. A patient with sudden shortness of breath is rescued by a visiting cardiologist, who quickly makes the diagnosis using pattern recognition and heuristics. Most medical errors are mistakes in thinking. Patients may be misdiagnosed due to the physician’s positive or negative social stereotypes. Errors may result from availability errors (recent experience) and confirmation bias (disregarding data inconsistent with one’s first guess), according to economics Nobel laureates Tversky and Kahneman. Patients can assume active roles by asking questions and helping to shape physician thinking. The time that medical gatekeepers (general internists, general pediatricians, and family practitioners) spend talking with patients is undervalued. Cognitive bias (matching the patient to a familiar prototype), zebra retreat (avoiding a rare diagnosis), and diagnosis momentum (passing along a first-impression diagnosis to peers) point physicians to a misdiagnosis. Physicians should accept uncertainty as an intrinsic feature of medical practice. The author-as-patient consulted 6 surgeons for a long-standing hand ailment, received 4 different diagnoses, and chose the surgeon with the best cognitive assessment. Current medical knowledge cannot entirely rely on evidence-based medicine and clinical trials for individual diagnoses and treatment. The author discusses errors that radiologists make in their practice, although similar errors are also made in other specialties. One problem may be an increase in the complexity and number of images that are …