To the Editor:
In their study of diagnostic reasoning in medical graduates, Sherbino et al1 have shown that diagnostic accuracy is inversely related to the speed of determining the diagnosis. They conclude that this challenges the dominant view in cognitive psychology that rapid intuitive reasoning is prone to error. However, we believe that these results are consistent with current models of clinical reasoning, in which decision making involves a combination of System 1 (intuitive) and System 2 (analytic) thinking and in which intuitive decisions are more vulnerable to error than those made in the analytic system.
An alternative interpretation of the study is that the time needed to solve the diagnostic problem—under the experimental conditions used by Sherbino et al—is simply a measure of problem difficulty, which is a reflection of the study participant’s knowledge base. For any problem that involves System 2 decision making, the finding that error is associated with a longer response time is not surprising. The exercises in the authors’ study are similar to those provided in exams, and it is likely that the participants primarily used System 2 in arriving at their diagnoses, as they would in exams. The relationship between the participants’ accuracy and their exam scores confirms this, and the authors state that “the longest RT [response time] was associated with the most difficult case (as judged by overall cohort accuracy).” Requiring a response within one minute does not necessarily cause participants to use System 1 in solving the cases, and faster responses are likely to occur when the participants have the knowledge base to quickly arrive at the correct conclusion using System 2. We know that System 2 reasoning is not error-free, especially with residents in training, and a System 2 failure is likely to be due to a deficiency in knowledge.
Because all pertinent findings were provided to the participants, the exercises in the authors’ experiment do not reflect the actual complex processes involved in patient assessment, where clinical decisions are made regarding which findings are pertinent for ruling diagnoses in or out. This type of study highlights the difficulty of conducting research that yields findings that can be meaningfully applied to the real world of clinical medicine, where context, patient factors, ambient conditions, human factors, team dynamics, and a variety of other influences prevail. It is likely through those complex processes that the majority of System 1 errors occur, and it would be dangerous to promote the notion to either trainees or practicing clinicians that speed increases accuracy.
The bottom line remains that in clinical practice, more errors are associated with System 1 than with System 2 thinking. Further, System 1 does not typically result in error; many intuitive decisions are correct, especially those made by experienced clinicians—System 1 thinking works most of the time. Of course, being right most of the time, no matter which system of thinking is used, is not acceptable in clinical medicine. Research to gain a deeper understanding of diagnostic processes in actual clinical situations may help us find ways to increase the rate of accurate diagnoses.
Pat Croskerry, MD, PhD
Professor, Department of Emergency Medicine, and director, Critical Thinking Program, Division of Medical Education, Faculty of Medicine, Dalhousie University, Halifax, Nova Scotia, Canada; firstname.lastname@example.org.
Gordon Tait, PhD
Assistant professor, Departments of Surgery and Anesthesia, University of Toronto Faculty of Medicine, and staff scientist, Department of Anesthesia and Pain Management, Toronto General Hospital, Toronto, Ontario, Canada.
1. Sherbino J, Dore KL, Wood TJ, et al. The relationship between response time and diagnostic accuracy. Acad Med.. 2012;87:785–791