A RTICLE The Importance of Cognitive Errors in Diagnosis and Strategies to Minimize Them Pat Croskerry, MD, PhD ABSTRACT In the area of patient safety, recent attention has focused a number of strategies for reducing them (‘‘cognitive on diagnostic error. The reduction of diagnostic error is debiasing’’). Principle among them is metacognition, an important goal because of its associated morbidity and a reflective approach to problem solving that involves potential preventability. A critical subset of diagnostic stepping back from the immediate problem to examine errors arises through cognitive errors, especially those and reflect on the thinking process. Further research associated with failures in perception, failed heuristics, effort should be directed at a full and complete and biases; collectively, these have been referred to as description and analysis of CDRs in the context of cognitive dispositions to respond (CDRs). Historically, medicine and the development of techniques for models of decision-making have given insufficient avoiding their associated adverse outcomes. Consider- attention to the contribution of such biases, and there able potential exists for reducing cognitive diagnostic has been a prevailing pessimism against improving errors with this approach. The author provides an cognitive performance through debiasing techniques. extensive list of CDRs and a list of strategies to reduce Recent work has catalogued the major cognitive biases diagnostic errors. in medicine; the author lists these and describes Acad. Med. 2003;78:775–780. he recent article by Graber et al.1 provides The no-fault and system-related categories of diagnostic a comprehensive overview of diagnostic errors in errors described1 certainly have the potential for reduction. medicine. There is, indeed, a long overdue and In fact, very simple changes to the system could result in Tpressing need to focus on this area. They raise a significant reduction in these errors. However, the many important points, several of which deserve extra greatest challenge, as they note, is the minimization of emphasis in the light of recent developments. They also cognitive errors, and specifically the biases and failed provide an important conceptual framework within which heuristics that underlie them. Historically, there has pre- strategies may be developed to minimize errors in this vailed an unduly negative mood toward tackling cognitive critical aspect of patient safety. Diagnostic errors are bias and finding ways to minimize or eliminate it. associated with a proportionately higher morbidity than is The cognitive revolution in psychology that took place the case with other types of medical errors.2–4 over the last 30 years gave rise to an extensive, empirical literature on cognitive bias in decision-making, but this Dr. Croskerry is associate professor, Departments of Emergency Medicine advance has been ponderously slow to enter medicine. and Medical Education, Dalhousie University Faculty of Medicine, Halifax, Decision-making theorists in medicine have clung to nor- Nova Scotia, Canada. He is also a member of the Center for Safety in Emergency Care, a research consortium of the University of Florida College mative, often robotic, models of clinical decision making of Medicine, Dalhousie University Faculty of Medicine, Northwestern that have little practical application in the real world of University The Feinberg School of Medicine, and Brown Medical School. decision making. What is needed, instead, is a systematic Correspondence and requests for reprints should be sent to Dr. Croskerry, analysis of what Reason5 has called ‘‘flesh and blood’’ Emergency Department, Dartmouth General Hospital Site, Capital District, decision-making. This is the real decision making that 325 Pleasant Street, Dartmouth, Nova Scotia, Canada B2Y 4G8; telephone: (902) 465-8491; fax: (902) 460-4148; e-mail: hxkerry@ occurs at the front line, when resources are in short supply, accesscable.neti. when time constraints apply, and when shortcuts are being Two responses to this article are printed after it. sought. When we look more closely at exactly what A CADEMIC M EDICINE,VOL. 78, NO .8/AUGUST 2003 775 C OGNITIVE E RRORS IN D IAGNOSIS, CONTINUED cognitive activity is occurring when these clinical decisions that can lead to diagnostic errors, but the most important are being made, we may be struck by how far it is removed strategy may well lie in familiarizing clinicians with the from what normative theory describes. Although it seems various types of CDRs that are out there, and how they certain we would be less likely to fail patients diagnostically might be avoided. I made a recent extensive trawl of medical when we follow rational, normative models of decision and psychological literature, which revealed at least 30 making, and although such models are deserving of ‘‘a CDRs,9 and there are probably more (List 1). This catalogue prominent place in Plato’s heaven of ideas,’’6 they are provides some idea of the extent of cognitive bias on impractical at the sharp end of patient care. Cognitive decision-making and gives us a working language to describe diagnostic failure is inevitable when exigencies of the clinical it. The failures to show improvement in decision support for workplace do not allow such Olympian cerebral approaches. clinical diagnosis that are noted by Graber et al.1 should Medical decision makers and educators have to do three come as no surprise. They are likely due to insufficient things: (1) appreciate the full impact of diagnostic errors in awareness of the influence of these CDRs, which is often medicine and the contribution of cognitive errors in subtle and covert.10 There appears to have been an historic particular; (2) refute the inevitability of cognitive diagnostic failure to fully appreciate, and therefore capture, where the errors; and (3) dismiss the pessimism that surrounds ap- most significant diagnostic failures are coming from. proaches for lessening cognitive bias. Not surprisingly, all CDRs are evident in emergency For the first, the specialties in which diagnostic un- medicine, a discipline that has been described as a ‘‘natural certainty is most evident and in which delayed or missed laboratory of error.’’11 In this milieu, decision-making is diagnoses are most likely are internal, family, and emer- often naked and raw, with its flaws highly visible. Nowhere gency medicine; this is borne out in findings from the in medicine is rationality more bounded by relatively poor benchmark studies of medical error.2–4 However, all spe- access to information and with limited time to process it, all cialties are vulnerable to this particular adverse event. The within a milieu renowned for its error-producing con- often impalpable nature of diagnostic error perhaps reflects ditions.12 It is where heuristics dominate, and without them why it does not appear in lists of serious reportable events.7 emergency departments would inexorably grind to a halt.13 For the second, there needs to be greater understanding of Best of all, for those who would like to study real decision the origins of the widespread inertia that prevails against making, it is where heuristics can be seen to catastrophically reducing or eliminating cognitive errors. This inertia may fail. Approximately half of all litigation brought against exist because such errors appear to be so predictable, so emergency physicians arises from delayed or missed diag- widespread among all walks of life, so firmly entrenched, noses.14 and, therefore, probably hardwired. Although the evolu- If we accept the pervasiveness and predictability of the tionary imperatives that spawned them may have served us CDRs that underlie diagnostic cognitive error, then we are well in earlier times, it now seems we are left with obliged to search for effective debiasing techniques. Despite cognitively vestigial approaches to the complex decision the prevailing pessimism, it has been demonstrated that, making required of us in the modern world. Although using a variety of strategies15,16 (Table 1), CDRs can be ‘‘cognitive firewalls’’ may have evolved to quarantine or overcome for a number of specific biases.16–23 It appears avoid cognitive errors, they are clearly imperfect8 and will that there are, indeed, cognitive pills for cognitive ills,22 require ontogenetic assistance (i.e., cognitive debiasing) to which makes intuitive sense. This is fortunate, for other- avoid their consequences. Accepting this, we should say less wise, how would we learn to avoid pitfalls, develop about biases and failed heuristics and more about cognitive expertise, and acquire clinical acumen, particularly if the dispositions to respond (CDRs) to particular situations in predisposition for certain cognitive errors is hardwired? various predictable ways. Removing the stigma of bias clears However, medical educators should be aware that if the the way toward accepting the capricious nature of decision- pills are not sufficiently sugared, they may not be swallowed. making, and perhaps goes some way toward exculpating Yates et al.24 have summarized some of the major clinicians when their diagnoses fail. impediments that have stood in the way of developing An understanding of why clinicians have particular effective cognitive debiasing strategies, and they are not CDRs in particular clinical situations will throw consider- insurmountable. The first step is to overcome the bias able light on cognitive diagnostic errors. The unmasking of against overcoming bias.