The most common cognitive biases involved in diagnostic errors in the emergency room were overconfidence, conformation, availability, and anchoring bias. Moreover, when initial and final diagnoses were examined, other diseases of the same organ or diseases of a nearby organ were often overlooked. To the best of our knowledge, this is the first study to quantify the types of cognitive bias that are likely to occur in emergency rooms in Japan.
In a study examining medical malpractice claims in Japan, diagnostic error-related claims (DERC) were significantly more common than non-diagnostic error-related claims (non-DERC) in the emergency room, and the mortality rate was significantly higher for DERC than for non-DERC [24]. Therefore, reducing diagnostic errors in emergency rooms is an important problem in Japan. The most common types of cognitive bias in the emergency room in previous studies were premature closure as well as anchoring, availability, and confirmation bias [1, 16, 25]. Moreover, overconfidence bias, which is overconfidence in one’s judgment, is a major factor that interferes with debiasing strategies and is reported to occur in about 15% of emergency room cases [9, 26]. The present study was consistent with the types of cognitive biases indicated in previous studies but with fewer premature closures. Premature closure is caused by stopping the thinking process after a diagnosis is made and failing to evaluate the hypothesis; it is the most common cognitive bias in diagnostic errors [27]. In this study, participants responded to the diagnostic error cases that they struggled to diagnose from their memories, which could explain lower instance of premature closure cases reported in this study. Along with these cognitive biases, aggregate bias, triage cueing, diagnosis momentum, representativeness restraint, search satisficing, psych-out error, visceral bias, posterior probability bias, gambler’s fallacy, blind-spot bias, and gender bias have also been reported as cognitive biases that are likely to occur in emergency rooms [9, 28]. However, the opposite opinion, that representativeness restraint and blind-spot bias are less likely to occur [29], has also been reported; therefore, further research is needed.
In this study, confirmation bias, premature closure, base-rate neglect, visceral bias, and Maslow’s hammer were more common during the night shift than day shift. In addition, rule bias, base-rate neglect, availability bias, visceral bias, hassle bias, and Maslow’s hammer were more common among non-emergency than emergency physicians. During the night shift, physicians are more prone to mental exhaustion due to fatigue and lack of sleep. This study also suggests that when non-emergency physicians are working in an emergency room, they may be more likely to be influenced by their own specialty or emotions from their diagnoses, or to follow incorrect rules or ignore disease frequency in an unfamiliar emergency room setting. Improving our understanding and awareness of cognitive biases is a practical first step in overcoming them [30]. A debiasing strategy is needed to overcome the cognitive biases that commonly occur in emergency rooms. Debiasing is having “adequate knowledge of alternative solutions and strategic rules for heuristic responses” and “the ability to disable System 1 processing” [31]. For premature closure, the worst-case scenario should be eliminated by asking, “What else might this be?’ [1, 28], for example, reviewing the differential diagnosis before admitting a patient or reviewing hand radiographs to look for a second fracture rather than assuming there is only one [32]. Anchoring and confirmation bias are closely related [25], and for the former, the diagnosis should be reviewed with new information and data, without preconceptions [28]. For confirmation bias, considering the opposite of the initial hypothesis [1, 28], revisiting the diagnosis if the data do not support it, and using metacognition, error theory, and cognitive coercion strategies [26] are recommended. For availability bias, it is useful to consider the objective reason for the diagnosis [28]. Other ways to reduce cognitive bias are to seek opinions outside of yourself, such as second opinions and decision support systems. A second opinion can be useful in identifying errors that might otherwise be missed and in interpreting test results. Decision support systems include checklists, flow charts, and visual aids. The availability of decision support systems and clinical information i for night shift and non-emergency physicians working the emergency room may reduce reliance on memory, and improve diagnostic reasoning performance under conditions such as stress and fatigue [32, 33].
The most common diseases that caused diagnostic errors include gastrointestinal, hepatobiliary, respiratory, cardiovascular, and infectious diseases as well as metabolic endocrinology, trauma, and malignant tumors [13, 16]. Moreover, in a study reviewing medical malpractice claims in Japan, non-hemorrhagic gastrointestinal diseases, such as gastroenteritis and intestinal obstruction were among the most common initial diagnoses in DERC cases [24]. The top four most common diagnostic error cases in this study were consistent with those of previous studies. In cases of headache, which was the fifth most common symptom, a diagnostic error for stroke was reported in 8.7% of cases [34]. Although subarachnoid hemorrhage occurs in 1–3% of patients with headaches [35], misdiagnosis or delayed diagnosis occurs in 12–51% of cases [36]. Furthermore, it has been reported that diagnostic discrepancies are associated with increased in-hospital mortality [37]. However, to the best of our knowledge, there are no studies that show the differences between the initial and final diagnostic labels; therefore, further research is needed.