RESUMEN
STUDY OBJECTIVE: Hindsight bias is the tendency to overestimate the predictability of an event after it has already occurred. We aimed to evaluate whether hindsight bias influences the retrospective interpretation of clinical scenarios in the field of anesthesiology, which relies on clinicians making rapid decisions in the setting of perioperative adverse events. DESIGN: Two clinical scenarios were developed (intraoperative hypotension and intraoperative hypoxia) with 3 potential diagnoses for each. Participants completed a crossover study reviewing one case without being informed of the supposed ultimate diagnosis (i.e., no 'anchor' diagnosis), referred to as their foresight case, and the other as a hindsight case wherein they were informed in the leading sentence of the scenario that 1 of the 3 conditions provided was the ultimate diagnosis (i.e., the diagnosis the participant might 'anchor' to if given this information at the start). Participants were randomly assigned to (1) which scenario (hypotension or hypoxia) was presented as the initial foresight case and (2) which of the 3 potential diagnoses for the second case (the hindsight case, which defaulted to whichever case the participant was not assigned for the first case) was presented as the ultimate diagnosis in the leading sentence in a 2 (scenario order) x 3 (hindsight case anchor) between-subjects factorial design (6 possible randomization assignments). SETTING: Two academic medical centers. PARTICIPANTS: Faculty, fellow, and resident anesthesiologists and certified nurse anesthetists (CRNAs). INTERVENTIONS: None. MEASUREMENTS: After reading each clinical scenario, participants were asked to rate the probability (%) of each of three potential diagnoses to have caused the hypotension or hypoxia. Compositional data analysis (CoDA) was used to compare whether diagnosis probabilities differ between the hindsight and the foresight case. MAIN RESULTS: 113 participants completed the study. 59 participants (52%) were resident anesthesiologists. Participants randomized to the hypotension scenario as a hindsight case were 2.82 times more likely to assign higher probability to the pulmonary embolus diagnosis if provided as an anchor (95% CI, 1.35-5.90; P = 0.006) and twice as likely to assign higher probability to the myocardial infarction diagnosis if provided as an anchor (95% CI, 1.12-3.58; P = 0.020). Participants randomized to the hypoxia scenario as a hindsight case were 1.78 times more likely to assign higher probability to the mainstem bronchus intubation diagnosis if provided in the anchor statement (95% CI, 1.00-3.14; P = 0.048) and 3.72 times more likely to assign higher probability to the pulmonary edema diagnosis if provided as an anchor (95% CI, 1.88-7.35; P < 0.001). CONCLUSIONS: Hindsight bias influences the clinical diagnosis probabilities assigned by anesthesia providers. Clinicians should be educated on hindsight bias in perioperative medicine and be cognizant of the effect of hindsight bias when interpreting clinical outcomes.
RESUMEN
Background: Patients are increasingly using artificial intelligence (AI) chatbots to seek answers to medical queries. Methods: Ten frequently asked questions in anaesthesia were posed to three AI chatbots: ChatGPT4 (OpenAI), Bard (Google), and Bing Chat (Microsoft). Each chatbot's answers were evaluated in a randomised, blinded order by five residency programme directors from 15 medical institutions in the USA. Three medical content quality categories (accuracy, comprehensiveness, safety) and three communication quality categories (understandability, empathy/respect, and ethics) were scored between 1 and 5 (1 representing worst, 5 representing best). Results: ChatGPT4 and Bard outperformed Bing Chat (median [inter-quartile range] scores: 4 [3-4], 4 [3-4], and 3 [2-4], respectively; P<0.001 with all metrics combined). All AI chatbots performed poorly in accuracy (score of ≥4 by 58%, 48%, and 36% of experts for ChatGPT4, Bard, and Bing Chat, respectively), comprehensiveness (score ≥4 by 42%, 30%, and 12% of experts for ChatGPT4, Bard, and Bing Chat, respectively), and safety (score ≥4 by 50%, 40%, and 28% of experts for ChatGPT4, Bard, and Bing Chat, respectively). Notably, answers from ChatGPT4, Bard, and Bing Chat differed statistically in comprehensiveness (ChatGPT4, 3 [2-4] vs Bing Chat, 2 [2-3], P<0.001; and Bard 3 [2-4] vs Bing Chat, 2 [2-3], P=0.002). All large language model chatbots performed well with no statistical difference for understandability (P=0.24), empathy (P=0.032), and ethics (P=0.465). Conclusions: In answering anaesthesia patient frequently asked questions, the chatbots perform well on communication metrics but are suboptimal for medical content metrics. Overall, ChatGPT4 and Bard were comparable to each other, both outperforming Bing Chat.
RESUMEN
ABSTRACT: In the United States, hospitals must meet eligibility criteria to receive federal funding. Regulatory bodies, such as the Joint Commission, are approved by the government to give, or withhold, accreditation to hospitals. This accreditation is a requisite to continue receiving funding. Hospitals are frequently cited for items such as inadequate wearing of boot covers or covering of facial hair in the operating rooms. There are very little, if any, data to support an improvement in patient safety when these items are complied with. There is, however, a large amount of data showing the negative consequences for patient safety when providers are burned out. We therefore propose that regulatory agencies such as the Joint Commission require that hospital systems measure burnout and reduce concerning levels of burnout in their employees to continue receiving certification. We briefly review evidence-based methods that hospital systems might consider to accomplish this goal.
Asunto(s)
Agotamiento Psicológico/epidemiología , Centers for Medicare and Medicaid Services, U.S./normas , Humanos , Pacientes , Estados UnidosRESUMEN
BACKGROUND: In our current digital age, textbooks have been supplemented or supplanted by multiple online modalities for knowledge acquisition. Trainees, often from a younger generation than their program directors (PDs), prefer asynchronous options such as podcasts, videos, and question banks. We sought to identify whether an educational gap exists between PDs and trainees regarding what is assigned and what is used. METHODS: A national cross-sectional survey was conducted in the United States in 2018-2019 to characterize anesthesiology resident and PD perceptions of academic knowledge acquisition. RESULTS: Of the 149 PDs, 85 completed the survey (57%). Of the 85 PDs, 36 forwarded the survey to residents. Of the 1414 residents who received the survey, 503 residents responded to the survey (36%). The PDs thought residents used didactics, assigned reading, and scheduled simulations more than residents reported (P < .001). Residents reported using self-directed learning more (P = .004). Most residents (74.1%) reported using textbooks or online reading materials. Those residents reporting >70th percentile on the In-Training Exam used textbooks or online materials more than those who reported low scores (<30th percentile; P = .001). CONCLUSIONS: There is a discrepancy between PD and resident views on where and how knowledge acquisition occurs. Asynchronous forms of education (especially podcasts) are popular, but they are rarely assigned by programs. Although residents have a wide variety of learning preferences, textbook and online reading may be associated with higher In-Training Exam scores (a common way that knowledge acquisition is measured). The PDs should consider providing multiple options for optimizing knowledge acquisition, including textbook reading, to meet resident preferences and maximize testing success.
RESUMEN
Residents inhabit an ambiguous world. They are no longer medical students but are still learners. They are not yet attendings but are still paid employees. This ambiguity leads to a misalignment of departmental incentives and trainee expectations. Trainees expect their learning and well-being to be prioritized, while departments are under pressure to meet staffing needs and cut costs. This sets up a fundamental disconnect between the "formal" Accreditation Council for Graduate Medical Education (ACGME) message of well-being and the dominant "hidden" workplace forces that pull in the opposite direction, possibly contributing to the epidemic of burnout in trainees. It is critical that all parties-health systems, graduate medical education (GME) programs, the ACGME, and residents-recognize this disconnect and collaborate to meaningfully implement current ACGME requirements to decompress work intensity and address well-being. Real change will require more than general directives. The ACGME will likely need to take the lead and consider taking a design-thinking approach to structuring regulations governing how and when residents work and how they are supported. It would also be worthwhile to revisit the Institute of Medicine recommendations on GME from 2014 related to funds flow and transformation initiatives. Taking a more comprehensive approach to residents as people, workers, and vital health care professionals is the right thing to do and may well improve retention, reduce burnout, decrease medical errors, and improve care.
Asunto(s)
Educación de Postgrado en Medicina/normas , Estado de Salud , Internado y Residencia/organización & administración , Facultades de Medicina/organización & administración , Estudiantes de Medicina/psicología , Recursos Humanos/organización & administración , Acreditación/normas , Humanos , Tolerancia al Trabajo Programado , Carga de TrabajoRESUMEN
BACKGROUND: Formative events during training help shape professional identity and may impact well-being. This study sought to identify formative experiences during anesthesia residency and measure their perceived impact on well-being. METHODS: A 24-item survey exploring the frequency and perceived impact of formative events was developed through a rigorous process involving a literature search, consultation with medical education experts, resident focus groups, graduate interviews, and pilot testing. All 80 anesthesiology residents at Johns Hopkins University were invited to participate. We measured the frequency of event exposure and perceived impact on well-being. RESULTS: Seventy-six residents (95%) completed the survey. Event exposure rate ranged from 56.6% to 100%. Events with greatest relative impact (RI) overall included leaving work earlier than expected (RIoverall = 85.8), a patient expresses genuine gratitude (RIoverall = 80.2), identifying a faculty role model (RIoverall = 75.7), and having a patient die under my care (RIoverall = 75.6). There was no statistically significant difference on RI for positive versus negative events. The perceived impact of events on well-being varied by gender. CONCLUSIONS: This work provides clarity for residency program leaders and educators about the commonly experienced formative events that have the greatest perceived impact on resident well-being. These results may inform curricular planning and can suggest times when trainees may need attention or support. Future research should evaluate the direct impact of formative events on well-being and the success of related interventions.
RESUMEN
BACKGROUND: Preoperative cognitive impairment is a major risk factor for postoperative delirium. We therefore investigated the prognostic significance and feasibility of administering a brief cognitive screen before surgery. METHODS: Patients > 65 yr of age undergoing hip, knee, or spine surgery were enrolled. A 60-sec cognitive screen, the animal fluency test (AFT), was administered preoperatively. Postoperative delirium was measured using a chart-based tool previously validated using criteria from the Confusion Assessment Method. RESULTS: Of the 362 patients satisfying the inclusion/exclusion criteria, 199 (55%) underwent the AFT. Among them, 57 patients (29%) had an AFT score < 15, and 38 patients (19%, 95% confidence interval [CI]: 14 to 25%) developed postoperative delirium as measured by chart review. Patients with scores of < 15 were more likely to develop postoperative delirium than those who scored ≥ 15 (54% vs 5%, P < 0.01). A multiple logistic regression, with postoperative delirium as the dependent variable, identified an AFT score of < 15 (odds ratio 20.1, 95% CI: 7.9 to 51.4) and high American Society of Anesthesiologists classification (odds ratio 3.5, 95% CI: 1.3 to 9.2) as independent predictors. CONCLUSIONS: The AFT is a potentially useful brief cognitive screen for identifying patients at risk of developing postoperative delirium. Limited participation by eligible participants in this study, however, raises questions about how useful and feasible systematic administration of the test is. Large studies using prospective measurement of postoperative delirium are indicated to validate our results.