Influence of a Large Language Model on Diagnostic Reasoning: A Randomized Clinical Vignette Study.

Importance: Diagnostic errors are common and cause significant morbidity. Large language models (LLMs) have shown promise in their performance on both multiple-choice and open-ended medical reasoning examinations, but it remains unknown whether the use of such tools improves diagnostic reasoning. Objective: To assess the impact of the GPT-4 LLM on physicians' diagnostic reasoning compared to conventional resources. Design: Multi-center, randomized clinical vignette study. Setting: The study was conducted using remote video conferencing with physicians across the country and in-person participation across multiple academic medical institutions. Participants: Resident and attending physicians with training in family medicine, internal medicine, or emergency medicine. Interventions: Participants were randomized to access GPT-4 in addition to conventional diagnostic resources or to just conventional resources. They were allocated 60 minutes to review up to six clinical vignettes adapted from established diagnostic reasoning exams. Main Outcomes and Measures: The primary outcome was diagnostic performance based on differential diagnosis accuracy, appropriateness of supporting and opposing factors, and next diagnostic evaluation steps. Secondary outcomes included time spent per case and final diagnosis. Results: 50 physicians (26 attendings, 24 residents) participated, with an average of 5.2 cases completed per participant. The median diagnostic reasoning score per case was 76.3 percent (IQR 65.8 to 86.8) for the GPT-4 group and 73.7 percent (IQR 63.2 to 84.2) for the conventional resources group, with an adjusted difference of 1.6 percentage points (95% CI -4.4 to 7.6; p=0.60). The median time spent on cases for the GPT-4 group was 519 seconds (IQR 371 to 668 seconds), compared to 565 seconds (IQR 456 to 788 seconds) for the conventional resources group, with a time difference of -82 seconds (95% CI -195 to 31; p=0.20). GPT-4 alone scored 15.5 percentage points (95% CI 1.5 to 29, p=0.03) higher than the conventional resources group. Conclusions and Relevance: In a clinical vignette-based study, the availability of GPT-4 to physicians as a diagnostic aid did not significantly improve clinical reasoning compared to conventional resources, although it may improve components of clinical reasoning such as efficiency. GPT-4 alone demonstrated higher performance than both physician groups, suggesting opportunities for further improvement in physician-AI collaboration in clinical practice.

Clinical Reasoning of a Generative Artificial Intelligence Model Compared With Physicians.

This cross-sectional study assesses the ability of a large language model to process medical data and display clinical reasoning compared with the ability of attending physicians and residents.

How Would You Prevent Subsequent Strokes in This Patient? Grand Rounds Discussion From Beth Israel Deaconess Medical Center.

Stroke is a major cause of morbidity, mortality, and disability. The American Heart Association/American Stroke Association recently published updated guidelines on secondary stroke prevention. In these rounds, 2 vascular neurologists use the case of Mr. S, a 75-year-old man with a history of 2 strokes, to discuss and debate questions in the guideline concerning intensity of atrial fibrillation monitoring in embolic stroke of undetermined source, diagnosis and management of moderate symptomatic carotid stenosis, and therapeutic strategies for recurrent embolic stroke of undetermined source in the setting of guideline-concordant therapy.

"Two Years Later I'm Still Just as Angry": A Focus Group Study of Emergency and Internal Medicine Physicians on Disrespectful Communication.

Phenomenon: Disrespectful behavior between physicians across departments can contribute to burnout, poor learning environments, and adverse patient outcomes. Approach: In this focus group study, we aimed to describe the nature and context of perceived disrespectful communication between emergency and internal medicine physicians (residents and faculty) at patient handoff. We used a constructivist approach and framework method of content analysis to conduct and analyze focus group data from 24 residents and 11 faculty members from May to December 2019 at a large academic medical center. Findings: We organized focus group results into four overarching categories related to disrespectful communication: characteristics and context (including specific phrasing that members from each department interpreted as disrespectful, effects of listener engagement/disengagement, and the tendency for communication that is not in-person to result in misunderstanding and conflict); differences across training levels (with disrespectful communication more likely when participants were at different training levels); the individual correspondent's tendency toward perceived rudeness; and negative/long-term impacts of disrespectful communication on the individual and environment (including avoidance and effects on patient care). Insights: In the context of predominantly positive descriptions of interdepartmental communication, participants described episodes of perceived disrespectful behavior that often had long-lasting, negative impacts on the quality of the learning environment and clinical work. We created a conceptual model illustrating the process and outcomes of these interactions. We make several recommendations to reduce disrespectful communication that can be applied throughout the hospital to potentially improve patient care, interdepartmental collaboration, and trainee and faculty quality of life.

Would You Screen This Patient for Cognitive Impairment? : Grand Rounds Discussion From Beth Israel Deaconess Medical Center.

Dementia, according to the American Psychiatric Association's Diagnostic and Statistical Manual of Mental Disorders, Fifth Edition, is defined by a significant decline in 1 or more cognitive domains that interferes with a person's independence in daily activities. Mild cognitive impairment (MCI) differs from dementia in that the impairment is not sufficient to interfere with independence. For the purposes of this discussion, cognitive impairment (CI) includes both dementia and MCI. Various screening tests are available for CI. These tests ask patients to perform a series of tasks that assess 1 or more domains of cognitive function or ask a caregiver to report on the patient's abilities. A positive result on a screening test does not equate to a diagnosis of CI; rather, it should lead to additional testing to confirm the diagnosis. On review of the evidence, the U.S. Preventive Services Task Force (USPSTF) concluded in 2020 that the evidence was insufficient to assess the balance of benefits and harms of screening for CI in older adults ("I statement"). The USPSTF did clarify that although there is insufficient evidence, there may be important reasons to identify CI. In this article, 2 experts review the available evidence to answer the following questions: What screening tools are available, and how effective are they in identifying patients with CI? What interventions are available for patients found to have CI, to what extent do they improve patient outcomes, and what, if any, negative effects occur? And, would they recommend screening for CI, and why or why not?

How Would You Manage This Patient With Clostridioides difficile Infection? : Grand Rounds Discussion From Beth Israel Deaconess Medical Center.

The Infectious Diseases Society of America/Society for Healthcare Epidemiology of America and the American College of Gastroenterology recently released updated guidelines on management of patients with Clostridioides difficile infection. Although these 2 guidelines generally agree, there are a few important differences in their advice to clinicians. In these rounds, 2 experts, an infectious diseases specialist and a gastroenterologist, discuss antibiotic treatment options for nonsevere disease, the role of fecal microbiota transplantation for fulminant disease, and the use of bezlotoxumab to prevent recurrence in the context of Ms. C, a 48-year-old woman with fulminant C difficile infection.

Accuracy of a Generative Artificial Intelligence Model in a Complex Diagnostic Challenge.

This study assesses the diagnostic accuracy of the Generative Pre-trained Transformer 4 (GPT-4) artificial intelligence (AI) model in a series of challenging cases.

How Would You Resuscitate This Patient With Septic Shock? : Grand Rounds Discussion From Beth Israel Deaconess Medical Center.

Sepsis is a potentially life-threatening systemic dysregulatory response to infection, and septic shock occurs when sepsis leads to systemic vasodilation and subsequent tissue hypoperfusion. The Surviving Sepsis Campaign published updated guidelines in 2021 on the management of sepsis and septic shock. Here, in the context of a patient with septic shock, 2 critical care specialists discuss and debate conditional guideline recommendations on using lactate to guide resuscitation, the use of balanced crystalloids versus normal saline, and the use of corticosteroids.

How Would You Screen This Patient for Colorectal Cancer? : Grand Rounds Discussion From Beth Israel Deaconess Medical Center.

Colorectal cancer (CRC) is the third leading cause of cancer death for men and women in the United States, with an estimated 52 580 people expected to die in 2022. Most frequently, CRC is diagnosed among persons aged 65 to 74 years. However, among persons younger than 50 years, incidence rates have been increasing since the mid-1990s. In 2021, partially because of the rising incidence, the U.S. Preventive Services Task Force (USPSTF) recommended CRC screening for adults aged 45 to 49 years (Grade B recommendation). Options for CRC screening include stool-based and direct visualization tests. The USPSTF did not recommend a specific screening test; rather, its guidance was to select a test after a discussion with the patient. Here, a primary care physician and a gastroenterologist discuss the recommendation to begin CRC screening at age 45, review options for CRC screening, and discuss how to choose among the available options.

How Would You Treat This Patient With Pulmonary Embolism? : Grand Rounds Discussion From Beth Israel Deaconess Medical Center.

Pulmonary embolism can be acutely life-threatening and is associated with long-term consequences such as recurrent venous thromboembolism and chronic thromboembolic pulmonary hypertension. In 2020, the American Society of Hematology published updated guidelines on the management of patients with venous thromboembolism. Here, a hematologist and a cardiology and vascular medicine specialist discuss these guidelines in the context of the care of a patient with pulmonary embolism. They discuss advanced therapies such as catheter-directed thrombolysis in the short-term management of patients with intermediate-risk disease, recurrence risk stratification at presentation, and ideal antithrombotic regimens for patients whose pulmonary embolism was associated with a transient minor risk factor.

Teaching Evidence-Based Physical Diagnosis: A Workshop for Hospitalists.

Introduction: Teaching on physical examination, especially evidence-based physical diagnosis, is at times lacking on general medicine rounds. We created a hospitalist faculty workshop on teaching evidence-based physical diagnosis. Methods: The workshop included a systematic approach to teaching evidence-based physical diagnosis, multiple teaching resources, and observed peer teaching. A long-term follow-up session was offered several months after the workshop. Participants completed questionnaires before and after the workshop as well as after the long-term follow-up session. Results: Four workshops were conducted and attended by 28 unique participants. Five hospitalists attended long-term follow-up sessions. Due to the COVID-19 pandemic, repeat sessions and long-term follow-up were limited. In paired analyses compared to preworkshop, respondents after the workshop reported a higher rate of prioritizing ( p = .008), having a systematic approach to ( p < .001), and confidence in ( p = .001) teaching evidence-based physical diagnosis. Compared to before the workshop, participants after the workshop were able to name more resources to inform teaching of evidence-based physical diagnosis ( p < .001). Informal feedback was positive. Respondents noted that the workshop could be improved by allowing more practice of the actual physical exam maneuvers and more observed teaching. Discussion: We created and implemented a workshop to train hospitalists in teaching evidence-based physical diagnosis. This workshop led to improvements in faculty attitudes and teaching skills. Long-term outcomes were limited by low participation due in part to the COVID-19 pandemic.

How Would You Treat This Patient With Acute and Chronic Pain From Sickle Cell Disease? : Grand Rounds Discussion From Beth Israel Deaconess Medical Center.

Sickle cell disease is prevalent in large numbers of patients in the United States and has a significant global impact. Its complications span numerous organs and lead to reduced life expectancy. Acute and chronic sickle cell pain is a common cause of patient suffering. The American Society of Hematology published updated guidelines on management of acute and chronic pain from sickle cell disease in 2019. Several of the recommendations are conditional and leave specific decisions to the treating physician. These include conditional recommendations about the use of ketamine for acute pain and the initiation and discontinuation of long-term opioid therapy for chronic pain. Here, 2 hematologists discuss these guidelines and make contrasting recommendations for the management of acute and chronic pain for a patient with sickle cell disease.

Providing social support for inpatients: Insights from a virtual medical student initiative.

BACKGROUND: Although the COVID-19 pandemic increased social isolation among hospitalised patients given isolation precautions, visitor restrictions and curtailed interactions with healthcare teams, medical students had limited opportunities for involvement in the care of inpatients. APPROACH: We designed a humanistic and narrative medicine intervention to engage medical students in combating social isolation in hospitalised patients during the COVID-19 pandemic at a tertiary care teaching hospital. In our programme, medical students provided virtual social support to hospitalised patients via phone by providing assistance connecting with family members, having informal conversations and check-ins and writing up patient life narratives. EVALUATION: From April 2020 to March 2021, we received 126 referrals of potentially isolated patients from inpatient medical teams. Fifty patients accepted and received our intervention, including 26 who completed life narratives. Feedback was positive, demonstrating benefit to medical students in learning about humanism and connecting with patients through their life stories. In addition, patients and medical teams felt more supported. We share key operational lessons and resources to facilitate the implementation of this intervention elsewhere. IMPLICATIONS: Our intervention allows medical students to meaningfully contribute to the care of inpatients, support beleaguered inpatient teams and learn important lessons about humanism in medicine. This educational and patient care intervention holds promise in other settings, including beyond the COVID-19 pandemic.

How Would You Treat This Patient Hospitalized With Community-Acquired Pneumonia? : Grand Rounds Discussion From Beth Israel Deaconess Medical Center.

Community-acquired pneumonia is a major cause of morbidity and mortality in the United States, leading to 1.5 million hospitalizations and at least 200 000 deaths annually. The 2019 American Thoracic Society/Infectious Diseases Society of America clinical practice guideline on diagnosis and treatment of adults with community-acquired pneumonia provides an evidence-based overview of this common illness. Here, 2 experts, a general internist who served as the co-primary author of the guidelines and a pulmonary and critical care physician, debate the management of a patient hospitalized with community-acquired pneumonia. They discuss disease severity stratification methods, whether to use adjunctive corticosteroids, and when to prescribe empirical treatment for multidrug-resistant organisms such as methicillin-resistant Staphylococcus aureus and Pseudomonas aeruginosa.

"Friction by Definition": Conflict at Patient Handover Between Emergency and Internal Medicine Physicians at an Academic Medical Center.

INTRODUCTION: Patient handoffs from emergency physicians (EP) to internal medicine (IM) physicians may be complicated by conflict with the potential for adverse outcomes. The objective of this study was to identify the specific types of, and contributors to, conflict between EPs and IM physicians in this context. METHODS: We performed a qualitative focus group study using a constructivist grounded theory approach involving emergency medicine (EM) and IM residents and faculty at a large academic medical center. Focus groups assessed perspectives and experiences of EP/IM physician interactions related to patient handoffs. We interpreted data with the matrix analytic method. RESULTS: From May to December 2019, 24 residents (IM = 11, EM = 13) and 11 faculty (IM = 6, EM = 5) from the two departments participated in eight focus groups and two interviews. Two key themes emerged: 1) disagreements about disposition (ie, whether a patient needed to be admitted, should go to an intensive care unit, or required additional testing before transfer to the floor); and 2) contextual factors (ie, the request to discuss an admission being a primer for conflict; lack of knowledge of the other person and their workflow; high clinical workload and volume; and different interdepartmental perspectives on the benefits of a rapid emergency department workflow). CONCLUSIONS: Causes of conflict at patient handover between EPs and IM physicians are related primarily to disposition concerns and contextual factors. Using theoretical models of task, process, and relationship conflict, we suggest recommendations to improve the EM/IM interaction to potentially reduce conflict and advance patient care.

Would You Recommend Aspirin to This Patient for Primary Prevention of Atherosclerotic Cardiovascular Disease? : Grand Rounds Discussion From Beth Israel Deaconess Medical Center.

Atherosclerotic cardiovascular disease (ASCVD) remains the leading cause of death in the United States. Reducing ASCVD risk through primary prevention strategies has been shown to be effective; however, the role of aspirin in primary prevention remains unclear. The decision to recommend aspirin has been limited by the difficulty clinicians and patients face when trying to balance benefits and harms. In 2016, the U.S. Preventive Services Task Force addressed this issue by determining the risk level at which prophylactic aspirin generally becomes more favorable, recommending aspirin above a risk cut point (>10% estimated ASCVD risk). In 2019, the American College of Cardiology and the American Heart Association issued a guideline on the primary prevention of CVD that recommends low-dose aspirin might be considered for the primary prevention of ASCVD among select adults aged 40 to 70 years who are at higher ASCVD risk but not at increased risk for bleeding. Here, 2 experts discuss how to apply this guideline in general and to a patient in particular while answering the following questions: How do you assess ASCVD risk, and what is the role, if any, of the coronary artery calcium score?; At what risk threshold of benefits and harms would you recommend aspirin or not?; and How do you help a patient come to a decision about starting or stopping aspirin therapy?