Katie E. Raffel

Background: Diagnostic errors are common and represent a leading cause of preventable harm in hospitalized medical patients. Despite growing recognition of this problem, the field lacks a systematic, evidence-based approach for developing, adapting, and selecting diagnostic excellence interventions at scale across multiple institutions. Methods: Using a structured, multi-phase process, the Achieving Diagnostic Excellence through Prevention and Teamwork (ADEPT) collaborative, a 14-site AHRQ-funded hospital medicine network: (1) conducted an environmental scan of existing diagnostic safety practices across sites; (2) constructed a conceptual framework organizing potential interventions by target (patient, provider/team, health system) and safety paradigm (Safety I and Safety II); (3) convened multidisciplinary working groups to develop detailed proposals for six intervention programs; (4) applied a pre-specified multi-criteria rubric to facilitate group deliberation, and (5) voted to select a single high-priority intervention for collaborative launch. Results: Six interventions were developed, selected, or adapted from prior work, spanning all cells of the target and safety paradigm framework. Finalized proposals were then shared at two separate ADEPT meetings – the first to solicit feedback on interventions and to coordinate activities across groups, and the second where a formal ranked-choice vote was held. This process resulted in selection of the Diagnostic Cross-Check (DXC) — a structured, peer second-opinion program triggered by rapid response team (RRT) events — as the highest priority intervention for collaborative-wide launch. The remaining interventions were designated as components of an ADEPT toolbox for future adaptation, adoption, and evaluation. Discussion: The ADEPT collaborative’s systematic, evidence-informed approach to diagnostic intervention development offers a replicable model for other multi-site quality improvement networks. The resulting portfolio of interventions provides a roadmap for the field of inpatient diagnostic excellence.

Background: The National Academies' Improving Diagnosis in Health Care report outlined eight goals for improving the diagnostic process and building a culture of diagnostic excellence. Despite a decade passing since these recommendations, development of programs remains inconsistent. Objectives: To explore diagnostic excellence improvement efforts across hospital medicine divisions in largely academic settings to identify opportunities. Methods: This study used rapid qualitative methods to analyze focus groups with hospitalist clinicians recruited from sites participating in an AHRQ‐funded study. After creating focus group summaries and an analytical matrix, thematic analysis was conducted using a mixed inductive and deductive method at the semantic level. Results: Seven focus groups were conducted with 30 participants across 14 hospitals. Participants described more than 80 interventions, programs, and tools to improve diagnostic processes. Five themes were identified: (1) current diagnostic excellence efforts were weighted toward reactive rather than proactive interventions; (2) successful diagnostic excellence programs need to focus on both systems issues and clinician performance; (3) clinicians desired greater support in the diagnostic process, especially when cognitive load impairs clinical reasoning; (4) clinicians desired a supportive culture for feedback; and (5) for new ideas to succeed, context of care and resource constraints will need to be considered. Discussion: Our findings highlight limited scope in diagnostic safety efforts, which remain largely centered on retrospective case reviews and conferences. Clinicians desire system‐level redesign encompassing improved team dynamics, reduced cognitive load, psychologically safe feedback, and optimization of the electronic health record. This shift is hindered by cultural, leadership, and resource‐related barriers.

Background: Diagnostic errors among hospitalized adults are prevalent and costly. Traditional approaches that focus on identifying and fixing mistakes (Safety‐I) may fall short in the dynamic and complex hospital environment. Instead, focusing on approaches that emphasize the understanding of how things go right and adapt to challenges (Safety‐II) may offer better insights into what contributes to accurate and timely diagnoses. Objectives: We sought to characterize hospitalists' perceptions of diagnostic excellence and the factors and adaptations needed to achieve it. Methods: Focus groups were conducted in partnership with the Hospital Medicine Reengineering Network (HOMERuN). We interviewed physicians, researchers, and patient representatives about (1) recognition of diagnostic excellence, (2) what makes diagnostic journeys go well, and (3) factors that facilitate diagnosis in difficult situations. Rapid qualitative methods including templated summaries and matrix analysis were applied to analyze these focus groups. Results: We conducted five focus groups with 36 individuals from 23 unique institutions. Participants shared three factors that facilitate a successful diagnostic journey: clinical reasoning that is structured yet adaptive, collaborative diagnosis through teamwork, and systems‐level support for diagnostic capacity. Participants readily recognized diagnostic excellence, though noted the challenge in systematically identifying these cases. Conclusions: While diagnostic excellence case identification is seen as a challenge, providers do note people, process, and technology opportunities to support better diagnosis. Optimization of diagnosis in hospital medicine may benefit from a focus on team and communication redesign, adaptive technology development, and workload management.

Team‐based care enhances accurate diagnosis in hospitalized patients. In addition to formal inter‐professional collaboration, such as specialty consultation, many hospitalists use informal strategies, such as peer diagnostic second opinions. Little is known about this practice. We conducted a cross‐sectional survey among hospital medicine physicians and Advanced Practice Providers at 13 hospitals. We assessed the frequency of requesting and providing peer second opinions, comfort discussing diagnostic uncertainty, and perceived barriers to second opinions. 338 hospitalists responded. Half (51.5%) reported routinely seeking a second opinion when facing diagnostic uncertainty. 98.2% felt psychologically safe discussing uncertainty, and 97.3% believed second opinions improved diagnostic accuracy. The most common barriers to seeking a peer‐second opinion were time constraints (64.3%) and fear of bothering colleagues (56.6%). Less experienced hospitalists were more likely to report vulnerability as a barrier (39.2% vs. 26.8%, p = .03). Strengthening systems that facilitate peer consultation and support junior clinicians could improve diagnostic collaboration and accuracy.

Background: Diagnostic errors (DEs) are common in patients who die or go to the intensive care unit (ICU), due to gaps in diagnostic processes, but it is unclear if this holds true for sepsis patients. Objectives: To determine whether patients with sepsis, as defined by International Classification of Disease‐10 (ICD‐10) code, have a different DE risk and/or underlying diagnostic process faults compared with other patients who die or are admitted to an ICU. Methods: We conducted a retrospective cohort study of two thousand four hundred and twenty‐eight adult patients discharged from 29 academic medical centers in 2019 were included in the vizient clinical database and the hospital medicine re‐engineering network, meeting centers for medicare & medicaid services criteria for a “medical” admission. Patients included either died in‐hospital or were admitted to the ICU during the studied admission. Safer Dx and diagnostic error evaluation and research (DEER) taxonomy defined DE and diagnostic process faults, respectively. Adjusted risk ratios (aRRs) for DE were estimated using modified Poisson regression clustered by site. Interaction terms tested differences between patients with and without sepsis. Results: Of 1218 sepsis patients (50.2%), DE rates were similar to nonsepsis patients (22.3% vs. 23.9%, p = .734). Adjusted analysis showed no association between sepsis and DE (aRR: 1.06, 95% confidence interval [CI]: 0.91–1.23). Sepsis was less strongly associated with DE if gaps in history‐taking (interaction‐term p = .02) or consultation problems (interaction‐term p = .005) were present. No other DEER interactions were significant. Conclusion: Sepsis was not associated with a higher adjusted risk for DEs or individual diagnostic process faults, suggesting diagnostic issues in sepsis are tied to broader clinical care problems rather than sepsis‐specific factors.

Given the recent expansion of US immigration and customs enforcement (ICE) detention facilities and the growing number of people in immigration detention, hospitalists are increasingly likely to provide care to immigrant detainees. These patients face distinct ethical, legal, and structural challenges that can affect clinical care. In this perspective piece, we outline practical strategies for hospitalists to support patient‐centered, equitable care for this population, emphasizing approaches that align with ethical and legal principles, mitigate bias, and respect patient autonomy.

Background: Behavioral escalation events have become commonplace in the acute care setting. Disruptive patient behaviors contribute to workplace injuries and can compromise patient care and safety. Despite frequently encountering behavioral escalation, limited research exists on hospitalists' practices and perspectives when addressing disruptive behaviors.Objective: To explore hospitalists' experiences and perspectives when navigating behavioral escalation events in adult acute care settings.Methods: We conducted a rapid qualitative study on February 10, 2023, with four semi‐structured virtual focus groups involving 27 hospital medicine participants across 19 US hospitals via the Hospital Medicine Reengineering Network (HOMERuN). Discussions examined hospitalist experiences with behavioral escalation and perspectives on response strategies.Results: Our rapid qualitative study identified four key themes. (1) Many hospitals lack strategies to identify patients or situations at risk of behavioral escalation. (2) Interdisciplinary collaboration is considered essential, yet dedicated response teams are often unavailable. (3) Lack of standardized approaches to disruptive behaviors and inadequate hospitalist training in de‐escalation may lead to varied responses. (4) Limited proactive measures to anticipate escalation and inconsistent use of EHR behavioral alerts may unintentionally stigmatize patients.Conclusion: Behavioral escalation events in the acute care setting pose significant risks to patient and staff safety while disrupting healthcare delivery. Participants highlighted limited strategies to anticipate behavioral escalation, inconsistent interdisciplinary team response structures, and minimal opportunity for debriefing or event review. Focus groups emphasized the need for systems that support team‐based de‐escalation training and incorporate bias and equity considerations into behavioral response practices.

Background: Workplace violence prevention programs are important for safety, but little is known about how they are perceived by clinicians or whether electronic health record (EHR) behavioral alerts are perceived as effective. Objective: To understand perspectives on the use of EHR behavioral alerts as part of workplace violence prevention programs in hospitals. Methods: Mixed methods study utilizing semi‐structured focus groups with a structured survey. Focus group participants were members of a national hopsitalist consortium. Results: Twenty‐eight individuals from 20 different organizations participated in focus groups, with 24 (86%) completing the survey. There was broad uncertainty in best practices for inpatient workplace violence prevention. There was also wide variation in EHR behavioral alert use across multiple domains, including how and why the alerts are placed and how they are used by the end user. Finally, focus groups had mixed sentiments on the potential impacts of these alerts; among participants who responded to surveys, half (50%) noted that alerts caused deviations in care and 21% indicated they witnessed instances where behavioral alerts led to adverse patient outcomes. Most (67%) survey respondents did not think EHR behavioral alerts prevented workplace violence. The majority (88%) of respondents also reported that patient demographic factors impacted whether an alert was placed. Limitations: The study focused on clinician experience at academic hospitals. Conclusions and Relevance: Reports of uncertainty in best practices for deploying EHR behavioral alerts, along with perceptions of variability in implementation and potential biases in alert usage, raise concerns about their effectiveness and potential for worsening disparities. Standardized, evidence‐based practices that safeguard healthcare workers without compromising patient care and equity are needed.

Patients who reside in areas of high neighborhood disadvantage have poorer health outcomes; the mechanisms for this disparity are complex. We sought to determine if there was an association between neighborhood disadvantage and diagnostic error among a cohort of adult inpatients who experienced either an ICU transfer or in‐hospital death. Using a sample of 527 patients from seven geographically diverse academic medical systems, we compared diagnostic error rates to patients' neighborhood disadvantage levels as measured by the Area Deprivation Index, a validated composite measure of socioeconomic status. In contrast to previous studies that found differences in hospital care based on socioeconomic status, we found no difference in diagnostic error rate between patients based on neighborhood disadvantage. Once a patient reaches the hospital, their risk of diagnostic error is not related to the neighborhood in which they live.

Background: Various strategies have attempted to address increased patient lengths of stay (LOS), but effectiveness varies. Factors related to work design and workforce experience may also play significant roles. Objective: Utilizing data from the Discharge in the A.M. trial, we aimed to validate an electronic measure of workload (i.e., note count) and assess the relationship of workload, patient complexity, and physician years of experience to LOS. Methods: Retrospective observational study at three large academic hospitals with hospital medicine physicians and patients they care for during the study. Measures: Workload as measured by electronic note count and physician years of experience; patient LOS. Results: From February 9, 2021 to July 31, 2021, 59 physicians completed daily surveys for 93% of 2318 scheduled shifts. We observed a moderate correlation ( r : .56) between starting morning census and note counts but no association with LOS. We observed an effect modification between note count and the Charlson Comorbidity Index (CCI), with LOS increasing by 2.3% (95% CI: 0.3%, 4.3%; p = .02) and 3.9% (95% CI: 2.0%, 5.9%; p < .0001) per patient for every 1 unit increase in note count for patients with a moderate CCI or severe CCI, respectively. Years since training was associated with a 0.7% decrease in LOS (95% CI: −1.3%, −0.1%, p = .03). Conclusion: Physician workload, as measured by note count, was associated with longer LOS with higher CCI, while more years of experience was associated with shorter LOS. Original Clinical Trial Registration: ClinicalTrials.gov number, NCT05370638.