Background. Burnout among medical students is associated with depression, fatigue, suicidal ideation, and impaired academic performance. Authors evaluated changes in burnout and well-being among medical students at the University of Kansas School of Medicine (KUSM) between 2019 and 2025 and explored students’ perspectives on promoting wellness.Methods. A repeated cross-sectional, anonymous online survey was administered to all KUSM medical students in 2025. Measures included burnout, depression, fatigue, quality of life, general health, suicidal ideation, and open-ended wellness questions. Quantitative findings were compared with a similar 2019 survey. Bivariate analyses, multivariable logistic regression, and thematic analysis were performed.Results. Of 808 eligible students, 295 participated (36.5%) and 288 completed the survey (response rate, 35.6%). Overall burnout prevalence remained similar to 2019 (43.4% vs. 48.0%), although depersonalization decreased significantly (16.3% vs. 26.5%; p=0.002). Depression remained common (40.4%), while fatigue worsened significantly (p=0.018). Overall, 7.6% of respondents reported suicidal ideation during the previous 12 months. Burnout was significantly associated with depression, fatigue, and suicidal ideation (all p <0.001). Third-year students had the highest rates of psychological distress and more than twice the odds of screening positive for depression than first-year students (aOR, 2.13; 95% CI, 1.04-4.36). Qualitative analysis identified 10 themes, with students attributing burnout primarily to insufficient time, limited flexibility, excessive workload, and reduced autonomy.Conclusions. Burnout remains highly prevalent among medical students and is closely associated with depression, fatigue, and suicidal ideation. Students emphasized the need for system-level interventions that increase flexibility, protected time, and autonomy to improve learner well-being.
Introduction. Large language models (LLMs) are being explored for multiple applications in medical research, including medical text classification. Authors of the study evaluated the performance of three off-the-shelf LLMs in classifying unstructured free-text CT pulmonary angiography reports for pulmonary embolism (PE)-related diagnostic labels.Methods. Authors assessed human-labeled CT reports using three LLMs (ChatGPT-4o, Llama 3.3 70B, and Llama 3.1 8B) with zero-shot simple and chain-of-thought (CoT) prompts. Classification tasks included acute PE (N = 1,025) and, among reports with PE (n = 660), bilateral PE and large PE.Results. LLMs demonstrated high classification performance for acute PE and bilateral PE but lower performance for large PE. With simple prompts, the highest F1 scores were 97.0% for acute PE (Llama 3.1 8B; sensitivity, 95.6%; specificity, 97.3%), 92.9% for bilateral PE (ChatGPT-4o; sensitivity, 98.0%; specificity, 88.6%), and 72.8% for large PE (ChatGPT-4o; sensitivity, 82.9%; specificity, 82.2%). CoT prompting did not meaningfully improve acute PE classification but provided slight improvements in subclassification performance for bilateral and large PE.Conclusions. Off-the-shelf LLMs can classify free-text CT reports using simple prompt engineering and may facilitate automated data extraction for medical research following appropriate evaluation of classification performance.
Introduction. Safe anesthesia requires technical and non-technical skills to prevent adverse events. Non-technical skills training is not routinely incorporated into formal anesthesia education in the United States. This quality educational improvement initiative aimed to provide clinical anesthesia (CA) residents with non-technical skills training.Methods. Three anesthesiologists completed virtual Anaesthetists’ Non-Technical Skills (ANTS) training in May 2023. An anesthesiologist developed two high-stress simulation scenarios. Fourteen residents were randomized into dyads for the first simulation/post-simulation survey, followed by a two-part didactic seminar on human factors based on the ANTS framework and a post-didactic survey. One hundred twenty days later, residents were re-randomized for a second simulation and another post-simulation survey. Two ANTS-trained anesthesiologists reviewed video recordings of both simulations.Results. Cohen’s kappa demonstrated moderate agreement (κ = 0.43; 95% CI, 0.36-0.49; p <0.001) between evaluators. Wilcoxon signed-rank tests indicated improved non-technical skill use among CA1 residents, while senior residents generally performed better during the first simulation. Fisher’s exact tests showed no significant associations in resident knowledge or confidence following training.Conclusions. Non-technical skills training may benefit residents early in CA training, while senior residents may have different learning needs. Simulation-based training appears feasible; however, formal, recurring, and consistent education may be needed to reinforce these skills throughout residency.
Introduction. JayDoc Community Clinic-Wichita is a student-run free clinic serving a predominantly Spanish-speaking patient population. Authors of this study evaluated whether interpreter staffing represents a potentially modifiable systems-level factor associated with clinic workflow.Methods. This retrospective observational quality improvement evaluation included 715 patient encounters nested within 128 clinic sessions at JayDoc from November 2023 to July 2025. Encounter-level workflow differences by Spanish-speaking status were evaluated using linear mixed-effects models accounting for clustering within clinic sessions and adjusting for clinic day. Clinic-session-level multivariable linear regression evaluated associations among Spanish-speaking patient volume, in-person interpreter staffing, and clinic workflow.Results. After accounting for clustering and clinic day, Spanish-speaking encounters were 5.3 minutes longer than non-Spanish-speaking encounters, although the difference was not statistically significant (p=0.065). Spanish-speaking encounters had significantly longer medical preparation time (adjusted mean difference, 2.57 minutes; 95% CI, 0.83 to 4.31; p=0.004) and wait-to-present time (adjusted mean difference, 3.46 minutes; 95% CI, 0.47 to 6.45; p=0.024). At the clinic-session level, each additional Spanish-speaking patient was associated with a 6.7-minute increase in clinic-session duration after controlling for total patient volume and interpreter staffing (p <0.001). Each 1-unit increase in the Spanish-speaking patient-to-interpreter ratio was associated with a 3.72-minute increase in clinic-session duration (p=0.052).Conclusions. Spanish-speaking encounters had longer medical preparation and wait-to-present times, while greater Spanish-speaking patient volume was associated with longer clinic-session duration. In-person interpreter staffing was not independently associated with session duration. These findings support further evaluation of strategies to better align interpreter resources with patient needs.
