Sukyo Lee

Background and Objectives: The prognostic relevance of early post-resuscitation interleukin-6 (IL-6) and tumor necrosis factor-α (TNF-α) remains uncertain. We evaluated their associations with discharge neurological outcome and combined discrimination. Materials and Methods: This prospective cohort study included adults achieving sustained return of spontaneous circulation (ROSC) after non-traumatic cardiac arrest. Plasma IL-6 and TNF-α were measured early after ROSC. Favorable neurological outcome was defined as Cerebral Performance Category 1–2 at hospital discharge. Analyses included receiver operating characteristic (ROC) curves and Firth regression adjusted for age, arrest location, and shockable initial rhythm. Results: Among 56 patients, 13 (23.2%) had favorable outcomes. The median ROSC-to-blood sampling interval was 29 min in both groups (p = 0.586). IL-6 concentrations were lower in patients with favorable than unfavorable outcomes (33.8 [15.9–52.6] vs. 66.2 [35.5–157.6] pg/mL), as were TNF-α concentrations (11.6 [4.9–15.3] vs. 27.6 [11.3–71.4] pg/mL; Holm-adjusted p = 0.034 for both). Areas under the ROC curves were 0.717 (95% confidence interval [CI], 0.541–0.867) for IL-6, 0.721 (0.552–0.862) for TNF-α, and 0.723 (0.553–0.865) for their combination, without significant pairwise differences. After adjustment, neither IL-6 (odds ratio per two-fold increase, 0.85; 95% CI, 0.49–1.53; p = 0.561) nor TNF-α (0.71; 95% CI, 0.43–1.11; p = 0.129) was independently associated with outcome. Conclusions: Early IL-6 and TNF-α concentrations differed between outcome groups and showed modest unadjusted discrimination, but neither demonstrated an independent association after adjustment. Given the few favorable outcomes, these exploratory findings are hypothesis-generating and require independent validation in larger cohorts.

Correction: An early presentation of neurogenic pulmonary edema in acute subarachnoid hemorrhage. Hyunju Kim,Hyunjoon Oh,Jonghak Park,Hanjin Cho,Sukyo Lee. Signa Vitae

Bentazone is a widely used herbicide considered a moderate hazard. Fatalities associated with its use are rare, with deaths reported only after intake ≥200 mL. In some studies, its use is accompanied by generalized rigidity. Malignant hyperthermia is a pharmacogenetic disease that presents a hypermetabolic response to anesthetic gases or depolarizing muscle relaxants due to calcium channel dysfunction. The classic symptoms of malignant hyperthermia include hyperthermia and muscle rigidity. We report the case of a 65-year-old man who died 4 hours after presenting to the emergency department after ingesting approximately 75 mL of Basagran M60 (bentazone 33.6%, 25.2 g). This is the smallest dose (364 mg/kg) reported in a fatal case to date. Electrocardiogram changes, including QRS widening and QT prolongation, were present, and hypocalcemia was confirmed. We propose that bentazone intoxication causes patient deterioration by a mechanism similar to that of malignant hyperthermia.

Objectives: Accurate triage in emergency departments (ED) is critical for appropriate resource allocation. While artificial intelligence (AI) has been explored for triage, prior models relied on summarized clinical scenarios. We aimed to develop and evaluate large language models (LLMs) trained on real-world clinical conversations to classify patient urgency. Materials and Methods: We used a nationally curated dataset of anonymized triage-level conversations from 3 tertiary Korean hospitals. Two BERT-based models were developed to classify urgency per the Korean Triage and Acuity Scale (KTAS) into urgent (KTAS 3) or non-urgent (KTAS 4-5). One model tokenized the entire conversation, while the other applied a hierarchical structure with sentence-level tokenization and speaker-role embeddings. Performance metrics included accuracy, precision, recall, and F1-score. We compared our models against ChatGPT GPT-4o and ClinicalBERT, and assessed explainability using SHapley Additive exPlanations (SHAP). Results: A total of 5244 clinical conversations, 1057 triage-level dialogues were used, with 950 for training and 107 for testing. Our model with hierarchical structure achieved accuracies of 75.94%, significantly outperforming ChatGPT (56.68%) or fine-tuned ClinicalBERT (69.42%). For urgent cases, the best model achieved a recall of 0.9610, outperforming ChatGPT (0.5352). SHapley Additive exPlanations analysis confirmed that our model focused on clinically relevant cues aligned with KTAS criteria. Conclusion: BERT-based LLMs trained on real-world ED conversations significantly outperform general-purpose models like ChatGPT in triage accuracy. This approach demonstrates the potential for enhancing clinical decision support with interpretable and efficient AI.

Background/Objectives: Minimizing interruptions in chest compressions is very important when resuscitating patients with cardiac arrest. Recently, research has analyzed electrocardiograms (ECGs) during chest compressions using convolutional neural networks (CNNs). This study aimed to compare the accuracy of deeper neural networks and more advanced structures. Methods: ECGs with chest compression artifacts were obtained from patients who visited the emergency department of Korea University Ansan Hospital from September 2019 to February 2024. We compared the accuracy of a deeper CNN, long short-term memory (LSTM), and a CNN with an attention mechanism and residual block against a reference model. The input of the model was 5 s ECG segments with compression artifacts, and the output was the probability that the ECG with the artifacts was a shockable rhythm. Results: A total of 1889 ECGs with compression artifacts from 172 patients were included in this study. There were 969 ECGs annotated as shockable and 920 as non-shockable. The area under the receiver operating characteristic curve (AUROC) of the reference model was 0.8672. The AUROCs of the deeper CNN for five and seven layers were 0.7374 and 0.6950, respectively. The AUROCs of LSTM and bidirectional LSTM were 0.7719 and 0.8287, respectively. The AUROC of the CNN with the attention mechanism and residual block was 0.7759. Conclusions: CNNs with deeper layers or those incorporating attention mechanisms, residual blocks, and LSTM architectures did not exhibit better accuracy. To improve the model accuracy, a larger dataset or advanced augmentation techniques may be required, rather than complicating the structure of the model.

Background: The early diagnosis and prompt treatment of sepsis can enhance clinical outcomes. This study aimed to assess the relationship between point-of-care testing (POCT) for lactate levels and both adherence to the Surviving Sepsis Campaign (SSC) guidelines and mortality rates among sepsis patients in the emergency department (ED). We hypothesized that bedside lactate POCT would lead to better clinical outcomes. Methods: We conducted a pre–post observational study utilizing data from a prospectively collected sepsis registry. Following the introduction of lactate POCT, lactate levels were determined using both the central laboratory pathway and a POCT device. We then compared the characteristics and clinical outcomes between the periods before and after the introduction of POCT lactate measurement. Results: The analysis included a total of 1191 patients. The introduction of bedside lactate POCT led to a significant reduction in the time taken to obtain lactate results (from 53 to 33 min) and an increase in the rate of subsequent lactate measurements (from 82.1% to 88.2%). Lactate POCT did not significantly affect adherence to the overall SSC guidelines bundle (47.5% vs. 45.0%) or reduce 30-day mortality rates (31.1% vs. 31.4%). However, bedside lactate POCT could decrease extremely delayed lactate measurements. Conclusions: Bedside lactate POCT successfully reduced the time to obtain lactate results. Although lactate POCT did not lead to improved adherence to the overall SSC guidelines bundle or affect short-term mortality rates in sepsis patients, it may have an advantage in a specific situation such as overcrowded ED where there are subsequent or multiple measurements required.

Ischemia-modified albumin (IMA) is produced during ischemia and reactive oxygen species production. This study aimed to evaluate the association between IMA and mortality in a larger population and the prognostic value of the combination of IMA and lactate for predicting mortality in septic shock patients in the emergency department. This retrospective observational study included adult septic shock patients between October 2019 and December 2021. A multivariable Cox proportional hazards model was performed. IMA was significantly higher in the non-surviving group than in the surviving group (89.1 ± 7.2 vs. 83.8 ± 6.2 U/mL, p < 0.001). IMA was independently associated with 28-day mortality after adjustments (adjusted hazard ratio [aHR]: 1.075, 95% confidence interval [CI]: 1.016–1.138, p = 0.012). The area under the ROC curve (AUROC) of IMA was 0.712 (95% CI: 0.648–0.775, p < 0.001) and was comparable to that of lactate. The AUROC of the combination of IMA and lactate was 0.838 (95% CI: 0.786–0.889, p < 0.001). The group with both high lactate and high IMA levels showed an extremely high risk of mortality than other groups (86.1%; aHR 8.956, 95% CI 4.071–19.70, p < 0.001). The elevation of IMA was associated with mortality in septic shock patients. The combination of IMA and lactate can be a helpful tool for early risk stratification of septic shock patients.

Objective Many studies have examined the July effect. However, little is known about the July effect in sepsis. We hypothesized that the July effect would result in worse outcomes for patients with sepsis. Methods Data from patients with sepsis, collected prospectively between January 2018 and December 2021, were analyzed. In Korea, the new academic year starts on March 1, so the “July effect” appears in March. The primary outcome was 30-day mortality. Secondary outcomes included adherence to the Surviving Sepsis Campaign bundle. Outcomes in March were compared to other months. A multivariate Cox proportional hazard regression was performed to adjust for confounders. Results We included 843 patients. There were no significant differences in sepsis severity. The 30-day mortality in March was higher (49.0% vs. 28.5%, P<0.001). However, there was no difference in bundle adherence in March (42.2% vs. 48.0%, P=0.264). The multivariate Cox proportional hazard regression showed that the July effect was associated with 30-day mortality in patients with sepsis (adjusted hazard ratio, 1.925; 95% confidence interval, 1.405–2.638; P<0.001). Conclusion The July effect was associated with 30-day mortality in patients with sepsis. However, bundle adherence did not differ. These results suggest that the increase in mortality during the turnover period might be related to unmeasured in-hospital management. Intensive supervision and education of residents caring for patients with sepsis is needed in the beginning of training.