Abstract Background Routine application of cervical spine collar for immobilisation is a common practice but remains controversial in the setting of gunshot wounds (GSW) to the neck. Objectives The aim of this study was to review the incidence of unstable cervical spine injuries (CSI) following GSW to the neck in a high volume, developing world setting. Methods A retrospective study conducted over a 10 year period from December 2012 to December 2022 at a major trauma centre in South Africa (Pietermaritzburg Metropolitan Trauma Service). Results A total of 221 patients with GSW were reviewed. Thirty two patients underwent immediate neck exploration and 189 (86%) patients underwent a computed tomography angiography (CTA). Of the 189 patients who underwent CTA of the neck, CSI was noted in 61 (32%). 13 of the 189 patients (7%) had a spinal cord injury. All 13 of these patients had evidence of neurological deficits documented on presentation. Of the 61 patients with CSI, only one patient had an injury considered to be an unstable CSI (comminuted fracture of C5) and underwent an anterior spinal fusion. For those with no neurological deficits on presentation, none developed neurological deficits during their hospital stay or after discharge. Conclusions CSI in the specific context of GSW to the neck is not rare, but the incidence of unstable CSI is a rare entity. There appears to be minimal benefit from the routine application of cervical collar in this setting, especially for patients presenting without neurological deficit.

Abstract Objective The objective of this study is to evaluate the diagnostic accuracy of Extended Focused Assessment with Sonography for Trauma (E‐FAST) performed by senior emergency medicine residents compared with computed tomography (CT) in patients with blunt thoracoabdominal trauma. Methods This prospective observational diagnostic accuracy study was conducted at a Level 1 trauma center between December 2023 and June 2024. Adult patients presenting with isolated blunt thoracoabdominal trauma who underwent both E‐FAST and thoracoabdominal CT were included. E‐FAST examinations were performed at the bedside by senior emergency medicine residents certified in ultrasonography. CT served as the reference standard. Sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV) of E‐FAST were calculated. Concordance between E‐FAST and CT findings was analyzed using Fisher's exact test wherever appropriate. Results A total of 170 patients were analyzed (124 males [72.9%] and mean age: 38.2 ± 14.7 years). E‐FAST demonstrated a sensitivity of 100%, specificity of 98.7%, PPV of 88.2%, and NPV of 100% compared to CT. Concordance between E‐FAST and CT findings was statistically significant for intra‐abdominal free fluid, pneumothorax, and hemothorax ( p < 0.001 for each). No pericardial effusion was detected by either modality. Conclusion E‐FAST performed by adequately trained emergency medicine residents is a rapid, reliable, and highly accurate diagnostic tool in the initial evaluation of blunt thoracoabdominal trauma. Incorporating E‐FAST into standard trauma assessment protocols can enhance diagnostic efficiency and reduce unnecessary CT utilization.

Abstract Background Existing mortality prediction models for intensive care unit (ICU) patients with heart failure (HF) often rely on variables collected at ICU admission and lacked adequate discriminative ability. This study aimed to develop a new prediction model based on ICU discharge variables to improve predictive performance in patients with acute HF. Methods This single‐center prospective observational study aimed to assess all‐cause mortality within 30 days of emergency ICU admission. Candidate variables with p < 0.05 in univariable analysis were dichotomized based on clinical characteristics or a restricted cubic spline model, and included in the multivariate logistic regression model. The clinical congestion score (CCS) score was revised using forward multivariable regression analysis. The discrimination, calibration, and clinical utility of CCS, lung ultrasound (LUS) score, age, LUS score combined with age, CCS combined with LUS score, CCS combined with age, the new prognostic model, and APACHE‐HF were assessed using the bootstrap method with 500 resamples. Results Seventy‐one participants were enrolled in this study, with a median age of 79 years (71.5–84), and 50.7% (36/71) were male. The multivariate logistic regression model revealed that a 1‐point increase in CCS (OR: 2.883, 95% CI: 1.358–6.121, p = 0.006), age ≥85 years (OR: 15.271, 95% CI: 2.434–95.803, p = 0.004), and LUS score ≥3 (OR: 4.646, 95% CI: 1.366–15.799, p = 0.014) were significantly associated with higher all‐cause mortality risk. Based on 500 bootstrap resamples, the new prognostic model (CCS combined with age and LUS score) had the highest AUC of 0.849 among prediction models. Furthermore, this new model demonstrated superior calibration and clinical utility compared to other scoring systems. Conclusion The CCS enhanced by age and LUS score enabled good prediction of all‐cause mortality within 30 days after admission to ICUs, potentially aiding clinical decision‐making for acute HF patients in ICUs including emergency intensive care units.

Abstract Objectives Large language models, such as GPT‐4o, have demonstrated potential in clinical decision‐making; however, their reliability in high‐stakes environments, including emergency department triage, remains uncertain. This study assessed the triage performance of GPT‐4o and GPT‐o1‐mini using raw Turkish trauma notes, focusing on accuracy, expert agreement, and severity of misclassification as determined by the Emergency Severity Index (ESI). Methods The GPT‐4o and GPT‐o1‐mini models were prompted with unstructured Turkish triage notes to retrospectively predict ESI levels. Model outputs were compared to expert‐labeled ESI scores. Performance was evaluated using F1 scores, accuracy, weighted error analysis, and interrater agreement (Cohen's kappa). Spearman correlation was used to assess associations between model predictions and ESI scores. Results A total of 5031 patients were analyzed based on age, sex, and raw triage notes. GPT‐4o achieved its best F1 scores in ESI‐3 (0.42) and ESI‐5 (0.57), whereas GPT‐o1‐mini performed better in ESI‐3 and ESI‐4. Both models achieved high overall accuracy in the binary classification of high‐acuity trauma cases (GPT‐4o: 91%, GPT‐o1‐mini: 93%). However, sensitivity was limited (GPT‐4o: 0.45; GPT‐o1‐mini: 0.29). GPT‐4o performed better in identifying critical cases (ESI 1–2), whereas GPT‐o1‐mini excelled in low‐acuity predictions (ESI 4–5). Agreement with expert ratings was fair ( κ ≈ 0.22–0.25). The area under the curve was 0.69 (95%CI: 0.645–0; p ≤ 0.001) for GPT‐4o and 0.62 (95%CI: 0.577–0.669; p ≤ 0.001) for GPT‐o1‐mini, suggesting limited clinical reliability. Weighted error analysis showed that GPT‐o1‐mini resulted in less severe misclassifications. Conclusion Neither GPT‐4o nor GPT‐o1‐mini achieved expert‐level performance in trauma triage, particularly for high‐acuity cases. Although GPT‐4o showed slightly better accuracy and GPT‐o1‐mini produced fewer severe errors, both models require substantial refinement before they can be considered for clinical use.

Abstract Background Blood transfusion plays a crucial role in the emergency care of trauma patients, significantly impacting their survival rates and prognoses, thereby saving millions of lives annually. Early and rapid recognition of transfusion needs in trauma patients is essential. This study aims to establish a predictive model for emergency transfusions in patients with severe trauma using machine learning (ML). Methods Data were obtained from a comprehensive and anonymized set of medical records. LASSO regression was employed for feature selection. Six ML algorithms were utilized to develop predictive models. The performance of these models was assessed based on their identification accuracy, calibration, and clinical utility. Additionally, the SHapley Additive exPlanations (SHAP) method was applied to visualize model features and predictions on an individual case basis. Results A total of 1716 trauma patients were included in the study and 278 (16.2%) receive blood transfusion after emergency room admission. A model with 11 variables was built, with XGBoost performing best, achieving an area under the curve of 0.884 (95% CI: 0.847–0.921) and brier score of 0.0878 (0.0734–0.1071). Key predictors included, shock index, systolic blood pressure, heart rate, traumatic brain injury, hepatic insufficiency, age, gender, respiratory rate, percutaneous arterial oxygen saturation, pelvic fracture, and femoral fracture. The model also showed robust net benefit across a threshold probability (0.1–0.75). Conclusion We developed a ML model to predict the need of transfusion in trauma patients and conducted a comprehensive assessment of 6 models in terms of discrimination, calibration, and clinical utility. The SHAP method was employed to visually interpret the influence of each variable, thereby enabling clinicians to better understand the underlying mechanisms of ML.

Abstract Background Endotracheal intubation (ETI) during massive hematemesis is challenging due to airway contamination. The suction‐assisted laryngoscopy and airway decontamination (SALAD) technique has shown promising results in manikin studies involving emergency doctors, but its effectiveness and feasibility with Hong Kong paramedics, for whom ETI is not a standard practice, remains unexplored. Objectives This study aimed to compare the efficacy of a simplified SALAD technique using a DuCanto catheter versus a conventional technique with a Yankauer suction catheter during paramedic‐performed ETI in a simulated massive hematemesis scenario, while assessing its feasibility for Hong Kong's prehospital system. Methods A randomized crossover manikin study was conducted involving 34 paramedics from the Fire Services Department, Hong Kong Special Administrative Region. Participants performed intubations using both techniques on a high‐fidelity manikin simulating massive hematemesis. Primary outcome was the first attempt intubation success rate. Secondary outcomes included intubation time, aspiration volume, esophageal intubation rate, and participants' perceptions via 5‐point Likert scales. Results SALAD‐1 demonstrated superior performance across multiple metrics including first attempt intubation success rate (73.5% vs. 52.9% and p = 0.078), intubation time (50.56 vs. 58.73 s and p = 0.033), and aspiration volume (22.06 vs. 286.32 mL and p < 0.001). Esophageal intubation occurred in three cases with the conventional method but in none with SALAD‐1, a difference that was not statistically significant ( p = 0.119). Participant feedback strongly favored SALAD‐1 across all evaluation domains (mean scores 4.5–4.79/5 vs. 2.91–3.44/5 and p < 0.001). Conclusions This study provides preliminary evidence that the SALAD technique can be rapidly acquired and effectively applied by Hong Kong paramedics in a simulated setting, demonstrating both clinical advantages and strong operator acceptance. These findings support the integration of SALAD into future paramedic training curricula and warrant further clinical research.

Abstract Background Territory‐wide epidemiological studies of acute Chinese medicine poisoning involving aconite after 2010 are lacking. Objectives To characterise the latest trends, presumed causes, clinical presentations, healthcare utilisation and patient outcomes of such poisoning cases in Hong Kong. Methods This was a retrospective observational study of consecutive patients reported to the Hong Kong Poison Information Centre for acute Chinese medicine poisoning by all public emergency departments (EDs) between 1 July 2008 and 30 June 2021. Aconite poisoning was defined by clinical toxicologist verification and laboratory detection of aconitine, related alkaloids or their metabolites in patient biological or herbal remnant samples. We analysed trends using a negative binomial model and characterised presentations using descriptive statistics. Results In total, 179 episodes of laboratory‐confirmed aconite poisoning occurred during the study period. The median annual incidence was 0.22 per 100,000 population (interquartile range 0.11–0.27), with a significant downward trend from 2009 to 2020 (relative risk 0.93, p = 0.037). Fuzi was the most frequently implicated herb (59.2%) and decoction was the most common herbal formulation. Most cases were related to the consumption of Schedule 2 processed aconite. Self‐purchase was common. Overdose accounted for 55.5% of cases with available dose information. Supportive treatment remained the mainstay of management. The mortality rate was 1.1%. Conclusions Despite a decline in incidence, Chinese medicine poisoning with aconite remains a recurring and life‐threatening problem in local EDs. Limiting the dose per sale in self‐purchase and strengthening public education on appropriate aconite use may further reduce such toxicities.

Abstract Background In‐hospital cardiac arrests present significant challenges for healthcare facilities. Basic life support (BLS) training for nonphysician employees is crucial to strengthening the in‐hospital chain of survival, yet this high‐stakes, infrequently used skill has received limited research attention. Objectives This study evaluated the learning effectiveness and learner experiences during a renewal of the institutional BLS course for nonphysician employees. Methods A 2‐year prospective, comparative, mixed‐methods study was conducted at a tertiary medical center in Taiwan. Participants included nurses, nonclinical healthcare providers, administrative staff, and research assistants. The renewed course incorporated flipped learning, team‐based practice, technology‐enhanced learning, and real‐time feedback. Follow‐up assessments at 3 months included written and skills tests, and participants also completed one‐on‐one interviews. Quantitative results were analyzed statistically, and qualitative data were thematically analyzed. Results From April 2021 to July 2023, 68 BLS courses trained 2052 participants. Follow‐up results showed a significant decline in knowledge and skills, but the renewed course improved retention in specific skills, including pulse check ( p = 0.03), chest compression location ( p = 0.04), and rechecking pulse after compressions ( p = 0.004). Interview themes highlighted factors facilitating learning, such as interactive methods and real‐time feedback, and revealed prevalent emotions including worry, nervousness, and hesitation. Conclusions Integrating evidence‐based teaching strategies can enhance skills’ retention in BLS training for nonphysician hospital staff. Given the emotional stress reported, future courses should address both technical competencies and emotional preparedness to improve confidence and performance in real emergencies.