Archives of Academic Emergency Medicine最新文献

Introduction: Various tools have been developed to determine the priority of radiography in trauma patients. This study aimed to investigate the role of machine learning models in predicting chest injuries following multiple trauma.

Methods: We used the database of a comprehensive cross-sectional survey conducted in 2015. Eight machine learning models were developed using demographic characteristics, physical exam findings, and radiologic results of 2860 patients.

Results: Area under the receiver operating characteristic curve (AUC) was greater than 0.96 in Random Forest, Gradient Boosting, XGBoost, Decision Tree, Support Vector Machine (SVM), Logistic Regression, K-Nearest Neighbors (KNN), and Neural Network models. The random forest model, XGBoost and Gradient Boosting had the highest accuracy (0.99). Sensitivity was also highest in the Gradient Boosting, XGBoost and KNN models (0.99). The specificity of all of the models in predicting chest radiography outcomes of multiple trauma patients was higher than 0.97, except for logistic regression and SVM (0.912 and 0.885 respectively).

Conclusions: Our study highlights the strong potential of machine learning models, especially Random Forest and Gradient Boosting, in predicting chest trauma outcomes with high accuracy and sensitivity.

Introduction: Prehospital Return of Spontaneous Circulation (P-ROSC), Utstein-Based Return of Spontaneous Circulation (UB-ROSC), and Return of Spontaneous Circulation After Cardiac Arrest (RACA) scores have been developed to estimate the likelihood of Return of Spontaneous Circulation (ROSC) in Out-of-hospital cardiac arrest (OHCA). This study aimed to validate and compare these three scoring systems.

Methods: A retrospective cohort study was conducted using electronic medical records of OHCA patients transported by Ramathibodi Emergency Medical Service (EMS) from January 2021 to October 2024. We included all OHCA patients aged >18 years who transported by Ramathibodi EMS. RACA, UB-ROSC, and P-ROSC scores were calculated, and ROSC was recorded. The area under the ROC curve (AUC) of each score were calculated to assess predictive accuracy.

Results: Among 336 OHCA cases, 94 (27.97%) patients achieved ROSC. The RACA score demonstrated the highest predictive accuracy, with an AUC of 0.77 (95% CI: 0.71-0.82). The UB-ROSC score followed with an AUC of 0.72 (95% CI: 0.66-0.78), while the P-ROSC score had the lowest predictive value with an AUC of 0.64 (95% CI: 0.58-0.70). Calibration analysis indicated that the RACA score aligned most closely with observed outcomes compared to the UB-ROSC and P-ROSC scores. The RACA score exhibited the best overall performance in terms of both discrimination and calibration.

Conclusions: Among the three predictive models assessed, the RACA and UB-ROSC scores demonstrated fair predictive accuracy for ROSC in OHCA patients, while the P-ROSC score had poor predictive value.

Introduction: Delays in accessing an appropriate level of care can lead to significant morbidity or even mortality of trauma patients. This study aimed to develop a simplified prehospital predictive model to determine the need for tertiary care trauma centers (TTC), enabling timely and appropriate transport decisions by emergency medical service (EMS) teams.

Methods: This is a retrospective cohort study conducted at the emergency department (ED) of Ramathibodi Hospital between January 2020 and April 2024. Prehospital trauma patients aged ≥15 years who were transported by EMS were included in the study. Patients were divided into two groups with and without the need for TTC, and the independent predictive factors of the need for TTC were explored using multivariable regression analysis.

Results: The study included 440 trauma patients, with 31.1% requiring TTC. The predictors of the need for TTC included age (coefficient (Coef.) -0.003; 95% confidence interval (CI): -0.018 to 0.012; P=0.693), traffic mechanism (Coef. 0.848; 95%CI: 0.150 to 1.546; P=0.017), respiratory rate (Coef. 0.044; 95%CI: -0.037 to 1.124; P=0.285), heart rate (Coef. -0.004; 95%CI: -0.020 to 0.012; P=0.610), and Glasgow Coma Scale (Coef. -0.312; 95%CI: -0.451 to -0.173; P<0.001). The predictive model categorized patients into low, moderate, and high-risk groups. Patients who were categorized in the high-risk group showed a positive likelihood ratio (LHR+) of 14.88 for requiring TTC. The model achieved an area under the receiver operating characteristic curve (AuROC) of 73%, indicating the good discriminative ability of this prediction model.

Conclusions: The predictive model classifies trauma patients into three risk groups based on five prognostic variables, which are able to predict the likelihood of requiring TTC. Internal validation has verified its high level of accuracy in trauma triage.

Introduction: Predicting the number of emergency medical team (EMT) consultations that are needed following a natural or man-made disaster can help improve decisions regarding the dispatch and withdrawal of these teams. This study aimed to predict the number of consultations by EMTs using the K value and constant attenuation model.

Methods: Data were collected using the Japan-Surveillance in Post-Extreme Emergencies and Disasters (J-SPEED) and Minimum Data Set (MDS) for five disasters in Japan and one disaster in Mozambique. We compared the number of consultations, which was predicted based on K value and constant attenuation model with actual data collected with J-SPEED/Minimum Data Set (MDS) tools.

Results: The total number of EMT consultations per disaster ranged from 684 to 18,468. The predicted curve and actual K data were similar for each of the disasters (R² from 0.953 to 0.997), but offset adjustments were needed for the Kumamoto earthquake and the Mozambique cyclone because their R² values were below 0.985. For the six disasters, the difference between the number of consultations predicted based on K values and the measured cumulative number of consultations ranged from ±1.0% to ± 4.1%.

Conclusions: The K value and constant attenuation model, although originally developed to predict the number of patients with COVID-19, provided reliable predictions of the number of EMT consultations required during six different disasters. This simple model may be useful for the coordination of future responses of EMTs during disasters.

Introduction: Emergency medical service (EMS) response time is a critical factor in managements of out-of-hospital cardiac arrest (OHCA) cases. This study aimed to investigate the effects of EMS response time on survival of OHCA patients.

Methods: This study employed a retrospective cohort design focused on prognosis research. Data was collected from the Erawan EMS Dispatch Center of the Bangkok Metropolitan Administration from January 2019 to December 2023. All OHCA cases visited by dispatched prehospital teams in Bangkok were included. Multivariable logistic regression was used to analyze the effect of response time on survival at scene, survival to emergency department (ED), and survival to hospital discharge of OHCA cases.

Results: Among the 5,433 OHCA patients included in the study, 29.17% achieved return of spontaneous circulation at the scene, 6.9% survived to ED, and 1% survived to hospital discharge. Each 1-minute increase in response time decreased the likelihood of survival at the scene by 6% (OR: 0.94, p < 0.001), survival to ED admission by 4% (OR: 0.96, p < 0.001), and survival to hospital discharge by 6% (OR: 0.94, p = 0.006). Response times under 8 minutes significantly improved outcomes, with survival at the scene increasing by 2.31 times (p < 0.001), survival to ED by 1.76 times (p < 0.001), and survival to hospital discharge by 2.09 times (p = 0.048).

Conclusions: A maximum response time of 8 minutes significantly enhances survival outcomes, including survival at the scene, survival to ED, and survival to hospital discharge. Furthermore, each 1-minute increase in response time is associated with a 6% reduction in the likelihood of survival to hospital discharge.

Introduction: Coronary artery bypass grafting (CABG) surgery requires an extended length of stay (LOS) in the intensive care unit (ICU). This study aimed to predict the factors affecting LOS in the ICU after CABG surgery using machine learning methods.

Methods: In this study, after extracting factors affecting the LOS of patients in the ICU after CABG surgery from the literature and confirming these factors by experts, the medical records of 605 patients at Farshchian Specialized Heart Hospital were reviewed between April 20 and August 9, 2024. Four machine learning models were trained and tested to predict the most desired factors, and finally, the performance of the models was evaluated based on the relevant criteria.

Results: The most important predictors of the LOS of CABG patients in the ICU were the length of intubation, body mass index (BMI), age, duration of surgery, and the number of postoperative transfusions of packed cells. The Random Forest model also performed best in predicting the effective factors (Mean square Error = 1.64, Mean absolute error = 0.93, and R² = 0.28).

Conclusion: The insights gained from the mashine learning model highlight the significance of demographic and clinical variables in predicting LOS in ICU. By understanding these predictors, healthcare professionals can better identify patients at higher risk for prolonged ICU stays.

Introduction: Traumatic injuries can affect respiration both directly and indirectly. This study aimed to evaluate the predictive factors of need for mechanical ventilation (MV) following traumatic injuries.

Methods: This Retrospective registry-based cross-sectional study comprised trauma patients admitted to a major referral trauma center in Iran, from March 28, 2019, to January 31, 2023, identified within the National Trauma Registry of Iran (NTRI). Logistic regression analysis was used to assess the association between demographic and clinical variables with the need for MV.

Results: A total of 2708 trauma patients with a mean age of 41.79 ± 21.84 (range:1-98) years were included (73.4% male). A total of 251 (9.3%) patients were admitted to the Intensive Care Unit (ICU); 113 (4.2%) experienced MV. The significant associated factors of need for MV based on the univariable analysis were age ≥ 65 years (p <0.001); penetrating trauma (p < 0.001) and falling (p = 0.01); private mode of transportation to ED (p < 0.001); site of injury (p < 0.001); heart rate ≥ 100/ minutes (p = 0.04); O2 saturation < 90 % on room air (p < 0.01); Glasgow Coma Scale (GCS) < 13 (p< 0.001); and injury Severity Score (ISS) ≥ 9 (p< 0.001). Based on the multivariate logistic regression analysis, the independent predictors of the need for MV in trauma patients were the site of injury (p < 0.001), GCS < 13 (p < 0.001), and ISS ≥ 9 (p < 0.001).

Conclusion: Based on the findings, ISS ≥ 9, GCS < 13, and site of injury were among the independent predictors of the need for MV following trauma.

Introduction: Despite the evident impact of ultrasonography on diagnosis in acute care settings, there is still a great deal of uncertainty regarding its accuracy. This study aimed to assess the diagnostic performance of lung ultrasonography (LUS) for the identification of acute heart failure in patients with suggestive manifestations.

Methods: Medline, Scopus, and Web of Science were comprehensively searched from their inception to November 2024 to identify original studies investigating accuracy of LUS for diagnosis of heart failure. Data extraction and quality assessment were performed by two independent reviewers. The statistical analysis for pooling the results of diagnostic performance parameters was conducted using Stata and Meta-DiSc softwares.

Results: Thirty-eight included studies in this meta-analysis were published between 2006 and 2024, encompassing a total of 6,783 patients. There was significant heterogeneity between included studies with respect to sensitivity (I²=92.51 and P<0.01) and specificity (I²=93.79 and P<0.01). The pooled sensitivity, specificity, and accuracy of LUS for detection of heart failure were 0.92 (95% CI, 0.87-0.95), 0.90 (95% CI, 0.86-0.93), and 0.96 (95% CI, 0.94-0.98), respectively. In addition, pooled positive likelihood ratio (PLR), negative likelihood ratio (NLR), and diagnostic odds ratio (DOR) were 7.87 (95% CI, 5.60-11.07), 0.14 (95% CI, 0.10-0.19), and 70.74 (95% CI, 41.98-119.21), respectively.

Conclusion: Our meta-analysis demonstrates that LUS is a highly practical imaging for diagnosing acute heart failure, with excellent sensitivity, specificity, and accuracy. It is particularly valuable for excluding the heart failure when the result is negative. However, the influence of outlier and influential studies warrants caution, and future studies should aim to further validate these findings in diverse clinical contexts.

Introduction: Hypokalemia, hyperkalemia, and acidosis are among the reversible causes of out-of-hospital cardiac arrest (OHCA) that can be promptly identified using point-of-care testing (POCT) for blood gas and electrolyte analysis. This study aimed to evaluate the efficacy of POCT in the prehospital setting for OHCA management.

Methods: In this cross-sectional study the management and outcomes of OHCA patients were compared before and after implementing the POCT for blood gas and electrolyte analysis by EMS in the prehospital setting of Ramathibodi Hospital, Thailand.

Results: 217 OHCA patients with a mean age of 61 ± 17.07 (range: 58.72-63.28) years were studied (64.06 % male). 148 (68.2%) patients received POCT in the prehospital setting. Patients in the POCT group received higher administration of sodium bicarbonate (p < 0.001) and calcium gluconate (p < 0.001) compared to those without POCT. Sustained ROSC was achieved in 25% of the POCT group, compared to 11.59% in the no POCT group (p = 0.030). POCT blood gas analysis was identified as an independent predictor of sustained ROSC based on multivariable analysis (adjusted Odds: 4.60, 95% CI: 1.35-15.69; p = 0.015).

Conclusions: It seems that POCT for blood gas and electrolyte analysis in the prehospital setting could improve sustained ROSC in OHCA patients by enabling rapid and targeted management of cardiac arrest's reversible causes.

Cardiopulmonary resuscitation (CPR) remains controversial with dismal outcomes for cardiac arrest (CA) victims. Inadequate organ perfusion and frequent CPR-related trauma most likely occur due to inappropriate adaptation to hemostatic conditions, electrophysiology, cardiotorsal anatomy, and thoracic biomechanics. Alternatively, we propose a new technique compromising chest compressions through the 5^th intercostal space while placing the victim in the left lateral decubitus position with wrapped abdomen and raised legs, allowing to: bypass the sternal barrier, refill the heart, and then recoil-rebound the chest (3R /CPR), within the axis of the cylindrical ribcage. Our goal is to evaluate the technique following its necessary application on two drowning victims. It seems that, 3R/CPR adapts the pathophysiological conditions of CA victims promoting a less traumatic return of spontaneous circulation (ROSC), making it worthy of further investigation and study.