Validation of the Taiwan triage and acuity scale: a new computerised five-level triage system

Prediction of high-risk emergency department revisits from a machine-learning algorithm: a proof-of-concept study

BACKGROUND

High-risk emergency department (ED) revisit is considered an important quality indicator that may reflect an increase in complications and medical burden. However, because of its multidimensional and highly complex nature, this factor has not been comprehensively investigated. This study aimed to predict high-risk ED revisit with a machine-learning (ML) approach.

METHODS

This 3-year retrospective cohort study assessed adult patients between January 2019 and December 2021 from National Taiwan University Hospital Hsin-Chu Branch with high-risk ED revisit, defined as hospital or intensive care unit admission after ED return within 72 hours. A total of 150 features were preliminarily screened, and 79 were used in the prediction model. Deep learning, random forest, extreme gradient boosting (XGBoost) and stacked ensemble algorithm were used. The stacked ensemble model combined multiple ML models and performed model stacking as a meta-level algorithm. Confusion matrix, accuracy, sensitivity, specificity and area under the receiver operating characteristic curve (AUROC) were used to evaluate performance.

RESULTS

Analysis was performed for 6282 eligible adult patients: 5025 (80.0%) in the training set and 1257 (20.0%) in the testing set. High-risk ED revisit occurred for 971 (19.3%) of training set patients vs 252 (20.1%) in the testing set. Leading predictors of high-risk ED revisit were age, systolic blood pressure and heart rate. The stacked ensemble model showed more favourable prediction performance (AUROC 0.82) than the other models: deep learning (0.69), random forest (0.78) and XGBoost (0.79). Also, the stacked ensemble model achieved favourable accuracy and specificity.

CONCLUSION

The stacked ensemble algorithm exhibited better prediction performance in which the predictions were generated from different ML algorithms to optimally maximise the final set of results. Patients with older age and abnormal systolic blood pressure and heart rate at the index ED visit were vulnerable to high-risk ED revisit. Further studies should be conducted to externally validate the model.

Use of Reverse Shock Index Multiplied by Simplified Motor Score in a Five-Level Triage System: Identifying Trauma in Adult Patients at a High Risk of Mortality

Background and Objectives: The Taiwan Triage and Acuity Scale (TTAS) is reliable for triaging patients in emergency departments in Taiwan; however, most triage decisions are still based on chief complaints. The reverse-shock index (SI) multiplied by the simplified motor score (rSI-sMS) is a more comprehensive approach to triage that combines the SI and a modified consciousness assessment. We investigated the combination of the TTAS and rSI-sMS for triage compared with either parameter alone as well as the SI and modified SI. Materials and Methods: We analyzed 13,144 patients with trauma from the Taipei Tzu Chi Trauma Database. We investigated the prioritization performance of the TTAS, rSI-sMS, and their combination. A subgroup analysis was performed to evaluate the trends in all clinical outcomes for different rSI-sMS values. The sensitivity and specificity of rSI-sMS were investigated at a cutoff value of 4 (based on previous study and the highest score of the Youden Index) in predicting injury severity clinical outcomes under the TTAS system were also investigated. Results: Compared with patients in triage level III, those in triage levels I and II had higher odds ratios for major injury (as indicated by revised trauma score < 7 and injury severity score [ISS] ≥ 16), intensive care unit (ICU) admission, prolonged ICU stay (≥14 days), prolonged hospital stay (≥30 days), and mortality. In all three triage levels, the rSI-sMS < 4 group had severe injury and worse outcomes than the rSI-sMS ≥ 4 group. The TTAS and rSI-sMS had higher area under the receiver operating characteristic curves (AUROCs) for mortality, ICU admission, prolonged ICU stay, and prolonged hospital stay than the SI and modified SI. The combination of the TTAS and rSI-sMS had the highest AUROC for all clinical outcomes. The prediction performance of rSI-sMS < 4 for major injury (ISS ≥ 16) exhibited 81.49% specificity in triage levels I and II and 87.6% specificity in triage level III. The specificity for mortality was 79.2% in triage levels I and II and 87.4% in triage level III. Conclusions: The combination of rSI-sMS and the TTAS yielded superior prioritization performance to TTAS alone. The integration of rSI-sMS and TTAS effectively enhances the efficiency and accuracy of identifying trauma patients at a high risk of mortality.

Interpretable Deep Learning System for Identifying Critical Patients Through the Prediction of Triage Level, Hospitalization, and Length of Stay: Prospective Study

BACKGROUND

Triage is the process of accurately assessing patients' symptoms and providing them with proper clinical treatment in the emergency department (ED). While many countries have developed their triage process to stratify patients' clinical severity and thus distribute medical resources, there are still some limitations of the current triage process. Since the triage level is mainly identified by experienced nurses based on a mix of subjective and objective criteria, mis-triage often occurs in the ED. It can not only cause adverse effects on patients, but also impose an undue burden on the health care delivery system.

OBJECTIVE

Our study aimed to design a prediction system based on triage information, including demographics, vital signs, and chief complaints. The proposed system can not only handle heterogeneous data, including tabular data and free-text data, but also provide interpretability for better acceptance by the ED staff in the hospital.

METHODS

In this study, we proposed a system comprising 3 subsystems, with each of them handling a single task, including triage level prediction, hospitalization prediction, and length of stay prediction. We used a large amount of retrospective data to pretrain the model, and then, we fine-tuned the model on a prospective data set with a golden label. The proposed deep learning framework was built with TabNet and MacBERT (Chinese version of bidirectional encoder representations from transformers [BERT]).

RESULTS

The performance of our proposed model was evaluated on data collected from the National Taiwan University Hospital (901 patients were included). The model achieved promising results on the collected data set, with accuracy values of 63%, 82%, and 71% for triage level prediction, hospitalization prediction, and length of stay prediction, respectively.

CONCLUSIONS

Our system improved the prediction of 3 different medical outcomes when compared with other machine learning methods. With the pretrained vital sign encoder and repretrained mask language modeling MacBERT encoder, our multimodality model can provide a deeper insight into the characteristics of electronic health records. Additionally, by providing interpretability, we believe that the proposed system can assist nursing staff and physicians in taking appropriate medical decisions.

A novel scale for triage assessment of frailty in the emergency department (ED-FraS): a prospective videotaped study

BACKGROUND

Rapid recognition of frailty in older patients in the ED is an important first step toward better geriatric care in the ED. We aimed to develop and validate a novel frailty assessment scale at ED triage, the Emergency Department Frailty Scale (ED-FraS).

METHODS

We conducted a prospective cohort study enrolling adult patients aged 65 years or older who visited the ED at an academic medical center. The entire triage process was recorded, and triage data were collected, including the Taiwan Triage and Acuity Scale (TTAS). Five physician raters provided ED-FraS levels after reviewing videos. A modified TTAS (mTTAS) incorporating ED-FraS was also created. The primary outcome was hospital admission following the ED visit, and secondary outcomes included the ED length of stay (EDLOS) and total ED visit charges.

RESULTS

A total of 256 patients were included. Twenty-seven percent of the patients were frail according to the ED-FraS. The majority of ED-FraS was level 2 (57%), while the majority of TTAS was level 3 (81%). There was a weak agreement between the ED-FraS and TTAS (kappa coefficient of 0.02). The hospital admission rate and charge were highest at ED-FraS level 5 (severely frail), whereas the EDLOS was longest at level 4 (moderately frail). The area under the Receiver Operating Characteristic curve (AUROC) in predicting hospital admission for the TTAS, ED-FraS, and mTTAS were 0.57, 0.62, and 0.63, respectively. The ED-FraS explained more variation in EDLOS (R2 = 0.096) compared with the other two methods.

CONCLUSIONS

The ED-Fras tool is a simple and valid screening tool for identifying frail older adults in the ED. It also can complement and enhance ED triage systems. Further research is needed to test its real-time use at ED triage internationally.

Correlates of the Veterans Visiting Emergency Departments in Taiwan: A Comparison Before and After the Coronavirus Disease 2019 Pandemic

BACKGROUND

Taiwan has a substantial number of veterans, but knowledge regarding their emergency department (ED) visits during the coronavirus disease 2019 (COVID-19) pandemic remains limited. This study examined the characteristics of veterans' ED visits during Taiwan's COVID-19 epidemic.

METHODS

This was a cross-sectional study conducted at the ED of a large veteran medical center located in Taipei, Taiwan, from May 2018 to October 2021. We analyzed the numbers and features of visits in summer and autumn according to the first wave of the COVID-19 epidemic in Taiwan in 2021.

RESULTS

Medical institutions were positively associated with veteran status. Emergency department complaints of trauma (adjusted odds ratio [AOR] = 1.15, 95% CI: 1.06-1.25; summer P < .01) and chest pain/tightness (AOR = 1.65, 95% CI: 1.45-1.87; summer P < .01; AOR = 1.4, 95% CI: 1.26-1.55; P < .01) were associated with increased odds of being a veteran. Triage levels above 2 were positively associated with veteran status in the autumn model (AOR = 1.14, 95% CI: 1.07-1.22; P < .01). Patients hospitalized after ED visits were associated with reduced odds of veteran status (P < .01). Those who spent a long time in the ED were more likely to be veterans than those who spent a shorter time in the ED (P < .01). Veterans were less likely to visit the ED regardless of the time frame of the study period (P < .01), except during the COVID-19 outbreak in the autumn (2019-2020).

CONCLUSIONS

The distinctions in ED visits highlighted the individuality of veterans' medical needs. Our findings suggest that the veteran medical system can add to the focus on improving senior-friendly care, fall prevention, quality of life of institutionalized veterans, access for homeless veterans, and care for ambulatory care-sensitive conditions.

The risk factors of the 72-h unscheduled return visit admission to emergency department in adults below 50 years old

BACKGROUND

An unscheduled return visit (URV) to the emergency department (ED) within 72-h is an indicator of ED performance. An unscheduled return revisit (URV) within 72-h was used to monitor adverse events and medical errors in a hospital quality improvement program. The study explores the potential factors that contribute to URV to the ED within 72-h and the unscheduled return revisit admission (URVA) in adults below 50 years old.

METHODS

The case-control study enrolled 9483 URV patients during 2015-2020 in National Cheng-Kung University Hospital. URVA and URV non-admission (URVNA) patients were analyzed. The Gini impurity index was calculated by decision tree (DT) to split the variables capable of partitioning the groups into URVA and URVNA. Logistic regression is applied to calculate the odds ratio (OR) of candidate variables. The α level was set at 0.05.

RESULTS

Among patients under the age of 50, the percentage of females in URVNA was 55.05%, while in URVA it was 53.25%. Furthermore, the average age of URVA patients was 38.20 ± 8.10, which is higher than the average age of 35.19 ± 8.65 observed in URVNA. The Charlson Comorbidity Index (CCI) of the URVA patients (1.59 ± 1.00) was significantly higher than that of the URVNA patients (1.22 ± 0.64). The diastolic blood pressure (DBP) of the URVA patients was 85.29 ± 16.22, which was lower than that of the URVNA (82.89 ± 17.29). Severe triage of URVA patients is 21.1%, which is higher than the 9.7% of URVNA patients. The decision tree suggests that the factors associated with URVA are "severe triage," "CCI higher than 2," "DBP less than 86.5 mmHg," and "age older than 34 years". These risk factors were verified by logistic regression and the OR of CCI was 2.42 (1.50-3.90), the OR of age was 1.84 (1.50-2.27), the OR of DBP less than 86.5 was 0.71 (0.58-0.86), and the OR of severe triage was 2.35 (1.83-3.03).

CONCLUSIONS

The results provide physicians with a reference for discharging patients and could help ED physicians reduce the cognitive burden associated with the diagnostic errors and stress.

Predictors of 72-h unscheduled return visits with admission in patients presenting to the emergency department with abdominal pain

BACKGROUND

Unscheduled return visits (URVs) to the emergency department (ED) constitute a crucial indicator of patient care quality.

OBJECTIVE

We aimed to analyze the clinical characteristics of patients who visited the ED with abdominal pain and to identify the risk of URVs with admission (URVAs) from URVs without admission (URVNAs).

METHODS

This retrospective study included adult patients who visited the ED of Taipei Medical University Hospital because of abdominal pain and revisited in 72 h over a 5-year period (January 1, 2014, to December 31, 2018). Multivariable logistic regression analysis was employed to identify risk factors for URVAs and receiver operating characteristic (ROC) curve analysis was performed to determine the efficacy of variables predicting URVAs and the optimal cut-off points for the variables. In addition, a classification and regression tree (CART)-based scoring system was used for predicting risk of URVA.

RESULTS

Of 702 eligible patients with URVs related to abdominal pain, 249 had URVAs (35.5%). In multivariable analysis, risk factors for URVAs during the index visit included execution of laboratory tests (yes vs no: adjusted odds ratio [AOR], 4.32; 95% CI 2.99-6.23), older age (≥ 40 vs < 40 years: AOR, 2.10; 95% CI 1.10-1.34), Level 1-2 triage scores (Levels 1-2 vs Levels 3-5: AOR, 2.30; 95% CI 1.26-4.19), and use of ≥ 2 analgesics (≥ 2 vs < 2: AOR, 2.90; 95% CI 1.58-5.30). ROC curve analysis results revealed the combination of these 4 above variables resulted in acceptable performance (area under curve: 0.716). The above 4 variables were used in the CART model to evaluate URVA propensity.

CONCLUSIONS

Elder patients with abdominal pain who needed laboratory workup, had Level 1-2 triage scores, and received ≥ 2 doses of analgesics during their index visits to the ED had higher risk of URVAs.

Machine Learning Approach for the Prediction of In-Hospital Mortality in Traumatic Brain Injury Using Bio-Clinical Markers at Presentation to the Emergency Department

BACKGROUND

Accurate prediction of in-hospital mortality is essential for better management of patients with traumatic brain injury (TBI). Machine learning (ML) algorithms have been shown to be effective in predicting clinical outcomes. This study aimed to identify predictors of in-hospital mortality in TBI patients using ML algorithms.

MATERIALS AND METHOD

A retrospective study was performed using data from both the trauma registry and electronic medical records among TBI patients admitted to the Hamad Trauma Center in Qatar between June 2016 and May 2021. Thirteen features were selected for four ML models including a Support Vector Machine (SVM), Logistic Regression (LR), Random Forest (RF), and Extreme Gradient Boosting (XgBoost), to predict the in-hospital mortality.

RESULTS

A dataset of 922 patients was analyzed, of which 78% survived and 22% died. The AUC scores for SVM, LR, XgBoost, and RF models were 0.86, 0.84, 0.85, and 0.86, respectively. XgBoost and RF had good AUC scores but exhibited significant differences in log loss between the training and testing sets (% difference in logloss of 79.5 and 41.8, respectively), indicating overfitting compared to the other models. The feature importance trend across all models indicates that aPTT, INR, ISS, prothrombin time, and lactic acid are the most important features in prediction. Magnesium also displayed significant importance in the prediction of mortality among serum electrolytes.

CONCLUSIONS

SVM was found to be the best-performing ML model in predicting the mortality of TBI patients. It had the highest AUC score and did not show overfitting, making it a more reliable model compared to LR, XgBoost, and RF.

A simplified scoring model for predicting bacteremia in the unscheduled emergency department revisits: The SADFUL score

BACKGROUND

Bacteremia is a severe complication of infectious disease. Patients with a high bacteremia risk in the emergency department (ED) but misidentified would lead to the unscheduled revisits. This study aimed to develop a simplified scoring model to predict bacteremia in patients with unscheduled ED revisits.

METHODS

Adult patients with unscheduled ED revisits within 72 h with a final diagnosis of infectious disease were retrospectively included. The development cohort included patients visiting the ED from January 1, 2019 to December 31, 2021. Internal validation was performed in patients visiting the ED from January 1, 2022 to March 31, 2022. Variables including demographics, pre-comorbidities, triage levels, vital signs, chief complaints, and laboratory data in the index visit were analyzed. Bacteremia was the primary outcome determined by blood culture in either index visits or revisits.

RESULTS

The SADFUL score for predicting bacteremia comprised the following predictors: "S"egmented neutrophil percentage (+3 points), "A"ge > 55 years (+1 point), "D"iabetes mellitus (+1 point), "F"ever (+2 points), "U"pper respiratory tract symptoms (-2 points), and "L"eukopenia (2 points). The area under receiver operating characteristic curve with 95% confidence interval in the development (1802 patients, 190 [11%] with bacteremia) and the validation cohort (134 patients, 17 [13%] with bacteremia) were 0.78 (0.74-0.81) and 0.79 (0.71-0.88), respectively.

CONCLUSIONS

The SADFUL score is a simplified useful tool for predicting bacteremia in patients with unscheduled ED revisits. The scoring model could help ED physicians decrease misidentification of patients at a high risk for bacteremia and potential complications.

An AI-Enabled Dynamic Risk Stratification for Emergency Department Patients with ECG and CXR Integration

Emergency department (ED) triage scale determines the priority of patient care and foretells the prognosis. However, the information retrieved from the initial assessment is limited, hindering the risk identification accuracy of triage. Therefore, we sought to develop a 'dynamic' triage system as secondary screening, using artificial intelligence (AI) techniques to integrate information from initial assessment data and subsequent examinations. This retrospective cohort study included 134,112 ED visits with at least one electrocardiography (ECG) and chest X-ray (CXR) in a medical center from 2012 to 2022. Additionally, an independent community hospital provided 45,614 ED visits as an external validation set. We trained an eXtreme gradient boosting (XGB) model using initial assessment data to predict all-cause mortality in 7 days. Two deep learning models (DLMs) using ECG and CXR were trained to stratify mortality risks. The dynamic triage levels were based on output from the XGB-triage and DLMs from ECG and CXR. During the internal and external validation, the area under the receiver operating characteristic curve (AUC) of the XGB-triage model was >0.866; furthermore, the AUCs of DLMs using ECG and CXR were >0.862 and >0.886, respectively. The dynamic triage scale provided a higher C-index (0.914-0.920 vs. 0.827-0.843) than the original one and demonstrated better predictive ability for 5-year mortality, 30-day ED revisit, and 30-day discharge. The AI-based risk scale provides a more accurate and dynamic stratification of mortality risk in ED patients, particularly in identifying patients who tend to be overlooked due to atypical symptoms.

Impact of the COVID-19 pandemic on medical-seeking behavior in older adults by comparing the presenting complaints of the emergency department visits

BACKGROUND

The outbreak of the coronavirus disease 2019 (COVID-19) has caused a catastrophic event worldwide. Since then, people's way of living has changed in terms of personal behavior, social interaction, and medical-seeking behavior, including change of the emergency department (ED) visiting patterns. The objective of this study was to analyze the impact of the COVID-19 pandemic on the ED visiting patterns of the older people to explore its variable expression with the intention of ameliorating an effective and suitable response to public health emergencies.

METHODS

This was a retrospective study conducted in three hospitals of the Cathay Health System in Taiwan. Patients aged ≥ 65 years who presented to the ED between January 21, 2020, and April 30, 2020 (pandemic stage), and between January 21, 2019, and April 30, 2019 (pre-pandemic stage) were enrolled in the study. Basic demographics, including visit characteristics, disposition, and chief complaints of the patients visiting the ED between these two periods of time, were compared and analyzed.

RESULTS

A total of 16,655 older people were included in this study. A 20.91% reduction in ED older adult patient visits was noted during the pandemic period. During the pandemic, there was a decrease in ambulance use among elderly patients visiting the ED, with the proportion decreasing from 16.90 to 16.58%. Chief complaints of fever, upper respiratory infections, psychological and social problems increased, with incidence risk ratios (IRRs) of 1.12, 1.23, 1.25, and 5.2, respectively. Meanwhile, the incidence of both non-life-threatening and life-threatening complaints decreased, with IRRs of 0.72 and 0.83, respectively.

CONCLUSION

Health education regarding life-threatening symptom signs among older adult patients and avocation of the proper timing to seek medical attention via ambulance were crucial issues during the pandemic.

Reliability and validity of Chengdu pediatric emergency triage criteria: case study of a single center in China

BACKGROUND

We aimed to examine the reliability and validity of Chengdu pediatric emergency triage criteria in order to provide a reference for the development of pediatric emergency triage within other hospitals.

METHODS

We developed Chengdu pediatric emergency triage criteria based on the conditions/symptom, vital signs, and the Pediatric Early Warning Score system within our hospital using the Delphi method in 2020. The simulation scenario triage and real-life triage which were conducted in our hospital during January - March 2021, and the retrospective study of triage records extracted from our hospital's health information system in February 2022, were used to measure the agreement in triage decisions between the triage nurses, and between the triage nurses and the expert team.

RESULTS

For the 20 simulation cases, the Kappa value of triage decisions between the triage nurses was 0.6 (95% CI 0.352-0.849), and the Kappa value of triage decisions between the triage nurses and the expert team was 0.73 (95% CI 0.540-0.911). For the 252 cases in the real-life triage, the Kappa value of triage decisions between the triage nurses and the expert team was 0.824 (95% CI 0.680-0.962). For the 20,540 cases selected for the retrospective study of triage records, the Kappa value of triage decisions between the triage nurses was 0.702 (95% CI 0.691-0.713); that between Triage Nurse 1 and the expert team was 0.634 (95% CI 0.623-0.647); and that between Triage Nurse 2 and the expert team was 0.725 (95% CI 0.713-0.736). The overall agreement rate in triage decisions between the triage nurses and the expert team in the simulation scenario triage was 80%; that between the triage nurses and the expert team in the real-life triage was 97.6%; and that between the triage nurses in the retrospective study was 91.9%. In the retrospective study, the agreement rates in triage decisions between Triage Nurse 1 and the expert team, and between Triage Nurse 2 and the expert team, were 88.0% and 92.3%, respectively.

CONCLUSION

Chengdu pediatric emergency triage criteria that developed within our hospital is reliable and valid, and can promote rapid and effective triage by triage nurses.

Integrating the clinical frailty scale with emergency department triage systems for elder patients: A prospective study

BACKGROUND

This prospective study investigated whether integrating the Clinical Frailty Scale (CFS) with a triage system would improve triage for older adult emergency department (ED) patients.

METHODS

We enrolled ED patients aged 65 years or older at 5 study sites in Taiwan between December 2020 and April 2021. All eligible patients were assigned a triage level by using the Taiwan Triage and Acuity Scale (TTAS) in accordance with usual practice. A CFS score was collected from them. The primary outcome was critical events, defined as ICU admission or in-hospital mortality. The secondary outcomes were ED medical expenditures, number of orders in the ED, and length of hospital stay (LOS). We applied a reclassification concept and integrated the CFS and TTAS to create the Triage Frailty Acuity Scale (TFAS). We compared the outcomes achieved between the TTAS and TFAS.

RESULTS

Of 1023 screened ED patients, 890 were enrolled. The majority were assigned to TTAS level 3 (73.26%) and had CFS scores of 4 to 9 (55.96%). The primary outcomes were better predicted by the TFAS than the TTAS (area under the curve [AUC] 0.82 vs. 064). Using multivariable approach, TTAS level 1 (odds ratio [OR], 4.8; 95% confidence interval [CI], 1.7-13.4) and CFS score (OR, 5.8; 95% CI, 1.9-17.2) were significantly associated with the primary outcomes. For older adults at the highest triage level, the TFAS was not associated with an increase in the primary outcomes compared with the TTAS; however, the TFAS was associated with a significant decrease in the number of older ED patients assigned to triage levels 3 to 5. In addition, TFAS had a longer average LOS but did not have a higher average number of orders or ED medical expenditures compared to TTAS.

CONCLUSIONS

The TFAS identified more older ED patients who had been triaged as less emergent but proceeded to need ICU admission or in-hospital death. Incorporating the CFS into triage may reduce the under-triage of older adults in the ED.

Validity of the Korean triage and acuity scale in older patients compared to the adult group

INTRODUCTION

While many patients visit the emergency department (ED) for various reasons, medical resources are limited. Therefore, various triage scale systems have been used to predict patient urgency and severity. South Korea has developed and used the Korean Triage and Accuracy Scale (KTAS) based on the Canadian classification tool. As the elderly population increases, the number of elderly patients visiting the ED also increases. However, in KTAS, there is no consideration for the elderly, and the same classification system as adults. The aim of this study is to verify the ability of KTAS to predict severity levels in the elderly group, compared to the adult group.

METHODS

This is a retrospective study for patients who visited the ED at two centers between February 1, 2018 and January 31, 2021. The initial KTAS level, changed level at ED discharge, general patient character, ED treatment results, in-hospital mortality, and lengths of hospital and ED stays were acquired. Area under the receiver operating characteristics (AUROC) was used to verify the severity prediction ability of the elderly group to KTAS, and logistic regression analysis was used for the prediction up-triage of KTAS.

RESULTS

The enrolled patients in the study were 87,220 in the adult group and 37,627 in the elderly group. The proportion of KTAS up-triage was higher in the elderly group (1.9 % vs. 1.2 %, p < 0.001). The AUROC for the overall admission rate was 0.686, 0.667 in the adult and elderly group, the AUROC for ICU admission was 0.842, 0.767, and the AUROC for in-hospital mortality prediction was 0.809, 0.711, indicating a decrease in the AUROC value in the elderly group. The independent factors of the up-triage predictors were old age, male gender, pulse, and ED length of stay, and old age was the most influential variable.

CONCLUSION

KTAS was poorly associated with severity in the elderly than in adults, and it was found that up-triaging was more likely to occur in the elderly. The severity and urgency of patients over 65 years of age should not be underestimated when initially determining the triage scale.

Taiwan's National Health Insurance at the Emergency Department following the COVID-19 outbreak

Taiwan's National Health Insurance (NHI) is a widely acclaimed universal healthcare system. In the past few years, particularly following the COVID-19 outbreak, challenges related to maintaining the NHI system have surfaced. Since 2020, NHI has faced a series of challenges, including excessive patient visits to the hospital emergency department, a lack of an effective primary care and referral system, and a high turnover rate of healthcare workers. We review major problems related to Taiwan's NHI, emphasizing input from frontline healthcare workers. We provide recommendations for potential policies addressing the concerns around NHI, for example, strengthening the role of primary care services under the NHI administration, reducing the high turnover rate of healthcare workers, and increase the premium and copayments. We hope that this policy analysis may allow policymakers and scholars to understand both the merits and critical problems related to NHI from the clinical perspective.

Machine learning to predict in-hospital cardiac arrest from patients presenting to the emergency department

In-hospital cardiac arrest (IHCA) in the emergency department (ED) is not uncommon but often fatal. Using the machine learning (ML) approach, we sought to predict ED-based IHCA (EDCA) in patients presenting to the ED based on triage data. We retrieved 733,398 ED records from a tertiary teaching hospital over a 7 year period (Jan. 1, 2009-Dec. 31, 2015). We included only adult patients (≥ 18 y) and excluded cases presenting as out-of-hospital cardiac arrest. Primary outcome (EDCA) was identified via a resuscitation code. Patient demographics, triage data, and structured chief complaints (CCs), were extracted. Stratified split was used to divide the dataset into the training and testing cohort at a 3-to-1 ratio. Three supervised ML models were trained and performances were evaluated and compared to the National Early Warning Score 2 (NEWS2) and logistic regression (LR) model by the area under the receiver operating characteristic curve (AUC). We included 316,465 adult ED records for analysis. Of them, 636 (0.2%) developed EDCA. Of the constructed ML models, Random Forest outperformed the others with the best AUC result (0.931, 95% CI 0.911-0.949), followed by Gradient Boosting (0.930, 95% CI 0.909-0.948) and Extra Trees classifier (0.915, 95% CI 0.892-0.936). Although the differences between each of ML models and LR (AUC: 0.905, 95% CI 0.882-0.926) were not significant, all constructed ML models performed significantly better than using the NEWS2 scoring system (AUC 0.678, 95% CI 0.635-0.722). Our ML models showed excellent discriminatory performance to identify EDCA based only on the triage information. This ML approach has the potential to reduce unexpected resuscitation events if successfully implemented in the ED information system.

Impact of interhospital transfer on emergency department timeliness of care and in-hospital outcomes of adult non-trauma patients

Background

Patients who present to the emergency department (ED) from interhospital transfer (IHT) and non-IHT are known to have differences in various clinical outcomes including mortality. The ED timeliness of care is an effective indicator of the quality of ED care and operational efficiency. The impact of IHT on ED timeliness of care remains unclear. We evaluated the association between IHT and ED timeliness of care or in-hospital outcomes in adult non-trauma patients.

Methods

Data of consecutive hospital admission of adult non-trauma patients who visited the ED of a medical center from January 2018 to Jun 2020 were retrospectively analyzed. The patients were divided into IHT and non-IHT cohorts. Various data were recorded. The ED length of stay (LOS) was measured as the outcome of ED timeliness of care, while hospital LOS and in-hospital death were measured as the in-hospital outcomes. Multiple regression analyses were performed using unmatched and propensity-matched cohorts. In the later analyses, both groups were propensity matched for sex, age, and other covariates that showed significant differences between two groups to achieve a 1:4 balanced cohort.

Results

Data on 1856 IHT patients and 16295 non-IHT patients were analyzed. IHT was associated with a shorter ED LOS, longer hospital LOS, and higher odds of in-hospital death compared with non-IHT in unmatched and propensity-matched analyses. The shorter ED LOS was due to the slightly longer interval of arrival to ED physicians (∼1 min) and considerably shorter intervals of ED physicians to decision (∼120 min) and decision to departure (∼105 min). Risk stratification revealed that IHT was associated with a shorter ED LOS in patients with all levels (1-5) of Taiwan Triage and Acuity Scale (TTAS) and associated with longer hospital LOS and higher odds of in-hospital death in patients with TTAS level ≥3.

Conclusions

IHT was associated with a shorter ED LOS, longer hospital LOS, and higher odds of in-hospital death in adult non-trauma patients compared with non-IHT. The expedited ED timeliness of care in the IHT cohort was due to considerably shorter intervals of both ED physicians to decision and decision to disposition.

Prediction of Prognosis in Patients with Trauma by Using Machine Learning

Background and Objectives: We developed a machine learning algorithm to analyze trauma-related data and predict the mortality and chronic care needs of patients with trauma. Materials and Methods: We recruited admitted patients with trauma during 2015 and 2016 and collected their clinical data. Then, we subjected this database to different machine learning techniques and chose the one with the highest accuracy by using cross-validation. The primary endpoint was mortality, and the secondary endpoint was requirement for chronic care. Results: Data of 5871 patients were collected. We then used the eXtreme Gradient Boosting (xGBT) machine learning model to create two algorithms: a complete model and a short-term model. The complete model exhibited an 86% recall for recovery, 30% for chronic care, 67% for mortality, and 80% for complications; the short-term model fitted for ED displayed an 89% recall for recovery, 25% for chronic care, and 41% for mortality. Conclusions: We developed a machine learning algorithm that displayed good recall for the healthy recovery group but unsatisfactory results for those requiring chronic care or having a risk of mortality. The prediction power of this algorithm may be improved by implementing features such as age group classification, severity selection, and score calibration of trauma-related variables.

Physician gestalt for emergency department triage: A prospective videotaped study

OBJECTIVE

Appropriate triage in patients presenting to the emergency department (ED) is often challenging. Little is known about the role of physician gestalt in ED triage. We aimed to compare the accuracy of emergency physician gestalt against the currently used computerized triage process.

METHODS

We conducted a prospective observational study in the ED at an academic medical center. Adult patients aged ≥20 years were included and underwent a standard triage protocol. The patients underwent system-based triage using the computerized software the Taiwan Triage and Acuity Scale. The entire triage process was recorded, and triage data were collected. Five physician raters provided triage levels (physician-based) according to their perceived urgency after reviewing videos. The primary outcome was hospital admission. The secondary outcomes were ED length of stay (EDLOS) and charges.

RESULTS

In total, 656 patients were recruited (mean age 52 years, 50% male). The median system-based triage level was 3. By contrast, the median physician-based triage level was 4. The physician raters tended to provide lower triage levels than the system, with an average difference of 1. There was modest concordance between the two triage methods (correlation coefficient 0.30), with a weighted kappa coefficient of 0.18. The area under the receiver operating curve for the system- and physician-based triage in predicting hospital admission were similar (0.635 vs. 0.631, p = 0.896). Attending physicians appeared to have better performance than residents in predicting admission. The variation explained (R² ) in EDLOS and charges were similar between the two triage methods (R²  = 3% for EDLOS, 7%-9% for charges).

CONCLUSIONS

Emergency physician gestalt for triage showed similar performance to a computerized system; however, physicians redistributed patients to lower triage levels. Physician gestalt has advantages for identifying low-risk patients. This approach may avoid undue time pressure for health care providers and promote rapid discharge.

Pain Assessment in the Emergency Department: A Prospective Videotaped Study

Introduction: Research suggests that pain assessment involves a complex interaction between patients and clinicians. We sought to assess the agreement between pain scores reported by the patients themselves and the clinician’s perception of a patient’s pain in the emergency department (ED). In addition, we attempted to identify patient and physician factors that lead to greater discrepancies in pain assessment. Methods: We conducted a prospective observational study in the ED of a tertiary academic medical center. Using a standard protocol, trained research personnel prospectively enrolled adult patients who presented to the ED. The entire triage process was recorded, and triage data were collected. Pain scores were obtained from patients on a numeric rating scale of 0 to 10. Five physician raters provided their perception of pain ratings after reviewing videos. Results: A total of 279 patients were enrolled. The mean age was 53 years. There were 141 (50.5%) female patients. The median self-reported pain score was 4 (interquartile range 0-6). There was a moderately positive correlation between self-reported pain scores and physician ratings of pain (correlation coefficient, 0.46; P <0.001), with a weighted kappa coefficient of 0.39. Some discrepancies were noted: 102 (37%) patients were rated at a much lower pain score, whereas 52 (19%) patients were given a much higher pain score from physician review. The distributions of chief complaints were different between the two groups. Physician raters tended to provide lower pain scores to younger (P = 0.02) and less ill patients (P = 0.008). Additionally, attending-level physician raters were more likely to provide a higher pain score than resident-level raters (P <0.001). Conclusion: Patients’ self-reported pain scores correlate positively with the pain score provided by physicians, with only a moderate agreement between the two. Under- and over-estimations of pain in ED patients occur in different clinical scenarios. Pain assessment in the ED should consider both patient and physician factors.

Revising Vital Signs Criteria for Accurate Triage of Older Adults in the Emergency Department

Objective

Because of physiologic changes in older adults, their vital signs need to be assessed differently. This study aimed to determine appropriate vital sign cut points for triage designation in older patients presented to the emergency department (ED).

Patients and Methods

Data from 78,524 ED visits of patients aged ≥65 years in Linkou Chang Gung Memorial Hospital (LCGMH) between 2016 and 2017 were collected. New cut points for vital signs (systolic blood pressure [SBP], heart rate [HR], body temperature [BT], and Glasgow Coma Scale [GCS]) were determined using the critical event rate (the composite of admission to ICU and mortality in hospital) for each vital sign. The newly proposed triage scale was then validated using two other databases (Chang Gung Research Database [CGRD] and Taipei City Hospital [TPECH] database). The Taiwan Triage and Acuity Scale (TTAS) was used in this study.

Results

In the LCGMH derivation group, older patients presenting with SBP < 80 mmHg, HR < 40 or > 140 beats per minute (bpm), BT < 35°C, and GCS score 3–8 had a critical event rate of >20% and were proposed to be uptriaged to TTAS level 1. Following a reclassification, a portion of older patients are uptriaged by the newly proposed TTAS, and increase in the critical event rate in TTAS level 1 and level 2 groups compared to the existing TTAS. The newly proposed TTAS exhibited comparable discriminatory ability for triage in older patients compared to the existing TTAS (the area under the receiver operating characteristics curve: CGRD, 0.76 vs 0.62; TPECH, 0.71 vs 0.59).

Conclusion

Revising the vital signs triage criteria for older patients could be a way to improve the identification of patients with critical event outcomes in high TTAS level, thereby improving triage accuracy among older patients visiting the ED.

The impact of COVID-19 on Emergency Department length of stay for urgent and life-threatening patients

OBJECTIVES

To determine the impact of the Coronavirus disease-2019 (COVID-19) pandemic on the length of stay (LOS) and prognosis of patients in the resuscitation area.

METHODS

A retrospective analysis of case data of patients in the resuscitation area during the early stages of the COVID-19 pandemic (January 15, 2020- January 14, 2021) was performed and compared with the pre-COVID-19 period (January 15, 2019 - January 14, 2020) in the First Affiliated Hospital of Soochow University. The patients' information, including age, sex, length of stay, and death, was collected. The Wilcoxon Rank sum test was performed to compare the LOS difference between the two periods. Fisher's Exact test and Chi-Squared test were used to analyze the prognosis of patients. The LOS and prognosis in different departments of the resuscitation area (emergency internal medicine, emergency surgery, emergency neurology, and other departments) were further analyzed.

RESULTS

Of the total 8278 patients, 4159 (50.24%) were enrolled in the COVID-19 pandemic period group, and 4119 (49.76%) were enrolled pre-COVID-19 period group. The length of stay was prolonged significantly in the COVID-19 period compared with the pre-COVID-19 period (13h VS 9.8h, p < 0.001). The LOS in the COVID-19 period was prolonged in both emergency internal medicine (15.3h VS 11.3h, p < 0.001) and emergency surgery (8.7h VS 4.9h, p < 0.001) but not in emergency neurology or other emergency departments. There was no significant difference in mortality between the two cohorts (4.8% VS 5.3%, p = 0.341).

CONCLUSION

The COVID-19 pandemic was associated with a significant increase in the length of resuscitation area stay, which may lead to resuscitation area crowding. The influence of the COVID-19 pandemic on patients of different departments was variable. There was no significant impact on the LOS of emergency neurology. According to different departments of the resuscitation area, the COVID-19 pandemic didn't significantly impact the prognosis of patients.

External validation of a triage tool for predicting cardiac arrest in the emergency department

Early recognition and prevention comprise the first ring of the Chain of Survival for in-hospital cardiac arrest (IHCA). We previously developed and internally validated an emergency department (ED) triage tool, Emergency Department In-hospital Cardiac Arrest Score (EDICAS), for predicting ED-based IHCA. We aimed to externally validate this novel tool in another ED population. This retrospective cohort study used electronic clinical warehouse data from a tertiary medical center with approximately 130,000 ED visits per year. We retrieved data from 268,208 ED visits over a 2-year period. We selected one ED visit per person and excluded out-of-hospital cardiac arrest or children. Patient demographics and computerized triage information were retrieved, and the EDICAS was calculated to predict the ED-based IHCA. A total of 145,557 adult ED patients were included. Of them, 240 (0.16%) developed IHCA. The EDICAS showed excellent discrimination with an area under the receiver operating characteristic (AUROC) of 0.88. The AUROC of the EDICAS outperformed those of other early warning scores (0.80 for Modified Early Warning Score [MEWS] and 0.83 for Rapid Emergency Medicine Score [REMS]) in the same ED population. An EDICAS of 6 or above (i.e., high-risk patients) corresponded to a sensitivity of 33%, a specificity of 97%, and a positive likelihood ratio of 12.2. In conclusion, we externally validated a tool for predicting imminent IHCA in the ED and demonstrated its superior performance over other early warning scores. The real-world impact of the EDICAS warning system with appropriate interventions would require a future prospective study.

Machine learning-based triage to identify low-severity patients with a short discharge length of stay in emergency department

BACKGROUND

Overcrowding in emergency departments (ED) is a critical problem worldwide, and streaming can alleviate crowding to improve patient flows. Among triage scales, patients labeled as "triage level 3" or "urgent" generally comprise the majority, but there is no uniform criterion for classifying low-severity patients in this diverse population. Our aim is to establish a machine learning model for prediction of low-severity patients with short discharge length of stay (DLOS) in ED.

METHODS

This was a retrospective study in the ED of China Medical University Hospital (CMUH) and Asia University Hospital (AUH) in Taiwan. Adult patients (aged over 20 years) with Taiwan Triage Acuity Scale level 3 were enrolled between 2018 and 2019. We used available information during triage to establish a machine learning model that can predict low-severity patients with short DLOS. To achieve this goal, we trained five models-CatBoost, XGBoost, decision tree, random forest, and logistic regression-by using large ED visit data and examined their performance in internal and external validation.

RESULTS

For internal validation in CMUH, 33,986 patients (75.9%) had a short DLOS (shorter than 4 h), and for external validation in AUH, there were 13,269 (82.7%) patients with short DLOS. The best prediction model was CatBoost in internal validation, and area under the receiver operating cha racteristic curve (AUC) was 0.755 (95% confidence interval (CI): 0.743-0.767). Under the same threshold, XGBoost yielded the best performance, with an AUC value of 0.761 (95% CI: 0.742- 0.765) in external validation.

CONCLUSIONS

This is the first study to establish a machine learning model by applying triage information alone for prediction of short DLOS in ED with both internal and external validation. In future work, the models could be developed as an assisting tool in real-time triage to identify low-severity patients as fast track candidates.

Using mobility status as a frailty indicator to improve the accuracy of a computerised five-level triage system among older patients in the emergency department

Background

Owing to societal ageing, the number of older individuals visiting emergency departments (EDs) has increased in recent years. For this patient population, accurate triage systems are required. This retrospective cohort study assessed the accuracy of a computerised five-level triage system, the Taiwan Triage and Acuity System (TTAS), by determining its ability to predict in-hospital mortality in older adult patients and compare it with the corresponding rate in younger adult patients presenting to EDs. The association between frailty, which the current triage system does not consider, was also investigated.

Methods

The medical records of adult patients admitted to a single ED between 2016 and 2017 were reviewed. Data collected included information on demographics, triage level, frailty status, in-hospital mortality, and medical resource utilisation. The patients were divided into four age groups: two older adult groups (older: 65–84 years and very old: ≥85 years) and two younger adult groups (young: 18–39 and middle-aged: 40–64 years).

Results

Our study included 265,219 ED adult patients, of whom 64,104 and 16,009 were in the older and very old groups, respectively. The in-hospital mortality rate at each triage level increased with age. The ability of the TTAS to predict in-hospital mortality decreased with age (area under the receiver operating characteristic curve [AUROC]: young: 0.86; middle-aged, 0.84; and older and very old: 0.79). Frailty was associated with in-hospital mortality (odds ratio, 2.20; 95% confidence interval, 2.03–2.38). Adding mobility status as a frailty indicator to TTAS only slightly improved its ability to predict in-hospital mortality (AUROC: 0.74–0.77) in patients ≥65 years of age.

Conclusions

The ability of the current triage system to predict in-hospital mortality decreases with age. Although frailty as mobility was associated with in-hospital mortality, its addition to the TTAS only slightly improved the accuracy with which in-hospital mortality in older patients presenting to EDs was predicted.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12873-022-00646-0.

A Computer-Assisted System for Early Mortality Risk Prediction in Patients with Traumatic Brain Injury Using Artificial Intelligence Algorithms in Emergency Room Triage

Traumatic brain injury (TBI) remains a critical public health challenge. Although studies have found several prognostic factors for TBI, a useful early predictive tool for mortality has yet to be developed in the triage of the emergency room. This study aimed to use machine learning algorithms of artificial intelligence (AI) to develop predictive models for TBI patients in the emergency room triage. We retrospectively enrolled 18,249 adult TBI patients in the electronic medical records of three hospitals of Chi Mei Medical Group from January 2010 to December 2019, and undertook the 12 potentially predictive feature variables for predicting mortality during hospitalization. Six machine learning algorithms including logistical regression (LR) random forest (RF), support vector machines (SVM), LightGBM, XGBoost, and multilayer perceptron (MLP) were used to build the predictive model. The results showed that all six predictive models had high AUC from 0.851 to 0.925. Among these models, the LR-based model was the best model for mortality risk prediction with the highest AUC of 0.925; thus, we integrated the best model into the existed hospital information system for assisting clinical decision-making. These results revealed that the LR-based model was the best model to predict the mortality risk in patients with TBI in the emergency room. Since the developed prediction system can easily obtain the 12 feature variables during the initial triage, it can provide quick and early mortality prediction to clinicians for guiding deciding further treatment as well as helping explain the patient’s condition to family members.

Interdisciplinary collaboration and computer-assisted home healthcare referral in the emergency department: a retrospective cohort study

Aim

Home healthcare (HHC) provides continuous care for disabled patients. However, HHC referral after the emergency department (ED) discharge remains unclear. Thus, this study aimed its clarification.

Methods

A computer-assisted HHC referral by interdisciplinary collaboration among emergency physicians, case managers, nurse practitioners, geriatricians, and HHC nurses was built in a tertiary medical center in Taiwan. Patients who had HHC referrals after ED discharge between February 1, 2020 and September 31, 2020, were recruited into the study. A non-ED HHC cohort who had HHC referrals after hospitalization from the ED was also identified. Comparison for clinical characteristics and uses of medical resources was performed between ED HHC and non-ED HHC cohorts.

Results

The model was successfully implemented. In total, 34 patients with ED HHC and 40 patients with non-ED HHC were recruited into the study. The female proportion was 61.8% and 67.5%, and the mean age was 81.5 and 83.7 years in ED HHC and non-ED HHC cohorts, respectively. No significant difference was found in sex, age, underlying comorbidities, and ED diagnoses between the two cohorts. The ED HHC cohort had a lower median total medical expenditure within 3 months (34,030.0 vs. 56,624.0 New Taiwan Dollars, p = 0.021) compared with the non-ED HHC cohort. Compared to the non-ED HHC cohort, the ED HHC cohort had a lower ≤ 1 month ED visit, ≤ 6 months ED visit, and ≤ 3 months hospitalization; however, differences were not significant.

Conclusion

An innovative ED HHC model was successfully implemented. Further studies with more patients are warranted to investigate the impact.

Supplementary Information

The online version contains supplementary material available at 10.1007/s40520-022-02109-9.

Development and Validation of Simple Age-Adjusted Objectified Korean Triage and Acuity Scale for Adult Patients Visiting the Emergency Department

PURPOSE

The study aimed to develop an objectified Korean Triage and Acuity Scale (OTAS) that can objectively and quickly classify severity, as well as a simple age-adjusted OTAS (S-OTAS) that reflects age and evaluate its usefulness.

MATERIALS AND METHODS

A retrospective analysis was performed of all adult patients who had visited the emergency department at three teaching hospitals. Sex, systolic blood pressure, diastolic blood pressure, pulse rate, respiratory rate, body temperature, O₂ saturation, and consciousness level were collected from medical records. The OTAS was developed with objective criterion and minimal OTAS level, and S-OTAS was developed by adding the age variable. For usefulness evaluation, the 30-day mortality, the rates of computed tomography scan and emergency procedures were compared between Korean Triage and Acuity Scale (KTAS) and OTAS.

RESULTS

A total of 44402 patients were analyzed. For 30-day mortality, S-OTAS showed a higher area under the curve (AUC) compared to KTAS (0.751 vs. 0.812 for KTAS and S-OTAS, respectively, p<0.001). Regarding the rates of emergency procedures, AUC was significantly higher in S-OTAS, compared to KTAS (0.807 vs. 0.830, for KTAS and S-OTAS, respectively, p=0.013).

CONCLUSION

S-OTAS showed comparative usefulness for adult patients visiting the emergency department as a triage tool compared to KTAS.

Trajectories of Vital Signs and Risk of In-Hospital Cardiac Arrest

Background: Little is known about the trajectories of vital signs prior to in-hospital cardiac arrest (IHCA), which could explain the heterogeneous processes preceding this event. We aimed to identify clinically relevant subphenotypes at high risk of IHCA in the emergency department (ED).

Methods: This retrospective cohort study used electronic clinical warehouse data from a tertiary medical center. We retrieved data from 733,398 ED visits over a 7-year period. We selected one ED visit per person and retrieved patient demographics, triage data, vital signs (systolic blood pressure [SBP], heart rate [HR], body temperature, respiratory rate, oxygen saturation), selected laboratory markers, and IHCA status. Group-based trajectory modeling was performed.

Results: There were 37,697 adult ED patients with a total of 1,507,121 data points across all vital-sign categories. Three to four trajectory groups per vital-sign category were identified, and the following five trajectory groups were associated with a higher rate of IHCA: low and fluctuating SBP, high and fluctuating HR, persistent hypothermia, recurring tachypnea, and low and fluctuating oxygen saturation. The IHCA-prone trajectory group was associated with a higher triage level and a higher mortality rate, compared to other trajectory groups. Except for the persistent hypothermia group, the other four trajectory groups were more likely to have higher levels of C-reactive protein, lactic acid, cardiac troponin I, and D-dimer. Multivariable analysis revealed that hypothermia (adjusted odds ratio [aOR], 2.20; 95% confidence interval [95%CI], 1.35–3.57) and recurring tachypnea (aOR 2.44; 95%CI, 1.24–4.79) were independently associated with IHCA.

Conclusions: We identified five novel vital-sign sub-phenotypes associated with a higher likelihood of IHCA, with distinct patterns in clinical course and laboratory markers. A better understanding of the pre-IHCA vital-sign trajectories may help with the early identification of deteriorating patients.

Serum Lactate for Predicting Cardiac Arrest in the Emergency Department

Objectives: Early recognition and prevention of in-hospital cardiac arrest (IHCA) play an increasingly important role in the Chain of Survival. However, clinical tools for predicting IHCA in the emergency department (ED) are scanty. We sought to evaluate the role of serum lactate in predicting ED-based IHCA. Methods: Data were retrieved from 733,398 ED visits over a 7-year period in a tertiary medical centre. We selected one ED visit per person and excluded out-of-hospital cardiac arrest, children, or those without lactate measurements. Patient demographics, computerised triage information, and serum lactate levels were extracted. The initial serum lactate levels were grouped into normal (≤2 mmol/L), moderately elevated (2 < lactate ≤ 4), and highly elevated (>4 mmol/L) categories. The primary outcome was ED-based IHCA. Results: A total of 17,392 adult patients were included. Of them, 342 (2%) developed IHCA. About 50% of the lactate levels were normal, 30% were moderately elevated, and 20% were highly elevated. In multivariable analysis, the group with highly elevated lactate had an 18-fold increased risk of IHCA (adjusted odds ratio [OR], 18.0; 95% confidence interval [CI], 11.5-28.2), compared with the normal lactate group. In subgroup analysis, the poor lactate-clearance group (<2.5%/h) was associated with a 7.5-fold higher risk of IHCA (adjusted OR, 7.5; 95%CI, 3.7-15.1) compared with the normal clearance group. Conclusions: Elevated lactate levels and poor lactate clearance were strongly associated with a higher risk of ED-based IHCA. Clinicians may consider a more liberal sampling of lactate in patients at higher risk of IHCA with follow-up of abnormal levels.

Factors associated with a high-risk return visit to the emergency department: a case-crossover study

BACKGROUND AND IMPORTANCE

Although factors related to a return emergency department (ED) visit have been reported, few studies have examined 'high-risk' return ED visits with serious adverse outcomes. Understanding factors associated with high-risk return ED visits may help with early recognition and prevention of these catastrophic events.

OBJECTIVES

We aimed to (1) estimate the incidence of high-risk return ED visits, and (2) to investigate time-varying factors associated with these revisits.

DESIGN

Case-crossover study.

SETTINGS AND PARTICIPANTS

We used electronic clinical warehouse data from a tertiary medical center. We retrieved data from 651 815 ED visits over a 6-year period. Patient demographics and computerized triage information were extracted.

OUTCOME MEASURE AND ANALYSIS

A high-risk return ED visit was defined as a revisit within 72 h of the index visit with ICU admission, receiving emergency surgery, or with in-hospital cardiac arrest during the return ED visit. Time-varying factors associated with a return visit were identified.

MAIN RESULTS

There were 440 281 adult index visits, of which 19 675 (4.5%) return visits occurred within 72 h. Of them, 417 (0.1%) were high-risk revisits. Multivariable analysis showed that time-varying factors associated with an increased risk of high-risk revisits included the following: arrival by ambulance, dyspnea, or chest pain on ED presentation, triage level 1 or 2, acute change in levels of consciousness, tachycardia (>90/min), and high fever (>39°C).

CONCLUSIONS

We found a relatively small fraction of discharges (0.1%) developed serious adverse events during the return ED visits. We identified symptom-based and vital sign-based warning signs that may be used for patient self-monitoring at home, as well as new-onset signs during the return visit to alert healthcare providers for timely management of these high-risk revisits.

Prediction of hospitalization using artificial intelligence for urgent patients in the emergency department

Timely assessment to accurately prioritize patients is crucial for emergency department (ED) management. Urgent (i.e., level-3, on a 5-level emergency severity index system) patients have become a challenge since under-triage and over-triage often occur. This study was aimed to develop a computational model by artificial intelligence (AI) methodologies to accurately predict urgent patient outcomes using data that are readily available in most ED triage systems. We retrospectively collected data from the ED of a tertiary teaching hospital between January 1, 2015 and December 31, 2019. Eleven variables were used for data analysis and prediction model building, including 1 response, 2 demographic, and 8 clinical variables. A model to predict hospital admission was developed using neural networks and machine learning methodologies. A total of 282,971 samples of urgent (level-3) visits were included in the analysis. Our model achieved a validation area under the curve (AUC) of 0.8004 (95% CI 0.7963–0.8045). The optimal cutoff value identified by Youden's index for determining hospital admission was 0.5517. Using this cutoff value, the sensitivity was 0.6721 (95% CI 0.6624–0.6818), and the specificity was 0.7814 (95% CI 0.7777–0.7851), with a positive predictive value of 0.3660 (95% CI 0.3586–0.3733) and a negative predictive value of 0.9270 (95% CI 0.9244–0.9295). Subgroup analysis revealed that this model performed better in the nontraumatic adult subgroup and achieved a validation AUC of 0.8166 (95% CI 0.8199–0.8212). Our AI model accurately assessed the need for hospitalization for urgent patients, which constituted nearly 70% of ED visits. This model demonstrates the potential for streamlining ED operations using a very limited number of variables that are readily available in most ED triage systems. Subgroup analysis is an important topic for future investigation.

Inpatient Outcomes Following a Return Visit to the Emergency Department: A Nationwide Cohort Study

Introduction

Emergency department (ED) revisits are traditionally used to measure potential lapses in emergency care. However, recent studies on in-hospital outcomes following ED revisits have begun to challenge this notion. We aimed to examine inpatient outcomes and resource use among patients who were hospitalized following a return visit to the ED using a national database.

Methods

This was a retrospective cohort study using the National Health Insurance Research Database in Taiwan. One-third of ED visits from 2012–2013 were randomly selected and their subsequent hospitalizations included. We analyzed the inpatient outcomes (mortality and intensive care unit [ICU] admission) and resource use (length of stay [LOS] and costs). Comparisons were made between patients who were hospitalized after a return visit to the ED and those who were hospitalized during the index ED visit.

Results

Of the 3,019,416 index ED visits, 477,326 patients (16%) were directly admitted to the hospital. Among the 2,504,972 patients who were discharged during the index ED visit, 229,059 (9.1%) returned to the ED within three days. Of them, 37,118 (16%) were hospitalized. In multivariable analyses, the inpatient mortality rates and hospital LOS were similar between the two groups. Compared with the direct-admission group, the return-admission group had a lower ICU admission rate (adjusted odds ratio, 0.78; 95% confidence interval [CI], 0.72–0.84), and lower costs (adjusted difference, −5,198 New Taiwan dollars, 95% CI, −6,224 to −4,172).

Conclusion

Patients who were hospitalized after a return visit to the ED had a lower ICU admission rate and lower costs, compared to those who were directly admitted. Our findings suggest that ED revisits do not necessarily translate to poor initial care and that subsequent inpatient outcomes should also be considered for better assessment.

Prior beta-blocker treatment improves outcomes in out-of-hospital cardiac arrest patients with non-shockable rhythms

The prognosis of out of cardiac arrest is poor and most cardiac arrest patients suffered from the non-shockable rhythm especially in patients without pre-existing cardiovascular diseases and medication prescription. Beta-blocker (ß-blocker) therapy has been shown to improve outcomes in cardiovascular diseases such as heart failure, ischemia related cardiac, and brain injuries. Therefore, we investigated whether prior ß-blockers use was associated with reduced mortality in patients with cardiac arrest and non-shockable rhythm. We conducted a population-based retrospective cohort study using multivariate propensity score-based regression to control for differences among patients with cardiac arrest. A total of 104,568 adult patients suffering a non-traumatic and non-shockable rhythm cardiac arrest between 2005 and 2011 were identified. ß-blocker prescription at least 30 days prior to the cardiac arrest event was defines as the ß-blockers group. We chose 12.5 mg carvedilol as the cut-off value and defined greater or equal to carvedilol 12.5 mg per day and its equivalent dose as high-dose group. After multivariate propensity score-based logistic regression analysis, patients with prior ß-blockers use were associated with better 1-year survival [adjusted odds ratio (OR), 1.15, 95% confidence interval (CI) 1.01-1.30; P = 0.031]. Compared to non-ß-blocker use group and prior low-dose ß-blockers use group, prior high-dose ß-blockers use group was associated with higher mechanical ventilator wean success rate (adjusted OR 1.19, 95% CI 1.01-1.41, P = 0.042). In conclusion, prior high dose ß-blockers use was associated with a better 1-year survival and higher weaning rate in patients with non-shockable cardiac arrest.

Impact of the COVID-19 Pandemic on Healthcare-Seeking Behaviors among Frequent Emergency Department Users: A Cohort Study

In 2020, Taiwan’s healthcare system faced a notable burden imposed by the coronavirus disease (COVID-19) pandemic. Emergency department (ED) is a high-risk area for severe acute respiratory syndrome coronavirus 2 transmission. The effect of COVID-19 on the utilization of ED services among frequent ED users remains unknown. This cohort study determined the impact of the COVID-19 pandemic on healthcare-seeking behaviors among frequent ED users at Taipei City Hospital, Taiwan. We included ED users aged ≥ 18 years admitted to Taipei City Hospital during February 2019–January 2020 (before the pandemic) and February 2020–January 2021 (during the pandemic). Frequent ED users were patients with four or more ED visits per year. Stepwise logistic regression was performed to identify predictors of frequent ED use during the COVID-19 pandemic. Frequent ED users had shorter hospital stays in the ED during the pandemic. After adjusting for sociodemographic factors and other covariates, patients with a triage status of level 4–5, pneumonia diagnosis, giddiness, or dyspnea were more likely frequent ED visitors during the COVID-19 pandemic. To reduce the risk of acquiring COVID-19, it is important to utilize territorial healthcare or telehealth to avoid inappropriate ED visits for patients with a low level of risk or chronic disease.

Developing a stroke alert trigger for clinical decision support at emergency triage using machine learning

BACKGROUND

Acute stroke is an urgent medical condition that requires immediate assessment and treatment. Prompt identification of patients with suspected stroke at emergency department (ED) triage followed by timely activation of code stroke systems is the key to successful management of stroke. While false negative detection of stroke may prevent patients from receiving optimal treatment, excessive false positive alarms will substantially burden stroke neurologists. This study aimed to develop a stroke-alert trigger to identify patients with suspected stroke at ED triage.

METHODS

Patients who arrived at the ED within 12 h of symptom onset and were suspected of a stroke or transient ischemic attack or triaged with a stroke-related symptom were included. Clinical features at ED triage were collected, including the presenting complaint, triage level, self-reported medical history (hypertension, diabetes, hyperlipidemia, heart disease, and prior stroke), vital signs, and presence of atrial fibrillation. Three rule-based algorithms, ie, Face Arm Speech Test (FAST) and two flavors of Balance, Eyes, FAST (BE-FAST), and six machine learning (ML) techniques with various resampling methods were used to build classifiers for identification of patients with suspected stroke. Logistic regression (LR) was used to find important features.

RESULTS

The study population consisted of 1361 patients. The values of area under the precision-recall curve (AUPRC) were 0.737, 0.710, and 0.562 for the FAST, BE-FAST-1, and BE-FAST-2 models, respectively. The values of AUPRC for the top three ML models were 0.787 for classification and regression tree with undersampling, 0.783 for LR with synthetic minority oversampling technique (SMOTE), and 0.782 for LR with class weighting. Among the ML models, logistic regression and random forest models in general achieved higher values of AUPRC, in particular in those with class weighting or SMOTE to handle class imbalance problem. In addition to the presenting complaint and triage level, age, diastolic blood pressure, body temperature, and pulse rate, were also important features for developing a stroke-alert trigger.

CONCLUSIONS

ML techniques significantly improved the performance of prediction models for identification of patients with suspected stroke. Such ML models can be embedded in the electronic triage system for clinical decision support at ED triage.

Analysis of the Adequacy of Prehospital Emergency Medical Services Use of Patients Who Visited Emergency Departments in Korea from 2016 to 2018: Data from the National Emergency Department Information System

Introduction

Proper ambulance use is important not only due to the patient's transport quality but also because of the need for efficient use of limited resources allotted by the system. Therefore, this study was conducted to check for overuse or underuse of the ambulance system by patients who visited the emergency department (ED).

Methods

In this study, a secondary data analysis was conducted using the existing database of the National Emergency Department Information System with all patients who visited EDs over the three-year study period from 2016 to 2018. The study subjects were classified into the following groups: (1) appropriate Emergency Medical Services (EMS) usage; (2) appropriate no EMS usage; (3) underuse; and (4) overuse groups.

Results

Of 18,298,535 patients, 11,668,581 (63.77%) were classified under the appropriate usage group, while 6,629,954 (36.23%) were classified under the inappropriate usage group. In the appropriate EMS usage group, there were 2,408,845 (13.16%) patients. In the appropriate no EMS usage group, there were 9,259,706 (50.60%) patients. As for the inappropriate usage group, there were 5,147,352 (28.13%) patients categorized under the underuse group. On the other hand, there were 1,482,602 (8.10%) patients under the overuse group.

Conclusion

There are many patients who use ambulances appropriately, but there are still many overuse and underuse. Guidelines on ambulance use are necessary for the efficient use of emergency medical resources and for the safety of patients.

Validity of the Taiwan Triage and Acuity Scale in mainland China: a retrospective observational study

OBJECTIVES

The Taiwan Triage and Acuity Scale (TTAS), developed for use in EDs, has been shown to be an excellent tool for triaging patients with high predictive performance, with an area under the receiver operating curve (AUROC) of 0.75. TTAS has been widely used in hospitals in Taiwan since 2010, but its utility has not been studied outside of Taiwan. Thus, the aim of this study was to evaluate the validity of using the TTAS in the ED of a tertiary hospital in mainland China to predict patient outcomes.

METHODS

A retrospective observational study was performed on patients 14 years of age or older attending the ED of a tertiary hospital in mainland China between 1 January 2016 and 31 March 2016. The validity of the TTAS in predicting hospital admission, intensive care unit (ICU) admission, death, ED length of stay (LOS) and ED resource utilisation was evaluated by determining the correlation of these outcomes with the TTAS, AUROC and test characteristics.

RESULTS

A total of 7843 patients were included in this study. There were significant differences between the TTAS categories in disposition, ED LOS and ED resource utilisation (p<0.0001). The TTAS was significantly correlated with patient disposition at discharge, hospital admission, ICU admission and death in the ED (Kendall rank correlations were 0.254, -0.254, -0.079 and -0.071, respectively; p=0.001). The AUROCs for the prediction of hospital admissions, ICU admissions and deaths in the ED were 0.749 (95% CI 0.732 to 0.765), 0.869 (95% CI 0.797 to 0.942) and 0.998 (95% CI 0.995 to 1.000), respectively. Our results demonstrated better performance using the TTAS for predictions of ICU admission and death.

CONCLUSIONS

The TTAS had good validity in predicting patient outcomes and ED resource utilisation in a tertiary hospital in mainland China. Compared with the performance of the TTAS in Taiwan, our results suggest that the TTAS can usefully be applied outside of Taiwan.

A modified cardiac triage strategy reduces door to ECG time in patients with ST elevation myocardial infarction

Timely performing electrocardiography (ECG) is crucial for early detection of ST-elevation myocardial infarction (STEMI). For shortening door-to-ECG time, a chief complaint-based “cardiac triage” protocol comprising (1) raising alert among medical staff with bedside triage tags, and (2) immediate bedside ECG after focused history-taking was implemented at the emergency department (ED) in a single tertiary referral center. All patients diagnosed with STEMI visiting the ED between November 2017 and January 2020 were retrospectively reviewed to investigate the effectiveness of strategy before and after implantation. Analysis of a total of 117 ED patients with STEMI (pre-intervention group, n = 57; post-intervention group, n = 60) showed significant overall improvements in median door-to-ECG time from 5 to 4 min (p = 0.02), achievement rate of door-to-ECG time < 10 min from 45 to 57% (p = 0.01), median door-to-balloon time from 81 to 70 min (p < 0.01). Significant trends of increase in achievement rates for door-to-ECG and door-to-balloon times (p = 0.032 and p = 0.002, respectively) was noted after strategy implementation. The incidences of door-to-ECG time > 10 min for those with initially underestimated disease severity (from 90 to 10%, p < 0.01) and walk-in (from 29.2 to 8.8%, p = 0.04) were both reduced. In conclusion, a chief complaint-based “cardiac triage” strategy successfully improved the quality of emergency care for STEMI patients through reducing delays in diagnosis and treatment.

Impact of the Coronavirus Disease 2019 Pandemic on an Emergency Department Service: Experience at the Largest Tertiary Center in Taiwan

Purpose

Severe Acute Respiratory Syndrome coronavirus 2 (SARS-CoV-2) is an emerging contagious pathogen that has caused community and nosocomial infections in many countries. This study aimed to evaluate the impact of Coronavirus disease 2019 (COVID-19) on emergency services of the largest medical center in Taiwan by comparing emergency department (ED) usage, turnover, and admission rates before the COVID-19 outbreak with those during the outbreak.

Materials and Methods

A retrospective cohort study was conducted in the ED of the largest tertiary medical center in Taiwan. Trends of adult, non-trauma patients who visited the ED during February-April 2019 were compared with those during February-April 2020. The number of visits, their dispositions, crowding parameters, and turnover rates were analyzed. The primary outcome was the change in ED attendance between the two periods. The secondary outcomes were changes in hospital admission rates, crowding parameters, and turnover rates.

Results

During the outbreak, there were decreased non-trauma ED visits by 33.45% (p < 0.001) and proportion of Taiwan Triage and Acuity Scale (TTAS) 3 patients (p=0.02), with increased admission rates by 4.7% (p < 0.001). Crowding parameters and turnover rate showed significant improvements.

Conclusion

Comparison of periods before and during the COVID-19 outbreak showed an obvious decline in adult, non-trauma ED visits. The reduction in TTAS 3 patient visits and the increased hospital admission rates provide references for future public-health policy-making to optimise emergency medical resource allocations globally.

A-qCPR risk score screening model for predicting 1-year mortality associated with hospice and palliative care in the emergency department

BACKGROUND

Evaluating the need for palliative care and predicting its mortality play important roles in the emergency department.

AIM

We developed a screening model for predicting 1-year mortality.

DESIGN

A retrospective cohort study was conducted to identify risk factors associated with 1-year mortality. Our risk scores based on these significant risk factors were then developed. Its predictive validity performance was evaluated using area under receiving operating characteristic analysis and leave-one-out cross-validation.

SETTING AND PARTICIPANTS

Patients aged 15 years or older were enrolled from June 2015 to May 2016 in the emergency department.

RESULTS

We identified five independent risk factors, each of which was assigned a number of points proportional to its estimated regression coefficient: age (0.05 points per year), qSOFA ⩾ 2 (1), Cancer (4), Eastern Cooperative Oncology Group Performance Status score ⩾ 2 (2), and Do-Not-Resuscitate status (3). The sensitivity, specificity, positive predictive value, and negative predictive value of our screening tool given the cutoff larger than 3 points were 0.99 (0.98-0.99), 0.31 (0.29-0.32), 0.26 (0.24-0.27), and 0.99 (0.98-1.00), respectively. Those with screening scores larger than 9 points corresponding to 64.0% (60.0-67.9%) of 1-year mortality were prioritized for consultation and communication. The area under the receiving operating characteristic curves for the point system was 0.84 (0.83-0.85) for the cross-validation model.

CONCLUSIONS

A-qCPR risk scores provide a good screening tool for assessing patient prognosis. Routine screening for end-of-life using this tool plays an important role in early and efficient physician-patient communications regarding hospice and palliative needs in the emergency department.

Developing and validating a model for predicting 7-day mortality of patients admitted from the emergency department: an initial alarm score by a prospective prediction model study

OBJECTIVES

To set up a prediction model for the 7-day in-hospital mortality of patients admitted from the emergency department (ED) because it is high but no appropriate initial alarm score is available.

DESIGN

This is a prospective cohort study for prediction model development.

SETTING

In a tertiary referred hospital in northern Taiwan.

PARTICIPANTS

ED-admitted medical patients in hospitalist care wards were enrolled during May 2010 to October 2016. Two-thirds of them were randomly assigned to a derivation cohort for development of the model and cross-validation was performed in the validation cohort.

PRIMARY OUTCOME MEASURED

7-day in-hospital mortality.

RESULTS

During the study period, 8649 patients were enrolled for analysis. The mean age was 71.05 years, and 51.91% were male. The most common admission diagnoses were pneumonia (36%) and urinary tract infection (20.05%). In the derivation cohort, multivariable Cox proportional hazard regression revealed that a low Barthel Index Score, triage level 1 at the ED, presence of cancer, metastasis and admission diagnoses of pneumonia and sepsis were independently associated with 7 days in-hospital mortality. Based on the probability developed from the multivariable model, the area under the receiver operating characteristic curve in the derivation group was 0.81 (0.79-0.85). The result in the validation cohort was comparable. The prediction score modified by the six independent factors had high sensitivity of 88.03% and a negative predictive value of 99.51% for a cut-off value of 4, whereas the specificity and positive predictive value were 89.61% and 10.55%, respectively, when the cut-off value was a score of 6.

CONCLUSION

The 7-day in-hospital mortality in the hospitalist care ward is 2.8%. The initial alarm score could help clinicians to prioritise or exclude patients who need urgent and intensive care.

Comprehensive geriatric assessment in the emergency department for the prediction of readmission among older patients: A 3-month follow-up study

BACKGROUND

Older people present to the emergency department (ED) with distinct patterns and emergency care needs. This study aimed to use comprehensive geriatric assessment (CGA) surveying the patterns of ED visits among older patients and determine frailty associated with the risk of revisits/readmission.

METHODS

This prospective study screened 2270 patients aged ≥75 years in the ED from August 2018 to February 2019. All patients underwent CGA. A 3-months follow-up was conducted to observe the hospital courses of admission and revisit/readmission.

RESULTS

A total of 270 older patients were enrolled. The independent predictors of admission at initial ED visit were the risk of nutritional deficit and instrumental activities of daily living (IADL). In the admission group, the independent predictors of revisit/readmission were a fall in the past year and mobility difficulties. In the discharge group, the independent predictors of revisit/readmission were frailty and insomnia. Regardless if older patients were either admitted or discharged at the initial ED visit, the independent predictor of revisit/readmission for older patients was frailty.

CONCLUSION

Our study showed that frailty was the only independent predictor for revisit/readmission after ED discharge during the 3-month follow up. For ED physicians, malnutrition and IADL were independent predictors in recognizing whether the older patient should be admitted to the hospital. For discharged older ED patients, frailty was the independent predictor for the integration of community services for older patients to decrease the rate of revisit/readmission in 3 months.

A Double Triage and Telemedicine Protocol to Optimize Infection Control in an Emergency Department in Taiwan During the COVID-19 Pandemic: Retrospective Feasibility Study

Background

Frontline health care workers, including physicians, are at high risk of contracting coronavirus disease (COVID-19) owing to their exposure to patients suspected of having COVID-19.

Objective

The aim of this study was to evaluate the benefits and feasibility of a double triage and telemedicine protocol in improving infection control in the emergency department (ED).

Methods

In this retrospective study, we recruited patients aged ≥20 years referred to the ED of the National Taiwan University Hospital between March 1 and April 30, 2020. A double triage and telemedicine protocol was developed to triage suggested COVID-19 cases and minimize health workers’ exposure to this disease. We categorized patients attending video interviews into a telemedicine group and patients experiencing face-to-face interviews into a conventional group. A questionnaire was used to assess how patients perceived the quality of the interviews and their communication with physicians as well as perceptions of stress, discrimination, and privacy. Each question was evaluated using a 5-point Likert scale. Physicians’ total exposure time and total evaluation time were treated as primary outcomes, and the mean scores of the questions were treated as secondary outcomes.

Results

The final sample included 198 patients, including 93 cases (47.0%) in the telemedicine group and 105 cases (53.0%) in the conventional group. The total exposure time in the telemedicine group was significantly shorter than that in the conventional group (4.7 minutes vs 8.9 minutes, P<.001), whereas the total evaluation time in the telemedicine group was significantly longer than that in the conventional group (12.2 minutes vs 8.9 minutes, P<.001). After controlling for potential confounders, the total exposure time in the telemedicine group was 4.6 minutes shorter than that in the conventional group (95% CI −5.7 to −3.5, P<.001), whereas the total evaluation time in the telemedicine group was 2.8 minutes longer than that in the conventional group (95% CI −1.6 to −4.0, P<.001). The mean scores of the patient questionnaire were high in both groups (4.5/5 to 4.7/5 points).

Conclusions

The implementation of the double triage and telemedicine protocol in the ED during the COVID-19 pandemic has high potential to improve infection control.

Emergency department disposition prediction using a deep neural network with integrated clinical narratives and structured data

BACKGROUND

Emergency department (ED) overcrowding has been a serious issue and demands effective clinical decision-making of patient disposition. In previous studies, emergency clinical narratives provide a rich context for clinical decisions. We aimed to develop the disposition prediction model using deep learning modeling strategy with the heterogeneous data, including the physicians' narratives.

METHODS

We constructed a retrospective cohort of all 104,083 ED visits of non-trauma adults during 2017-18 from an academically affiliated ED in Taiwan. 18,308 visits were excluded based on the completeness of each record and the unpredictable dispositions, such as out-of-hospital cardiac arrest, against-advice discharge, and escapes. We integrated subjective section of the first physicians' clinical narratives and structured data (e.g., demographics, triage vital signs, etc.) as available predictors at the first physician-patient encounter. To predict final patient disposition (i.e., hospitalization or discharge), a deep neural network (DNN) model was developed with word embedding, a common natural language processing method. We compared the proposed model to a reference model using the Rapid Emergency Medicine Score, a logistic regression model with structured data, and a DNN model with paragraph vectors. F1 score was used to measure the predictive performance for each model.

RESULTS

The F1 score (with 95 % CI) for the proposed model, the reference model, the logistic regression model with structured data, and the DNN model with paragraph vectors were 0.674 (0.669-0.679), 0.474 (0.469-0.479), 0.547 (0.543-0.551), and 0.602 (0.596-0.607), respectively. While analyzing the relationship between context length and predictive performance under the proposed model, the F1 score at 95th percentile of the word counts was higher than that at 25th percentile of the word counts in chief complaint [0.634 (0.629-0.640) vs. 0.624 (0.620-0.628)] and in present illness [0.671 (0.667-0.674) vs. 0.654 (0.651-0.658)], but not in past medical history [0.674 (0.669-0.679) vs. 0.673 (0.666-0.679)].

CONCLUSIONS

The proposed deep learning model with the usage of the first physicians' clinical narratives and structured data based on natural language processing outperformed the commonly used ones in terms of F1 score. It also evidenced the importance of the subjective section of clinical narratives, which serve as vital predictors for ED clinical decision-making.

Improving Emergency Department Efficiency by Patient Scheduling Using Deep Reinforcement Learning

Emergency departments (ED) in hospitals usually suffer from crowdedness and long waiting times for treatment. The complexity of the patient’s path flows and their controls come from the patient’s diverse acute level, personalized treatment process, and interconnected medical staff and resources. One of the factors, which has been controlled, is the dynamic situation change such as the patient’s composition and resources’ availability. The patient’s scheduling is thus complicated in consideration of various factors to achieve ED efficiency. To address this issue, a deep reinforcement learning (RL) is designed and applied in an ED patients’ scheduling process. Before applying the deep RL, the mathematical model and the Markov decision process (MDP) for the ED is presented and formulated. Then, the algorithm of the RL based on deep Q-networks (DQN) is designed to determine the optimal policy for scheduling patients. To evaluate the performance of the deep RL, it is compared with the dispatching rules presented in the study. The deep RL is shown to outperform the dispatching rules in terms of minimizing the weighted waiting time of the patients and the penalty of emergent patients in the suggested scenarios. This study demonstrates the successful implementation of the deep RL for ED applications, particularly in assisting decision-makers under the dynamic environment of an ED.

The Korean Triage and Acuity Scale: associations with admission, disposition, mortality and length of stay in the emergency department

OBJECTIVE

The Korean Triage and Acuity Scale (KTAS) was implemented in our emergency department (ED) in May 2016 and is fully integrated into the electronic medical record (EMR) system. Our objective was to determine whether the KTAS is associated with changes in admissions to the hospital, admission disposition, inpatient mortality and length of stay (LOS).

DESIGN

Quasi-experimental, uncontrolled before-and-after study.

SETTING

The urban tertiary teaching hospital with 1100 beds and receives approximately annual 90 000 ED visits.

PARTICIPANTS

122 370 patients who visited the ED during the before-and-the after period.

INTERVENTIONS

ED staff were educated on the KTAS for 1 month, after which the KTAS evaluation period began. Admission, disposition, mortality and LOS were compared between the 'before' period (1 June 2015 to 30 April 2016) and the 'after' period (1 June 2016 to 30 April 2017).

MAIN OUTCOME MEASURES

Admissions to the hospital, admission disposition, inpatient mortality and LOS.

RESULTS

A total of 59 220 and 63 150 patients were included in the before-and-after periods of KTAS implementation, respectively. The pattern of admission and disposition changed significantly after implementation of the KTAS. The mean LOS was 343 min (standard deviation [SD] = 432 min) during the before period, which significantly decreased to 289 min (SD = 333 min) after implementation (P < 0.001). The total mortality rate was significantly reduced after implementation of the KTAS (213 (0.36%) vs. 179 (0.28%), P = 0.020).

CONCLUSION

Implementation of the KTAS changed admission and disposition patterns and reduced the LOS and mortality in the ED.