Association between independent practice time and patient outcomes in the emergency department: a retrospective study of residents in three urban hospitals in Taiwan.

BACKGROUND: The purpose of the study was to investigate the relationship between the independent practice time of residents and the quality of care provided in the Emergency Department (ED) across three urban hospitals in Taiwan. The study focused on non-pediatric and non-obstetric complaints, aiming to provide insights into the optimal balance between resident autonomy and patient safety. METHODS: A comprehensive retrospective study was conducted using de-identified electronic health records (EHRs) from the hospital's integrated medical database (iMD) from August 2015 to July 2019. The independent practice time was defined as the duration from the first medical order by a resident to the first modifications by the attending physician. The primary outcome was revisits to the ED within 72 h following discharge. Statistical analysis was conducted using RStudio and pyGAM. RESULTS: The study identified several factors associated with shorter independent practice times (< 30 minutes), including older patient age, male sex, higher body temperature, higher heart rate, lower blood pressure, and the presence of certain comorbidities. Residents practicing independently for 30-120 minutes were associated with similar adjusted odds of patient revisits to the ED (OR 1.034, 95% CI 0.978-1.093) and no higher risk of 7-day mortality (OR 0.674, 95% CI 0.592-0.767) compared to the group with less autonomy. However, independent practice times exceeding 120 minutes were associated with higher odds of revisiting the ED within 72 h. For the group with 120-210 minutes of independent practice time, the OR was 1.113 (95% CI: 1.025-1.208, p = 0.011). For the group with > 210 minutes, the OR was 1.259 (95% CI: 1.094-1.449, p = 0.001), indicating an increased risk of adverse outcomes as the independent practice time increasing. CONCLUSIONS: The study concludes that while providing residents an independent practice time between 30 to 120 minutes may be beneficial, caution should be exercised when this time exceeds 120 minutes. The findings underscore the importance of optimal supervision in enhancing patient care quality and safety. Further research is recommended to explore the long-term effects of different levels of resident autonomy on patient outcomes and the professional development of the residents themselves.

The Effectiveness of Online-Only Blended Cardiopulmonary Resuscitation Training: Static-Group Comparison Study.

BACKGROUND: Basic life support (BLS) education is essential for improving bystander cardiopulmonary resuscitation (CPR) rates, but the imparting of such education faces obstacles during the outbreak of emerging infectious diseases, such as COVID-19. When face-to-face teaching is limited, distance learning-blended learning (BL) or an online-only model-is encouraged. However, evidence regarding the effect of online-only CPR training is scarce, and comparative studies on classroom-based BL (CBL) are lacking. While other strategies have recommended self-directed learning and deliberate practice to enhance CPR education, no previous studies have incorporated all of these instructional methods into a BLS course. OBJECTIVE: This study aimed to demonstrate a novel BLS training model-remote practice BL (RBL)-and compare its educational outcomes with those of the conventional CBL model. METHODS: A static-group comparison study was conducted. It included RBL and CBL courses that shared the same paradigm, comprising online lectures, a deliberate practice session with Little Anne quality CPR (QCPR) manikin feedback, and a final assessment session. In the main intervention, the RBL group was required to perform distant self-directed deliberate practice and complete the final assessment via an online video conference. Manikin-rated CPR scores were measured as the primary outcome; the number of retakes of the final examination was the secondary outcome. RESULTS: A total of 52 and 104 participants from the RBL and CBL groups, respectively, were eligible for data analysis. A comparison of the 2 groups revealed that there were more women in the RBL group than the CBL group (36/52, 69.2% vs 51/104, 49%, respectively; P=.02). After adjustment, there were no significant differences in scores for QCPR release (96.9 vs 96.4, respectively; P=.61), QCPR depth (99.2 vs 99.5, respectively; P=.27), or QCPR rate (94.9 vs 95.5, respectively; P=.83). The RBL group spent more days practicing before the final assessment (12.4 vs 8.9 days, respectively; P<.001) and also had a higher number of retakes (1.4 vs 1.1 times, respectively; P<.001). CONCLUSIONS: We developed a remote practice BL-based method for online-only distant BLS CPR training. In terms of CPR performance, using remote self-directed deliberate practice was not inferior to the conventional classroom-based instructor-led method, although it tended to take more time to achieve the same effect. TRIAL REGISTRATION: Not applicable.

The Prevalence and Impact of Fake News on COVID-19 Vaccination in Taiwan: Retrospective Study of Digital Media.

BACKGROUND: Vaccination is an important intervention to prevent the incidence and spread of serious diseases. Many factors including information obtained from the internet influence individuals' decisions to vaccinate. Misinformation is a critical issue and can be hard to detect, although it can change people's minds, opinions, and decisions. The impact of misinformation on public health and vaccination hesitancy is well documented, but little research has been conducted on the relationship between the size of the population reached by misinformation and the vaccination decisions made by that population. A number of fact-checking services are available on the web, including the Islander news analysis system, a free web service that provides individuals with real-time judgment on web news. In this study, we used such services to estimate the amount of fake news available and used Google Trends levels to model the spread of fake news. We quantified this relationship using official public data on COVID-19 vaccination in Taiwan. OBJECTIVE: In this study, we aimed to quantify the impact of the magnitude of the propagation of fake news on vaccination decisions. METHODS: We collected public data about COVID-19 infections and vaccination from Taiwan's official website and estimated the popularity of searches using Google Trends. We indirectly collected news from 26 digital media sources, using the news database of the Islander system. This system crawls the internet in real time, analyzes the news, and stores it. The incitement and suspicion scores of the Islander system were used to objectively judge news, and a fake news percentage variable was produced. We used multivariable linear regression, chi-square tests, and the Johnson-Neyman procedure to analyze this relationship, using weekly data. RESULTS: A total of 791,183 news items were obtained over 43 weeks in 2021. There was a significant increase in the proportion of fake news in 11 of the 26 media sources during the public vaccination stage. The regression model revealed a positive adjusted coefficient (ß=0.98, P=.002) of vaccine availability on the following week's vaccination doses, and a negative adjusted coefficient (ß=-3.21, P=.04) of the interaction term on the fake news percentage with the Google Trends level. The Johnson-Neiman plot of the adjusted effect for the interaction term showed that the Google Trends level had a significant negative adjustment effect on vaccination doses for the following week when the proportion of fake news exceeded 39.3%. CONCLUSIONS: There was a significant relationship between the amount of fake news to which the population was exposed and the number of vaccination doses administered. Reducing the amount of fake news and increasing public immunity to misinformation will be critical to maintain public health in the internet age.

Characteristics, prognostic factors, and chronological trends of out-of-hospital cardiac arrests with shockable rhythms in Taiwan - A 7-year observational study.

BACKGROUND: The study aimed to explore the characteristics, predictors, and chronological trends of outcomes for adult out-of-hospital cardiac arrests (OHCAs) with shockable rhythms. METHODS: A 7-year, community-wide observational study using an Utstein-style registry was conducted. Patients who were not transported, those who experienced trauma and those who lacked electronic electrocardiography data were excluded; those with initial shockable rhythms of ventricular fibrillation (VF) or pulseless ventricular tachycardia (pVT) were included. Outcomes were survival of discharge (SOD) and favorable neurological status (CPC 1-2). The outcome predictors, chronological trends, and their relationship with system interventions were analyzed. RESULTS: Of the 1544 shockable OHCAs (incidence 12.6%) included, 97.6% had VF and 2.4% had pVT. VF showed better outcomes than pVT. Predictors for both outcomes (SOD; CPC 1-2) were chronological change (adjusted odds ratio [aOR]: 1.133; 1.176), younger age (aOR: 0.973; 0.967), shorter response time (aOR: 0.998; 0.999), shorter scene time (aOR: 0.999; 0.999), witnessed collapse (aOR: 1.668; 1.670), and bystander cardiopulmonary resuscitation (BCPR) (aOR: 1.448; 1.576). Predictors for only SOD were public location (aOR: 1.450) and successful prehospital defibrillation (aOR: 3.374). The use of the supraglottic airway was associated with adverse outcomes. Chronologically with system interventions, BCPR rate, the proportion of shockable OHCA, and improved neurological outcomes increased over time. CONCLUSION: The incidence of shockable OHCA remained low in Asian community. VF showed better outcomes than pVT. Over time, the incidence of shockable rhythm, BCPR rate and patient outcomes did improve with health system interventions. The number of prehospital defibrillations did not predict outcomes.

Attitude and behavior toward bystander cardiopulmonary resuscitation during COVID-19 outbreak.

BACKGROUND: Outbreaks of emerging infectious diseases, such as COVID-19, have negative impacts on bystander cardiopulmonary resuscitation (BCPR) for fear of transmission while breaking social distancing rules. The latest guidelines recommend hands-only cardiopulmonary resuscitation (CPR) and facemask use. However, public willingness in this setup remains unknown. METHODS: A cross-sectional, unrestricted volunteer Internet survey was conducted to assess individuals' attitudes and behaviors toward performing BCPR, pre-existing CPR training, occupational identity, age group, and gender. The raking method for weights and a regression analysis for the predictors of willingness were performed. RESULTS: Among 1,347 eligible respondents, 822 (61%) had negative attitudes toward performing BCPR. Healthcare providers (HCPs) and those with pre-existing CPR training had fewer negative attitudes (p < 0.001); HCPs and those with pre-existing CPR training and unchanged attitude showed more positive behaviors toward BCPR (p < 0.001). Further, 9.7% of the respondents would absolutely refuse to perform BCPR. In contrast, 16.9% would perform BCPR directly despite the outbreak. Approximately 9.9% would perform it if they were instructed, 23.5%, if they wore facemasks, and 40.1%, if they were to perform hands-only CPR. Interestingly, among the 822 respondents with negative attitudes, over 85% still tended to perform BCPR in the abovementioned situations. The weighted analysis showed similar results. The adjusted predictors for lower negative attitudes toward BCPR were younger age, being a man, and being an HCP; those for more positive behaviors were younger age and being an HCP. CONCLUSIONS: Outbreaks of emerging infectious diseases, such as COVID-19, have negative impacts on attitudes and behaviors toward BCPR. Younger individuals, men, HCPs, and those with pre-existing CPR training tended to show fewer negative attitudes and behaviors. Meanwhile, most individuals with negative attitudes still expressed positive behaviors under safer measures such as facemask protection, hands-only CPR, and available dispatch instructions.

Disease Concept-Embedding Based on the Self-Supervised Method for Medical Information Extraction from Electronic Health Records and Disease Retrieval: Algorithm Development and Validation Study.

BACKGROUND: The electronic health record (EHR) contains a wealth of medical information. An organized EHR can greatly help doctors treat patients. In some cases, only limited patient information is collected to help doctors make treatment decisions. Because EHRs can serve as a reference for this limited information, doctors' treatment capabilities can be enhanced. Natural language processing and deep learning methods can help organize and translate EHR information into medical knowledge and experience. OBJECTIVE: In this study, we aimed to create a model to extract concept embeddings from EHRs for disease pattern retrieval and further classification tasks. METHODS: We collected 1,040,989 emergency department visits from the National Taiwan University Hospital Integrated Medical Database and 305,897 samples from the National Hospital and Ambulatory Medical Care Survey Emergency Department data. After data cleansing and preprocessing, the data sets were divided into training, validation, and test sets. We proposed a Transformer-based model to embed EHRs and used Bidirectional Encoder Representations from Transformers (BERT) to extract features from free text and concatenate features with structural data as input to our proposed model. Then, Deep InfoMax (DIM) and Simple Contrastive Learning of Visual Representations (SimCLR) were used for the unsupervised embedding of the disease concept. The pretrained disease concept-embedding model, named EDisease, was further finetuned to adapt to the critical care outcome prediction task. We evaluated the performance of embedding using t-distributed stochastic neighbor embedding (t-SNE) to perform dimension reduction for visualization. The performance of the finetuned predictive model was evaluated against published models using the area under the receiver operating characteristic (AUROC). RESULTS: The performance of our model on the outcome prediction had the highest AUROC of 0.876. In the ablation study, the use of a smaller data set or fewer unsupervised methods for pretraining deteriorated the prediction performance. The AUROCs were 0.857, 0.870, and 0.868 for the model without pretraining, the model pretrained by only SimCLR, and the model pretrained by only DIM, respectively. On the smaller finetuning set, the AUROC was 0.815 for the proposed model. CONCLUSIONS: Through contrastive learning methods, disease concepts can be embedded meaningfully. Moreover, these methods can be used for disease retrieval tasks to enhance clinical practice capabilities. The disease concept model is also suitable as a pretrained model for subsequent prediction tasks.

Modified Bidirectional Encoder Representations From Transformers Extractive Summarization Model for Hospital Information Systems Based on Character-Level Tokens (AlphaBERT): Development and Performance Evaluation.

BACKGROUND: Doctors must care for many patients simultaneously, and it is time-consuming to find and examine all patients' medical histories. Discharge diagnoses provide hospital staff with sufficient information to enable handling multiple patients; however, the excessive amount of words in the diagnostic sentences poses problems. Deep learning may be an effective solution to overcome this problem, but the use of such a heavy model may also add another obstacle to systems with limited computing resources. OBJECTIVE: We aimed to build a diagnoses-extractive summarization model for hospital information systems and provide a service that can be operated even with limited computing resources. METHODS: We used a Bidirectional Encoder Representations from Transformers (BERT)-based structure with a two-stage training method based on 258,050 discharge diagnoses obtained from the National Taiwan University Hospital Integrated Medical Database, and the highlighted extractive summaries written by experienced doctors were labeled. The model size was reduced using a character-level token, the number of parameters was decreased from 108,523,714 to 963,496, and the model was pretrained using random mask characters in the discharge diagnoses and International Statistical Classification of Diseases and Related Health Problems sets. We then fine-tuned the model using summary labels and cleaned up the prediction results by averaging all probabilities for entire words to prevent character level-induced fragment words. Model performance was evaluated against existing models BERT, BioBERT, and Long Short-Term Memory (LSTM) using the Recall-Oriented Understudy for Gisting Evaluation (ROUGE) L score, and a questionnaire website was built to collect feedback from more doctors for each summary proposal. RESULTS: The area under the receiver operating characteristic curve values of the summary proposals were 0.928, 0.941, 0.899, and 0.947 for BERT, BioBERT, LSTM, and the proposed model (AlphaBERT), respectively. The ROUGE-L scores were 0.697, 0.711, 0.648, and 0.693 for BERT, BioBERT, LSTM, and AlphaBERT, respectively. The mean (SD) critique scores from doctors were 2.232 (0.832), 2.134 (0.877), 2.207 (0.844), 1.927 (0.910), and 2.126 (0.874) for reference-by-doctor labels, BERT, BioBERT, LSTM, and AlphaBERT, respectively. Based on the paired t test, there was a statistically significant difference in LSTM compared to the reference (P<.001), BERT (P=.001), BioBERT (P<.001), and AlphaBERT (P=.002), but not in the other models. CONCLUSIONS: Use of character-level tokens in a BERT model can greatly decrease the model size without significantly reducing performance for diagnoses summarization. A well-developed deep-learning model will enhance doctors' abilities to manage patients and promote medical studies by providing the capability to use extensive unstructured free-text notes.

Using a smartwatch with real-time feedback improves the delivery of high-quality cardiopulmonary resuscitation by healthcare professionals.

AIM: Cardiopulmonary resuscitation (CPR) quality affects survival after cardiac arrest. We aimed to investigate if a smartwatch with real-time feedback can improve CPR quality by healthcare professionals. METHODS: An app providing real-time audiovisual feedback was developed for a smartwatch. Emergency Department (ED) professionals were recruited and randomly allocated to either the intervention group wearing a smartwatch with the preinstalled app, or to a control group. All participants were asked to perform a two-minute CPR on a manikin at a 30:2 compression-ventilation ratio. Primary outcomes were the mean CCR and CCD measured on the manikin. A secondary outcome was the percentage of chest compressions meeting both the guideline-recommended rate (100-120 min-1) and depth (50-60 mm) of high-quality CPR during a 2-min period. Differences between groups were evaluated with t-test, Chi-Square test, or Mann-Whitney U test depending on the distribution. RESULTS: Eighty participants were recruited. 40 people were assigned to the intervention and 40 to the control group. The compression rates (mean ± SD, min-1) were significantly faster (but above the guideline recommendation, P < 0.001) in the control (129.1 ± 14.9) than in the intervention group (112.0 ± 3.5). The compression depths (mean ± SD, mm) were significantly deeper (P < 0.001) in the intervention (50.9 ± 6.6) than in the control group (39.0 ± 8.7). The percentage (%) of high-quality CPR was significantly higher (P < 0.001) in the intervention (median 39.4, IQR 27.1-50.1) than in the control group (median 0.0, IQR 0.0-0.0). CONCLUSION: Without real-time feedback, chest compressions tend to be too fast and too shallow. CPR quality can be improved with the assistance of a smartwatch providing real-time feedback.

A novel depth estimation algorithm of chest compression for feedback of high-quality cardiopulmonary resuscitation based on a smartwatch.

INTRODUCTION: High-quality cardiopulmonary resuscitation (CPR) is a key factor affecting cardiac arrest survival. Accurate monitoring and real-time feedback are emphasized to improve CPR quality. The purpose of this study was to develop and validate a novel depth estimation algorithm based on a smartwatch equipped with a built-in accelerometer for feedback instructions during CPR. METHODS: For data collection and model building, researchers wore an Android Wear smartwatch and performed chest compression-only CPR on a Resusci Anne QCPR training manikin. We developed an algorithm based on the assumptions that (1) maximal acceleration measured by the smartwatch accelerometer and the chest compression depth (CCD) are positively correlated and (2) the magnitude of acceleration at a specific time point and interval is correlated with its neighboring points. We defined a statistic value M as a function of time and the magnitude of maximal acceleration. We labeled and processed collected data and determined the relationship between M value, compression rate and CCD. We built a model accordingly, and developed a smartwatch app capable of detecting CCD. For validation, researchers wore a smartwatch with the preinstalled app and performed chest compression-only CPR on the manikin at target sessions. We compared the CCD results given by the smartwatch and the reference using the Wilcoxon Signed Rank Test (WSRT), and used Bland-Altman (BA) analysis to assess the agreement between the two methods. RESULTS: We analyzed a total of 3978 compressions that covered the target rate of 80-140/min and CCD of 4-7 cm. WSRT showed that there was no significant difference between the two methods (P = 0.084). By BA analysis the mean of differences was 0.003 and the bias between the two methods was not significant (95% CI: -0.079 to 0.085). CONCLUSION: Our study indicates that the algorithm developed for estimating CCD based on a smartwatch with a built-in accelerometer is promising. Further studies will be conducted to evaluate its application for CPR training and clinical practice.