Use of a Large Language Model to Assess Clinical Acuity of Adults in the Emergency Department.

Importance: The introduction of large language models (LLMs), such as Generative Pre-trained Transformer 4 (GPT-4; OpenAI), has generated significant interest in health care, yet studies evaluating their performance in a clinical setting are lacking. Determination of clinical acuity, a measure of a patient's illness severity and level of required medical attention, is one of the foundational elements of medical reasoning in emergency medicine. Objective: To determine whether an LLM can accurately assess clinical acuity in the emergency department (ED). Design, Setting, and Participants: This cross-sectional study identified all adult ED visits from January 1, 2012, to January 17, 2023, at the University of California, San Francisco, with a documented Emergency Severity Index (ESI) acuity level (immediate, emergent, urgent, less urgent, or nonurgent) and with a corresponding ED physician note. A sample of 10 000 pairs of ED visits with nonequivalent ESI scores, balanced for each of the 10 possible pairs of 5 ESI scores, was selected at random. Exposure: The potential of the LLM to classify acuity levels of patients in the ED based on the ESI across 10 000 patient pairs. Using deidentified clinical text, the LLM was queried to identify the patient with a higher-acuity presentation within each pair based on the patients' clinical history. An earlier LLM was queried to allow comparison with this model. Main Outcomes and Measures: Accuracy score was calculated to evaluate the performance of both LLMs across the 10 000-pair sample. A 500-pair subsample was manually classified by a physician reviewer to compare performance between the LLMs and human classification. Results: From a total of 251 401 adult ED visits, a balanced sample of 10 000 patient pairs was created wherein each pair comprised patients with disparate ESI acuity scores. Across this sample, the LLM correctly inferred the patient with higher acuity for 8940 of 10 000 pairs (accuracy, 0.89 [95% CI, 0.89-0.90]). Performance of the comparator LLM (accuracy, 0.84 [95% CI, 0.83-0.84]) was below that of its successor. Among the 500-pair subsample that was also manually classified, LLM performance (accuracy, 0.88 [95% CI, 0.86-0.91]) was comparable with that of the physician reviewer (accuracy, 0.86 [95% CI, 0.83-0.89]). Conclusions and Relevance: In this cross-sectional study of 10 000 pairs of ED visits, the LLM accurately identified the patient with higher acuity when given pairs of presenting histories extracted from patients' first ED documentation. These findings suggest that the integration of an LLM into ED workflows could enhance triage processes while maintaining triage quality and warrants further investigation.

Early Cold Stored Platelet Transfusion Following Severe Injury: A Randomized Clinical Trial.

OBJECTIVE: To determine the feasibility, efficacy, and safety of early cold stored platelet transfusion compared to standard care resuscitation in patients with hemorrhagic shock. SUMMARY BACKGROUND DATA: Data demonstrating the safety and efficacy of early cold stored platelet transfusion are lacking following severe injury. METHODS: A phase 2, multicenter, randomized, open label, clinical trial was performed at five U.S. trauma centers. Injured patients at risk of large volume blood transfusion and the need for hemorrhage control procedures were enrolled and randomized. The intervention was the early transfusion of a single apheresis cold stored platelet unit, stored for up to 14 days vs. standard care resuscitation. The primary outcome was feasibility and the principal clinical outcome for efficacy and safety was 24-hour mortality. RESULTS: Mortality at 24 hours was 5.9% in patients who were randomized to early cold stored platelet transfusion compared to 10.2% in the standard care arm (difference, -4.3%; 95% CI, -12.8% to 3.5%; P=0.26). No significant differences were found for any of the prespecified ancillary outcomes. Rates of arterial and/or venous thromboembolism and adverse events did not differ across treatment groups. CONCLUSIONS AND RELEVANCE: In severely injured patients, early cold stored platelet transfusion is feasible, safe and did not result in a significant lower rate of 24-hour mortality. Early cold stored platelet transfusion did not result in a higher incidence of arterial and/or venous thrombotic complications or adverse events. The storage age of the cold stored platelet product was not associated with significant outcome differences.

Evaluating Large Language Models for Drafting Emergency Department Discharge Summaries.

Importance: Large language models (LLMs) possess a range of capabilities which may be applied to the clinical domain, including text summarization. As ambient artificial intelligence scribes and other LLM-based tools begin to be deployed within healthcare settings, rigorous evaluations of the accuracy of these technologies are urgently needed. Objective: To investigate the performance of GPT-4 and GPT-3.5-turbo in generating Emergency Department (ED) discharge summaries and evaluate the prevalence and type of errors across each section of the discharge summary. Design: Cross-sectional study. Setting: University of California, San Francisco ED. Participants: We identified all adult ED visits from 2012 to 2023 with an ED clinician note. We randomly selected a sample of 100 ED visits for GPT-summarization. Exposure: We investigate the potential of two state-of-the-art LLMs, GPT-4 and GPT-3.5-turbo, to summarize the full ED clinician note into a discharge summary. Main Outcomes and Measures: GPT-3.5-turbo and GPT-4-generated discharge summaries were evaluated by two independent Emergency Medicine physician reviewers across three evaluation criteria: 1) Inaccuracy of GPT-summarized information; 2) Hallucination of information; 3) Omission of relevant clinical information. On identifying each error, reviewers were additionally asked to provide a brief explanation for their reasoning, which was manually classified into subgroups of errors. Results: From 202,059 eligible ED visits, we randomly sampled 100 for GPT-generated summarization and then expert-driven evaluation. In total, 33% of summaries generated by GPT-4 and 10% of those generated by GPT-3.5-turbo were entirely error-free across all evaluated domains. Summaries generated by GPT-4 were mostly accurate, with inaccuracies found in only 10% of cases, however, 42% of the summaries exhibited hallucinations and 47% omitted clinically relevant information. Inaccuracies and hallucinations were most commonly found in the Plan sections of GPT-generated summaries, while clinical omissions were concentrated in text describing patients' Physical Examination findings or History of Presenting Complaint. Conclusions and Relevance: In this cross-sectional study of 100 ED encounters, we found that LLMs could generate accurate discharge summaries, but were liable to hallucination and omission of clinically relevant information. A comprehensive understanding of the location and type of errors found in GPT-generated clinical text is important to facilitate clinician review of such content and prevent patient harm.

Physical restraint use in children with mental and behavioral health emergencies in the prehospital setting.

Objective: Emergency medical services (EMS) transport for mental and behavioral health (MBH) emergencies occurs frequently in children, yet little is understood regarding prehospital physical restraint use despite the potential for serious adverse events. We aim to describe restraint use prevalence and primary impressions among children with MBH emergencies. Methods: This is a retrospective cross-sectional study of children with MBH emergencies evaluated by Alameda County (ALCO), California EMS from January 1, 2012 to December 31, 2018. Patient demographics and clinical variables were collected from the EMS records including sex, age at time of encounter, year of encounter, transport destination, medication use, and primary impression(s). The primary outcome was the use of physical restraints. Descriptive statistics were used to characterize the primary outcome and associated demographic and diagnostic features, as well as temporal use patterns. Sex and age were compared between restrained and non-restrained youth using chi-square analysis. Results: Over the 7-year study period, ALCO EMS transported 9775 children with MBH emergencies. Of these transports, 1205 (12.3%) were physically restrained. Most children restrained had the primary impression of "behavioral/psychiatric crisis" (51.1%), "psychiatric crisis" (27.4%), and "behavioral-other" (12.4%) and the remaining children (9.1%) had a non-psychiatric/behavioral health primary impression. Over time, there was no statistically significant change in either number of children with MBH emergencies transported or physical restraint rate. Conclusions: More than 1 in 8 children with MBH emergencies are being physically restrained during EMS transport. Restraint rate did not substantially change over time. Further studies to understand existing restraint rates and EMS resources available to address acute agitation in children are needed to inform quality and care enhancing initiatives.

Association of SARS-CoV-2 nucleocapsid viral antigen and the receptor for advanced glycation end products with development of severe disease in patients presenting to the emergency department with COVID-19.

Introduction: There remains a need to better identify patients at highest risk for developing severe Coronavirus Disease 2019 (COVID-19) as additional waves of the pandemic continue to impact hospital systems. We sought to characterize the association of receptor for advanced glycation end products (RAGE), SARS-CoV-2 nucleocapsid viral antigen, and a panel of thromboinflammatory biomarkers with development of severe disease in patients presenting to the emergency department with symptomatic COVID-19. Methods: Blood samples were collected on arrival from 77 patients with symptomatic COVID-19, and plasma levels of thromboinflammatory biomarkers were measured. Results: Differences in biomarkers between those who did and did not develop severe disease or death 7 days after presentation were analyzed. After adjustment for multiple comparisons, RAGE, SARS-CoV-2 nucleocapsid viral antigen, interleukin (IL)-6, IL-10 and tumor necrosis factor receptor (TNFR)-1 were significantly elevated in the group who developed severe disease (all p<0.05). In a multivariable regression model, RAGE and SARS-CoV-2 nucleocapsid viral antigen remained significant risk factors for development of severe disease (both p<0.05), and each had sensitivity and specificity >80% on cut-point analysis. Discussion: Elevated RAGE and SARS-CoV-2 nucleocapsid viral antigen on emergency department presentation are strongly associated with development of severe disease at 7 days. These findings are of clinical relevance for patient prognostication and triage as hospital systems continue to be overwhelmed. Further studies are warranted to determine the feasibility and utility of point-of care measurements of these biomarkers in the emergency department setting to improve patient prognostication and triage.

Effect of the Extended Focused Assessment With Sonography for Trauma on the Screening Performance of the National Emergency X-Radiography Utilization Study Chest Decision Instrument.

STUDY OBJECTIVE: Developed to decrease unnecessary thoracic computed tomography use in adult blunt trauma patients, the National Emergency X-Radiography Utilization Study (NEXUS) Chest clinical decision instrument does not include the extended Focused Assessment with Sonography in Trauma (eFAST). We assessed whether eFAST improves the NEXUS Chest clinical decision instrument's diagnostic performance and may replace the chest radiograph (CXR) as a predictor variable. METHODS: We performed a secondary analysis of prospective data from 8 Level I trauma centers from 2011-2014. We compared performance of modified clinical decision instruments that (1) added eFAST as a predictor (eFAST-added clinical decision instrument), and (2) replaced CXR with eFAST (eFAST-replaced clinical decision instrument), in screening for blunt thoracic injuries. RESULTS: One thousand nine hundred fifty-seven patients had documented computed tomography, CXR, clinical NEXUS criteria, and adequate eFAST; 624 (31.9%) patients had blunt thoracic injuries, and 126 (6.4%) had major injuries. Compared to the NEXUS Chest clinical decision instrument, the eFAST-added clinical decision instrument demonstrated unchanged screening performance for major injury (sensitivity 0.98 [0.94 to 1.00], specificity 0.28 [0.26 to 0.30]) or any injury (sensitivity 0.97 [0.95 to 0.98], specificity 0.21 [0.19 to 0.23]). The eFAST-replaced clinical decision instrument demonstrated unchanged sensitivity for major injury (sensitivity 0.93 [0.87 to 0.97], specificity 0.31 [0.29 to 0.34]) and decreased sensitivity for any injury (0.93 [0.91 to 0.951] versus 0.97 [0.953 to 0.98]). CONCLUSION: In our secondary analysis, adding eFAST as a predictor variable did not improve the diagnostic screening performance of the original NEXUS Chest clinical decision instrument; eFAST cannot replace the CXR criterion of the NEXUS Chest clinical decision instrument.

Toxic Exposures Among Young Children One Year into the COVID-19 Pandemic: A Retrospective Review of Three San Francisco Bay Area Emergency Departments.

BACKGROUND: Daycare and school closures prompted by shelter-in-place orders may have increased opportunities for unintentional ingestions among young children. OBJECTIVES: We examined emergency department (ED) presentations for toxic exposures among young children during the COVID-19 pandemic in the San Francisco Bay Area, which had some of the strictest and most prolonged shelter-in-place policies in the United States. METHODS: We performed a retrospective cross-sectional study of children 0 to 5 years of age who presented with an ED International Statistical Classification of Diseases and Related Health Problems, Tenth Revision diagnosis code of toxic exposure within a tertiary care hospital system between March 16, 2016 and March 15, 2021. We considered the period after March 16, 2020 to represent the pandemic. RESULTS: During the pandemic, the absolute number of poisonings among young children remained stable. Overall, ED encounters within this cohort decreased by 55%, which doubled the relative toxic exposure rate per 1000 ED encounters from 4.99 (95% confidence interval [CI] 3.19-5.90) to 9.79 (95% CI 8.09-11.49). Rates of admission, severe medical complications, operating room case requests, and length of stay were not significantly different. Shelter-in-place was associated with significantly higher odds of cannabis ingestion (odds ratio = 2.70, 95% CI 1.60-4.49). CONCLUSION: Despite dramatic decreases in overall ED patient volumes, the absolute number and severity of toxic exposures were similar during the pandemic compared with previous years. © 2022 Elsevier Inc.

Establishing the international research priorities for pediatric emergency medicine point-of-care ultrasound: A modified Delphi study.

BACKGROUND: The Pediatric Emergency Medicine (PEM) Point-of-care Ultrasound (POCUS) Network (P2Network) was established in 2014 to provide a platform for international collaboration among experts, including multicenter research. The objective of this study was to use expert consensus to identify and prioritize PEM POCUS topics, to inform future collaborative multicenter research. METHODS: Online surveys were administered in a two-stage, modified Delphi study. A steering committee of 16 PEM POCUS experts was identified within the P2Network, with representation from the United States, Canada, Italy, and Australia. We solicited the participation of international PEM POCUS experts through professional society mailing lists, research networks, social media, and "word of mouth." After each round, responses were refined by the steering committee before being reissued to participants to determine the ranking of all the research questions based on means and to identify the high-level consensus topics. The final stage was a modified Hanlon process of prioritization round (HPP), which emphasized relevance, impact, and feasibility. RESULTS: Fifty-four eligible participants (16.6%) provided 191 items to Survey 1 (Round 1). These were refined and consolidated into 52 research questions by the steering committee. These were issued for rating in Survey 2 (Round 2), which had 45 participants. At the completion of Round 2, all questions were ranked with six research questions reaching high-level consensus. Thirty-one research questions with mean ratings above neutral were selected for the HPP round. Highly ranked topics included clinical applications of POCUS to evaluate and manage children with shock, cardiac arrest, thoracoabdominal trauma, suspected cardiac failure, atraumatic limp, and intussusception. CONCLUSIONS: This consensus study has established a research agenda to inform future international multicenter PEM POCUS trials. This study has highlighted the ongoing need for high-quality evidence for PEM POCUS applications to guide clinical practice.

Accuracy and Interrater Reliability of Point-of-Care Ultrasonography Image Interpretation for Intussusception.

OBJECTIVES: The aim of this study was to determine the accuracy and interrater reliability of (1) point-of-care ultrasound (POCUS) image interpretation for identification of intussusception and (2) reliability of secondary signs associated with intussusception among experts compared with novice POCUS reviewers. METHODS: We conducted a planned secondary analysis of a prospective, convenience sample of children aged 3 months to 6 years who were evaluated with POCUS for intussusception across 17 international pediatric emergency departments between October 2018 and December 2020. A random sample of 100 POCUS examinations was reviewed by novice and expert POCUS reviewers. The primary outcome was identification of the presence or absence of intussusception. Secondary outcomes included intussusception size and the presence of trapped free fluid or echogenic foci. Accuracy was summarized using sensitivity and specificity, which were estimated via generalized mixed effects logistic regression. Interrater reliability was summarized via Light's κ statistics with bootstrapped standard errors (SEs). Accuracy and reliability of expert and novice POCUS reviewers were compared. RESULTS: Eighteen expert and 16 novice POCUS reviewers completed the reviews. The average expert sensitivity was 94.5% (95% confidence interval [CI], 88.6-97.5), and the specificity was 94.3% (95% CI, 90.3-96.7), significantly higher than the average novice sensitivity of 84.7% (95% CI, 74.3-91.4) and specificity of 80.4% (95% CI, 72.4, 86.7). κ was significantly greater for expert (0.679, SE 0.039) compared with novice POCUS reviewers (0.424, SE 0.044; difference 0.256, SE 0.033). For our secondary outcome measure of intussusception size, κ was significantly greater for experts (0.661, SE 0.038) compared with novices (0.397, SE 0.041; difference 0.264, SE 0.029). Interrater reliability was weak for expert and minimal for novice reviewers regarding the detection of trapped free fluid and echogenic foci. CONCLUSIONS: Expert POCUS reviewers demonstrate high accuracy and moderate interrater reliability when identifying intussusception via image interpretation and perform better than novice reviewers.

Injury-Related Pediatric Emergency Department Visits in the First Year of COVID-19.

OBJECTIVES: To describe the epidemiology of pediatric injury-related visits to children's hospital emergency departments (EDs) in the United States during early and later periods of the severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) pandemic. METHODS: We conducted a cross-sectional study using the Pediatric Health Information System, an administrative database to identify injury-related ED visits at 41 United States children's hospitals during the SARS-CoV-2 pandemic period (March 15, 2020 to March 14, 2021) and a 3 year comparator period (March 15-March 14, 2017-2020). For these 2 periods, we compared patient characteristics, injury type and severity, primary discharge diagnoses, and disposition, stratified by early (March 15, 2020 to June 30, 2020), middle (July 1, 2020 to October 31, 2020), and late (November 1, 2020 to March 14, 2021) pandemic periods. RESULTS: Overall, ED injury-related visits decreased by 26.6% during the first year of the SARS-CoV-2 pandemic, with the largest decline observed in minor injuries. ED injury-related visits resulting in serious-critical injuries increased across the pandemic (15.9% early, 4.9% middle, 20.6% late). Injury patterns with the sharpest relative declines included superficial injuries (41.7% early) and sprains/strains (62.4% early). Mechanisms of injury with the greatest relative increases included (1) firearms (22.9% early; 42.8% middle; 37% late), (2) pedal cyclists (60.4%; 24.9%; 32.2%), (3) other transportation (20.8%; 25.3%; 17.9%), and (4) suffocation/asphyxiation (21.4%; 20.2%; 28.4%) and injuries because of suicide intent (-16.2%, 19.9%, 21.8%). CONCLUSIONS: Pediatric injury-related ED visits declined in general. However, there was a relative increase in injuries with the highest severity, which warrants further investigation.

Diagnostic accuracy of point-of-care ultrasonography for intussusception in children: A systematic review and meta-analysis.

OBJECTIVES: Ileocolic intussusception can be challenging to diagnose due to vague complaints, but rapid diagnosis and treatment can help prevent morbidity and mortality. Prior research has focused on radiologic ultrasound, with more recent studies focusing on point-of-care ultrasonography (POCUS). This systematic review and meta-analysis assesses the diagnostic accuracy of POCUS for children with suspected ileocolic intussusception. METHODS: PubMed, Embase, CINAHL, LILACS, the Cochrane databases, Google Scholar, conference abstracts, and bibliographies of selected articles were searched for studies evaluating the accuracy of POCUS for the diagnosis of intussusception in children. Data were dual extracted into a predefined worksheet, and quality analysis was performed with the QUADAS-2 tool. Data were summarized, and a meta-analysis was performed. RESULTS: Eleven studies (n = 2400 children) met our inclusion criteria. Overall, 14.4% of children had intussusception. POCUS was 95.1% (95% CI: 90.3% to 97.2%) sensitive and 98.1% (95% CI: 95.8% to 99.2%) specific with a positive likelihood ratio of 50 (95% CI: 23 to 113) and a negative likelihood ratio of 0.05 (95% CI: 0.03 to 0.09). CONCLUSIONS: POCUS has excellent diagnostic accuracy for intussusception in children presenting to the emergency department.

Development of a Consensus-Based Definition of Focused Assessment With Sonography for Trauma in Children.

Importance: The wide variation in the accuracy and reliability of the Focused Assessment With Sonography for Trauma (FAST) and the extended FAST (E-FAST) for children after blunt abdominal trauma reflects user expertise. FAST and E-FAST that are performed by experts tend to be more complete, better quality, and more often clinically valuable. Objective: To develop definitions of a complete, high-quality, and accurate interpretation for the FAST and E-FAST in children with injury using an expert, consensus-based modified Delphi technique. Design, Setting, and Participants: This consensus-based qualitative study was conducted between May 1 to June 30, 2021. It used a scoping review and iterative Delphi technique and involved 2 rounds of online surveys and a live webinar to achieve consensus among a 26-member panel. This panel consisted of international experts in pediatric emergency point-of-care ultrasonography. Main Outcomes and Measures: Definitions of complete, high-quality, and accurate FAST and E-FAST studies for children after injury. Results: Of the 29 invited pediatric FAST experts, 26 (15 men [58%]) agreed to participate in the panel. All 26 panelists completed the 2 rounds of surveys, and 24 (92%) participated in the live and asynchronous online discussions. Consensus was reached on FAST and E-FAST study definitions, and the panelists rated these 5 anatomic views as important and appropriate for a complete FAST: right upper-quadrant abdominal view, left upper-quadrant abdominal view, suprapubic views (transverse and sagittal), and subxiphoid cardiac view. For E-FAST, the same FAST anatomic views with the addition of the lung or pneumothorax view were deemed appropriate and important. In addition, the panelists rated a total of 32 landmarks as important for assessing completeness. Similarly, the panelists rated 14 statements on quality and 20 statements on accurate interpretation as appropriate. Conclusions and Relevance: This qualitative study generated definitions for complete FAST and E-FAST studies with high image quality and accurate interpretation in children with injury. These definitions are similar to those in adults with injury and may be used for future education, quality assurance, and research. Future research may focus on interpretation of trace volumes of abdominal free fluid and the use of serial FAST.

Child Opportunity Index and Changes in Pediatric Acute Care Utilization in the COVID-19 Pandemic.

BACKGROUND: Pediatric acute care utilization decreased dramatically during the coronavirus disease 2019 (COVID-19) pandemic. This study examined the association between the Child Opportunity Index (COI), a multidimensional neighborhood measure of childhood opportunity, and changes in acute care utilization at US pediatric hospitals during the COVID-19 pandemic compared with the previous 3 years. METHODS: This observational study used administrative data across 41 US-based pediatric hospitals. Children aged 0 to 17 years with emergency department (ED) encounters during the study period were included. The COVID-19 pandemic time period (March 15, 2020-March 14, 2021) was the primary exposure. The primary outcome was the relative volume drop in ED encounters and observation/inpatient admissions through the ED by COI quintile. RESULTS: Of 12 138 750 encounters, 3 705 320 (30.5%) were among the very low COI quintile. Overall, there was a 46.8% relative volume reduction in the pandemic period compared with the prepandmic period. This drop in volume occurred disproportionately among the very low COI quintile (51.1%) compared with the very high COI quintile (42.8%). The majority of clinical diagnosis groups demonstrated larger relative volume drops among the very low COI quintile. CONCLUSIONS: Acute care utilization decreased the most among children from very low COI neighborhoods, narrowing previously described acute care utilization disparities. Additional study of patient perspectives on health care needs and access during this period is required to understand these changes.

SARS COV-2 anti-nucleocapsid and anti-spike antibodies in an emergency department healthcare worker cohort: September 2020 - April 2021.

BACKGROUND: Emergency department (ED) workers have an increased seroprevalence of SARS-CoV-2 antibodies. However, breakthrough infections in ED workers have led to a reduced workforce within a strained healthcare system. By measuring levels of IgG antibodies to the SARS-CoV-2 nucleocapsid and spike antigens in ED workers, we determined the incidence of infection and described the course of antibody levels. We also measured the antibody response to vaccination and examined factors associated with immunogenicity. METHODS: We conducted a prospective cohort study of ED workers conducted at a single ED from September 2020-April 2021. IgG antibodies to the SARS-CoV-2 nucleocapsid antigen were measured at baseline, 3, and 6 months, and IgG antibodies to the SARS-CoV-2 spike antigen were measured at 6 months. RESULTS: At baseline, we found 5 out of 139 (3.6%) participants with prior infection. At 6 months, 4 of the 5 had antibody results below the test manufacturer's positivity threshold. We identified one incident case of SARS-COV-2 infection out of 130 seronegative participants (0.8%, 95% CI 0.02-4.2%). In 131 vaccinated participants (125 BNT162b2, 6 mRNA-1273), 131 tested positive for anti-spike antibodies. We identified predictors of anti-spike antibody levels: time since vaccination, prior COVID-19 infection, age, and vaccine type. Each additional week since vaccination was associated with an 11.1% decrease in anti-spike antibody levels. (95% CI 6.2-15.8%). CONCLUSION: ED workers experienced a low incidence of SARS-CoV-2 infection and developed antibodies in response to vaccines and prior infection. Antibody levels decreased markedly with time since infection or vaccination.

Development and Validation of a Deep Learning Strategy for Automated View Classification of Pediatric Focused Assessment With Sonography for Trauma.

OBJECTIVE: Pediatric focused assessment with sonography for trauma (FAST) is a sequence of ultrasound views rapidly performed by clinicians to diagnose hemorrhage. A technical limitation of FAST is the lack of expertise to consistently acquire all required views. We sought to develop an accurate deep learning view classifier using a large heterogeneous dataset of clinician-performed pediatric FAST. METHODS: We developed and conducted a retrospective cohort analysis of a deep learning view classifier on real-world FAST studies performed on injured children less than 18 years old in two pediatric emergency departments by 30 different clinicians. FAST was randomly distributed to training, validation, and test datasets, 70:20:10; each child was represented in only one dataset. The primary outcome was view classifier accuracy for video clips and still frames. RESULTS: There were 699 FAST studies, representing 4925 video clips and 1,062,612 still frames, performed by 30 different clinicians. The overall classification accuracy was 97.8% (95% confidence interval [CI]: 96.0-99.0) for video clips and 93.4% (95% CI: 93.3-93.6) for still frames. Per view still frames were classified with an accuracy: 96.0% (95% CI: 95.9-96.1) cardiac, 99.8% (95% CI: 99.8-99.8) pleural, 95.2% (95% CI: 95.0-95.3) abdominal upper quadrants, and 95.9% (95% CI: 95.8-96.0) suprapubic. CONCLUSION: A deep learning classifier can accurately predict pediatric FAST views. Accurate view classification is important for quality assurance and feasibility of a multi-stage deep learning FAST model to enhance the evaluation of injured children.

The P2Network-Advancing Pediatric Emergency Care With Point-of-Care Ultrasound.

ABSTRACT: Over the last 2 decades, the use of point-of-care ultrasound (POCUS) in pediatric emergency medicine (PEM) has grown exponentially. In 2014, a group of PEM POCUS leaders met and formed the P2Network. The P2Network provides a platform to build collaborative relationships and share expertise among members from various countries and practice settings. It works with educators and researchers within and outside of the field to advance POCUS practice in PEM. As an organization, the P2Network promotes the evidence-based application of POCUS to facilitate and improve care in the PEM setting and addresses issues related to integration of the PEM POCUS practitioner in this nascent field. The P2Network is building and augmenting its infrastructure for PEM POCUS research and education and has already made some progress in the areas, with published manuscripts and ongoing clinical research studies under its sponsorship. Future goals include developing a PEM POCUS research agenda, formalizing teaching and assessment of PEM POCUS skills, and implementing multicenter research studies on potentially high impact applications.

Predictability and stability testing to assess clinical decision instrument performance for children after blunt torso trauma.

OBJECTIVE: The Pediatric Emergency Care Applied Research Network (PECARN) has developed a clinical-decision instrument (CDI) to identify children at very low risk of intra-abdominal injury. However, the CDI has not been externally validated. We sought to vet the PECARN CDI with the Predictability Computability Stability (PCS) data science framework, potentially increasing its chance of a successful external validation. MATERIALS & METHODS: We performed a secondary analysis of two prospectively collected datasets: PECARN (12,044 children from 20 emergency departments) and an independent external validation dataset from the Pediatric Surgical Research Collaborative (PedSRC; 2,188 children from 14 emergency departments). We used PCS to reanalyze the original PECARN CDI along with new interpretable PCS CDIs developed using the PECARN dataset. External validation was then measured on the PedSRC dataset. RESULTS: Three predictor variables (abdominal wall trauma, Glasgow Coma Scale Score <14, and abdominal tenderness) were found to be stable. A CDI using only these three variables would achieve lower sensitivity than the original PECARN CDI with seven variables on internal PECARN validation but achieve the same performance on external PedSRC validation (sensitivity 96.8% and specificity 44%). Using only these variables, we developed a PCS CDI which had a lower sensitivity than the original PECARN CDI on internal PECARN validation but performed the same on external PedSRC validation (sensitivity 96.8% and specificity 44%). CONCLUSION: The PCS data science framework vetted the PECARN CDI and its constituent predictor variables prior to external validation. We found that the 3 stable predictor variables represented all of the PECARN CDI's predictive performance on independent external validation. The PCS framework offers a less resource-intensive method than prospective validation to vet CDIs before external validation. We also found that the PECARN CDI will generalize well to new populations and should be prospectively externally validated. The PCS framework offers a potential strategy to increase the chance of a successful (costly) prospective validation.

Diagnostic Accuracy of Point-of-Care Ultrasound for Intussusception: A Multicenter, Noninferiority Study of Paired Diagnostic Tests.

STUDY OBJECTIVE: To determine the diagnostic accuracy of point-of-care ultrasound (POCUS) performed by experienced clinician sonologists compared to radiology-performed ultrasound (RADUS) for detection of clinically important intussusception, defined as intussusception requiring radiographic or surgical reduction. METHODS: We conducted a multicenter, noninferiority, observational study among a convenience sample of children aged 3 months to 6 years treated in tertiary care emergency departments across North and Central America, Europe, and Australia. The primary outcome was diagnostic accuracy of POCUS and RADUS with respect to clinically important intussusception. Sample size was determined using a 4-percentage-point noninferiority margin for the absolute difference in accuracy. Secondary outcomes included agreement between POCUS and RADUS for identification of secondary sonographic findings. RESULTS: The analysis included 256 children across 17 sites (35 sonologists). Of the 256 children, 58 (22.7%) had clinically important intussusception. POCUS identified 60 (23.4%) children with clinically important intussusception. The diagnostic accuracy of POCUS was 97.7% (95% confidence interval [CI] 94.9% to 99.0%), compared to 99.3% (95% CI 96.8% to 99.9%) for RADUS. The absolute difference between the accuracy of RADUS and that of POCUS was 1.5 percentage points (95% CI -0.6 to 3.6). Sensitivity for POCUS was 96.6% (95% CI 87.2% to 99.1%), and specificity was 98.0% (95% CI 94.7% to 99.2%). Agreement was high between POCUS and RADUS for identification of trapped free fluid (83.3%, n=40/48) and decreased color Doppler signal (95.7%, n=22/23). CONCLUSION: Our findings suggest that the diagnostic accuracy of POCUS performed by experienced clinician sonologists may be noninferior to that of RADUS for detection of clinically important intussusception. Given the limitations of convenience sampling and spectrum bias, a larger randomized controlled trial is warranted.

Diagnostic Applications of Point-of-Care Ultrasound in Pediatric Emergency Medicine.

Point-of-care ultrasound has become an essential part of pediatric emergency medicine training and practice. It can have significant clinical benefits, including improving diagnostic accuracy and decreasing length of stay, and does not require radiation exposure for patients. In this review, we summarize the current diagnostic point-of-care ultrasound applications in pediatric emergency medicine, their evidence, and techniques.

Evaluating a Web-based Point-of-care Ultrasound Curriculum for the Diagnosis of Intussusception.

OBJECTIVES: Intussusception is a pediatric medical emergency that can be difficult to diagnose. Radiology-performed ultrasound is the diagnostic study of choice but may lead to delays due to lack of availability. Point-of-care ultrasound for intussusception (POCUS-I) studies have shown excellent accuracy and reduced lengths of stay, but there are limited POCUS-I training materials for pediatric emergency medicine (PEM) providers. METHODS: We performed a prospective cohort study assessing PEM physicians undergoing a primarily Web-based POCUS-I curriculum. We developed the POCUS-I curriculum using Kern's six-step model. The curriculum included a Web-based module and a brief, hands-on practice that was developed with a board-certified pediatric radiologist. POCUS-I technical skill, knowledge, and confidence were determined by a direct observation checklist, multiple-choice test, and a self-reported Likert-scale survey, respectively. We assessed participants immediately pre- and postcourse as well as 3 months later to assess for retention of skill, knowledge, and confidence. RESULTS: A total of 17 of 17 eligible PEM physicians at a single institution participated in the study. For the direct observation skills test, participants scored well after the course with a median (interquartile range [IQR]) score of 20 of 22 (20-21) and maintained high scores even after 3 months (20 [20-21]). On the written knowledge test, there was significant improvement from 57.4% (95% CI = 49.8 to 65.2) to 75.3% (95% CI = 68.1 to 81.6; p < 0.001) and this improvement was maintained at 3 months at 81.2% (95% CI = 74.5 to 86.8). Physicians also demonstrated improved confidence with POCUS-I after exposure to the curriculum, with 5.9% reporting somewhat or very confident prior to the course to 76.5% both after the course and after 3 months (p < 0.001). CONCLUSION: After a primarily Web-based curriculum for POCUS-I, PEM physicians performed well in technical skill in POCUS-I and showed improvement in knowledge and confidence, all of which were maintained over 3 months.