Publications

Space travel has transformed in the past several years. Given the burgeoning market for space tourism, in-flight medical emergencies are likely to be expected. Ultrasound is one of the few diagnostic and therapeutic modalities available for astronauts in space. However, while point-of-care ultrasound (POCUS) is available, there is no current standard of training for astronaut preparation. We suggest an organized and structured methodology by which astronauts should best prepare for space with the medical equipment available on board. As technology continues to evolve, the assistance of other artificial intelligence and augmented reality systems are likely to facilitate training and dynamic real-time needs during space emergencies. Summary: As space tourism continues to evolve, an organized methodology for POCUS use is advised to best prepare astronauts for space.

BACKGROUND: Machine learning (ML) models can yield faster and more accurate medical diagnoses; however, developing ML models is limited by a lack of high-quality labeled training data. Crowdsourced labeling is a potential solution but can be constrained by concerns about label quality.

OBJECTIVE: This study aims to examine whether a gamified crowdsourcing platform with continuous performance assessment, user feedback, and performance-based incentives could produce expert-quality labels on medical imaging data.

METHODS: In this diagnostic comparison study, 2384 lung ultrasound clips were retrospectively collected from 203 emergency department patients. A total of 6 lung ultrasound experts classified 393 of these clips as having no B-lines, one or more discrete B-lines, or confluent B-lines to create 2 sets of reference standard data sets (195 training clips and 198 test clips). Sets were respectively used to (1) train users on a gamified crowdsourcing platform and (2) compare the concordance of the resulting crowd labels to the concordance of individual experts to reference standards. Crowd opinions were sourced from DiagnosUs (Centaur Labs) iOS app users over 8 days, filtered based on past performance, aggregated using majority rule, and analyzed for label concordance compared with a hold-out test set of expert-labeled clips. The primary outcome was comparing the labeling concordance of collated crowd opinions to trained experts in classifying B-lines on lung ultrasound clips.

RESULTS: Our clinical data set included patients with a mean age of 60.0 (SD 19.0) years; 105 (51.7%) patients were female and 114 (56.1%) patients were White. Over the 195 training clips, the expert-consensus label distribution was 114 (58%) no B-lines, 56 (29%) discrete B-lines, and 25 (13%) confluent B-lines. Over the 198 test clips, expert-consensus label distribution was 138 (70%) no B-lines, 36 (18%) discrete B-lines, and 24 (12%) confluent B-lines. In total, 99,238 opinions were collected from 426 unique users. On a test set of 198 clips, the mean labeling concordance of individual experts relative to the reference standard was 85.0% (SE 2.0), compared with 87.9% crowdsourced label concordance (P=.15). When individual experts' opinions were compared with reference standard labels created by majority vote excluding their own opinion, crowd concordance was higher than the mean concordance of individual experts to reference standards (87.4% vs 80.8%, SE 1.6 for expert concordance; P<.001). Clips with discrete B-lines had the most disagreement from both the crowd consensus and individual experts with the expert consensus. Using randomly sampled subsets of crowd opinions, 7 quality-filtered opinions were sufficient to achieve near the maximum crowd concordance.

CONCLUSIONS: Crowdsourced labels for B-line classification on lung ultrasound clips via a gamified approach achieved expert-level accuracy. This suggests a strategic role for gamified crowdsourcing in efficiently generating labeled image data sets for training ML systems.

BACKGROUND: Opioid prescribing patterns, including those after surgery, have been implicated as a significant contributor to the US opioid crisis. A plethora of interventions-from nudges to reminders-have been deployed to improve prescribing behavior, but reasons for persistent outlier behavior are often unknown.

STUDY DESIGN: Our institution employs multiple prescribing resources and a near real-time, feedback-based intervention to promote appropriate opioid prescribing. Since 2019, an automated system has emailed providers when a prescription exceeds the 75th percentile of typical opioid consumption for a given procedure-as defined by institutional data collection. Emails include population consumption metrics and an optional survey on rationale for prescribing. Responses were analyzed to understand why providers choose to prescribe atypically large discharge opioid prescriptions. We then compared provider prescriptions against patient consumption.

RESULTS: During the study period, 10,672 eligible postsurgical patients were discharged; 2,013 prescriptions (29.4% of opioid prescriptions) exceeded our institutional guideline. Surveys were completed by outlier prescribers for 414 (20.6%) encounters. Among patients where both consumption data and prescribing rationale surveys were available, 35.2% did not consume any opioids after discharge and 21.5% consumed <50% of their prescription. Only 93 (39.9%) patients receiving outlier prescriptions were outlier consumers. Most common reasons for prescribing outlier amounts were attending preference (34%) and prescriber analysis of patient characteristics (34%).

CONCLUSIONS: The top quartile of opioid prescriptions did not align with, and often far exceeded, patient postdischarge opioid consumption. Providers cite assessment of patient characteristics as a common driver of decision-making, but this did not align with patient usage for approximately 50% of patients.

BACKGROUND: Work Relative Value Units (wRVUs) are a component of many compensation models, and a proxy for the effort required to care for a patient. Accurate prediction of wRVUs generated per patient at triage could facilitate real-time load balancing between physicians and provide many practical operational and clinical benefits.

OBJECTIVE: We examined whether deep-learning approaches could predict the wRVUs generated by a patient's visit using data commonly available at triage.

METHODS: Adult patients presenting to an urban, academic emergency department from July 1, 2016-March 1, 2020 were included. Deidentified triage information included structured data (age, sex, vital signs, Emergency Severity Index score, language, race, standardized chief complaint) and unstructured data (free-text chief complaint) with wRVUs as outcome. Five models were examined: average wRVUs per chief complaint, linear regression, neural network and gradient-boosted tree on structured data, and neural network on unstructured textual data. Models were evaluated using mean absolute error.

RESULTS: We analyzed 204,064 visits between July 1, 2016 and March 1, 2020. The median wRVUs were 3.80 (interquartile range 2.56-4.21), with significant effects of age, gender, and race. Models demonstrated lower error as complexity increased. Predictions using averages from chief complaints alone demonstrated a mean error of 2.17 predicted wRVUs per visit (95% confidence interval [CI] 2.07-2.27), the linear regression model: 1.00 wRVUs (95% CI 0.97-1.04), gradient-boosted tree: 0.85 wRVUs (95% CI 0.84-0.86), neural network with structured data: 0.86 wRVUs (95% CI 0.85-0.87), and neural network with unstructured data: 0.78 wRVUs (95% CI 0.76-0.80).

CONCLUSIONS: Chief complaints are a poor predictor of the effort needed to evaluate a patient; however, deep-learning techniques show promise. These algorithms have the potential to provide many practical applications, including balancing workloads and compensation between emergency physicians, quantify crowding and mobilizing resources, and reducing bias in the triage process.

Yield-stress materials, which require a sufficiently large forcing to flow, are currently ill-understood theoretically. To gain insight into their yielding transition, we study numerically the rheology of a suspension of deformable droplets in 2D. We show that the suspension displays yield-stress behavior, with droplets remaining motionless below a critical body-force. In this phase, droplets jam to form an amorphous structure, whereas they order in the flowing phase. Yielding is linked to a percolation transition in the contacts of droplet-droplet overlaps and requires strict conservation of the droplet area to exist. Close to the transition, we find strong oscillations in the droplet motion that resemble those found experimentally in confined colloidal glasses. We show that even when droplets are static, the underlying solvent moves by permeation so that the viscosity of the composite system is never truly infinite, and its value ceases to be a bulk material property of the system.

STUDY OBJECTIVE: We compare intubation first-attempt success with the direct laryngoscope, hyperangulated video laryngoscope, and standard geometry video laryngoscope among emergency medicine residents at various postgraduate years (PGY) of training.

METHODS: We analyzed prospective data from emergency department (ED) patients enrolled in the National Emergency Airway Registry from January 1, 2016 to December 31, 2018 using mixed-effects logistic regression to assess the association between PGY of training and first-attempt success by the device.

RESULTS: Among 15,204 intubations performed by emergency medicine trainees, first-attempt success for PGY-1, PGY-2, and PGY3+ residents, respectively were: 78.8% (95% CI, 75.0 to 82.2%), 81.3% (79.4 to 83.0), and 83.6% (95% CI, 82.1 to 85.1) for direct laryngoscope; 87.2% (95% CI, 84.2 to 89.7), 90.4% (95% CI, 88.8 to 91.9%), and 91.2% (95% CI, 89.8 to 92.5%) for hyperangulated video laryngoscope; and 88.7% (95% CI, 86.1 to 90.9), 90.2% (95% CI, 88.7 to 91.5%), and 94.6% (95% CI 93.9 to 95.3%) for standard geometry video laryngoscope. Direct laryngoscope first-attempt success improved for PGY-2 (adjusted odds ratio [aOR],1.41; 95% CI, 1.09 to 1.82) and PGY-3+ (aOR, 1.76; 1.36 to 2.27) trainees compared to PGY-1. Hyperangulated video laryngoscope success also improved for PGY-2 (aOR, 1.51; 1.1 to 2.05) and PGY-3+ (aOR, 1.56; 1.15 to 2.13) trainees compared to PGY-1. For the standard geometry video laryngoscope, only PGY-3+ (aOR, 1.72; 1.25 to 2.36) was associated with improved first-attempt success compared to PGY-1.

CONCLUSION: Each laryngoscopy device class was associated with improvement in first-attempt success as training progressed. The video laryngoscope outperformed the direct laryngoscope for all operator groups, and PGY-1 trainees achieved higher first-attempt success using a standard geometry video laryngoscope than PGY-3+ trainees using a direct laryngoscope. These findings support the conjecture that in adult patients, a direct laryngoscope should not be routinely used for the first intubation attempt unless clinical circumstances, such as the presence of a soiled airway, would favor its success. These findings need to be validated with prospective randomized clinical trials.

STUDY OBJECTIVE: For patients with hemodynamic instability undergoing rapid sequence intubation, experts recommend reducing the sedative medication dose to minimize the risk of further hemodynamic deterioration. Scant data support this practice for etomidate and ketamine. We sought to determine if the dose of etomidate or ketamine was independently associated with postintubation hypotension.

METHODS: We analyzed data from the National Emergency Airway Registry from January 2016 to December 2018. Patients aged 14 years or older were included if the first intubation attempt was facilitated with etomidate or ketamine. We used multivariable modeling to determine whether drug dose in milligrams per kilogram of patient weight was independently associated with postintubation hypotension (systolic blood pressure < 100 mm Hg).

RESULTS: We analyzed 12,175 intubation encounters facilitated by etomidate and 1,849 facilitated by ketamine. The median drug doses were 0.28 mg/kg (interquartile range [IQR] 0.22 mg/kg to 0.32 mg/kg) for etomidate and 1.33 mg/kg (IQR 1 mg/kg to 1.8 mg/kg) for ketamine. Postintubation hypotension occurred in 1,976 patients (16.2%) who received etomidate and in 537 patients (29.0%) who received ketamine. In multivariable models, neither the etomidate dose (adjusted odds ratio [aOR] 0.95, 95% confidence interval [CI] 0.90 to 1.01) nor ketamine dose (aOR 0.97, 95% CI 0.81 to 1.17) was associated with postintubation hypotension. Results were similar in sensitivity analyses excluding patients with preintubation hypotension and including only patients intubated for shock.

CONCLUSION: In this large registry of patients intubated after receiving either etomidate or ketamine, we observed no association between the weight-based sedative dose and postintubation hypotension.

BACKGROUND: Cricothyrotomy is a critical technique for rescue of the failed airway in the emergency department (ED). Since the adoption of video laryngoscopy, the incidence of rescue surgical airways (those performed after at least one unsuccessful orotracheal or nasotracheal intubation attempt), and the circumstances where they are attempted, has not been characterized.

OBJECTIVE: We report the incidence and indications for rescue surgical airways using a multicenter observational registry.

METHODS: We performed a retrospective analysis of rescue surgical airways in subjects ≥14 years of age. We describe patient, clinician, airway management, and outcome variables.

RESULTS: Of 19,071 subjects in NEAR, 17,720 (92.9%) were ≥14 years old with at least one initial orotracheal or nasotracheal intubation attempt, 49 received a rescue surgical airway attempt, an incidence of 2.8 cases per 1000 (0.28% [95% confidence interval 0.21 to 0.37]). The median number of airway attempts prior to rescue surgical airways was 2 (interquartile range 1, 2). Twenty-five were in trauma victims (51.0% [36.5 to 65.4]), with neck trauma being the most common traumatic indication (n = 7, 14.3% [6.4 to 27.9]).

CONCLUSION: Rescue surgical airways occurred infrequently in the ED (0.28% [0.21 to 0.37]), with approximately half performed due to a trauma indication. These results may have implications for surgical airway skill acquisition, maintenance, and experience.

INTRODUCTION: Airway management is a critical component of the management of emergency department (ED) patients. The ED airway literature primarily focuses upon endotracheal intubation; relatively less is known about the ED use of extraglottic devices (EGDs). The goal of this study was to describe the frequency of use, success, and complications for EGDs among ED patients.

METHODS: The National Emergency Airway Registry (NEAR) is a prospective, multi-center, observational registry. It captures data on all ED patients at participating sites requiring airway management. Intubating clinicians entered all data into an online system as soon as practical after each encounter. We conducted a secondary analysis of these data for all ED encounters in which EGD placement occurred. We used descriptive statistics to characterize these encounters.

RESULTS: Of 19,071 patients undergoing intubation attempts, 56 (0.3%) underwent EGD placement. Of 25 participating sites, 13 reported no cases undergoing EGD placement; the median number of EGDs placed per site was 2 (interquartile range 1-2.5, range 1-31). Twenty-nine (54%) patients had either hypotension or hypoxia prior to the start of airway management. Clinicians reported anticipation of a difficult airway in 55% and at least one difficult airway characteristic in 93% of these patients. Forty-one encounters entailed placement of a laryngeal mask airway (LMA®) Fastrach™, 33 of whom underwent subsequent successful intubation through the EGD and 7 of whom underwent intubation by alternative methods. An additional 10 encounters utilized a standard LMA® device. Providers placed 34 (61%) EGDs during the first intubation attempt. Seventeen EGD patients (30%) experienced peri-procedure adverse events, including 14 (25%) experiencing hypoxemia. None of these patients expired due to failed airways.

CONCLUSIONS: EGD use was rare in this multi-center ED registry. EGD occurred predominantly in patients with difficult airway characteristics with favorable airway management outcomes. Clinicians should consider this emergency airway device for patients with a suspected difficult airway.