Publications

OBJECTIVE: Linking emergency medical services (EMS) electronic patient care reports (ePCRs) to emergency department (ED) records can provide clinicians access to vital information that can alter management. It can also create rich databases for research and quality improvement. Unfortunately, previous attempts at ePCR and ED record linkage have had limited success. In this study, we use supervised machine learning to derive and validate an automated record linkage algorithm between EMS ePCRs and ED records.

MATERIALS AND METHODS: All consecutive ePCRs from a single EMS provider between June 2013 and June 2015 were included. A primary reviewer matched ePCRs to a list of ED patients to create a gold standard. Age, gender, last name, first name, social security number, and date of birth were extracted. Data were randomly split into 80% training and 20% test datasets. We derived missing indicators, identical indicators, edit distances, and percent differences. A multivariate logistic regression model was trained using 5-fold cross-validation, using label k-fold, L2 regularization, and class reweighting.

RESULTS: A total of 14 032 ePCRs were included in the study. Interrater reliability between the primary and secondary reviewer had a kappa of 0.9. The algorithm had a sensitivity of 99.4%, a positive predictive value of 99.9%, and an area under the receiver-operating characteristic curve of 0.99 in both the training and test datasets. Date-of-birth match had the highest odds ratio of 16.9, followed by last name match (10.6). Social security number match had an odds ratio of 3.8.

CONCLUSIONS: We were able to successfully derive and validate a record linkage algorithm from a single EMS ePCR provider to our hospital EMR.

INTRODUCTION: Transfers of skilled nursing facility (SNF) residents to emergency departments (ED) are linked to morbidity, mortality and significant cost, especially when transfers result in hospital admissions. This study investigated an alternative approach for emergency care delivery comprised of SNF-based telemedicine services provided by emergency physicians (EP). We compared this on-site emergency care option to traditional ED-based care, evaluating hospital admission rates following care by an EP.

METHODS: We conducted a retrospective, observational study of SNF residents who underwent emergency evaluation between January 1, 2017-January 1, 2018. The intervention group was comprised of residents at six urban SNFs in the Northeastern United States, who received an on-demand telemedicine service provided by an EP. The comparison group consisted of residents of SNFs that did not offer on-demand services and were transferred via ambulance to the ED. Using electronic health record data from both the telemedicine and ambulance transfers, our primary outcome was the odds ratio (OR) of a hospital admission. We also conducted a subanalysis examining the same OR for the three most common chronic disease-related presentations found among the telemedicine study population.

RESULTS: A total of 4,606 patients were evaluated in both the SNF-based intervention and ED-based comparison groups (n=2,311 for SNF based group and 2,295 controls). Patients who received the SNF-based acute care were less likely to be admitted to the hospital compared to patients who were transferred to the ED in our primary and subgroup analyses. Overall, only 27% of the intervention group was transported to the ED for additional care and presumed admission, whereas 71% of the comparison group was admitted (OR for admission = 0.15 [9% confidence interval, 0.13-0.17]).

CONCLUSION: The use of an EP-staffed telemedicine service provided to SNF residents was associated with a significantly lower rate of hospital admissions compared to the usual ED-based care for a similarly aged population of SNF residents. Providing SNF-based care by EPs could decrease costs associated with hospital-based care and risks associated with hospitalization, including cognitive and functional decline, nosocomial infections, and falls.

STUDY OBJECTIVE: Triage quickly identifies critically ill patients, facilitating timely interventions. Many emergency departments (EDs) use emergency severity index (ESI) or abnormal vital sign triggers to guide triage. However, both use fixed thresholds, and false activations are costly. Prior approaches using machinelearning have relied on information that is often unavailable during the triage process. We examined whether deep-learning approaches could identify critically ill patients only using data immediately available at triage.

METHODS: We conducted a retrospective, cross-sectional study at an urban tertiary care center, from January 1, 2012-January 1, 2020. De-identified triage information included structured (age, sex, initial vital signs) and textual (chief complaint) data, with critical illness (mortality or ICU admission within 24 hours) as the outcome. Four progressively complex deep-learning models were trained and applied to triage information from all patients. We compared the accuracy of the models against ESI as the standard diagnostic test, using area under the receiver-operator curve (AUC).

RESULTS: A total of 445,925 patients were included, with 60,901 (13.7%) critically ill. Vital sign thresholds identified critically ill patients with AUC 0.521 (95% confidence interval [CI] = 0.519-0.522), and ESI <3 demonstrated AUC 0.672 (95% CI = 0.671-0.674), logistic regression classified patients with AUC 0.803 (95% CI = 0.802-0.804), 2-layer neural network with structured data with AUC 0.811 (95% CI = 0.807-0.815), gradient tree boosting with AUC 0.820 (95% CI = 0.818-0.821), and the neural network model with textual data with AUC 0.851 (95% CI = 0.849-0.852). All successive increases in AUC were statistically significant.

CONCLUSION: Deep-learning techniques represent a promising method of augmenting triage, even with limited information. Further research is needed to determine if improved predictions yield clinical and operational benefits.

Medical documentation is one of the primary methods by which physicians share clinical information and impressions over time with one another. As the adage says, "a picture is worth a thousand words," and physicians have started leveraging consumer mobile technology to share images with one another. However, image sharing often uses short message service texting and similar methods, which can be noncompliant with privacy regulations and can also limit the ability to communicate information longitudinally and across specialties. Sharing of such information is increasingly important, however, as smaller practices are joining to create large geographically spread out health care networks. To this end, we developed an application to acquire and store images via smartphone and seamlessly transfer into the patient's electronic medical record (EMR) to enable digital consults and longitudinal evaluation in a private and compliant method.

BACKGROUND: Studies have shown the value of CT brain imaging in adults with first-time seizures, but there are no recommendations regarding emergent brain CTs in persons with an established seizure disorders. Our study aimed to derive a clinical decision instrument (CDI) to determine which patients with status epilepticus (SE) require emergent brain imaging.

METHODS: This was a retrospective chart review of patients who presented to our emergency department with SE between 2010 and 2018. Patients with first-time seizures were excluded. A priori, we defined high risk criteria for emergent imaging as well as positive findings on brain CT. High risk criteria included known malignancy, trauma, and immunosuppression. Positive CT scans included findings such as intracranial hemorrhage (ICH) and mass.

RESULTS: We identified 214 patients who met inclusion criteria Of the 181 patients without high risk criteria, 3% had positive CT scans. Of the 33 patients with high risk criteria, 10% had positive CT scans. The sensitivity, specificity, PPV, and NPV for our initial CDI were 38%, 85%, 9%, and 97%. Adding the criterion of prior ICH would have lowered our miss rate to 0.6%. Modifying our CDI to 1) History of ICH, 2) Malignancy, 3) Immunosuppression, and 4) Trauma would result in a CDI with sensitivity, specificity, PPV, and NPV of 87.5%, 87.4%, 21.2%, and 99.5%.

CONCLUSIONS: By using four criteria to identify high risk patients, we can defer CT scanning in the vast majority of patients with SE and known seizure disorders. This CDI should be prospectively validated before adoption.

BACKGROUND: Previous work has suggested low rates of post-intubation sedation in patients undergoing endotracheal intubation (ETI) in the emergency department (ED) with limited data examining factors associated with sedation use. Utilizing a national database; we sought to determine the frequency of post-intubation sedation and associated factors.

METHODS: We performed a retrospective analysis of a prospectively collected database (National Emergency Airway Registry (NEAR) from 25 EDs from January 1, 2016 to December 31, 2017). Patients were considered to have received post-intubation sedation if they received any of the following medications within 15 min of ETI completion; propofol, midazolam, diazepam, ketamine, etomidate, fentanyl, and morphine. We calculated odds ratios for post-intubation sedation.

RESULTS: Of the 11,748 eligible intubations, 9099 received post-intubation sedation (77.5%) while 2649 did not (22.5%). Pre-intubation hypotension (odds ratio; 95% confidence Interval) (0.27; 0.24-0.31) and post-intubation hypotension (0.27; 0.24-0.31) were associated with lower odds of post-intubation sedation. Patients with a medical indication compared to a traumatic indication for ETI had higher odds of receiving post-intubation sedation (1.16; 1.05-1.28) as did those that underwent rapid sequence intubation (15.15; 13.56-16.93). Use of succinylcholine was associated with a higher odd of post-intubation sedation compared to a long-acting neuromuscular blocking agent (i.e. rocuronium or vecuronium) (1.89; 1.68-2.12).

CONCLUSION: Post-intubation sedation rates in NEAR are higher than previously reported and multiple factors including the indication for intubation and succinylcholine use, are associated with higher odds of receiving post-intubation sedation.

OBJECTIVES: With respect to first-attempt intubation success, the pediatric literature demonstrates either clinical equipoise or superiority of direct laryngoscopy (DL) when compared to video laryngoscopy (VL). Furthermore, it is unknown how VL compares to DL, when DL is "augmented" by maneuvers, such as optimal external laryngeal manipulation (OELM), upright or ramped positioning, or the use of the bougie. The objective of our study was to compare first-attempt success between VL and all DL, including "augmented DL" for pediatric intubations.

METHODS: We analyzed the National Emergency Airway Registry database of intubations of patients < 16 years. Variables collected included patient demographics, body habitus, impression of airway difficulty, intubating position, reduced neck mobility, airway characteristics, device, medications, and operator characteristics, adjusted for clustering by center. Primary outcome was the difference in first-attempt success for DL and augmented DL versus VL. Secondary outcomes included adverse events. In a planned sensitivity analysis, a propensity-adjusted analysis for first-attempt success and a subgroup analysis of children < 2 years was also performed.

RESULTS: Of 625 analyzable pediatric encounters, 294 (47.0%, 95% confidence interval [CI] = 25.1% to 69.0%) were DL; 332 (53.1%, 95% CI = 31.0% to 74.9%) were VL. Median age was 4 years (interquartile range = 1 to 10 years); 225 (36.0%, 95% CI = 30.8% to 41.2%) were < 2 years. Overall first-pass success was 79.6% (95% CI = 74.1% to 84.9%). VL first-pass success was 278/331 (84.0%) versus 219/294 for DL (74.5%), adjusted for clustering (odds ratio [OR] = 1.7, 95% CI = 1.3 to 2.5). Multivariable regression showed that VL yielded a higher odds of first-attempt success than DL augmented by OELM or use of a bougie (adjusted OR = 5.5, 95% CI = 1.7 to 18.1). Propensity-adjusted analyses supported the main results. Subgroup analysis of age < 2 years also demonstrated VL superiority (OR = 2.0, 95% CI = 1.1 to 3.3) compared with DL. Adverse events were comparable in both univariate and multivariable analysis.

CONCLUSIONS: When compared to DL, VL was associated with higher first-pass success in this pediatric population, even in the subgroup of patients < 2 years, as well as when DL was augmented. There were no differences in adverse effects between DL and VL.

OBJECTIVE: The objective was to compare first-attempt intubation success using direct laryngoscopy augmented by laryngeal manipulation, ramped patient positioning, and use of a bougie (A-DL) with unaided video laryngoscopy (VL) in adult emergency department (ED) intubations.

METHODS: This study was a secondary analysis of a multicenter prospective observational database of ED intubations from the National Emergency Airway Registry (NEAR). We compared all VL procedures to seven exploratory permutations of A-DL using multivariable regression models. We further stratified by blade shape into hyperangulated VL (HA-VL) and standard-geometry VL (SG-VL). We report differences in first-attempt intubation success and peri-intubation adverse events with cluster-adjusted odds ratios (ORs) with 95% confidence intervals (CIs). We report univariate comparisons in patient characteristics, difficult airway attributes, and intubation methods using descriptive statistics and OR with 95% CI.

RESULTS: We analyzed 11,714 intubations performed from January 1, 2016, through December 31, 2017. Of these encounters, 6,938 underwent orotracheal intubation with either A-DL or unaided VL on first attempt. A-DL was used first in 3,936 (56.7%, 95% CI = 46.9 to 66.5) versus unaided VL in 3,002 (43.3%, 95% CI = 33.5 to 53.1). Of the A-DL first intubations 1,787 (45.4%) employed ramped positioning alone, 1,472 (37.4%) had external laryngeal manipulation (ELM), and 365 (9.3%) used a bougie. Rapid sequence intubation (RSI) was the most common method used in 5,602 (80.8%, 95% CI = 77.0 to 84.5) cases. First-attempt success was significantly higher with all VL (90.9%, 95% CI = 88.7 to 93.1) versus all A-DL (81.1%, 95% CI = 78.7 to 83.5) despite the VL group having more patients with reduced mouth opening, neck immobility, and an initial impression of airway difficult. Multivariable regression analyses controlling for indication, method, operator specialty and year of training, center clustering, and all registry-recorded difficult airway predictors revealed first-attempt success was higher with all unaided VL compared with any A-DL (adjusted OR [AOR] = 2.8, 95% CI = 2.4 to 3.3), DL with bougie (AOR = 2.7, 95% CI = 2.1 to 3.5), DL with ELM (AOR = 1.8, 95% CI = 1.5 to 2.2), DL with ramped positioning (AOR = 2.8, 95% CI = 2.3 to 3.3), or DL with ELM plus bougie (AOR = 2.8, 95% CI = 2.3 to 3.3). Subgroup analyses of HA-VL and SG-VL compared with any A-DL yielded similar results (AOR = 3.2, 95% CI = 2.6 to 3.0; and AOR = 2.4, 95% CI = 1.9 to 3.0, respectively). The propensity score-adjusted odds for first-attempt success with VL was also 2.8 (95% CI = 2.4 to 3.3). Fewer esophageal intubations were observed in the VL cohort (0.4% vs. 1.3%, AOR = 0.2, 95% CI = 0.1 to 0.5).

CONCLUSIONS: Video laryngoscopy used without any augmenting maneuver, device, or technique results in higher first-attempt success than does DL that is augmented by use of a bougie, ELM, ramping, or combinations thereof.

Tracheal intubation is commonly performed in critically ill patients. Unfortunately, this procedure also carries a high risk of complications; half of critically ill patients with difficult airways experience life-threatening complications. The high complication rates stem from difficulty with laryngoscopy and tube placement, consequences of physiologic derangement, and human factors, including failure to recognize and reluctance to manage the failed airway. The last 10 years have seen a rapid expansion in devices available that help overcome anatomic difficulties with laryngoscopy and provide rescue oxygenation in the setting of failed attempts. Recent research in critically ill patients has highlighted other important considerations for critically ill patients and evaluated interventions to reduce the risks with repeated attempts, desaturation, and cardiovascular collapse during emergency airway management. There are three actions that should be implemented to reduce the risk of danger: 1) preintubation assessment for potential difficulty (e.g., MACOCHA score); 2) preparation and optimization of the patient and team for difficulty-including using a checklist, acquiring necessary equipment, maximizing preoxygenation, and hemodynamic optimization; and 3) recognition and management of failure to restore oxygenation and reduce the risk of cardiopulmonary arrest. This review describes the history of emergency airway management and explores the challenges with modern emergency airway management in critically ill patients. We offer clinically relevant recommendations on the basis of current evidence, guidelines, and expert opinion.

STUDY OBJECTIVE: It is unclear whether laryngoscopy using a standard-geometry blade shape, able to obtain both direct and indirect views, is associated with different first-attempt success or adverse events during emergency intubation compared with using a hyperangulated blade capable of indirect laryngoscopy only. We sought to compare first-attempt intubation success between patients intubated with a standard geometry video laryngoscope versus a hyperangulated video laryngoscope.

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 attempt at oral intubation was performed with a standard-geometry or hyperangulated video laryngoscope. We used multiple logistic regression to determine whether blade shape was independently associated with first-attempt intubation success.

RESULTS: During the study period, 11,927 of 19,071 intubation encounters met inclusion criteria, including 7,255 (61%) with a standard blade and 4,672 (39%) with a hyperangulated blade. Unadjusted analysis revealed higher success with a standard-geometry blade, 91.9% versus 89.2% (absolute difference 2.7% [95% confidence interval 1.6% to 3.8%]; odds ratio for standard-geometry laryngoscope compared with hyperangulated laryngoscope 1.37 [95% confidence interval 1.21 to 1.55]). The logistic regression model, however, demonstrated no association between blade shape and first-attempt success (adjusted odds ratio for standard-geometry laryngoscopy compared with hyperangulated laryngoscopy 1.32 [95% confidence interval 0.81 to 2.17]).

CONCLUSION: In this large registry of patients intubated with video laryngoscopy in the emergency department, we observed no association between blade shape (standard-geometry versus hyperangulated laryngoscope) and first-attempt intubation success after adjusting for confounding variables.