Abstract 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.
Abstract Objectives Despite decades of literature recognizing racial disparities (RDs) in emergency medicine (EM), published curricula dedicated to addressing them are sparse. We present details of our novel RD curriculum for EM clerkships and its educational outcomes. Methods We created a 30‐min interactive didactic module on the topic designed for third‐ and fourth‐year medical students enrolled in our EM clerkships. Through a modified Delphi process, education faculty and content experts in RD developed a 10‐question multiple‐choice test of knowledge on RD that the students completed immediately prior to and 2 weeks following the activity. Students also completed a Likert‐style learner satisfaction survey. Median pre‐ and posttest scores were compared using a paired Wilcoxon signed‐rank test and presented using medians and 95% confidence intervals (CIs). Satisfaction survey responses were dichotomized into favorable and neutral/not favorable. Results For the 36 students who completed the module, the median pretest score was 40% (95% CI 36%–50%) and the posttest score was 70% (95% CI 60%–70%) with a p‐value of <0.001. Thirty‐five of the 36 students improved on the posttest with a mean increase of 24.2% (95% CI 20.2–28.2). The satisfaction survey also showed a positive response, with at least 83% of participants responding favorably to all statements (overall mean favorable response 93%, 95% CI 90%–96%). Conclusions This EM‐based module on RD led to improvement in students' knowledge on the topic and positive reception by participants. This is a feasible option for educating students in EM on the topic of RD.
Introduction: Chest pain is a common emergency department (ED) presentation accounting for 8-10million visits per year in the United States. Physician-level factors such as risk tolerance are predictive ofadmission rates. The recent advent of accelerated diagnostic pathways and ED observation units mayhave an impact in reducing variation in admission rates on the individual physician level.Methods: We conducted a single-institution retrospective observational study of ED patients with adiagnosis of chest pain as determined by diagnostic code from our hospital administrative database.We included ED visits from 2012 and 2013. Patients with an elevated troponin or an electrocardiogram(ECG) demonstrating an ST elevation myocardial infarction were excluded. Patients were divided into twogroups: "admission" (this included observation and inpatients) and "discharged." We stratified physiciansby age, gender, residency location, and years since medical school. We controlled for patient- andhospital-related factors including age, gender, race, insurance status, daily ED volume, and lab values.Results: Of 4,577 patients with documented dispositions, 3,252 (70.9%) were either admitted to thehospital or into observation (in an ED observation unit or in the hospital), while 1,333 (29.1%) weredischarged. Median number of patients per physician was 132 (interquartile range 89-172). Averageadmission rate was 73.7±9.5% ranging from 54% to 96%. Of the 3,252 admissions, 2,638 (81.1%) wereto observation. There was significant variation in the admission rate at the individual physician level withadjusted odds ratio ranging from 0.42 to 5.8 as compared to the average admission. Among physicians'characteristics, years elapsed since finishing medical school demonstrated a trend towards associationwith a higher admission probability.Conclusion: There is substantial variation among physicians in the management of patients presentingwith chest pain, with physician experience playing a role.
The present study examined current and lifetime psychiatric morbidity, chest pain, and health care utilization in 229 patients with noncardiac chest pain (NCCP), angina-like pain in the absence of cardiac etiology. Diagnostic interview findings based on the Diagnostic and Statistical Manual of Mental Disorders (4th ed.; DSM-IV; American Psychiatric Association, 1994) revealed a psychiatrically heterogeneous sample of whom 44% had a current Axis I psychiatric disorder. A total of 41% were diagnosed with a current anxiety disorder, and 13% were diagnosed with a mood disorder. Overall, 75% of patients had an Axis I clinical or subclinical disorder. Lifetime diagnoses of anxiety (55%) and mood disorders (44%) were also prevalent, including major depressive disorder (41%), social phobia (25%), and panic disorder (22%). Patients with an Axis I disorder reported more frequent and more painful chest pain compared with those without an Axis I disorder. Presence of an Axis I disorder was associated with increased life interference and health care utilization. Findings reveal that varied DSM-IV Axis I psychiatric disorders are prevalent among patients with NCCP, and this psychiatric morbidity is associated with a less favorable NCCP presentation. Implications for early identification of psychiatric disorders are discussed.
