Altered mental status (AMS) is frequently associated with poor outcomes across a wide spectrum of conditions including infections. This study aims to identify whether AMS in emergency department (ED) patients with COVID-19 is independently associated with in-hospital mortality. This was a retrospective multicenter cohort study. We included all patients with a positive SARS-CoV-2 PCR within 2 weeks of presentation, who were admitted from the ED of three hospitals in the greater Boston area between March and August 2020. The primary covariate of interest was ED AMS at ED arrival and the primary outcome was in-hospital mortality. The ED charts were abstracted for demographics, comorbid conditions, symptoms, laboratory testing, and radiology testing along with in-hospital outcomes. AMS was defined by documentation of changes in mental status from baseline. We used logistic regression modeling with backwards elimination to determine an adjusted estimate for the independent association of AMS with mortality. We included 824 visits with 51% male, a mean age was 67.1 (SD 17.0) and 153 (18.6%) had AMS. There were 132 deaths for an overall mortality rate of 16.1%. Patients with AMS had in-hospital mortality of 38.2% (95% CI 30.4%-46.4%), compared to 11.1% (8.8%-13.7%) for patients without AMS (p<0.0001). After adjusting for potential confounders, visits by patients with AMS during their stay at the ED had 3.1 (95% CI, 2.1-5.9) times the odds of death compared to those without AMS. Among patients with COVID-19, AMS in the ED was associated with three-fold increase in mortality compared to patients without AMS.

Patients frequently visit the emergency department with conditions that place them at risk of worse outcomes when accompanied by coagulopathy. Routine tests of coagulation—prothrombin time, partial thromboplastin time, platelets, and fibrinogen—have shortcomings that limit their use in providing emergency care. One alternative is to investigate coagulation disturbance with viscoelastic monitoring (VEM), a coagulation test that measures the timing and strength of blood clot development in real time. VEM is widely used and studied in cardiac surgery, liver transplant surgery, anesthesia, and trauma. In this article, we review the technique of VEM and the biologic rationale of using it in addition to routine tests of coagulation in emergency clinical situations. Then, we review the evidence (or lack thereof) for using VEM in the diagnosis and treatment of specific conditions. Finally, we describe the limitations of the test and future directions for clinical use and research in emergency medicine. Patients frequently visit the emergency department with conditions that place them at risk of worse outcomes when accompanied by coagulopathy. Routine tests of coagulation—prothrombin time, partial thromboplastin time, platelets, and fibrinogen—have shortcomings that limit their use in providing emergency care. One alternative is to investigate coagulation disturbance with viscoelastic monitoring (VEM), a coagulation test that measures the timing and strength of blood clot development in real time. VEM is widely used and studied in cardiac surgery, liver transplant surgery, anesthesia, and trauma. In this article, we review the technique of VEM and the biologic rationale of using it in addition to routine tests of coagulation in emergency clinical situations. Then, we review the evidence (or lack thereof) for using VEM in the diagnosis and treatment of specific conditions. Finally, we describe the limitations of the test and future directions for clinical use and research in emergency medicine.

Patients frequently visit the emergency department with conditions that place them at risk of worse outcomes when accompanied by coagulopathy. Routine tests of coagulation-prothrombin time, partial thromboplastin time, platelets, and fibrinogen-have shortcomings that limit their use in providing emergency care. One alternative is to investigate coagulation disturbance with viscoelastic monitoring (VEM), a coagulation test that measures the timing and strength of blood clot development in real time. VEM is widely used and studied in cardiac surgery, liver transplant surgery, anesthesia, and trauma. In this article, we review the technique of VEM and the biologic rationale of using it in addition to routine tests of coagulation in emergency clinical situations. Then, we review the evidence (or lack thereof) for using VEM in the diagnosis and treatment of specific conditions. Finally, we describe the limitations of the test and future directions for clinical use and research in emergency medicine.

A crucial step within secondary analysis of electronic health records (EHRs) is to identify the patient cohort under investigation. While EHRs contain medical billing codes that aim to represent the conditions and treatments patients may have, much of the information is only present in the patient notes. Therefore, it is critical to develop robust algorithms to infer patients' conditions and treatments from their written notes. In this paper, we introduce a dataset for patient phenotyping, a task that is defined as the identification of whether a patient has a given medical condition (also referred to as clinical indication or phenotype) based on their patient note. Nursing Progress Notes and Discharge Summaries from the Intensive Care Unit of a large tertiary care hospital were manually annotated for the presence of several high-context phenotypes relevant to treatment and risk of re-hospitalization. This dataset contains 1102 Discharge Summaries and 1000 Nursing Progress Notes. Each Discharge Summary and Progress Note has been annotated by at least two expert human annotators (one clinical researcher and one resident physician). Annotated phenotypes include treatment non-adherence, chronic pain, advanced/metastatic cancer, as well as 10 other phenotypes. This dataset can be utilized for academic and industrial research in medicine and computer science, particularly within the field of medical natural language processing.

Sepsis is a common and deadly clinical syndrome that affects many patients presenting to the emergency department (ED). Sepsis-induced inflammation leads to abnormal coagulation. Additionally, one potential mechanism for abnormal coagulation and organ dysfunction in sepsis is injury to the endothelial glycocalyx; the glycocalyx contains heparans which are released during degradation and may cause mild coagulopathy. We hypothesize that coagulation abnormalities detected by bedside viscoelastic monitoring (VEM), such as thromboelastography, are associated with organ dysfunction and death (suggesting abnormal coagulation as a mediator).

Objective In septic patients, multiple retrospective studies show an association between large volumes of fluids administered in the first 24 h and mortality, suggesting a benefit to fluid restrictive strategies. However, these studies do not directly estimate the causal effects of fluid-restrictive strategies, nor do their analyses properly adjust for time-varying confounding by indication. In this study, we used causal inference techniques to estimate mortality outcomes that would result from imposing a range of arbitrary limits (“caps”) on fluid volume administration during the first 24 h of intensive care unit (ICU) care. Design Retrospective cohort study Setting ICUs at the Beth Israel Deaconess Medical Center, 2008–2012 Patients One thousand six hundred thirty-nine septic patients (defined by Sepsis-3 criteria) 18 years and older, admitted to the ICU from the emergency department (ED), who received less than 4 L fluids administered prior to ICU admission Measurements and main results Data were obtained from the Medical Information Mart for Intensive Care III (MIMIC-III). We employed a dynamic Marginal Structural Model fit by inverse probability of treatment weighting to obtain confounding adjusted estimates of mortality rates that would have been observed had fluid resuscitation volume caps between 4 L–12 L been imposed on the population. The 30-day mortality in our cohort was 17%. We estimated that caps between 6 and 10 L on 24 h fluid volume would have reduced 30-day mortality by − 0.6 to − 1.0%, with the greatest reduction at 8 L (− 1.0% mortality, 95% CI [− 1.6%, − 0.3%]). Conclusions We found that 30-day mortality would have likely decreased relative to observed mortality under current practice if these patients had been subject to “caps” on the total volume of fluid administered between 6 and 10 L, with the greatest reduction in mortality rate at 8 L.

BackgroundUnderstanding factors that drive admissions is critical to containing cost and optimising hospital operations. We hypothesised that, due to multiple factors, emergency physicians would be more likely to admit a patient seen later in their shift.MethodsRetrospective study examining all patient visits at a large academic hospital from July 2010 to July 2016. Patients with missing data (n=191) were excluded. 294 031 emergency department (ED) visits were included in the final analysis. The exposure of interest was the time during the shift at which a patient was first evaluated by the clinician, and outcome was hospital admission. We used a generalised estimating equation with physician as the clustering level to adjust for patient age, gender, Emergency Severity Index (ESI, 1=most severe illness, 5=least severe illness) and 24 hours clock time. We also conducted a stratified analysis by three ESI categories.ResultsFrom the 294 031 ED visits, 5977 were seen in the last hour of the shift. Of patients seen in the last shift hour, 43% were admitted versus 39% seen at any other time during the shift. There was a significant association between being evaluated in the last hour (RR 1.03, 95% CI 1.01 to 1.06) and last quarter (RR 1.02, 1.01 to 1.03) of shift and the likelihood of admission. Patients with an ESI Score of 4–5 saw the largest effect sizes (RR 1.62, 0.996–2.635 for last hour and RR 1.24, 0.996–1.535 for last quarter) but these were not statistically significant. Additionally, there was a trend towards increased likelihood of admission later in shift; the relative risk of admission was 1.04 in hour 6, (1.02–1.05), 1.03 in hour 7 (1.01–1.05), 1.04 in hour 8 (1.01–1.06) and 1.06 in hour 9 (1.013–1.101).ConclusionsThere is a small but significant association between a patient being evaluated later in an emergency physician’s shift and their likelihood of being admitted to the hospital.

Intravenous fluids, an essential component of sepsis resuscitation, may paradoxically worsen outcomes by exacerbating endothelial injury. Preclinical models suggest that fluid resuscitation degrades the endothelial glycocalyx, a heparan sulfate-enriched structure necessary for vascular homeostasis. We hypothesized that endothelial glycocalyx degradation is associated with the volume of intravenous fluids administered during early sepsis resuscitation. We used mass spectrometry to measure plasma heparan sulfate (a highly sensitive and specific index of systemic endothelial glycocalyx degradation) after 6 h of intravenous fluids in 56 septic shock patients, at presentation and after 24 h of intravenous fluids in 100 sepsis patients, and in two groups of non-infected patients. We compared plasma heparan sulfate concentrations between sepsis and non-sepsis patients, as well as between sepsis survivors and sepsis non-survivors. We used multivariable linear regression to model the association between volume of intravenous fluids and changes in plasma heparan sulfate. Consistent with previous studies, median plasma heparan sulfate was elevated in septic shock patients (118 [IQR, 113–341] ng/ml 6 h after presentation) compared to non-infected controls (61 [45–79] ng/ml), as well as in a second cohort of sepsis patients (283 [155–584] ng/ml) at emergency department presentation) compared to controls (177 [144–262] ng/ml). In the larger sepsis cohort, heparan sulfate predicted in-hospital mortality. In both cohorts, multivariable linear regression adjusting for age and severity of illness demonstrated a significant association between volume of intravenous fluids administered during resuscitation and plasma heparan sulfate. In the second cohort, independent of disease severity and age, each 1 l of intravenous fluids administered was associated with a 200 ng/ml increase in circulating heparan sulfate (p = 0.006) at 24 h after enrollment. Glycocalyx degradation occurs in sepsis and septic shock and is associated with in-hospital mortality. The volume of intravenous fluids administered during sepsis resuscitation is independently associated with the degree of glycocalyx degradation. These findings suggest a potential mechanism by which intravenous fluid resuscitation strategies may induce iatrogenic endothelial injury.

Importance:Laboratory data are frequently collected throughout the care of critically ill patients. Currently, these data are interpreted by comparison with values from healthy outpatient volunteers. Whether this is the most useful comparison has yet to be demonstrated. Objectives:To understand how the distribution of intensive care unit (ICU) laboratory values differs from the reference range, and how these distributions are related to patient outcomes. Design, Setting, and Participants:Cross-sectional study of a large critical care database, the Medical Information Mart for Intensive Care database, from January 1, 2001, to October 31, 2012. The database is collected from ICU data from a large tertiary medical center in Boston, Massachusetts. The data are collected from medical, cardiac, neurologic, and surgical ICUs. All patients in the database from all ICUs for 2001 to 2012 were included. Common laboratory measurements over the time window of interest were sampled. The analysis was conducted from March to June 2017. Main Outcomes and Measures:The overlapping coefficient and Cohen standardized mean difference between distributions were calculated, and kernel density estimate visualizations for the association between laboratory values and the probability of death or quartile of ICU length of stay were created. Results:Among 38 605 patients in the ICU (21 852 [56.6%] male; mean [SD] age, 74.5 [55.1] years), 8878 (23%) had the best outcome (ICU survival, shortest quartile length of stay) and 3090 (8%) had the worst outcome (ICU nonsurvival). Distribution curves based on ICU data differed significantly from the hospital standard range (mean [SD] overlapping coefficient, 0.51 [0.32-0.69]). All laboratory values for the best outcome group differed significantly from those in the worst outcome group. Both the best and worst outcome group curves revealed little overlap with and marked divergence from the reference range. Conclusions and Relevance:The standard reference ranges obtained from healthy volunteers differ from the analogous range generated from data from patients in intensive care. Laboratory data interpretation may benefit from greater consideration of clinically contextual and outcomes-related factors.

One challenge of contemporary medical education is that shorter lengths of stay and time-limited clerkships often interrupt a student's relationship with a patient before a diagnosis is made or treatment is completed, limiting the learning experience. Medical students sometimes use electronic health records (EHRs) to overcome these limitations. EHRs provide access to patients' future medical records, enabling students to track former patients across care venues to audit their diagnostic impressions and observe outcomes. While this activity has potential to improve clinical training, there is a risk of unintended harm to patients through loss of privacy. Students need guidance on how to perform this activity appropriately. This article describes an ethical framework for tracking using an "educational registry," a list of former patients housed within the EHR that one follows longitudinally for educational purposes. Guiding principles include obtaining permission from patients, having legitimate educational intent, and restricting review of records to those essential for training. This framework could serve as a foundation for institutions seeking to develop a policy on tracking former patients, and may facilitate research on the use of EHRs to improve medical education, such as reducing diagnostic error and promoting self-directed learning.

Objectives: Interhospital transfer, a common intervention, may be subject to healthcare disparities. In mechanically ventilated patients with sepsis, we hypothesize that disparities not disease related would be found between patients who were and were not transferred. Design: Retrospective cohort study. Setting: Nationwide Inpatient Sample, 2006–2012. Patients: Patients over 18 years old with a primary diagnosis of sepsis who underwent mechanical ventilation. Interventions: None. Measurements and Main Results: We obtained age, gender, length of stay, race, insurance coverage, do not resuscitate status, and Elixhauser comorbidities. The outcome used was interhospital transfer from a small- or medium-sized hospital to a larger acute care hospital. Of 55,208,382 hospitalizations, 46,406 patients met inclusion criteria. In the multivariate model, patients were less likely to be transferred if the following were present: older age (odds ratio, 0.98; 95% CI, 0.978–0.982), black race (odds ratio, 0.79; 95% CI, 0.70–0.89), Hispanic race (odds ratio, 0.79; 95% CI, 0.69–0.90), South region hospital (odds ratio, 0.79; 95% CI, 0.72–0.88), teaching hospital (odds ratio, 0.31; 95% CI, 0.28–0.33), and do not resuscitate status (odds ratio, 0.19; 95% CI, 0.15–0.25). Conclusions: In mechanically ventilated patients with sepsis, we found significant disparities in race and geographic location not explained by medical diagnoses or illness severity.

Background: Acute appendicitis is common in the adult emergency department (ED). Computed tomography (CT) scan is frequently used to diagnose this condition, but ultrasound (US)-commonly used in pediatric diagnosis-may also have a role. Objectives: Review the clinical utility and define the frequency and diagnostic accuracy of US to diagnose appendicitis in an adult population in the ED setting. Methods: Retrospective cohort study of patients who underwent appendiceal US in an academic, tertiary ED from July 2013-October 2015. Results: There were 174 patients included, of which 39 (22%) had pathology-confirmed appendicitis. There were 25 patients who had an US scan that was positive for appendicitis, 146 (84%) were indeterminate, and 3 (1.7%) were negative. Among patients with a positive US, 25/25 (100%, 95% confidence interval [CI] 84-100%) had appendicitis, 32/146 (22%, 95% CI 16-29%) with an indeterminate US had appendicitis, and 0/3 (0%, 95% CI 0-6.2%) with a negative US had appendicitis. In the 28 definitive cases, US had a sensitivity of 64%, specificity of 2%, positive predictive value of 100%, and negative predictive value of 100%. The likelihood ratio positive and negative were 173 and 0, respectively. Conclusion: Our initial data suggest that an US that shows appendicitis seems to be reliable; however, a high prevalence of indeterminate studies limits the diagnostic utility as a universal approach in adult patients in the ED setting. Larger studies are needed to identify which patient populations would benefit from US as the initial imaging modality, what factors contribute to the large numbers of indeterminate results, and if any interventions may reduce the number of indeterminate results. (C) 2018 Elsevier Inc. All rights reserved.

Objectives: The outcomes of critically ill patients who undergo interhospital transfer are not well understood. Physicians assume that patients who undergo interhospital transfer will receive more advanced care that may translate into decreased morbidity or mortality relative to a similar patient who is not transferred. However, there is little empirical evidence to support this assumption. We examined country-level U.S. data from the Nationwide Readmissions Database to examine whether, in mechanically ventilated patients with sepsis, interhospital transfer is associated with a mortality benefit. Design: Retrospective data analysis using complex survey design regression methods with propensity score matching. Setting: The Nationwide Readmissions Database contains information about hospital admissions from 22 States, accounting for roughly half of U.S. hospitalizations; the database contains linkage numbers so that admissions and transfers for the same patient can be linked across 1 year of follow-up. Patients: From the 2013 Nationwide Readmission Database Sample, 14,325,172 hospital admissions were analyzed. There were 61,493 patients with sepsis and on mechanical ventilation. Of these, 1,630 patients (2.7%) were transferred during their hospitalization. A propensity-matched cohort of 1,630 patients who did not undergo interhospital transfer was identified. Interventions: None. Measurements and Main Results: The exposure of interest was interhospital transfer to an acute care facility. The primary outcome was hospital mortality; the secondary outcome was hospital length of stay. The propensity score included age, gender, insurance coverage, do not resuscitate status, use of renal replacement therapy, presence of shock, and Elixhauser comorbidities index. After propensity matching, interhospital transfer was not associated with a difference in in-hospital mortality (12.3% interhospital transfer vs 12.7% non–interhospital transfer; p = 0.74). However, interhospital transfer was associated with a longer total hospital length of stay (12.8 d interquartile range, 7.7–21.6 for interhospital transfer vs 9.1 d interquartile range, 5.1–17.0 for non–interhospital transfer; p < 0.01). Conclusions: Patients with sepsis requiring mechanical ventilation who underwent interhospital transfer did not have improved outcomes compared with a cohort with matched characteristics who were not transferred. The study raises questions about the risk-benefit profile of interhospital transfer as an intervention.

In secondary analysis of electronic health records, a crucial task consists in correctly identifying the patient cohort under investigation. In many cases, the most valuable and relevant information for an accurate classification of medical conditions exist only in clinical narratives. Therefore, it is necessary to use natural language processing (NLP) techniques to extract and evaluate these narratives. The most commonly used approach to this problem relies on extracting a number of clinician-defined medical concepts from text and using machine learning techniques to identify whether a particular patient has a certain condition. However, recent advances in deep learning and NLP enable models to learn a rich representation of (medical) language. Convolutional neural networks (CNN) for text classification can augment the existing techniques by leveraging the representation of language to learn which phrases in a text are relevant for a given medical condition. In this work, we compare concept extraction based methods with CNNs and other commonly used models in NLP in ten phenotyping tasks using 1,610 discharge summaries from the MIMIC-III database. We show that CNNs outperform concept extraction based methods in almost all of the tasks, with an improvement in F1-score of up to 26 and up to 7 percentage points in area under the ROC curve (AUC). We additionally assess the interpretability of both approaches by presenting and evaluating methods that calculate and extract the most salient phrases for a prediction. The results indicate that CNNs are a valid alternative to existing approaches in patient phenotyping and cohort identification, and should be further investigated. Moreover, the deep learning approach presented in this paper can be used to assist clinicians during chart review or support the extraction of billing codes from text by identifying and highlighting relevant phrases for various medical conditions.

Objective: We investigate whether deep learning techniques for natural language processing (NLP) can be used efficiently for patient phenotyping. Patient phenotyping is a classification task for determining whether a patient has a medical condition, and is a crucial part of secondary analysis of healthcare data. We assess the performance of deep learning algorithms and compare them with classical NLP approaches. Materials and Methods: We compare convolutional neural networks (CNNs), n-gram models, and approaches based on cTAKES that extract pre-defined medical concepts from clinical notes and use them to predict patient phenotypes. The performance is tested on 10 different phenotyping tasks using 1,610 discharge summaries extracted from the MIMIC-III database. Results: CNNs outperform other phenotyping algorithms in all 10 tasks. The average F1-score of our model is 76 (PPV of 83, and sensitivity of 71) with our model having an F1-score up to 37 points higher than alternative approaches. We additionally assess the interpretability of our model by presenting a method that extracts the most salient phrases for a particular prediction. Conclusion: We show that NLP methods based on deep learning improve the performance of patient phenotyping. Our CNN-based algorithm automatically learns the phrases associated with each patient phenotype. As such, it reduces the annotation complexity for clinical domain experts, who are normally required to develop task-specific annotation rules and identify relevant phrases. Our method performs well in terms of both performance and interpretability, which indicates that deep learning is an effective approach to patient phenotyping based on clinicians' notes.

Advancement of Artificial Intelligence (AI) capabilities in medicine can help address many pressing problems in healthcare. However, AI research endeavors in healthcare may not be clinically relevant, may have unrealistic expectations, or may not be explicit enough about their limitations. A diverse and well-functioning multi-disciplinary team (MDT) can help identify appropriate and achievable AI research agendas in healthcare, and advance medical AI technologies by developing AI algorithms as well as addressing the shortage of appropriately labeled datasets for machine learning. In this paper, our team of engineers, clinicians and machine learning experts share their experience and lessons learned from their two-year-long collaboration on a natural language processing (NLP) research project. We highlight specific challenges encountered in cross-disciplinary teamwork, dataset creation for NLP research, and expectation setting for current medical AI technologies.

Department of Medicine, Beth Israel Deaconess Medical Center, Boston, MA Division of Pulmonary, Critical Care and Sleep Medicine, Beth Israel Deaconess Medical Center, Boston, MA; and Laboratory for Computational Physiology, Institute for Medical Engineering and Science, Massachusetts Institute of Technology, Cambridge, MA *See also p. 1344. Dr. Celi is funded by the National Institute of Health through the National Institute of Biomedical Informatics and Bioimaging grant R01 EB017205-01A1. The Laboratory for Computational Physiology receives research funding from Philips. Dr. Tyler has disclosed that he does not have any potential conflicts of interest.

PURPOSE:Surgical resection of colorectal liver metastases is not achievable in more than 70% of the cases. Although the liver directed therapies have become a part of the stand of care, lack of a preclinical model impedes the assessment of toxicity and therapeutic benefits attributed several candidate drugs or treatment regimens that can be designed. In the present study we aim develop and characterize a rat colorectal liver metastasis model. MATERIALS AND METHODS:Growth characteristics of CC-531 cells were determined in vitro followed by subcapsular liver implantation in syngeneic WAG/Rij rats. Tumor growth progression was followed over 3 weeks by ultrasound (US) and magnetic resonance imaging (MRI). Growth characteristics were also assessed by histopathology and immunohistochemistry in harvested tumor tissues. RESULTS:The doubling time of CC-531 cells was found be under 24hrs and all the implanted rats grew tumors. US imaging showed hypoechoic masses and MRI showed contrast enhancement representing complex tumor microenvironments. Hematoxylin and Eosin staining confirmed tumor growth and uniform CD31 staining in tumor confirmed even vessel density. CONCLUSION:CC-531 can be used as a metastatic rat tumor colorectal liver metastases model with well-defined characteristics that can be readily followed by imaging whilst having a therapeutic window for interventions.