Search

Background In-field triage tools for trauma patients are limited by availability of information, linear risk classification, and a lack of confidence reporting. We therefore set out to develop and test a machine learning algorithm that can overcome these limitations by accurately and confidently making predictions to support in-field triage in the first hours after traumatic injury. Methods Using an American College of Surgeons Trauma Quality Improvement Program-derived database of truncal and junctional gunshot wound (GSW) patients (aged 16-60 years), we trained an information-aware Dirichlet deep neural network (field artificial intelligence triage). Using supervised training, field artificial intelligence triage was trained to predict shock and the need for major hemorrhage control procedures or early massive transfusion (MT) using GSW anatomical locations, vital signs, and patient information available in the field. In parallel, a confidence model was developed to predict the true-class probability (scale of 0-1), indicating the likelihood that the prediction made was correct, based on the values and interconnectivity of input variables. Results A total of 29,816 patients met all the inclusion criteria. Shock, major surgery, and early MT were identified in 13.0%, 22.4%, and 6.3% of the included patients, respectively. Field artificial intelligence triage achieved mean areas under the receiver operating characteristic curve of 0.89, 0.86, and 0.82 for prediction of shock, early MT, and major surgery, respectively, for 80/20 train-test splits over 1,000 epochs. Mean predicted true-class probability for errors/correct predictions was 0.25/0.87 for shock, 0.30/0.81 for MT, and 0.24/0.69 for major surgery. Conclusion Field artificial intelligence triage accurately identifies potential shock in truncal GSW patients and predicts their need for MT and major surgery, with a high degree of certainty. The presented model is an important proof of concept. Future iterations will use an expansion of databases to refine and validate the model, further adding to its potential to improve triage in the field, both in civilian and military settings. Level of evidence Prognostic, Level III.

BACKGROUND: We sought to describe characteristics, multisystem outcomes, and predictors of mortality of the critically ill COVID-19 patients in the largest hospital in Massachusetts. METHODS: This is a prospective cohort study. All patients admitted to the intensive care unit (ICU) with reverse-transcriptase-polymerase chain reaction-confirmed severe acute respiratory syndrome coronavirus 2 infection between March 14, 2020, and April 28, 2020, were included; hospital and multisystem outcomes were evaluated. Data were collected from electronic records. Acute respiratory distress syndrome (ARDS) was defined as PaO2/FiO2 ratio of ≤300 during admission and bilateral radiographic pulmonary opacities. Multivariable logistic regression analyses adjusting for available confounders were performed to identify predictors of mortality. RESULTS: A total of 235 patients were included. The median (interquartile range [IQR]) Sequential Organ Failure Assessment score was 5 (3-8), and the median (IQR) PaO2/FiO2 was 208 (146-300) with 86.4% of patients meeting criteria for ARDS. The median (IQR) follow-up was 92 (86-99) days, and the median ICU length of stay was 16 (8-25) days; 62.1% of patients were proned, 49.8% required neuromuscular blockade, and 3.4% required extracorporeal membrane oxygenation. The most common complications were shock (88.9%), acute kidney injury (AKI) (69.8%), secondary bacterial pneumonia (70.6%), and pressure ulcers (51.1%). As of July 8, 2020, 175 patients (74.5%) were discharged alive (61.7% to skilled nursing or rehabilitation facility), 58 (24.7%) died in the hospital, and only 2 patients were still hospitalized, but out of the ICU. Age (odds ratio [OR], 1.08; 95% confidence interval [CI], 1.04-1.12), higher median Sequential Organ Failure Assessment score at ICU admission (OR, 1.24; 95% CI, 1.06-1.43), elevated creatine kinase of ≥1,000 U/L at hospital admission (OR, 6.64; 95% CI, 1.51-29.17), and severe ARDS (OR, 5.24; 95% CI, 1.18-23.29) independently predicted hospital mortality.Comorbidities, steroids, and hydroxychloroquine treatment did not predict mortality. CONCLUSION: We present here the outcomes of critically ill patients with COVID-19. Age, acuity of disease, and severe ARDS predicted mortality rather than comorbidities. LEVEL OF EVIDENCE: Prognostic, level III.

Patients in the US receive disproportionally higher amounts of opioids after operations compared with their non-US counterparts. We aimed to assess the relationship between perceived pain severity after operation and the amount of opioid medications prescribed at discharge in US vs non-US patients.We conducted a post-hoc analysis of the International Patterns of Opioid Prescribing multicenter study. Patients 16 years and older who underwent appendectomy, cholecystectomy, or inguinal herniorrhaphy in 1 of 14 participating hospitals across 8 countries between October 2016 and March 2017 were included. In hospitals where pain severity was assessed using a 0 to 10 visual analog scale before hospital discharge, patients were stratified into the following groups, depending on the pain severity: none, mild (1 to 3), moderate (4 to 6), and severe (7 to 10). The number of opioid prescriptions, total number of pills, and oral morphine equivalents prescribed were calculated for each group and US and non-US patients were compared.A total of 2,024 patients were included. Eighty-three percent of US patients without pain were prescribed opioids compared with 8.7% of non-US patients without pain (p0.001). The number of opioid prescriptions, number of pills, and oral morphine equivalents prescribed were similar across the 4 pain severity groups in US patients (p0.05). In contrast, the number of opioid prescriptions, number of opioid pills, and oral morphine equivalents prescribed among non-US patients were incrementally higher as the pain severity progressed from no pain to severe pain (all, p0.05).US patients are prescribed opioids at high rates and doses regardless of pain severity. Additional efforts should be directed toward tailoring opioid prescriptions to patients' needs.

BACKGROUND The Emergency Surgery Score (ESS) is a point-based scoring system validated to predict mortality and morbidity in emergency general surgery (EGS). In addition to demographics and comorbidities, ESS accounts for the acuity of disease at presentation. We sought to examine whether ESS can predict the destination of discharge of EGS patients, as a proxy for quality of life at discharge. METHODS Using the 2007 to 2017 American College of Surgeons National Surgical Quality Improvement Program database, we identified all EGS patients. EGS cases were defined as per American College of Surgeons National Surgical Quality Improvement Program as those performed by a general surgeon within a short interval from diagnosis or the onset of related symptomatology, when the patient's well-being and outcome may be threatened by unnecessary delay and patient's status could deteriorate unpredictably or rapidly. Emergency Surgery Score patients were then categorized by their discharge disposition to home versus rehabilitation or nursing facilities. All patients with missing ESS or discharge disposition and those discharged to hospice, senior communities, or separate acute care facilities were excluded. Emergency Surgery Score was calculated for each patient. C statistics were used to study the correlation between ESS and the destination of discharge. RESULTS Of 6,485,915 patients, 84,694 were included. The mean age was 57 years, 51% were female, and 79.6% were discharged home. The mean ESS was 5. Emergency Surgery Score accurately and reliably predicted the discharge destination with a C statistic of 0.83. For example, ESS of 1, 10, and 20 were associated with 0.9%, 56.5%, and 100% rates of discharge to a rehabilitation or nursing facility instead of home. CONCLUSION Emergency Surgery Score accurately predicts which EGS patients require discharge to rehabilitation or nursing facilities and can thus be used for preoperatively counseling patients and families and for improving early discharge preparations, when appropriate. LEVEL OF EVIDENCE Prognostic and epidemiological, level III.