IMPORTANCE Many patients with coronavirus disease 2019 (COVID-19) are critically ill and require care in the intensive care unit (ICU). OBJECTIVE To evaluate the independent risk factors associated with mortality of patients with COVID-19 requiring treatment in ICUs in the Lombardy region of Italy. DESIGN, SETTING, AND PARTICIPANTS This retrospective, observational cohort study included 3988 consecutive critically ill patients with laboratory-confirmed COVID-19 referred for ICU admission to the coordinating center (Fondazione IRCCS [Istituto di Ricovero e Cura a Carattere Scientifico] Ca' Granda Ospedale Maggiore Policlinico, Milan, Italy) of the COVID-19 Lombardy ICU Network from February 20 to April 22, 2020. Infection with severe acute respiratory syndrome coronavirus 2 was confirmed by real-time reverse transcriptase-polymerase chain reaction assay of nasopharyngeal swabs. Follow-up was completed on May 30, 2020. EXPOSURES Baseline characteristics, comorbidities, long-term medications, and ventilatory support at ICU admission. MAIN OUTCOMES AND MEASURES Time to death in days from ICU admission to hospital discharge. The independent risk factors associated with mortality were evaluated with a multivariable Cox proportional hazards regression. RESULTS Of the 3988 patients included in this cohort study, the median age was 63 (interquartile range [IQR] 56-69) years; 3188 (79.9%; 95% CI, 78.7%-81.1%) were men, and 1998 of 3300 (60.5%; 95% CI, 58.9%-62.2%) had at least 1 comorbidity. At ICU admission, 2929 patients (87.3%; 95% CI, 86.1%-88.4%) required invasive mechanical ventilation (IMV). The median follow-up was 44 (95% CI, 40-47; IQR, 11-69; range, 0-100) days; median time from symptoms onset to ICU admission was 10 (95% CI, 9-10; IQR, 6-14) days; median length of ICU stay was 12 (95% CI, 12-13; IQR, 6-21) days; and median length of IMV was 10 (95% CI, 10-11; IQR, 6-17) days. Cumulative observation time was 164 305 patient-days. Hospital and ICU mortality rates were 12 (95% CI, 11-12) and 27 (95% CI, 26-29) per 1000 patients-days, respectively. In the subgroup of the first 1715 patients, as of May 30, 2020, 865 (50.4%) had been discharged from the ICU, 836 (48.7%) had died in the ICU, and 14 (0.8%) were still in the ICU; overall, 915 patients (53.4%) died in the hospital. Independent risk factors associated with mortality included older age (hazard ratio [HR], 1.75; 95% CI, 1.60-1.92), male sex (HR, 1.57; 95% CI, 1.31-1.88), high fraction of inspired oxygen (FiO 2) (HR, 1.14; 95% CI, 1.10-1.19), high positive end-expiratory pressure (HR, 1.04; 95% CI, 1.01-1.06) or low PaO 2 :FiO 2 ratio (HR, 0.80; 95% CI, 0.74-0.87) on ICU admission, and history of chronic obstructive pulmonary disease (HR, 1.68; 95% CI, 1.28-2.19), hypercholesterolemia (HR, 1.25; 95% CI, 1.02-1.52), and type 2 diabetes (HR, 1.18; 95% CI, 1.01-1.39). No medication was independently associated with mortality (angiotensin-converting enzyme inhibitors HR, 1.17; 95% CI, 0.97-1.42; angiotensin receptor blockers HR, 1.05; 95% CI, 0.85-1.29). CONCLUS...

Backgrounds The COVID-19 pandemic drastically strained the health systems worldwide, obligating the reassessment of how healthcare is delivered. In Lombardia, Italy, a Regional Emergency Committee (REC) was established and the regional health system reorganized, with only three hospitals designated as hubs for trauma care. The aim of this study was to evaluate the effects of this reorganization of regional care, comparing the distribution of patients before and during the COVID-19 outbreak and to describe changes in the epidemiology of severe trauma among the two periods. Methods A cohort study was conducted using retrospectively collected data from the Regional Trauma Registry of Lombardia (LTR). We compared the data of trauma patients admitted to three hub hospitals before the COVID-19 outbreak (September 1 to November 19, 2019) with those recorded during the pandemic (February 21 to May 10, 2020) in the same hospitals. Demographic data, level of pre-hospital care (Advanced Life Support-ALS, Basic Life Support-BLS), type of transportation, mechanism of injury (MOI), abbreviated injury score (AIS, 1998 version), injury severity score (ISS), revised trauma score (RTS), and ICU admission and survival outcome of all the patients admitted to the three trauma centers designed as hubs, were reviewed. Screening for COVID-19 was performed with nasopharyngeal swabs, chest ultrasound, and/or computed tomography. Results During the COVID-19 pandemic, trauma patients admitted to the hubs increased (46.4% vs 28.3%, p < 0.001) with an increase in pre-hospital time (71.8 vs 61.3 min, p < 0.01), while observed in hospital mortality was unaffected. TRISS, ISS, AIS, and ICU admission were similar in both periods. During the COVID-19 outbreak, we observed substantial changes in MOI of severe trauma patients admitted to three hubs, with increases of unintentional (31.9% vs 18.5%, p < 0.05) and intentional falls (8.4% vs 1.2%, p < 0.05), whereas the pandemic restrictions reduced road- related injuries (35.6% vs 60%, p < 0.05). Deaths on scene were significantly increased (17.7% vs 6.8%, p < 0.001). Conclusions The COVID-19 outbreak affected the epidemiology of severe trauma patients. An increase in trauma patient admissions to a few designated facilities with high level of care obtained satisfactory results, while COVID-19 patients overwhelmed resources of most other hospitals.

Background Reverse Transcription-Polymerase Chain Reaction (RT-PCR) for Severe Acute Respiratory Syndrome Coronavirus 2 (SARS-COV-2) diagnosis currently requires quite a long time span. A quicker and more efficient diagnostic tool in emergency departments could improve management during this global crisis. Our main goal was assessing the accuracy of artificial intelligence in predicting the results of RT-PCR for SARS-COV-2, using basic information at hand in all emergency departments. Methods This is a retrospective study carried out between February 22, 2020 and March 16, 2020 in one of the main hospitals in Milan, Italy. We screened for eligibility all patients admitted with influenza-like symptoms tested for SARS-COV-2. Patients under 12 years old and patients in whom the leukocyte formula was not performed in the ED were excluded. Input data through artificial intelligence were made up of a combination of clinical, radiological and routine laboratory data upon hospital admission. Different Machine Learning algorithms available on WEKA data mining software and on Semeion Research Centre depository were trained using both the Training and Testing and the K-fold cross-validation protocol. Results Among 199 patients subject to study (median [interquartile range] age 65 [46–78] years; 127 [63.8%] men), 124 [62.3%] resulted positive to SARS-COV-2. The best Machine Learning System reached an accuracy of 91.4% with 94.1% sensitivity and 88.7% specificity. Conclusion Our study suggests that properly trained artificial intelligence algorithms may be able to predict correct results in RT-PCR for SARS-COV-2, using basic clinical data. If confirmed, on a larger-scale study, this approach could have important clinical and organizational implications.

Background: Reverse Transcription-Polymerase Chain Reaction (RT-PCR) for Severe Acute Respiratory Syndrome Coronavirus 2 (SARS-COV-2) diagnosis currently requires quite a long time span. A quicker and more efficient diagnostic tool in emergency departments could improve management during this global crisis. Our main goal was assessing the accuracy of artificial intelligence in predicting the results of RT-PCR for SARS-COV-2, using basic information at hand in all emergency departments.Methods: This is a retrospective study carried out between February 22, 2020 and March 16, 2020 in one of the main hospitals in Milan, Italy. We screened for eligibility all patients admitted with influenza-like symptoms tested for SARS-COV-2. Patients under 12 years old and patients in whom the leukocyte formula was not performed in the ED were excluded. Input data through artificial intelligence were made up of a combination of clinical, radiological and routine laboratory data upon hospital admission. Different Machine Learning algorithms available on WEKA data mining software and on Semeion Research Centre depository were trained using both the Training and Testing and the K-fold cross-validation protocol.Results: Among 199 patients subject to study (median [interquartile range] age 65 [46-78] years; 127 [63.8%] men), 124 [62.3%] resulted positive to SARS-COV-2. The best Machine Learning System reached an accuracy of 91.4% with 94.1% sensitivity and 88.7% specificity. Conclusion: Our study suggests that properly trained artificial intelligence algorithms may be able to predict correct results in RT-PCR for SARS-COV-2, using basic clinical data. If confirmed, on a larger-scale study, this approach could have important clinical and organizational implications.