Machine learning identifies ICU outcome predictors in a multicenter COVID-19 cohort.

BACKGROUND: Intensive Care Resources are heavily utilized during the COVID-19 pandemic. However, risk stratification and prediction of SARS-CoV-2 patient clinical outcomes upon ICU admission remain inadequate. This study aimed to develop a machine learning model, based on retrospective & prospective clinical data, to stratify patient risk and predict ICU survival and outcomes. METHODS: A Germany-wide electronic registry was established to pseudonymously collect admission, therapeutic and discharge information of SARS-CoV-2 ICU patients retrospectively and prospectively. Machine learning approaches were evaluated for the accuracy and interpretability of predictions. The Explainable Boosting Machine approach was selected as the most suitable method. Individual, non-linear shape functions for predictive parameters and parameter interactions are reported. RESULTS: 1039 patients were included in the Explainable Boosting Machine model, 596 patients retrospectively collected, and 443 patients prospectively collected. The model for prediction of general ICU outcome was shown to be more reliable to predict "survival". Age, inflammatory and thrombotic activity, and severity of ARDS at ICU admission were shown to be predictive of ICU survival. Patients' age, pulmonary dysfunction and transfer from an external institution were predictors for ECMO therapy. The interaction of patient age with D-dimer levels on admission and creatinine levels with SOFA score without GCS were predictors for renal replacement therapy. CONCLUSIONS: Using Explainable Boosting Machine analysis, we confirmed and weighed previously reported and identified novel predictors for outcome in critically ill COVID-19 patients. Using this strategy, predictive modeling of COVID-19 ICU patient outcomes can be performed overcoming the limitations of linear regression models. Trial registration "ClinicalTrials" (clinicaltrials.gov) under NCT04455451.

Comparison of isoflurane and propofol sedation in critically ill COVID-19 patients-a retrospective chart review.

PURPOSE: In this retrospective study, we compared inhaled sedation with isoflurane to intravenous propofol in invasively ventilated COVID-19 patients with ARDS (Acute Respiratory Distress Syndrome). METHODS: Charts of all 20 patients with COVID-19 ARDS admitted to the ICU of a German University Hospital during the first wave of the pandemic between 22/03/2020 and 21/04/2020 were reviewed. Among screened 333 days, isoflurane was used in 97 days, while in 187 days, propofol was used for 12 h or more. The effect and dose of these two sedatives were compared. Mixed sedation days were excluded. RESULTS: Patients' age (median [interquartile range]) was 64 (60-68) years. They were invasively ventilated for 36 [21-50] days. End-tidal isoflurane concentrations were high (0.96 ± 0.41 Vol %); multiple linear regression yielded the ratio (isoflurane infusion rate)/(minute ventilation) as the single best predictor. Infusion rates were decreased under ECMO (3.5 ± 1.4 versus 7.1 ± 3.2 ml∙h-1; p < 0.001). In five patients, the maximum recommended dose of propofol of 4 mg∙hour-1∙kg-1ABW was exceeded on several days. On isoflurane compared to propofol days, neuro-muscular blocking agents (NMBAs) were used less frequently (11% versus 21%; p < 0.05), as were co-sedatives (7% versus 31%, p < 0.001); daily opioid doses were lower (720 [720-960] versus 1080 [720-1620] mg morphine equivalents, p < 0.001); and RASS scores indicated deeper levels of sedation (- 4.0 [- 4.0 to - 3.0] versus - 3.0 [- 3.6 to - 2.5]; p < 0.01). CONCLUSION: Isoflurane provided sufficient sedation with less NMBAs, less polypharmacy and lower opioid doses compared to propofol. High doses of both drugs were needed in severely ill COVID-19 patients.