ABSTRACT
INTRODUCTION: Given the novelty of COVID-19 induced ARDS, well-established traditional ARDS management strategies were initially used to treat these patients. In this report we compared treatment and outcomes for COVID-19 patients with severe ARDS with a similar cohort of patients seen in 2019. METHODS: We retrospectively identified patients with a diagnosis of severe ARDS (P/F ratio of <100) admitted in 2019, and COVID-19 patients with severe ARDS admitted 3/1-5/31/2020. Statistical comparisons were made for demographics, measures of illness severity [APACHE and Discharge Readiness Score (DRS, Philips) scores], treatments [including vasopressors, inhaled epoprostenol (EPO), and ECMO], and outcomes [ICU mortality, and ICU and hospital lengths of stay (LOS)]. RESULTS: 60 patients with severe ARDS admitted in 2019 and 66 COVID-19 positive patients with severe ARDS admitted between 3/1-5/31/2020 were identified. The COVID-19 patients were older (63.3 vs 53.9 yrs, respectively, p<0.01), with a higher proportion of African Americans (74.2 vs 20.0%, p<0.001). Illness severity was higher for the COVID-19 cohort using peak DRS (86.7 vs 76.5, p<0.05) but lower using APACHE (98.7 vs 80.2, p<0.001). There was no difference in vasopressor use (98.0 vs 95.5%, p=0.45), but COVID-19 patients were more likely to have received inhaled EPO (37.9 vs 16.7%, p=0.014), and less likely to have received ECMO (10.6 vs 26.7%, p=0.04). COVID-19 patients had higher ICU Mortality (48.5 vs 36.7%) and hospital LOS (28.8 vs 21.8 days), but these differences were not statistically significant (p=0.14 and 0.08, respectively). ICU LOS was similar (16.7 vs 16.5 days). CONCLUSIONS: The 66 COVID-19 patients with severe ARDS seen in the first 3 months of the pandemic exceeded the 60 patients with severe ARDS seen in 2019 at our large urban academic medical center. COVID-19 patients were older and more likely to be of African American ethnicity. There was a notable shift involving more inhaled EPO and less ECMO as rescue therapy in the COVID-19 cohort. Interestingly, while ICU mortality for COVID-19 patients was higher (albeit not significantly in this small cohort), measures of disease severity on ICU admission (APACHE) appeared less able to capture the increased mortality risk compared to peak DRS measured continuously.
ABSTRACT
INTRODUCTION: In the COVID-19 pandemic, novel solutions were desperately needed to protect healthcare workers while continuing essential bedside care, multiprofessional rounds, specialist consultation, and family visitation. A strategy was implemented to harness the power of an established Tele ICU program to mitigate patient contact isolation using expanded access for bedside providers and family to the Tele ICU software and cameras. The objective of this study was to quantify the use and added value of this expanded access program. METHODS: Data from the BJC Healthcare system Tele ICU program covering 13 ICUs in 9 hospitals was compiled from the Philips eICU Total Data Access Kit (TDAK) database, and video usage by Tele ICU and bedside providers was quantified over the period March 12 through June 30, 2020. A separate query of the Tele ICU Caregility database was done to estimate family visits. Instances where direct patient contact was replaced with a video session were estimated from these data. Supply chain costs for personal protective equipment (PPE) were used to estimate PPE cost savings and added PPE availability. RESULTS: 78,766 individual video sessions totaling 3,713 hours were identified in the TDAK database including 72,053 sessions (1625 hours, 43.8%) attributed to Tele ICU staff and 6,713 sessions (2088 hours, 56.2%) attributed to bedside staff. An additional 2240 family video sessions totaling 140 hours were identified in the Caregility database. Together, the 8,953 instances in which Tele ICU video sessions were used in lieu of direct exposure of bedside providers or family to COVID patients resulted in an estimated cost savings of $24,000 related to decreased PPE use, and a substantial increase in the supply of PPE. CONCLUSIONS: This report highlights an unforeseen benefit of Tele ICU in the COVID pandemic resulting from our expanded access strategy allowing virtual visits between patients and bedside providers and their families. In addition to savings related to less PPE use, there was increased PPE availability and reduced exposure of clinicians and family to infected patients. While difficult to quantify, decreased transmission of the disease to staff and minimizing workforce loss are additional potential benefits.
ABSTRACT
INTRODUCTION: Vital signs (VS) are important indicators of disease severity and clinical deterioration. However, the predictive scope of VS for ICU mortality is unknown and there are no validated system for early and real-time prediction of ICU mortality from VS data alone. In this study we aimed to develop and validate a Machine Learning (ML) classifier to predict ICU mortality from continuous VS data. METHODS: We used de-identified patient VS data obtained from our eSearch (Philips Healthcare) database to encode 7 continuous VS time series and use of 5 VS monitoring devices. Mean, standard deviation, autocorrelation, and the trend of the mean were used to encode VS time series variations and were adjusted to the entire ICU stay, and 6, 12, or 24 hours before death. Our approach did not encode diagnoses but agnostically classified based on VS features. Performance of the models was determined on a naïve cohort and an independent sample of patients with COVID-19. RESULTS: A total of 19,266 ICU stays prior to COVID were studied including 17,339 in the training cohort, and 1,927 in the naïve validation cohort with ICU mortalities of 9%. An independent sample of 548 patients with COVID-19 with mortality of 22% was also used for validation. For the entire stay, and 6-, 12-, and 24-hours in advance, the ML classifier achieved AUCs and PRCs of 0.97 - 0.81 and 0.78 - 0.40, respectively in the naïve population obtained prior to COVID, and AUCs and PRCs of 0.92 - 0.80 and 0.81 - 0.58, respectively, for the COVID cohort. Notably, a differential ranking of features was found for mortality predictions in the COVID-19 sample, as well as in 9 other specific diagnoses. The effectiveness of this approach compared favorably with six other ML methods and with the DRS (Philips) mortality predictions. CONCLUSIONS: A data-driven ML algorithm developed from composite vital sign data alone made ICU mortality predictions with model performance on a naïve ICU test population, as well as on a COVID-19 patient population, that rivals other prediction models using more complex data domains. Shapley Additive exPlanations provided interpretability and clinical validation of the ML model related to the specific features in the ICU subpopulations.