Sum Score to Predict Need for Mechanical Ventilation in Patients with COVID-19

Abstract

Introduction/Rationale: The novel coronavirus disease 2019 (COVID-19) created an unprecedented healthcare crisis and has put enormous strain on hospital systems across the world. The unpredictability of this disease has led to critical care shortages such as ICU beds, ventilator availability and staffing. To our knowledge, a novel scoring criteria is not available that can assist clinicians in predicting who may decompensate and eventually require mechanical ventilation and the highest level of available care. Such a scoring criteria would be beneficial in times of surge capacity, in which the score could be applied to patients upon admission and assist in determining where resources may need to be allocated. Methods: The electronic medical records of the first 150 patients to present to a large, tertiary referral center in the Southeastern US with COVID-19 pneumonia were reviewed. A multivariable logistic regression model was used to determine odds of requiring mechanical ventilation after admission using demographic and clinical characteristics. Adjusted odds ratios (aOR) and 95% confidence intervals (95% CI) were calculated. A prognostic index from aOR was created and validated with one-leave-out cross validation method. SAS v9.4 was used for data management and statistical data analysis. Results: Three variables were found to be directly linked with the need for mechanical ventilation in patients with COVID-19 pneumonia. An increased number of comorbidities (obesity, hypertension, diabetes mellitus, chronic lung disease or cardiovascular disease) was associated with a two-fold risk for mechanical ventilation (aOR 1.955 [95% CI=1.27-3.011]). A decreased SpO2/FiO2 ratio compared to normal range was associated with a two-fold risk in need for mechanical ventilation ]. An increase in neutrophil/lymphocyte ratio compared to a normal range was associated with an aOR of 1.783 (95% CI=1.142-2.783). Conclusion: Our proposed scoring system is a sum score for number of comorbidities, neutrophil/lymphocyte ratio, and oxygen saturation/fraction of inspired oxygen ratio in patients with COVID-19 pneumonia. As each of these variables increase, the patients are assigned an increasing patient score based on the values found on admission. A sum score greater than eight was found to have high predictive value for requiring mechanical ventilation, including a sensitivity of 77.1%, specificity of 83.1%, positive predictive value of 71.1% and negative predictive value of 87.1%. Our score was internally validated, accurately predicting mechanical ventilation in 81% of patients, but will have to be applied to a larger sample size prospectively for external validation before clinical application is considered.