A method for predicting the effectiveness of glucocorticoid therapy in patients with moderate COVID-19 based on simple clinical and laboratory data

Abstract

Background: In patients hospitalized with coronavirus infection (COVID-19), methods for predicting the effectiveness of anti-inflammatory therapy have important practical implications for optimizing treatment and outcomes. To date, a number of indicators of COVID-19 patients (age, comorbidities, laboratory criteria for the intensity of inflammation) have been identified that indicate a high probability of a severe course and a risk of an adverse outcome. However, the problem of predicting the effectiveness of anti-inflammatory therapy in patients with moderate COVID-19 is not well understood. Aims: to develop a predictive model to determine the effectiveness/failure of glucocorticosteroid (GCS) monotherapy in patients with moderate COVID-19. Methods. Retrospective analysis of electronic medical record data of all patients admitted consecutively from October 1, 2020 to January 31, 2021. The study included 71 patients with a probable (clinically confirmed) and confirmed (laboratory) case of COVID-19 of moderate course, with characteristic changes in the lungs according to computed tomography of the chest organs (CT-CCT). Given the severity of the course, all patients in this sample were prescribed GCS in accordance with the current version of the Interim Guidelines of the Ministry of Health of the Russian Federation. Results. A total of 71 patients were studied, 53 (74.7%) of them did not require an escalation of anti-inflammatory therapy, which was regarded as an effective use of corticosteroids in the form of monotherapy (group 1). In the remaining 18 patients, the use of corticosteroids for an average of 5.5 (from 3 to 6) days did not have a definite clinical effect and required the additional use of monoclonal antibodies (MCA) to interleukin-6 (IL-6) or to its receptor (group 2). Using logistic regression analysis and ROC analysis, a mathematical model was developed and evaluated to predict the outcome of anti-inflammatory corticosteroid therapy in patients with moderate COVID-19. As risk factors, indicators were selected that had significant differences in the studied groups before the appointment of GCS: the number of lymphocytes, platelets and body temperature.The quality of the constructed model is assessed as very good, the optimal cutoff point is 0.697. The sensitivity index of the model is 81.1%, the specificity index is 72.2%. Conclusions. The mathematical model makes it possible to predict the effectiveness of GCS therapy according to the number of lymphocytes, platelets and body temperature. The mathematical model is adequate, has a high sensitivity and specificity.