A novel evidence-based algorithm to predict thromboembolism in patients with COVID-19: preliminary data from a single-centre cohort.

Abstract

BACKGROUND\nSevere acute respiratory syndrome coronavirus-2 (SARS-CoV-2)-related disease (COVID-19) is an infectious disease characterised by systemic inflammation, which might enhance baseline thrombotic risk, especially in hospitalised patients. Little is, however, known about predictors of thrombotic complications in patients with COVID-19.\n\n\nMETHODS\nWe prospectively followed up 180 hospitalised COVID-19 patients. Demographics, clinical and laboratory features at presentation and past medical history were tested as predictors of the first thrombotic complication through multivariate Cox regression analysis and a categorical score generated based on the results.\n\n\nRESULTS\nSixty-four thromboses were recorded in 54 patients, of whom seven with thrombosis on admission and 47 with thrombosis during hospitalisation. Patients with thrombosis were mainly Caucasian and diabetic, had marked baseline signs of inflammation and organ damage, lower PaO2/FiO2 ratio, higher D-dimer levels and history of major haemorrhages. The latter three variables were independently associated to thrombotic complications and concurred to a 0-5 score, which accounted for 80% of the total sample variability. Patients with three or more points of the newly generated score were at higher risk for thrombotic complications (HR=4.9, p<0.001). Patients with thrombotic complications were more likely to be admitted to intensive care and/or to die (HR=1.9, p=0.036). Five of 180 patients were diagnosed with disseminated intravascular coagulation and three of them died. Eleven minor and no major bleeding events were observed.\n\n\nCONCLUSIONS\nPatients with COVID-19 are at increased risk for thrombosis and might be stratified on admission based lower Pao2/FiO2 ratio, higher D-dimer levels and history of major haemorrhages.