Background
Previous studies that assessed risk factors for venous thromboembolism (VTE) in COVID-19 patients have shown inconsistent results. Our aim was to investigate VTE predictors by both logistic regression (LR) and machine learning (ML) approaches, due to their potential complementarity.
Methods
This substudy of a large Brazilian COVID-19 Registry included COVID-19 adult patients from 16 hospitals. Symptomatic VTE was confirmed by objective imaging. LR analysis, tree-based boosting and bagging were used to investigate the association of variables upon hospital presentation with VTE.
Results
Among 4,120 patients (55·5% men, 39·3% critical patients), VTE was confirmed in 6·7%. In multivariate LR analysis, obesity (OR 1·50, 95%CI 1·11-2·02); being an ex-smoker (OR 1·44, 95%CI 1·03-2·01); surgery ≤ 90 days (OR 2·20, 95%CI 1·14-4·23); axillary temperature (OR 1·41, 95%CI 1·22-1·63); D-dimer ≥ 4 times above the upper limit of reference value (OR 2·16, 95%CI 1·26-3·67), lactate (OR 1·10, 95%CI 1·02-1·19), C-reactive protein levels (CRP, OR 1·09, 95% CI 1·01-1·18); and neutrophil count (OR 1·04, 95%CI 1·005-1·075) were independent predictors of VTE. Atrial fibrillation, peripheral oxygen saturation/inspired oxygen fraction (SF) ratio and prophylactic use of anticoagulants were protective. Temperature at admission, SF ratio, neutrophil count, D-dimer, CRP and lactate levels were also identified as predictors by ML methods.
Conclusion
By using ML and LR analyse, we showed that D-dimer, axillary temperature, neutrophil count, CRP and lactate levels are risk factors for VTE in COVID-19 patients.