Introduction: Patients with COVID-19 and cardiovascular disease risk factors (CVDRF) have been reported to develop coagulation abnormalities frequently. However, there are limitations in conventional predictive models for the occurrence of thromboembolism in patients with COVID-19 and CVDRF. Methods: Among data on 1518 hospitalized patients with COVID-19 registered with CLAVIS-COVID, a Japanese nationwide cohort study, 693 patients with CVDRF were subjected to least absolute shrinkage and selection operator (LASSO) analysis; a method of shrinking coefficients for reducing variance and minimizing bias to increase predictive accuracy. LASSO analysis was performed to identify risk factors for systemic thromboembolic events; occurrence of arterial and venous thromboembolism during the index hospitalization as the primary endpoint. Results: LASSO analysis identified a prior systemic thromboembolism, male sex, hypoxygenemia requiring invasive mechanical ventilation support, C-reactive protein levels and D-dimer levels at admission, and congestion on chest X-ray at admission as potential risk factors for the primary endpoint. The developed risk model consisting of these risk factors showed good discriminative performance (AUC-ROC: 0.83, 95 % confidence interval [CI]: 0.77–0.90), which was significantly better than that shown by D-dimer (AUC-ROC: 0.70, 95 % CI: 0.60–0.80) (p < 0.001). Furthermore, systemic embolic events were independently associated with in-hospital mortality (adjusted odds ratio: 3.29; 95 % CI: 1.31–8.00). Conclusions: Six parameters readily available at the time of admission were identified as risk factors for thromboembolic events, and these may be capable of stratifying the risk of in-hospital thromboembolic events, which are associated with in-hospital mortality, in patients with COVID-19 and CVDRF.
ASJC Scopus subject areas