An optimized D-dimer cut-off value to predict pulmonary embolism in COVID-19 patients: a retrospective study

Abstract

<b>Background:</b> Pulmonary embolism (PE) is a common complication in COVID-19 patients. Elevated D-dimer levels are observed even in the absence of PE, reducing its discriminative ability as a screening test. <b>Aim:</b> To determine the optimal D-dimer cut-off value to predict PE in COVID-19 patients. We hypothesized the cut-off value to be higher than the cut-off values implemented in the YEARS algorithm (van der Hulle et al. Lancet 2017;390:289-297). <b>Methods:</b> Confirmed COVID-19 patients with a Computed Tomography Pulmonary Angiography (CTPA) performed ≤5 days after admission between 03-2020 and 02-2021, at Medisch Spectrum Twente, Netherlands, were retrospectively analysed. The association between PE and D-dimer levels prior to CTPA, and other potential predictors, was analysed using logistic regression analyses. The optimal cut-off value was identified using ROC curve analyses. <b>Results:</b> In 142 patients, PE prevalence was 20.4%. D-dimer level prior to CTPA and white blood cell (WBC) count were associated with PE (OR 1.05; 95%-CI:1.03–1.07; p&lt;0.001 and OR 1.10; 95%-CI:1.00–1.22; p=0.081, respectively). The final regression model contained solely D-dimer level prior to CTPA (AUC 0.83) due to a marginal difference in diagnostic accuracy with the model including WBC count (AUC 0.84). The optimal cut-off value was 750 ng/ml (sensitivity 100%; specificity 19.5%; NPV 100%; PPV 24.2%). In total, 13.3% of the patients without PE had a 500≤D-dimer&lt;750 ng/ml. <b>Conclusion:</b> A D-dimer level &lt;750 ng/ml rules out PE with 100% certainty. Compared to the YEARS 500 ng/ml cut-off value, 13.3% less patients are in need for a CTPA, with similar sensitivity. Future research is required for external validation.