Simple demographic, laboratory and chest radiograph variables can identify COVID-19 patients with pulmonary thromboembolism: a retrospective multicentre United Kingdom study

Abstract

Objectives: To (1) identify discriminatory demographic, laboratory and initial CXR findings; (2) explore correlation between D-dimer and radiographic severity scores; and (3) assess accuracy of published D-dimer thresholds to identify pulmonary thromboembolism (PTE) in COVID-19 patients. Methods: Retrospective study including all COVID-19 patients admitted from 1st to 30th April 2020 meeting inclusion criteria from 25 United Kingdom hospitals. Demographics, blood results, CXR and CTPA findings were compared between positive and negative PTE cohorts using uni- and multivariable logistic regression. Published D-dimer cut-offs were applied. Results: 389 patients were included [median age 63; 237 males], of which 26.2% had a PTE. Significant univariable discriminators for PTE were peak D-dimer, sex, neutrophil count at the time of the D-dimer and at admission, abnormal CXR, and CXR zonal severity score. Only neutrophil count at peak D-dimer remained significant for predicting PTE on multivariable analysis (p = 0.008). When compared with the published literature, sensitivity for PTE were lower than those published at all cut-off values; however, specificity at different cut-offs was variable. Conclusions: In this multicentre COVID-19 cohort, univariable admission factors that could indicate pulmonary thromboembolism were male sex, high neutrophil count and abnormal CXR with a greater CXR zonal severity score. The accuracy levels of published D-dimer thresholds were not reproducible in our population. Advances in knowledge: This is a large multicentre study looking at the discriminatory value of simple variables to determine if a patient with COVID-19 has PTE or not, in addition to comparing D-dimer cut-off values against published values.