Wells' prediction rules for pulmonary embolism: valid in all clinical subgroups?

Abstract

Pulmonary embolism is major cause of hospital death. Clinical prediction rules such as Wells’ prediction rules can help in selection of at-risk patients who need further testing for pulmonary embolism. We evaluated the usefulness of such criteria for detection of patients with diagnosed pulmonary embolism. Patients enrolled in National Research Institute of Tuberculosis and Lung Disease (NRITLD) deep venous thrombosis (DVT) registry were evaluated and those with objective data about presence or absence of pulmonary embolism were selected for this study. Diagnosis of pulmonary embolism was based on computed tomography pulmonary angiography (CTPA). We calculated the embolic burden in those with CTPA-confirmed pulmonary embolism. Eighty-six patients entered the study (58 males, 28 females, mean age = 54.39 ± 1.74 years). Fifty-four cases had coexisting pulmonary embolism (embolic burden score: 10.77 ± 1.181). Embolic burden score was correlated to presence of massive pulmonary embolism (Pearson rho: 0.43, P = 0.002). There was no association between Wells’ pulmonary embolism score and the occurrence of pulmonary embolism (Spearman's rho: 0.085, P = 0.51). Dividing the patients into two, or three, risk groups according to Wells’ model did not reveal an association with occurrence of pulmonary embolism either (P = 0.99 and P = 0.261, respectively). Tachycardia and hemoptysis were the only parameters from the Wells’ pulmonary embolism score correlated to presence of pulmonary embolism (Spearman's rho: 0.373, P < 0.000 and Spearman's rho: 0.297, P = 0.005, correspondingly). Wells’ pulmonary embolism score could not predict the occurrence of pulmonary embolism in DVT patients suspected of having coexisting pulmonary embolism. Until further studies shed light on this patient subset, overreliance on Wells’ prediction rules as the solo decision making tool should be cautioned.