Assessment of clinical probability scores for pulmonary embolism diagnosis during pregnancy and postpartum in women with a history of venous thromboembolism: a Highlow ancillary study.

Background: The value of pretest clinical probability scores in the diagnosis of pulmonary embolism (PE) during pregnancy and postpartum is unknown in women with a history of venous thromboembolism (VTE). Objectives: We evaluate the modified Wells, revised Geneva, and pregnancy-adapted Geneva (PAG) scores for the diagnosis of PE during pregnancy and the postpartum period in women with a history of VTE. Methods: Data from a multicenter randomized trial (Highlow) including 1110 pregnant women with a history of VTE and treated with either weight-adjusted intermediate-dose or fixed low-dose low-molecular-weight heparin subcutaneously once daily until 6 weeks postpartum were used. The modified Wells, revised Geneva, and PAG scores were calculated retrospectively in all women with a clinical suspicion of PE, and their discriminative capacity was assessed. Receiver operating characteristic (ROC) curve analysis was performed for quantitative variables and the optimal threshold defined. Results: There were 102 suspected cases of PE, of which 12 were confirmed events. During pregnancy, the ROC curves showed an area under the curve of 0.68, 0.33, and 0.36 for the Wells, Geneva, and PAG scores, respectively. During postpartum, the ROC curves showed an area under the curve of 0.75, 0.55, and 0.52 for the Wells, Geneva, and PAG scores, respectively. Conclusion: The 3 pretest clinical scores have modest discriminatory power, during both the antepartum and the postpartum period, to classify patients into 3 categories of pretest clinical probability. Further work is required to develop clinical-decision tools to exclude imaging in pregnant women with prior VTE with suspected PE in pregnancy.