Accuracy of PE rule-out strategies in pregnancy: secondary analysis of the DiPEP study prospective cohort

OBJECTIVE: Recent studies suggest that combinations of clinical probability assessment (the YEARS algorithm or Geneva score) and D-dimer can safely rule out suspected pulmonary embolism (PE) in pregnant women. We performed a secondary analysis of the DiPEP (Diagnosis of Pulmonary Embolism in Pregnancy) study data to determine the diagnostic accuracy of these strategies. METHODS: The DiPEP study prospectively recruited and collected data and blood samples from pregnant/postpartum women with suspected PE across 11 hospitals and retrospectively collected data from pregnant/postpartum women with diagnosed PE across all UK hospitals (15 February 2015 to 31 August 2016). We selected prospectively recruited pregnant women who had definitive diagnostic imaging for this analysis. We used clinical data and D-dimer results to determine whether the rule out strategies would recommend further investigation. Two independent adjudicators used data from imaging reports, treatments and adverse events up to 30 days to determine the reference standard. RESULTS: PEs were diagnosed in 12/219 (5.5%) women. The YEARS/D-dimer strategy would have ruled out PE in 96/219 (43.8%) but this would have included 5 of the 12 with PEs. Sensitivity for PE was 58.3% (95% CI 28.6% to 83.5%) and specificity 44.0% (37.1% to 51.0%). The Geneva/D-dimer strategy would have ruled out PE in 46/219 (21.0%) but this would have included three of the 12 with PE. Sensitivity was 75.0% (95% CI 42.8% to 93.3%) and specificity 20.8% (95% CI 15.6% to 27.1%). Administration of anticoagulants prior to blood sampling may have reduced D-dimer sensitivity for small PE. CONCLUSION: Strategies using clinical probability and D-dimer have limited diagnostic accuracy and do not accurately rule out all PE in pregnancy. It is uncertain whether PE missed by these strategies lead to clinically important consequences.