Recent studies suggest that combinations of clinical probability assessment and D-dimer can safely rule out suspected PE in pregnant women. Van der Pol (NEJM 2019) reported that a pregnancy-modified YEARS algorithm and D-dimer ruled out PE in 195/498 (39%) and Righini (Ann Intern Med 2018) reported that the Geneva score and D-dimer ruled out PE in 46/395 (12%) without adverse outcome. We undertook a secondary analysis of pregnant women with suspected PE prospectively recruited to the DiPEP study to determine the diagnostic accuracy of these strategies.

The DiPEP study collected data and blood samples from pregnant/postpartum women with suspected PE across 11 UK hospitals and with diagnosed PE from all UK hospitals over 18 months. We selected prospectively recruited pregnant women who had definitive diagnostic imaging for analysis. We used clinical data and D-dimer results to determine whether the van der Pol and Righini strategies would recommend further investigation and imaging results to determine whether this would detect or miss PE.

We analysed 219 prospectively enrolled patients, including 12 (4.6%) with PE. The van der Pol strategy indicated no PE in 96/219 (43.8%), but this would have included 5/12 false negative cases with PE. Sensitivity for PE was 58.3% (95% CI 28.6–83.5%) and specificity 44.0% (37.1–51.0%). The Righini strategy indicated no PE in 46/219 (21.0%) but this would have missed 3/12 cases with PE. Sensitivity was 75.0% (21.9–98.7%) and specificity 20.8% (15.6–27.1%).

Strategies using clinical probability and D-dimer do not accurately rule out PE in pregnancy. The absence of adverse events in the published management studies may reflect lack of statistical power to detect clinically important adverse event rates. We therefore recommend against using clinical probability assessment and D-dimer testing to rule out suspected PE in pregnancy.