Pulmonary embolism (PE) is responsible for maternal death in approximately one of every 100,000 deliveries in Western countries. Diagnosing PE is problematic because characteristic symptoms such as shortness of breath and leg edema are common in normal pregnancy, and frequently used imaging tests such as computed tomographic pulmonary angiography (CTPA) expose mother and fetus to radiation. Algorithms using clinical prediction rules and blood testing for D-dimer level have been shown to reduce the need for CTPA in patients with suspected PE.1,2 However, the safety of this approach in pregnant women has not been confirmed.
Dr. Liselotte M. van der Pol and colleagues performed a multicenter, prospective cohort study that used the YEARS criteria (clinical signs of deep-vein thrombosis [DVT], hemoptysis, PE as the most likely diagnosis) and a D-dimer level to direct management of pregnant patients with new onset or worsening chest pain or dyspnea. Women who had none of the YEARS criteria plus a D-dimer level less than 1,000 ng/mL or one to three YEARS criteria plus a D-dimer level less than 500 ng/mL did not undergo CTPA, and anticoagulant therapy was withheld. Women with any other combination of results underwent a CTPA (or compression ultrasound for signs and symptoms consistent with DVT). The primary outcome measure was the proportion of women who had anticoagulant therapy withheld based on the algorithm who were diagnosed with symptomatic PE or DVT at 90 days.
A total of 498 pregnant women (mean age of 30 years), the majority in their third trimester (46%), were enrolled. Of 195 women who had PE ruled out based on the algorithm, one developed a DVT (YEARS 0, D-dimer 480 ng/mL, 2nd trimester) during follow-up, yielding a total missed venous thromboembolism (VTE) rate of 0.51 percent (95% CI, 0.09%-2.9%). Of the entire study population, PE was diagnosed in 3 percent, and DVT in 1 percent. CTPA was not indicated in 65 percent of patients who presented within the first trimester, which was also the time frame with the highest incidence of confirmed PE (6.8%). In conclusion, PE was safely excluded using the algorithm in 39 percent of pregnant women without the need for CTPA.
In Brief
This study confirms the value of a diagnostic strategy that uses a validated clinical prediction rule to dictate the D-dimer threshold for excluding VTE.3,4 The premise of this approach is that a high D-dimer threshold is used in settings with low VTE prevalence (e.g., YEARS = 0), whereas a lower D-dimer threshold is used when the prevalence is higher (YEARS = 1-3). Using either a clinical prediction rule or D-dimer testing alone is not as accurate and may partly explain the discrepant results of another pregnancy-associated PE diagnostic study.5
Pregnancy-adapted YEARS algorithm for the management of suspected acute pulmonary embolism in pregnant patients. CT, computed tomography.
Pregnancy-adapted YEARS algorithm for the management of suspected acute pulmonary embolism in pregnant patients. CT, computed tomography.
Limitations of this study include selection bias (all women had at least a moderate suspicion of PE) and the D-dimer level for some patients was known prior to (and may have influenced) determination of the YEARS score (Figure). It is also worthy of note that the subgroup with highest incidence of PE (first trimester) was also the smallest subgroup in terms of numbers (n=74), which resulted in an upper limit of the CI reaching 15 percent. This subgroup will need to be watched carefully as the study algorithm becomes more widely used. Overall, given the reduction in need for radiation exposure and the difficulties in conducting research in pregnant women, this study is a laudable accomplishment.
In Brief
Limitations of this study include selection bias (all women had at least a moderate suspicion of PE) and the D-dimer level for some patients was known prior to (and may have influenced) determination of the YEARS score (Figure). It is also worthy of note that the subgroup with highest incidence of PE (first trimester) was also the smallest subgroup in terms of numbers (n=74), which resulted in an upper limit of the CI reaching 15 percent. This subgroup will need to be watched carefully as the study algorithm becomes more widely used. Overall, given the reduction in need for radiation exposure and the difficulties in conducting research in pregnant women, this study is a laudable accomplishment.
