Should we look for pulmonary emboli in patients with syncope?

The standard evaluation of patients with syncope has changed little over the years, with considerable continuity between American Family Physician reviews published in 2005, 2011, and this March. But the field received an unexpected jolt last October, when Dr. Paolo Prandoni and colleagues published a cross-sectional study in the New England Journal of Medicine that calculated a surprisingly high prevalence of pulmonary embolism (PE) of 17.3% in patients hospitalized for a first episode of syncope. In this study, 230 of 560 patients at 11 Italian hospitals who did not have a low pretest probability of PE by the Wells rule and negative D-dimer assay underwent computed tomographic pulmonary angiography (CTPA) or ventilation-perfusion lung scans. 97 of these patients had evidence of PE. Overall, the investigators identified PE in 25% of patients with no alternative explanation for syncope and 13% of patients with an alternative explanation.

Algorithms for syncope evaluation focus on identifying cardiac and neurally mediated causes and do not include routine testing for PE. The question raised by the Italian study is if all inpatients with syncope need evaluation, since nearly 1 in 6 patients may have PE. There are good reasons not to jump to this conclusion. First, we don't know if the prevalence of PE in this study was representative of the general population of adults hospitalized for syncope, as it only included patients admitted from emergency departments (and not patients initially evaluated in primary care settings). Second, the study did not determine if PE was the etiology of syncope, as opposed to an incidental finding. This is important because evidence suggests that the enhanced sensitivity of CTPA for detecting small, subsegmental PE is increasing overdiagnosis and overtreatment of clinically insignificant clots. Finally, the absence of a comparison group means that it is not known if the systematic workup for PE improved the patients' prognoses or just added additional cost and risk to the evaluation.

A retrospective cross-sectional study in JAMA Internal Medicine recently used clinical and administrative data from 4 hospitals in Toronto to estimate the prevalence of PE in 1305 patients with a first episode of syncope who were not receiving anticoagulation at the time of admission. 120 of these patients received CTPA, ventilation-perfusion scan, and/or compression ultrasonography, leading to the diagnosis of PE in 18 patients. As opposed to the Italian study, where all persons received at least a D-dimer test, the decision to evaluate for PE was driven by clinical judgment. Nonetheless, of 146 patients in the Toronto study who received any test for PE (presumably those who were most likely to have one), only 12% were positive, and the overall prevalence of PE was a mere 1.4%. The authors concluded that "there is little, if any, justification for routine testing for [venous thromboembolism] in all patients hospitalized for a first episode of syncope."

Based on both studies, I agree that evidence does not support routinely looking for PEs in patients with syncope. In patients with possible signs or symptoms, another clinical decision rule may be used to confidently rule out the diagnosis without resorting to the less sensitive Wells rule or less specific D-dimer test, where a positive result often leads to unnecessary CTPA.

**

A slightly different version of this post first appeared on the AFP Community Blog.