“Have you thought about pulmonary embolism [PE]? I’ll come see the patient, but would you mind putting in for a CTA? You know, almost a fifth of patients with syncope have PE.”

Welcome to the hospital admission conversations of the future.

To great commotion and kerfuffle, the New England Journal of Medicine has delivered a new shiny pearl of medical evidence regarding the prevalence of PE in patients with syncope. The study in question, the Pulmonary Embolism in Syncope Italian Trial (PESIT), prospectively enrolled patients admitted from the emergency department with a first-time diagnosis of syncope.¹ These authors were concerned PE was an underappreciated cause of syncope in this population and undertook a systematic approach to evaluate its diagnosis. First, every patient admitted was risk-stratified using Wells criteria for PE. Then patients who were stratified as “PE unlikely” were tested using D-dimer. Finally, patients with positive D-dimer results and those who stratified as “PE likely” underwent imaging, either computed tomography pulmonary angiograms (CTPA) or ventilation-perfusion lung scanning.

The shock and awe: 560 patients were included in the study, 230 underwent imaging or autopsy, and 97 received a diagnosis of PE. As summarized by the authors of the study: “In conclusion, among patients who were hospitalized for a first episode of syncope and who were not receiving anticoagulation therapy, pulmonary embolism was confirmed in 17.3 percent (approximately one of every six patients).”

This number is immediately eye-catching because it’s entirely discordant with the 2 to 3 percent prevalence of PE in syncope from previously published literature.^2,3 The authors contend their number is more accurate because of their systematic evaluation of consecutive first-time admissions regardless of associated symptoms or suspected underlying etiology. They conclude they have found these “occult” PEs as true missed causes of syncope. This is where your conversations with your admitting consultant may become increasingly frustrating. Chances are, they haven’t delved into the details.

The first consideration is with regard to selection bias. These patients are higher-risk for PE than otherwise might be expected in the syncope population just based on their initial sorting. Of the 2,584 patients receiving a diagnosis of syncope in the emergency department, 1,867 were discharged home. We aren’t given any information on these patients nor a description of specific testing performed in the emergency department other than a vague accounting of the suspected underlying benign etiology of syncope. However, in this greater context, the diagnosis of PE is made in 97 patients out of 2,584, or 3.8 percent of cases, which is much closer to the familiar 2 to 3 percent range. The cohort these authors are systematically evaluating, just based on their initial clinical selection, is more likely to have a serious underlying etiology for syncope, including PE.

In the syncope patient without any clinical symptoms relating to PE, it is reasonable to hypothesize some of these radiographic findings are unrelated and of uncertain clinical significance.

Another important factor grossly inflates the number of PEs identified in this inpatient cohort: the many patients whose PE should have been identified in the emergency department prior to study inclusion. There were 60 patients admitted to the hospital documented as having clear and obvious signs of deep vein thrombosis (DVT) in their lower extremities but for whom apparently no objective evaluation for PE was performed. Following the authors’ protocol subsequent to admission, two-thirds of these patients proved to have PE. Then an additional 77 patients exhibited tachypnea on admission, and another 107 exhibited tachycardia. Similar numbers for other risk factors for PE, such as previous DVT or PE, immobility, recent surgery, or active cancer were noted. Between 30 and 60 percent of these patients with physiologic changes or risk factors for PE were ultimately diagnosed with PE. It is legitimate to question whether these patients received an adequate ED evaluation considering PE as a diagnosis, since many appear clinically obvious. Indeed, the authors report only 24 of their 97 patients with PE were absent clinical signs or symptoms, suggesting their prevalence estimate is inflated not by occult PE but simply by PE missed by the emergency department.

Furthermore, the elephant in everyone’s room is the troubling prevalence of false-positive imaging for PE, particularly when the pretest likelihood is low.⁴ In this cohort, 24 of 72 patients undergoing CTPA had found PE in segmental or subsegmental locations. Between one-quarter and one-half of PEs in these distal branches were found to be false-positives on subsequent subspecialty radiologist review.⁵ Another 24 patients received diagnosis of PE based on ventilation-perfusion scanning, and half of those found perfusion defects in between 1 and 25 percent of the area of both lungs. Again, as the size of the defect decreases, the likelihood of false-positives increases.

Lastly, we enter into the complicated question of whether these PEs are acutely related to the syncopal episode or whether they represent chronic or incidental findings. In the syncope patient without any clinical symptoms relating to PE, it is reasonable to hypothesize some of these radiographic findings are unrelated and of uncertain clinical significance. Attributing temporary global cerebral hypoperfusion to PE requires a complicated cascade of vascular obstruction, vasoconstriction, and right ventricular afterload. In the instance that these are incidental or represent overdiagnosis, it is reasonable to be concerned about the bleeding risks of long-term anticoagulation in this mostly elderly cohort. PE is not a zero-miss diagnosis for this precise reasoning. The risks of treatment outweigh the benefits for some patients diagnosed with PE.

At the end of the day, the important takeaway is this: These data don’t generalize to our typical emergency department, which pursues a reasonable explanation for vital sign abnormalities and clinical signs suspicious for DVT. It is important to consider the diagnosis of PE in patients with syncope, but most of the patients with clinically important or true-positive PE will have obvious signs pointing to a need for objective testing. There is no indication this study reflects an otherwise unexpected population of PE in syncope following admission to the hospital. I fear this study will lead to an increase in harmful low-value overtesting.

References

1. Prandoni P, Lensing AW, Prins MH, et al. Prevalence of pulmonary embolism among patients hospitalized for syncope. N Engl J Med. 2016;375:1524-1531.
2. Cook OG, Mukarram MA, Rahman OM, et al. Reasons for hospitalization among emergency department patients with syncope. Acad Emerg Med. July 18, 2016; [ePub ahead of print]
3. Blanc JJ, L‘her C, Touiza A, et al. Prospective evaluation and outcome of patients admitted for syncope over a 1 year period. Eur Heart J. 2002;23(10):815-820.
4. Stein PD, Fowler SE, Goodman LR, et al. Multidetector computed tomography for acute pulmonary embolism. N Engl J Med. 2006;354(22):2317-2327.
5. Hutchinson BD, Navin P, Marom EM, et al. Overdiagnosis of pulmonary embolism by pulmonary CT angiography. AJR Am J Roentgenol. 2015;205:271-277.