Prevalence of Pulmonary Embolism in Patients Presenting With Syncope

Empiric diagnostic testing is not warranted in all syncope patients

Benefits in NNT

1 in 125 emergency department patients with syncope were found to have a PE
1 in 100 hospitalized patients with syncope were found to have a PE
0.8% of ED patients with syncope had PE
1.0% of hospitalized patients were diagnosed with PE

Harms in NNT

Not applicable: data on harms were not available
Not applicable: data on harms were not available
View As:


Oqab Z, Ganshorn H, Sheldon R. Prevalence of pulmonary embolism in patients presenting with syncope. A systematic review and meta-analysis. Am J Emerg Med 2018;36:551–5.

Study Population: 6,608 ED patients (nine studies) and 975 hospitalized patients (three studies) presenting with syncope

Efficacy Endpoints

Prevalence of PE in patients presenting to the ED with syncope

Harm Endpoints

No harm endpoints were assessed


Syncope accounts for 1% to 3% of emergency department (ED) visits and 1% to 6% of hospital admissions.1, 2 There are numerous etiologies, ranging from relatively benign vasovagal syncope to dangerous dysrhythmias. The ED evaluation and management of syncope is composed of history, examination, and typically an electrocardiogram, with further investigation dependent on clinical decision making and suspected conditions.2 Previously, pulmonary embolism (PE) was thought to account for a small minority of patients with syncope. However, a recent study by Prandoni and colleagues3 reported a high prevalence of PE in admitted patients with syncope (3.8% of ED patients and 17.3% of hospitalized patients). Evaluating for PE in all patients with syncope carries significant risks including radiation exposure, contrast-induced nephropathy, and adverse events from anticoagulation therapy.4, 5 In this evidence-based review, we summarize and critically appraise a published meta-analysis that evaluated the overall prevalence of PE in patients presenting with syncope to provide guidance to clinicians regarding testing decisions in this population.6

This meta-analysis included studies evaluating patients with syncope who presented to the ED or were admitted to the hospital that reported underlying etiologies, which included PE. There were no limitations on age, language, time, or setting, and to assess methodologic quality, authors modified an existing quality scale.6 The authors identified 1,920 studies, of which 12 papers (excluding Prandoni et al.) met inclusion criteria. Nine studies (n = 6,608 patients) took place in the ED, and three studies (n = 975 patients) occurred in the hospital environment. Weighted median age in ED patients was 61.5 years, compared to 67.1 years in hospitalized patients. PE was confirmed through computerized tomography angiography (CTA) of chest, ventilation perfusion scan, pulmonary angiography, or autopsy.

Results of the current meta-analysis suggest a low prevalence of PE in patients presenting with syncope: 0.8% (95% confidence interval [CI] = 0.5%–1.3%) in ED patients (number needed to screen = 125) and 1.0% (95% CI = 0.5%–1.9%) in hospitalized patients (number needed to screen = 100), with an overall prevalence of 0.9% (95% CI = 0.6%–1.3%).6


The meta-analysis discussed here had several important limitations. First, the authors included both prospective and retrospective data. Additionally, only four of the included studies discussed specific diagnostic strategies for PE in this meta-analysis.6 Another concern is that the authors utilized their own modified scale to assess methodologic quality, rather than using one of the more established tools, such as QUADAS-2 or the Newcastle-Ottawa criteria.7 Moreover, the decision to order CTA was mostly based on clinician judgment. Finally, the presenting symptoms, patient characteristics, and rationale for obtaining the CTA were not discussed in most of the included trials. While CTA of chest with contrast possesses high sensitivity and specificity for diagnosis of PE in low pretest probability patients, test characteristics decrease in patients with high pretest probability.8 Discordance among radiologists for diagnosis of PE can also be severe, with poor interreader reliability.9

A second important consideration is that syncope has a significant number of potential etiologies, and determining a specific cause can be difficult. Therefore, as expected, clinical heterogeneity among the included studies was significant.6 Since the studies did not systematically screen for PE, it is unclear how many cases may have been missed. Follow-up for patients discharged from the ED to ensure they did not have PE was unclear in the majority of studies. Studies also demonstrated variable patient populations and baseline characteristics.6 Most importantly, whether identifying these positive cases of PE affected long-term outcomes (e.g., mortality) of the patients is not known. PE can be asymptomatic and/or an incidental finding. A significant portion of patients demonstrate incidental PE at the time of autopsy, with rates ranging from 9% to 63%.5 Thus, PE may occur and resolve without clinical effect.

Another major caveat for diagnostic evaluation of syncope patients for PE is establishing causality. To cause syncope, a pulmonary blood clot must result in dysrhythmia, acute right ventricular failure, or a Bezold-Jarisch reflex.10 The literature suggests only PE located in the main pulmonary or lobar arteries are associated with syncope.10 However, in the study by Prandoni et al.,3 approximately one-third of PE were segmental or subsegmental, which would be unlikely to result in syncope. Therefore, it is unclear whether the diagnosed PEs were associated with the syncope or incidental findings. Additionally, it is unclear how many cases were false positives due to imaging artifact.

Based on the low prevalence of PE in patients with syncope in this meta-analysis (low-quality evidence), dedicated testing for PE in all syncope patients is not recommended. Overtesting for PE may result in risks from the testing itself, as well as side effects from anticoagulation given in cases with false-positive test results or clinically insignificant cases. We assign testing for PE in all syncope patients red (harm > benefit). While consideration of PE in patients with syncope is warranted, the decision to trigger diagnostic evaluation for PE should be guided by proper risk stratification using history and physical examination.

The original manuscript was published in Academic Emergency Medicine as part of the partnership between and AEM.


Brit Long, MD; Alex Koyfman, MD; Michael Gottlieb, MD, RDMS
Supervising Editor: Shahriar Zehtabchi, MD


October 1, 2019