Up until 2016, all was well in the world of syncope and pulmonary embolism (PE) risk. Then Prandoni et al [1] published their multicenter study that used a mandated protocol and found a prevalence of PE 17.3% (95% CI 14.2 – 20.5%) in patients hospitalized for syncope. Based on their numbers, I calculated an ED prevalence of 4.1% (95% CI 3.4 – 5.0%), with a worst-case scenario of 7% (95% CI 6.1 – 8.1%). (84 patients refused admission or declined participation.) The study had a mandated protocol of applying Wells, PERC, and D-Dimer and/or imaging testing based on calculated risk using Wells score. With PE rates reported this high, clinicians may feel compelled to test all patients for PE who present to the ED with syncope.

In direct response to Prandoni et al [1], Oqab et al [2] performed a systematic review and meta-analysis that found a lower prevalence of 0.8% (95% CI 0.5 – 1.3%) in ED patients. In this meta-analysis, only 5 of 12 studies were prospective, methodologies were incomplete, and they excluded Prandoni despite it meeting their inclusion criteria! Importantly, in contrast to Prandoni et al [1], they were all “pragmatic” – without pre-determined mandated protocolized evaluations.

 

Thiruganasambandamoorthy et al [3] pooled two large multisite (one in the US and one in Canada) prospective trials and reported an ED prevalence of 0.6% (95% CI 0.5 – 0.8%). These findings were consistent with the meta-analysis by Oqab [2], but not with Prandoni [1]. As the authors note, both trials in Thiruganasambandamoorthy et al [3] were also pragmatic, so not every patient presenting with syncope underwent systematic approach testing for PE.

And finally, most recently, Badertscher et al [4] performed a multicenter, multinational, prospective study with 2-year follow up for all ED patients presenting with syncope within 12 hours. As in Prandoni et al[1], there was a pre-specified mandated protocol for the evaluation of all patients, however, this protocol was not designed to workup all patients with syncope for PE. The protocol was based on the 2018 ESC Guidelines for the diagnosis and management of syncope [5], however, patients that didn’t receive D-Dimer testing, for any reason, were excluded. It is unclear in the paper why some patients did or didn’t receive D-Dimer testing – these patients might have “PERC’d out” after initial risk assessment with Wells. The authors found an ED prevalence of PE of 1.4% (95% CI: 0.87% – 2.11%), and incidence of new PEs and/or cardiovascular death during 2-year follow-up was 0.9% (95% CI: 0.5% – 1.5%) in patients deemed high risk or with positive D-dimers who did not receive imaging. They also reported a PE prevalence in patients hospitalized for syncope of 2.3% (95% CI: 1.4% – 3.7%) and also reported from their 2-year follow-up an incidence of new PEs and cardiovascular death of 0.9% (95% CI: 0.4% – 2.0%).

It’s important for us to look at how a prospective, protocolized mandated approach to this research question compares to a pragmatically designed trial. In a prospective, protocolized approach, all patients that meet inclusion criteria are worked up following the same prespecified pathway. In Prandoni et al[1], this meant applying Wells score and then performing a D-dimer – not age-adjusted – and then performing either a VQ scan or CT pulmonary angiography if the Wells score placed them as high risk or the D-dimer was positive. This approach is meant to remove physician bias from the workup. Problems with these types of studies are generally enrollment-based, as patients are only enrolled when a research staff member is present. Or, if they rely on EPs, then enrollment is often not uniformly applied.

In pragmatic studies, like Thiruganasambandamoorthy et al[3], you risk verification bias because the workup is dependent on current clinical practice. As such, a patient whom an EP thinks doesn’t have a PE won’t get any testing for PE. This can lead to verification bias, where patients in whom no PE is suspected are counted in the “no PE” group. This might underestimate the total number of PEs. In Thiruganasambandamoorthy et al[3] seven patients (0.1% of discharged patients) returned to the ED and were later found to have PEs. These patients made up 12.5% of all diagnosed PEs. However, there could have been other patients who were lost to follow up and presented to other non-participating hospitals and were found to have a PE or patients who had an event after the 30-day follow up in the study.

A protocolized study is likely to find more PEs, many of which may be clinically insignificant. A pragmatic study is more likely to undercount PEs (verification bias), but with adequate follow-up, it will hopefully account for the clinically significant ones.

As one can see, there is huge variability in these studies. Prandoni et al[1] is clearly an outlier, but it is actually one of the better-designed trials. As with all studies, your PICO (population, intervention, control, and outcomes) question and your protocol/methodology greatly contribute to the outcomes of the study. Based on the preponderance of the evidence, the prevalence of PE in ED patients presenting with asymptomatic syncope is likely lower than in Prandoni et al[1], but the exact answer is unclear. Below is a table summarizing all of the studies with their strengths and incidence rates. These studies, when applied to the right population, can help you better understand a patient who presents with syncope and their risk of PE and guide your decision making.