Pulmonary Embolism Evaluation in Syncope: or How I Learned to Stop Worrying and Love the CT Scanner

Hi all! Bill here. Today’s piece is written by Aalap Shah, a man with the dubious honor of being one of my closest medical school friends. Aalap is a third year emergency medicine resident at the University of Cincinnati Medical Center. Here is an embarrassing picture of him! Excited to have an EM perspective on our IM blog given how thoroughly our specialties rely on one another and how limited our exchange of ideas is in comparison to that reliance. I really enjoyed his take on a piece I had read before but not understood very well, and I think you will too! Without further ado:

The PESIT Trial

Before we begin, I know what you’re thinking: “Great, here comes the ED doc with some article that says we should get CT pulmonary angiograms (CTPA’s) on everybody.” To those people I would say, joke’s on you! Obviously we don’t need evidence-based reasons to scan people!

But seriously, we are here to talk about a study that got some big time press, in a (the?) high impact journal of our time. The PESIT trial by Prandoni et al. set out to determine the prevalence of PE among patients hospitalized for syncope. Interestingly, this was a clinical question that had largely not been investigated in the literature, though we do know from prior studies that approximately 6-10% of patients with PE are found to have an initial presentation of syncope. Still, the authors felt that PE had been underinvestigated as a potential cause of syncope and set out to determine the value of testing in this specific scenario.

If You’re Only Going to Read One Paragraph

In this study, the prevalence of PE in a relatively high risk population was 17.3% (or 1 in 6) patients who were admitted to the hospital with syncope. The study was cross-sectional in nature and did not specifically set out to determine a cause and effect relationship between the two. They did not follow up with patients (“Typical.” – Bill) and did not note changes in outcome or clinical significance based on identification. Therefore, we likely can’t draw any conclusions from this study except that there is good reason to consider PE in the syncope patient with appropriate presentation/risk factors.


This was a multicenter cross-sectional study which took place across 11 Italian hospitals from 2012-2014. They evaluated all patients >18 years of age who presented to the ED with first time syncope and were admitted to a medical ward. Indications for admission were left to discretion of the ED provider. Patients were excluded if they had a history of prior syncope, were anticoagulated, or were pregnant. Seems pretty straightforward, right?

From there, they had 48 hours for the patient to undergo a series of tests as per the 2014 guidelines of the European Society of Cardiology. Basically, this meant that they had an H&P performed which identified any clear prodromal symptoms of vasovagal or neurocardiogenic syncope, as well as identified any recent changes in hemodynamically active drugs. Patients were also asked about risk factors for thrombotic disease and symptoms concerning for DVT/PE, a la the ‘Wells Criteria’ that have become so near and dear to our hearts. They also had their vitals taken and any abnormalities noted. Hooray, that’s some good doctoring.

Patients were NOT assigned to any sort of standardized syncope workup while in the hospital; rather, this was left to individual providers. They do note that all patients did have CXR, EKG, ABG, and “routine blood testing.” They do not specify whether patients had further testing such as telemetry or holter monitoring, echocardiograms, etc. The only protocolized part of the study was applying a Simplified Wells Score to all patients and drawing a D-dimer. If they were determined to be low risk and had a negative D-dimer (based on standard laboratory range, not age-adjusted or provider determined), they had no further testing and were deemed PE free. If they had either moderate/high risk features by Wells, positive D-dimer, or both, they had further workup by way of CTPA preferentially, or V/Q scan if unable to complete CTPA. A PE was defined as intraluminal filling defect on CTPA or perfusion defect >75% on VQ (AKA high probability V/Q scan).

As a cross-sectional study, there was no comparator group. The primary (and only) outcome was prevalence of PE based on CTPA, V/Q, or in 1 case, autopsy.

A Bit of Context

There is no algorithmic method for evaluating the syncope patient — it’s something that has so far eluded physicians. Part of the difficulty is in narrowing down the many causes of syncope and the often limited history that patients are able to provide. (“Hey man, go easy on them, they were unconscious.” -Bill) Prior recommendations for syncope evaluation have hinged on risk stratifying the patient based on clinical and historical factors. As a result, most syncope guidelines rely on history and physical to guide further evaluation.

In fact, the only consistent diagnostic recommendation among guidelines is a resting ECG. If you don’t believe me, or want to get further into the weeds on the current guidelines, feel free to peruse the ACC/AHA or the ESC guidelines. Fair warning, most of it centers around determining which people are at risk for arrhythmogenesis or structural heart disease — and this is exactly why cardiac workup has been the ED physician and Internal Medicine provider’s focus to date. As a comparison, the concept of PE is mentioned a grand total of 2 times in the ~80 page AHA guideline document, which basically states, “think about PE when clinically indicated, and, uh, maybe get a CT if you want.”


A total of 2584 patients were seen in the ED for syncope during the study period. Of those, only 717 were admitted to the hospital. Of those admitted, 560 met inclusion criteria. They were able to rule out PE by low risk criteria and negative D-dimer in 330 of those patients (58.9%). Of the remaining 230, PE was diagnosed in 97 patients by either CTPA, VQ, or autopsy. They report this as 17.3% prevalence of PE in their population (which they defined as the 560 patients who met inclusion criteria, NOT the 2584 patients who were seen in the ED for syncope).

They went on to further characterize the clot burden in the PEs they diagnosed, and while this to some degree may be a clever way to suggest causation, it does help us better understand their results. Specifically, they found that 17 out of 97 patients (12 with VQ scans and 5 with CTPA) likely suffered from subsegmental PE (SSPE). This is important because, although controversial, we can infer from prior studies that SSPE can be found incidentally in approx 5% of patients (although as high as 17% in patients age >80), are possibly not even clinically significant, and have a high false positive rate, (likely ranging 10-15%). While this should probably make you take the numbers from this study with a grain of salt, it still doesn’t explain the unexpectedly high number of large PEs the investigators found.

They included in their discussion some postulates for the mechanism of syncope secondary to PE, including transient decrease in cardiac output or arrhythmogenesis from intermittent intracardiac clot burden, which, while interesting, remains unsubstantiated.

So what does this all mean? Well, on the surface, this could mean that as many as one out of six of patients admitted for syncope have a PE, and that these are potentially going undiagnosed. What is important to note, however, is that despite the temptation to extrapolate the results of this study, we cannot use the study data to make any assumptions on the cause of the patients’ syncope. The coexistence of PE and syncope is not proof of causality.

There are some issues with the study which warrant unpacking as well. Firstly, despite being a relatively large, multicenter trial, with appropriate patient selection and exclusion criteria, this study is poorly generalizable. Not only was it performed in only Italian hospitals, the patient demographics likely do not fit that of a typical US hospital. The average age of patients in the study was 76 years old, with a median of 80. Only about 6% were obese. Of the patients included in the initial ED evaluation for syncope, 72% were discharged and not followed up. These numbers are likely not typical for our standard of practice in the US. Further breakdown of the numbers reveals that the study group was likely a particularly high risk population with 5% having history of VTE and 11% with active malignancy.

Looking at the clinical characteristics reported in the study, those that were shown to have a statistically significant correlation with finding of +PE are as follows: prior VTE, undetermined cause of syncope, tachypnea, tachycardia, signs of DVT on exam, and active malignancy. So, not exactly rocket science. Unless you think reading the Wells criteria counts as rocket science, in which case please do not build any rockets.

Does this mean that we need to now scan all patients with syncope for potential PE? I’m not sure I would go that far. However, based on the results of this study, it does seem reasonable to seriously consider PE in these patients, and perhaps to give it a bit more thought than you might have otherwise. We should especially consider PE in patients who are elderly, have significant risk factors for VTE, have vital sign abnormalities, or lack any apparent alternative causes of syncope. Essentially this supports current practices, but if you’re one of the holdouts who doesn’t believe in PE as a cause of syncope, this study is evidence that it should be on your differential.

Tomorrow on IM HEAT

No post tomorrow! Take a day off from evidence and read something lighter. Meet us back here on Friday for more Hot Evidence and Trials!

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s