Skip to main content


Prandoni P et al. Prevalence of Pulmonary Embolism among Patients Hospitalised for Syncope. NEJM 375:16 Oct 20 2016.

This is a big study in the NEJM, that a lot of people will pay attention to. The finding in a nutshell, was that Pulmonary Embolism(PE) occurs in 1 in 6 patients hospitalised with the first episode of syncope.  WHAT??

PE, according to this paper, is much more prevalent than we thought…..OR IS IT? Let’s take a closer look.

A Summary

This was a cross-sectional study, that looked to determine the prevalence of PE amongst patients presenting with the first episode of syncope. It was conducted across 11 hospitals in Italy.

A total of 2584 patients presented to emergency departments with syncope. Of these, 717 were admitted to hospitals. Following further exclusions, 560 patients were included in the study. These patients had a work-up, which included CXR, ECG, blood tests and a D-dimer. Of the 560 patients, 330 had PE ruled out either by low clinical pretest probability or a negative D-dimer.

Of the 230 left, who were D-dimer positive, or had a high clinical suspicion of PE, 229 underwent a CTPA or V/Q scan. 96 patients had a proven PE. One patient died and had a PE found at autopsy.

The conclusion of the study was that 97 of the 230(42%) patients thought to have PE, did in fact have a PE. The prevalence of PE in those patients admitted was 97/560 ie., 17.32%.

 Let’s Look at The Details

This paper basically had a lot of sick, older patients and looked, not at all patients presenting to the Emergency Department with Syncope, but at those being admitted. There was selection bias, as those patients admitted were sicker and had symptoms of tachypnoea, tachycardia, clinical symptoms of DVT and active malignancy. The same symptoms that increased the investigator’s suspicion of a PE.

In effect what was shown in this paper was that if you work up patients with symptoms of a PE, you find a higher prevalence of PE in that group.

The real frequency of PE in patients presenting with syncope in this group was  97/2584, or if we remove those with recurrent syncope 97/2549 = 3.8% Again this group was more elderly and had a significant symptoms.

Looking at the admitted patients, the vitals and rate of malignancy are shown below:

PE Confirmed PE Ruled Out
RR>20bpm 44/97=45% 33/463= 7%
HR>100bpm 32/97=33% 75/436= 16%
SBP<110mmHg 35/97=36% 106/463=23%
Clinical Symptoms of DVT 39/97=40% 21/463=  5%
Active Malignancy 19/97=21% 46/463=  10%

I won’t talk much about the approach of doing D-dimers on everyone, because this is something we’ve been fighting against for so long. This is not the way the D-dimer was meant to be used and shouldn’t be used in this way.

The types of emboli detected is also important


This study, doesn’t really change my mind on PE prevalence in patients with syncope. It doesn’t change my approach to the syncope patient, as I am always suspicious of a PE.

I approach the patient with syncope by asking 2 questions:

  1. Is this cardiac? A cardiac cause of syncope has been shown to have the worst outcomes. I specifically look at the ECG for my minimum findings i.e.., my ECG’s of Syncope as discussed in the Cardiac Bootcamp last year:            
  2. If it is not cardiac, is it one of the other conditions that may lead to significant mortality and morbidity:
  • Subarachnoid Heamorrhage or other bleed
  • Thoracic Aortic Dissection
  • AAA
  • Pulmonary Embolism – So it is one of those diagnoses I always consider

I look for specific PE findings clinically, such as tachypnoea, dyspnoea, low saturations, and risk factors. I look at the  ECG especially for:

  • Sinus tachycardia or Tachyarrhythmias such as AF or SVT
  • S1Q3T3
  • T wave inversion in the Inferior and Anterior leads
  • Right Axis deviation
  • Transition Point shift
  • P pulmonale
  • Dominant R wave in V1

If there are no symptoms to indicate a PE, I consider ruling out the patient with the PERC rule(no patient in this study of those admitted could have been ruled out using PERC). If not and they are low risk, I then do a D-dimer. If its positive I investigate.

I don’t think my clinical practice will change, EXCEPT, perhaps I might look for it more, in the elderly, like I do for AAA.

So that’s me. Let’s see what some other had to say:

Greg Treston: Director Emergency Department Mater Hospital Brisbane


2584 patients

45 of 2054 had PE “proven” when there was no suspicion of PE or strong alternative diagnosis. (although to be very fair 2000 or so weren’t investigated because they had what might be called low chance or no chance of PE)

So I would call that 2.2%

I have to suspect that this would be the baseline rate of PE on imaging if you got together a bunch of people of similar age.  Of course, when you don’t check the vast majority of them, that percentage might change, but you see my point.

Standardized Protocol?

If you see on article page 1529 on the bottom left the text states “…all participating hospitals used a standardized protocol for the diagnostic work-up of syncope that was based on international guidelines.”  I can only assume that the reviewers did not check these references, because when I did there was no “standardized protocol” in either article referenced, at least not as we would understand it in Emergency Medicine terms.


It is always tough to read an article without knowing the conflict of interest statement.  The NEJM has a statement on its website regarding this article.  They note:

Disclosure forms provided by the authors are available with the full text of this article at
No potential conflict of interest relevant to this article was reported
I looked at everyones disclosures, nill.  Absolutely none.  Not even research grants.  However some of the authors are linked to a place called the Thrombosis Research Institute, and a study called Garfield-AF.
What is the Thrombosis Research Institute?

Excerpt from their website:
Both Thrombosis Research Institutes in London and Bangalore are independent, free standing organisations and recognized charities undertaking a multidisciplinary approach to research by bringing together clinicians and scientists involved in basic research.

Our two organisations are supported by peer reviewed grants and charitable donations:
Over its lifetime, the London Institute has received endowments from the British Heart Foundation, Department of Health, United Kingdom, Emmanuel Kaye Foundation, Garfield Weston Foundation, Medical Research Council, Wellcome Trust and the Wolfson Foundation, among other charitable bodies, individuals and pharmaceutical companies.

The GARFIELD-AF registry is funded by an unrestricted research grant from Bayer Pharma AG, Berlin, Germany.

So Bayer is funding the Thrombosis Research Institute, and Bayer make Xarelto.

James Edwards: Acting Director Emergency Department Royal Prince Alfred Hospital

Syncope is a common presenting symptom to the Emergency Department (ED). Pulmonary embolus (PE) is a common ED diagnosis. But is PE a common cause of syncope? Not in my clinical experience. Has the recent PESIT study changed my approach? Yes and no

My current approach to a patient presenting with syncope to the ED is to search for an underlying cause.  For syncope that remains unexplained following a clinical exam, ECG and sometimes some a blood workup, my next step is to risk stratify.  This risk stratification is focused on not missing an underlying cardiac cause such as an arrhythmia. But should excluding PE, with more liberal d-dimer and CTPA ordering, form part of my diagnostic workup?

There has a flurry of comments following the release of the PESIT paper discussing the limitation of the study including selection bias and over-diagnosis of PE as a consequence of increasing false positive CTPA results. I will concentrate on how it will impact on my clinical approach

Firstly, I have assumed that if a PE is large enough to result in syncope, there will be still be some persistent signs (a few breadcrumbs) that would provide some clues to its presence. The group later diagnosed with a PE after being admitted with syncope was more likely to be tachycardic, tachypneic and have an underlying malignancy in the PESIT study. This reassures me to a degree that patients with a clinical suspicion of PE are more likely to have a PE.

I could say that I will ignore the paper. However, the most likely influence on my clinical practice will be the following

  1. PE will move closer to the top of my list of differential diagnosis secondary to the availability heuristic
  2. I will look for subtle clinical signs such as tachycardia and tachypnea in patients presenting with syncope
  3. By looking closer for clinical signs and risk factors for PE, I will probably investigate for a PE more often in patients presenting with syncope
  4. After a period of time and a number of negative CTPAs, common sense will return and I will return to my usual practice.

Luke Lawton: Director Emergency Department Townsville Emergency Department

I read this paper and for a moment I was tempted to start ordering CTPAs on every patient I admit with syncope. Why? Because the incidence of PE in this patient population is high enough to – according to our conventional model – invalidate the negative predictive value of a d-dimer for PE. (I make it about 97%, compared with >99% in most published standards). Additionally, of the PEs in this particular study about 60% looked like they were “PEs that matter”. BUT, when I drill down I’m of the opinion the paper doesn’t really give a valid description of the incidence of PEs in the population studied. If the d-dimer doesn’t work, how did we use a d-dimer to exclude a PE in half the patients??

So, where does that leave us?
1. I think we need to do a bit more work to flesh out how likely a population of patients with syncope is to have a PE – i.e. the true incidence in these people
2. We need to have a think about our risk scoring models in older people – remembering that d-dimer rises with age.
3. We need to work out how important diagnosis of PEs actually is. I spoke about this at EMCORE 2015, and I’m still no closer to getting a serious answer….
For the moment I’m going to leave the blanket CTPA approach in the basket. But I am going to think very carefully about PEs in syncope, rather than dismiss it out of hand as a possible diagnosis.
I hope you enjoyed the review.
Peter Kas

One Comment

  • Rodney says:

    Wow, superb weblog layout! How long have you been blogging for?

    you make running a blog look easy. The entire look of your site is excellent,
    let alone the content!

Leave a Reply