eFAST for Pneumothorax

This is a bit of a rant about a few things that I am interested in lately.

I was watching the FOAMed feed on Twitter the other day when I saw a paper mentioned by Dr Jonathan Henry ( @EmergencyEcho ).  Thanks for the post pub, peer review. Here was the tweet that caught my eye:

Here is the paper:

Detection of pneumothoraces in patients with multiple blunt trauma: use and limitations of eFAST published in the EMJ this week, lead author is Dr Sauter.  The study is a retrospective, cross-sectional analysis of trauma patients with pneumothorax who underwent both an eFAST exam and a CT.  It was done in a Bernese ED in Switzerland.

The highlight statistic here was “59%”, a really low sensitivity for eFAST detection of pneumothorax.  This caught my eye as it is one of the lowest sensitivities I have ever seen for the US detection of traumatic pneumothorax.  Now, we must note that this is a small, retrospective, single-centre data set.  There area  lot of potential problems with this sort of study.  However, I would like to dive a little deeper into the paper – there are some things we can learn from this data.  So onto the podcast to hear what I really think about these numbers.


Let me know what you think below.

See you in Berlin, Brisbane or Broome soon



  1. Hi Casey!

    Thanks for posting this.

    Firstly, yes -- let’s be clear that looking for lung sliding in trauma is mainly intended for detecting pneumothorax as a cause of instability in a hemodynamically unstable patient (tension pneumo). Not surprisingly, none of those were missed in this study. But let’s go ahead and take it a step further and look at test characteristics for “clinically significant” pneumothorax.

    “Clinically significant” meant they “needed a chest tube”, which is based on clinical judgement and obviously allows for discrepancy -- and these circumstances were not documented.

    Overall, 106 patients were ultimately included. 13 “clinically significant” pneumos missed. This means that the actual sensitivity for detecting a clinically significant pneumo with respect to all patients with CT documented pneumothorax was 93/106 = ~88%.

    This is pretty good and consistent with what I would expect.

    They also excluded patients who’s pneumos were detected by U/S, but got immediate chest tube. They could have chosen to include these as confirmed by clinical “rush of air”, etc, but they did not. If you include these patients, U/S sensitivity would be even higher.

    They also don’t tell us how much time went by from the initial scan to the follow up CT. We all know neumos can grow over time, especially in intubated patients undergoing PPV (they didn’t include that data either).

    Looking at the numbers and considering the above limitations, and then realizing that this was a single center, retrospective study with a relatively small n, I would interpret their numbers as very reassuring regarding the sensitivity of U/S for pneumothorax in blunt trauma.

    Lastly -- as you said Casey -- really wanna increase the sensitivity? Just move the probe around a bit on the chest!

    I hope you’re doing well my friend.


Speak Your Mind