A 16 year old girl presented with altered mental status, probably due to alcohol intoxication, but also with possible overdose.
As part of her workup, she had an ECG:
|The computer interpreted this as Left Bundle Branch Block|
The overreading physician confirmed this read.
2 subsequent identical ECGs were confirmed as left bundle branch block, by different physicians.
What do you think?
This ECG was texted to me in real time by the resident, asking what it was.
My immediate reply: WPW
Notice the very short PR interval and the profound delta waves. This is clearly NOT left bundle branch block.
The patient awoke from her intoxication without complications. She and her mother denied ever having palpitations, tachycardia, chest pain, or shortness of breath.
The management of asymptomatic WPW is beyond the scope of this blog. Suffice it to say that there is a very high incidence of serious events in untreated children with this ECG finding.
See this publication:
The Natural History of Asymptomatic Ventricular Pre-ExcitationVolume 53, Issue 3, 20 January 2009, Pages 275–28.
Such cases should all be referred to cardiology, pediatric cardiology, or an electrophysiologist, avoid exercise until follow up, and call 911 for chest pain, SOB, or palpitations.
The computer is often wrong, but leads to serious diagnostic momentum. What would the overreading physicians have interpreted had the computer:
1) Given some other diagnosis?
2) Given no diagnosis at all?
3) Or if the physicians read the ECG first, then looked at the computer interpretation?
I don't know. How did you interpret it? Correctly? Or as LBBB?
I will be presenting an abstract at SAEM on the diagnosis of atrial dysrhythmias by computer and by overread, compared to a new neural network, machine-learning algorithm. When the standard diagnostic algorithm falsely diagnosed atrial fibrillation, the physician corrected it only half the time. By the way, the new algorithm performed far better.
1. Blind yourself to the computer interpretation until you make your own.
2. Only then look at the computer interpretation.
3. Then look back at the ECG if the computer sees something (accurate or not) that you did not.