Friday, March 9, 2018

A patient with a "seizure" and a completely "normal" ECG

I happened on this ECG while walking by, and read it with no clinical information, remarked on it, and discussed it with the physicians caring for the patient.
The computer read the ECG as completely normal.

I heard this clinical information: This patient presented with a "seizure," and was to be worked up by neurology for new onset of seizures.

What do you think?

The computer measured the QT at 420 ms, and QTc at 445 ms.

Let's magnify V2 and V3:

Here I put lines in at the onset of the QRS and end of the T-wave (or is it a U-wave?)
Approximately 595 ms

Here I point out an apparent U-wave in V1
So it is possible that the T-wave is really a U-wave
A long QU interval is just as dangerous.

The K returned at 2.5 mEq/L and it turns out that the patient had been having diarrhea.

There was a previous ECG recorded with K = 3.6 mEq/L
Much more normal, although has some nonspecific ST depression

The patient did not likely have a seizure, but more likely had an episode of VT (Torsades de Pointes) which resulted in tonic clonic activity.

Even after the K returned low, no clinician went back to look at the ECG to see if there were any findings.  No clinician questioned the symptom of "seizure."   The faculty physician involved is one of the smartest and best clinicians I've ever worked with, and very good at ECG interpretation.  Yet he did not scrutinize the ECG.

This is the problem with computerized interpretations.  People tend to trust them implicitly, and turn off their critical faculties.

Learning Points:

1. Always look at the QT interval.  If it looks long, you must measure it.  Do not trust the computer.  On the other hand, when it looks normal, it is usually accurate.  (That is why it works in the formula for differentiating subtle LAD occlusion from normal variant ST elevation.)

2.  A long QT could be a long QU interval (prominent U-wave), but the danger of polymorphic ventricular tachycardia (Torsades de Pointes) is just as high.

3.  Even when the computer states the ECG is completely normal, it may be very abnormal.

There was a paper last year contending that if the computer reads the ECG as normal, the physician need not look at it, at least not in triage.

I discuss this paper here:

A middle-aged woman with chest pain and a "normal" ECG in triage

Read these:

In depth on QT correction and QT in general:

QT Correction Formulas Compared to The Rule of Thumb ("Half the QT")

I just posted this case yesterday.  Both of these cases were this week.


  1. Hello, in which lead should QT be measured, II, V5, V6? Or in the leads with the longest QT?

    1. It is the longest QT of the 12 leads, usually II, V2, or V5

  2. Nice case! Lots of artifact on the initial ECG! — but we see complexes reasonably well in leads V4,V5,V6 — and what struck me is a rounder (broader) than-usual T wave. As per Dr. Smith — the QTc appears to be somewhat long. The “eyeball estimation method” — shows the QT in V5 to be clearly more than half the R-R interval in this lead. Using calipers — I measured the actual QT interval in lead V5 = 440 msec. At a heart rate of 70-75/minute — this corrects to a value clearly above the normal range (ie, >450msec). My favorite “List” are the causes of a long QTc = i) Drugs; ii) Lytes (ie, low K+/Mg++/Ca++); and/or iii) CNS catastrophes. Clearly, other things (ie, acute ischemia/infarction, BBB) can lengthen the QTc — but these other factors aren’t operative in this case. While true that this patient just had a seizure that might alter the QTc — the beauty of this list (in my experience) — is that it prompts the interpreter to consider all 3 entities likely to produce QTc lengthening as the sole ECG abnormality — which in this case should prompt checking serum electrolytes. THANKS to Dr. Smith for presenting this illustration of how an ECG may help in a noncardiac disorder!


Recommended Resources