Wednesday, December 19, 2018

The computer calls this a “normal ECG”. We'll just keep making this point.

This case was contributed by Brooks Walsh, an emergency physician in Connecticut.

A middle-aged woman had an acute onset of chest pain and dyspnea. The symptoms improved somewhat after the paramedic gave her nitroglycerin. 

The pain had almost resolved by the time an ECG was obtained in the ED:
Here is the computer diagnosis 
What do you think?

Perhaps you might think these changes are too subtle to immediately call for the cath lab. But do you think the ECG is “normal?” Well, the computer thought so!

Note leads III and aVF. There is subtle ST segment elevation in those leads. Although it is not high in absolute terms, it is quite high relative to the low QRS voltage in those leads. In combination with the subtle ST depression in aVL, these changes are diagnostic for, or at least nearly so, for acute occlusion of a coronary artery, probably with some reperfusion, as inferior T-waves are inverted and the T-wave in aVL is reciprocally upright.

Despite the computer interpretation, the ED physician was concerned about the ECG, and planned for serial troponins.

The first level was undetectable.  Three hours later, however, the troponin I was elevated at 0.08 ng/ml (99th percentile 0.01 ng/ml). The patient was verified to be symptom-free, and a repeat ECG was performed.

Note that the ST elevation in the inferior leads has basically resolved, especially compared with the restored QRS voltage. The ST depression in aVL is also resolved.

This confirms that there were dynamic signs of ischemia on the initial ECG. It was not normal!

A third troponin I returned at 0.27 ng/ml, a three-fold increase from troponin #2, also confirming acute myocardial infarction.

Angiography, however, did not reveal significant obstruction of the RCA, LAD, or circumflex. No residual signs of a culprit lesion could be identified.

No echo was done.

Management: dual antiplatelet therapy (DAPT) with aspirin and a P2Y12 inhibitior (clopidogrel, ticagrelor or prasugrel)

Smith comment: This is another unusual case in which the T-wave in lead III looks like a reperfusion T-wave but is present during pain and then normalizes later.  
This is similar to this case which I posted last week:

Dynamic, Reversible, Ischemic T-wave inversion mimics Wellens'. All trops negative.

Like that last case, the upright T-wave does NOT represent pseudonormalization due to re-occlusion, but rather true normalization. This upright T-wave is associated with absence of chest pain and an open artery (pseudonormalization is associated with recurrence of chest pain and a closed artery). But in contrast to that last case, this case had a positive troponin. This is particularly unusual, as T-wave normalization is usually unstable angina (negative troponin).

Learning Point:

Do not trust the computer when it interprets the ECG as "normal".

Here are many other cases in which the computer algorithm called the ECG completely normal when it was in fact dangerously abnormal.

Comment by KEN GRAUER, MD (12/19/2018):
Illustrative case by Dr. Brooks Walsh regarding a number of important concepts! For clarity — I’ve put the 2 ECGs done on this patient together (Figure-1).
Figure-1: The 2 ECGs in this case (See text).
As per Dr. Walsh — despite computer interpretation of the initial ECG done in the ED ( = ECG #1) as showing,“Sinus rhythm; Normal ECG” — the ED physician was appropriately concerned by a number of abnormal findings. In the interest of academic discussion — I’ll add the following thoughts:
  • The computer interpretation was obviously wrong. That said, this is not the “fault” of the computer. The computer is good at some things — but less good at others. Computerized ECG interpretations are generally very accurate for making measurements (ie, of rate, intervals, axis). However, the computer is not good at assessing rhythms other than sinus — and, it is not good at recognizing subtle ST-T wave changes of acute ischemia or infarction. As a result — We should not expect the computer to be accurate in assessing non-sinus rhythms or subtle ST-T wave abnormalities. Instead, I believe it is the clinicians fault if he/she accepts the computer interpretation as “normal” without first overreading the tracing. We should not be blaming the computer if it doesn’t pick up on subtle infarctions ...
NOTE: Optimal use of computerized ECG interpretations depends on a number of factors, including: iExpertise of the interpreter; iiAwareness of the strengths and weaknesses of the computer interpretation program being used at your institution; andiiiRemembering what you are looking for on the ECG of the patient in front of you.
  • For “My Take” on optimal use of computerized ECG interpretation — please CLICK HERE 
While good that the ED physician recognized abnormal findings on the initial ECG — an elevated troponin should not be needed in this case before contacting the cardiologist on call. That’s because ECG #1 shows abnormal ECG findings that have to be assumed acute until you prove otherwise.
  • As emphasized numerous times on this ECG Blog — troponin values may be initially normal in cases of brief coronary occlusion in which there is prompt spontaneous reperfusion. As a result, even if the first one or two troponin values in this case were to come back normal (and the 1st troponin in this case was undetectable) — nothing is proven. Instead, time is lost ... This patient should be admitted to the hospital — and, at the least — should be cathed at some time during her hospital stay before she goes home. Making cardiology aware sooner (rather than later) that they have a consult in the ED with new-onset chest pain and recent (if not acute) ECG changes on her initial ECG, will hopefully serves to expedite decision-making.
  • History is important. The fact that this middle-aged patient had new-onset chest pain that was severe enough to call EMS — in association with the initial ECG shown in Figure-1 — significantly increases the likelihood of an acute event even before you look at the initial ECG. In contrast, the initial ECG in Figure-1 would be less suspicious if the history was unconvincing.
In a patient with new-onset chest pain — the findings in ECG #1 are of definite concern:
  • Acute inferior MIs often manifest no more than low-amplitude ST-T wave changes. As a result — shape rather than amount of ST segment deviation is most important. Hyperacute-looking ST-T waves with slight-but-real ST elevation is seen in all 3 of the inferior leads of ECG #1. Given the tiny height of the R wave in lead aVF — the amount of ST elevation in this lead proportionately is considerable (ie, almost 1/3 the height of the R wave!).
  • There is a small-but-real q wave in lead II.
  • There is a QS complex in lead III (at least for the first 2 beats in lead III).
  • Although T wave inversion in lead III is not necessarily an abnormal finding when the QRS complex in this lead is predominantly negative — the coved shape of the ST segment seen here in lead III is never normal.
  • Finally, the slight-but-real coved ST depression in lead aVL is the exact mirror-image opposite shape of the coved ST elevation in lead III. This qualifies as reciprocal ST depression.
  • Subtlety — In additional to some baseline artifact, there is some beat-to-beat variation in QRST morphology in lead III of ECG #1. For example, the ST-T wave of the 1st complex in lead III looks the most abnormal. Is this the “real” shape of the ST-T wave in this lead? Given how subtle ST-T wave changes are in ECG #1 — this beat-to-beat variation in one or more leads makes accurate assessment that much more difficult. SUGGESTION: When the initial ECG is equivocal — and especially when there is some beat-to-beat variation in morphology — obtaining one or more follow-up ECGs much sooner than 3 hours should help to clarify if there are acute changes.
  • P.S. Some time ( ? how much) had passed since the onset of symptoms and the time when ECG #1 was recorded in the ED — and, the patient's chest pain was almost gone by the time ECG #1 was recorded. As a result, we might expect ECG findings in ECG #1 to be more subtle than if symptoms were still ongoing ...

QUESTION: Has there been a change in frontal plane axis between ECG #1 and ECG #2? If so — How might this alter your ability to assess serial change between these 2 tracings?

ANSWER: The frontal plane axis in ECG #1 is about +15 degrees. The axis in ECG #2 is about +45 degrees.
  • Lead-to-lead comparison of the 2 ECGs in this case facilitates recognizing this change in frontal plane axis. That is, the QRS complex is predominantly negative in lead III of ECG #1 — but predominantly positive in lead III of ECG #2. While I agree that ST segment deviation has clearly improved in ECG #2 — it should be appreciated that the shift in frontal plane axis makes comparison and detection of serial ST-T wave change in these 2 tracings that much more challenging.


  1. With the first EKG being in the ED I'm curious what the initial from EMS looked like

    1. FIrst response is that that ECG was described by EMS as "normal," but was not scanned into the EMR.

  2. Dr Stephen smith
    What are those inverted waves after t wave in v4 to v6?

    1. That blip between the last 2 QRS complexes appear to be artifact. Unlikely to be, e.g. inverted U wave if only occurred with one beat!


Recommended Resources