As in Incident command...

IC established!
We'll have several different sections reporting in - recent research, local topics, or highlighting areas of the Sponsor Hospital Council of Greater Bridgeport protocols.

*** Keep in mind - this website does not replace your protocols, and these posts do not reflect SHCGB or Bridgeport Hospital policies. This is a place to discuss research, controversies, or discuss possible future protocols. When in doubt, check your current protocols through the official source.

Thursday, May 9, 2013

Two ECGs - which goes to the cath lab?

A recent study suggests that a computer interpretation of the ECG can be extremely specific for diagnosing a STEMI; i.e. if the computer reads 

*** ACUTE MI  ***

you can take that to the bank.

This hasn't fit with my experience, and so I was very interested in Peter Canning's latest post, since it validated my suspicions. He found that the ECGs his system are acquiring show surprisingly poor sensitivity and specificity for STEMI, if you simply rely on the computer to diagnose. 

As an illustration of this point, I submit ECGs from 2 patients. 

(For more pairs of ECGs that show the problem with relying on the computer diagnosis of STEMI, click on the label "Paramedics need to read ECGs..." on the right.)

Case 1
Let's say that this was a middle-aged female, who started having substernal chest pain about 15 minutes ago. The EMS 12-lead shows:

Aside from diagnosing the patient as "borderline," anything else look suspicious?

Case 2
Again, a middle-aged female, this time with pleuritic chest pain and wheezing. An ECG obtained 5 minutes after arrival in the ED shows:

What's an appropriate next step? Call in the (cardiology) cavalry, or do a little sleuthing?

Call for a bat-stent? (source)
If you look closely at  few of the leads, especially V3, you can see small spikes preceding the QRS. Since the computer hadn't seemed to notice, I adjusted the settings to recognize pacemakers. A second ECG then showed pretty much the same complexes, but a very different interpretation.

She turned out to have a fairly ordinary case of COPD.

How about case #1?

Evidently the patient was first transported to a non-PCI capable hospital. About 2 hours later she was on her way to a different hospital for an urgent cardiac catheterization. This gave EMS a unique opportunity to capture the evolution of the ECG over a time frame that we don't often find in urban/suburban EMS.

Frankly, I'm inclined to agree with the computer this time! But what did the computer "miss" on the first ECG?

Hyperacute T waves
As Peter found after analysis of his system's STEMIs, computers aren't good at recognizing the earliest sign of an MI on an ECG, the hyperacute T-wave. These are transient features, before the ST segment has had a chance to elevate, and EMS is in a unique position to find these on their initial ECG.

Stephen Smith has some great examples, some of which look very similar to case #2 here. For instance, this ECG was acquired by EMS, and was instrumental in suggesting ACS to the emergency physician:

Dr Smith's ECG Blog - 6/2011
Another case involved an anterior MI that was misdiagnosed as hyperkalemia because of the magnitude of the hyperacute T-waves:

Dr Smith's ECG Blog - 2/2009
Very similar to our patient #2!

The Bottom Line
For more teaching on hyperacute T-waves, follow the links above to the blogs written by Peter Canning or Dr Smith, or check out this review.

And remember - sometimes you have to treat the monitor, not just the patient. Just make sure you're not treating a mistaken computer!


  1. Are there set variables that define "hyperacute T waves" IE; t wave amplitude > 1/2 height of QRS in lead "x", or is this entirely based on clinician gestalt?

    1. The definition of hyperacute T waves hasn't been well-worked out. Many references use an absolute cut-off; Marriott's uses > 1.4 mV in V2 or V3.

      This seems incomplete, since so many other ECG definitions use relative definitions: the degree of expected STE in LVH, the "Smith criteria" for STEMI in LBBB, as well as ST/T amplitude in pericarditis.

      One study suggests that relative cut-offs might be useful in this area as well, and I talk about a single retrospective study, so-far unvalidated, that derived some such criteria. See my FB post that goes over this study:

      Of note, though, ECG #1 in this post only meets 2 out of the 4 criteria that the study derived. With that in mind, I think clinician gestalt still has a huge role. (Or 20/20 hindsight!)