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.

Monday, April 9, 2012

An Alternative Method of ECG Interpretation

Just when a paramedic student has started to feel somewhat confident about rhythm interpretation, she is introduced to the other 11 leads.

First off, the leads are organized even worse than the QWERTY keyboard. Inferior leads are the left of anterior, the lateral leads are in two different places, and aVR sits there all by itself, like a chump.

Then there are all the depressions and elevations, T waves flipping around, ischemia vs infarct. And then someone shows you how to pick up on a posterior MI by flipping the paper over. Madness, I tell you.

In particular, identifying a STEMI can be difficult, even if ST segment elevation is clearly seen. For example, the following ECGs all show ST segment elevation, but...
1. Not a STEMI

2. Not a STEMI either

3. Nope.
 But with a fairly undramatic ECG like:

4. Bingo - Occlusion of the proximal LAD
The first first three ECGs demonstrate 3 common cause of ST elevation that we see in EMS or the ED, so-called "mimics" of STEMI. Now, there are a host of rules and criteria to help you diagnose each of these mimics, but it's hard to learn all of these, and to feel confident about them.

Is there a simpler way to achieve ECG excellence? Some short-cut to Jedi-level ECG mastery other than slogging through hundreds of tracings?

Perhaps a training montage?
Well, no.

But there are a few different ways to develop pattern recognition, and switching up the methods can put things in perspective. Hartman and colleagues have helped the novice ECG student tremendously with a new, focused approach to ECG interpretation. While this does not replace experience, practice, and feedback on interpretations, it's a good alternative way to tackle ECGs.

Abstract. If you want a pdf, message me at Facebook.

The rule has 4 steps, and we'll tackle them in that order

1. Is there ST elevation in at least 2 related leads?

The first rule specifies a minimum amount of elevation: 1-2 mm in two anatomically related leads.

It doesn't take long before a paramedic student identifies their first patient with ST elevation. Okay, granted, it's usually not an actual STEMI that they find, since the majority of ST elevation found in the ED or by EMS is not a STEMI. Typically, ST elevation will be due to any number of "mimics," such as left bundle branch block (LBBB), left ventricular hypertrophy (LVH), early repolarization (ER), as well as a number of other conditions. Surprisingly, if you look at all the patients who come into the ED with ST elevation, only about 1 in 7 patients have a true STEMI!

On the other hand, if you don't have some ST elevation, the patient probably doesn't have a STEMI. (Yeah, we're going to miss a true posterior or a proximal left main. This rule is for the novice reader, okay?)

No ST elevation, so not a STEMI.

2. Is the QRS a normal height?

The heart, over a period of years, responds to hypertension by bulking up and adding muscle mass. This process results in LVH, which, in the long run, isn't good. It shows up on the ECG as deep S-waves in V1 and V2, and high R-waves in V5 and V6.

In the short term, though, it mainly serves to distract us, as it can produce ECG findings that can look a lot like a STEMI. If we look at ECG #1 above, we see ST elevations in leads V2 and V3. Could these represent a STEMI?

Likely no, for several reasons. Now, a lot of the reasons involve interpretation of subtle, qualitative signs - the morphology of the ST segments and T waves, "notching" of the J-point,  reciprocal changes, etc. it just doesn't "look" like a STEMI, but you need to read hundreds of ECGs to feel comfortable with those.

It is far simpler to count the big boxes. Rule #2 boils down 3 sub-steps:
  • First, look at the S-waves in V1 and V2. Pick the deepest one, and count the big boxes.
  • Next, look at the R-waves in V5 and V6. Pick the highest one, and count the big boxes.
  • Last, add those two numbers. If it is over 7 big boxes, the ST elevation is probably due to LVH
7 big boxes equals 35 little boxes, or 35 mm. Count the small boxes if you prefer, or if the you're near the cutoff. Looking at ECG #1 as an example, and counting the little boxes, we find:

So, about 40 mm, or 8 big boxes, so likely not a STEMI.

3. Is the QRS a normal width?

Rule #3 is simple -  If the QRS is over 0.12 seconds long, don't call a STEMI.

Probably the most common cause of dramatic ST elevation is the LBBB, as in ECG #3 above. You can also see the same pattern if the the patient has a pacemaker.

Now, the experienced and sophisticated paramedic knows that there is a way to interpret the LBBB for signs of STEMI, but even the "simplified" rules for determining STEMI in LBBB are somewhat complicated. Many paramedics are familiar with the rule, but the new paramedic shouldn't be expected to make this call. If the patient has a pacemaker, it's even more unreliable to interpret the ECG.

4. Is there ST depression in at least 1 lead?

Rule #4 - if there is no ST depression, do not call a STEMI.

Most students have learned that you should look for reciprocal ST depression in a STEMI. Unfortunately, because of the non-intuitive, non-anatomic way that the ECG is arranged, it isn't clear which leads are "opposite" each other. And the patterns of depression can vary a lot, depending on which coronary artery is occluded. For example, an "inferior" STEMI may or may not have depressions in I and aVL; it depend on whether the culprit artery is the RCA or the obtuse marginal.

A much simpler criterion for reciprocal depression is any ST depression on the ECG. This would eliminate, for example, ECG #2 above. Although the computer interpretation was STEMI, it is a classic example of early repolarization, or possibly pericarditis (less likely, as the ECG did not evolve). Another example from my ED is this ECG:

27 y.o., prior dx of pericarditis
Just like ECG #2, there is diffuse ST elevation without any ST depression. Not a STEMI.

Applying the rule

Let's take another look at ECG #4:

Okay, going through the rules:

  • Rule #1 - Over 1 mm of ST elevation is seen in both V1 and V2, which are anatomically contiguous.
  • Rule #2 - The S-wave in V1 is about 1 big box deep, while the R-wave in V5 is 3 big boxes high. That's a total of 4, so the QRS height is normal.
  • Rule #3 - The QRS looks narrow, about 0.100 seconds wide.
  • Rule #4 - There are ST depressions in the lateral leads, most notably in V5.
So we see that this simple 4-step rule, intended to assist the novice paramedic, actually picks up a STEMI that the computer missed!

The Bottom Line

This elegant method of ECG interpretation, although intended for the student, can be very useful for the experienced paramedic as well.


  1. Thanks for the great post! Interesting ECGs...

    Regarding ECG #2, which also meets criteria for LVH... while it has classic features of early repol (such as upward concave STE in precordials, short QTc, notched j point...), how are you explaining the poor R wave progression in the right precordials, the STE in I and aVL with reciprocal depression in III and aVF (with T wave inversion)?

    In the setting of chest pain, i feel this ECG has features atypical of early repol and would be considered "confounding"...


    Dave B

    1. Ya know, you got a point. I was trying to use only ECGs that I had collected myself, but I hardly have the library of, say, Dr. Smith. I'm going to replace it with a better example, so that it doesn't distract from the overall message of the post.

      In reality, I thought this was BER when I first saw it (that's my signature on the ECG), but I have the advantage also of knowing that the serial troponins and echo were normal, and that the ECG was stable over about 12 hours!

      Flipped T waves in III never seem to indicate anything, so I didn't pay it much heed. The J-point in III and aVF doesn't seem depressed, at least to my eye. The elevation pattern in I and aVL are certainly not classic! I'll get a more classic one.

    2. Replaced ECG #2 with a more classic example.

      It also shows changes consistent with LVH (voltage in V2 and V5), but this is pretty common to find both in the same patient.

      Anyway, I was just trying to illustrate the article - I'm going to leave the finer points of electrocardiography to and

      Thanks for the feedback!

  2. Here's my problem (well, I have a lot of problems -- what's the deal with this mole? -- but the germane one anyway): are we willing to say, or teach, that no patient who has a baseline bundle branch block, or LVH, or a pacemaker/ventricular rhythm, or WPW, OR who develops any of those things acutely SECONDARY to his gigantic MI... none of these patients ever have heart attacks?

    I'm not trying to complicate things, or wave my hands and say "there are no tricks, only way to ECG mastery is ten years meditation upon morphology while seated on column." But it seems a little bit scorched-earth to simply decide that in order to simplify the issue of diagnosis we'll just put a bunch of at-risk patients in the negative column right off the bat.

  3. Brandon -

    This is a point I think I should have made clearer, so thank you for giving me a second chance!

    Clearly a number of people with comorbid heart problems will have further heart problems! The problem is that, while you & I are sussing out the latest insight of Dr. Smith regarding the finer points of Wellen's sign, the new guy is trying to remember if T-waves are always flipped in V1. That new gal ain't going to use Sgarbossa criteria even if you tattooed them to her hand - too much data and criteria, not enough context and experience.

    The authors of the paper are clear to point out that the algorithim is meant to ASSIST the NOVICE interpreter of ECGs. It is really meant to prevent most of the inappropriate cath lab activations, while triggering a recognition of a STEMI in most of the appropriate cases.

    It is expected and desired that the ECG will subsequently be interpreted in the ED by a (hopefully) expert.

    1. Yeah -- I remember we hashed over this when the study first came out. I'm actually very sympathetic to the importance of reducing false positives, but I think we'd all agree that it shouldn't come at the expense of any significant increase in false negatives; we should begin from the place of people calling most elevation a STEMI (which is not uncommon), and try to gradually draw them back from there. I wasn't being completely rhetorical, because I don't know how many of the STEMIs do occur in these comorbid populations (we get a somewhat biased view when we spend a lot of our time peering at "cool" 12-leads, because there's a reason they're cool!), so I don't really know how big that hit would truly be. (That might be a worthwhile study.) Quite frankly, I think if you're looking for a plug-and-play method to reduce false positives, you could do a lot worse than incorporating the computerized interpretation in an intelligent way; the Marquette is high specificity and works well in that way.

    2. Well, the computer interpretations are always interesting, and several systems or EMS studies have used them as the sole element of "paramedic interpretation." I'll leave it to folks like Chris Watsford to discuss the finer points. Nonetheless, the methods here are very useful are pedagogic tools, and as an alternative way to view a challenging ECG.

    3. As Brandon noted, I actually did look into this and am still looking into it (hopefully soon with cath lab registry data). Basically the algorithm is no better and in many cases worse than a computer interpretation! Using a cohort of cases from EMS 12-Lead and Harvard WaveMaven the algorithm is only 40% sensitive and 88% specific. The mix of ECGs is not the same as their figures, but its worth seeing how the algorithm handles a real world cohort.

      It does a really bad job with anterior STEMIs that have no clear ST-depression. And we're not talking "hard" ones, blatant in your face Anterior STEMIs. It also excludes RBBB-STEMI which isn't hard to spot either. Posterior MI's are excluded if you choose not to acquire V7-V9. Borderline STEMIs are also excluded.

      I think using the Brady 4-step approach glosses over the education of reading 12-Leads for ACS. It simply furthers the many "myths" or "half truths" of what you see on the 12-Lead. EKG education in my area, born out of the STEMI system in place, focuses a bit more on why you see changes rather than what changes you're looking for. Where this makes a difference is say LBBB, because if you know that LBBB has secondary ST/T-wave changes then you know LBBB isn't a STEMI. LBBB simply has ST-E normally, and doesn't really mimic it at all unless you use algorithms that emphasize features rather than processes.

      The one thing the Brady 4-Step does well is reduce false activations, and my opinion is it does so at the expense of E2B times. If you're a hospital administrator, this is probably what you want done. If you're a patient, you probably want a healthy amount of overtriage.

      I'm in an area where medics have "entry criteria", but can activate at their discretion. ED physicians make the call based on our alert. No transmission is done. Why our system works without a high false positive rate is the QA/QI loop. Every provider (EMT thru Cardiologist) gets feedback on each case.

      If you miss a STEMI or if you activate one and it isn't a STEMI, you get feedback (and it comes quick, within 24-48 hours usually). Emphasis is placed on education and early identification with clinical context rather than absolute adherence. This obviously requires more buy-in to use, as the hospital and EMS systems have to agree to work together and educate one another; but when an entire state can show an 85% "true positive" activation rate, it's obvious the system is feasible.

      Really this algorithm adds little to what we already have, especially when it is no better than the computer. An interesting aside is you can increase the sensitivity AND specificity of the Brady 4-step by altering the last step to be: 4. Is there ST-segment depression present in at least 1 lead or does the computer read *** ACUTE MI SUSPECTED ***?

      Perhaps this algorithm could be used to introduce STEMI activation to areas which don't have field activation, but any system not built around a feedback loop will fail. And once this algorithm gets into a feedback loop, its deficiency in sensitivity will be glaring. Paramedics need to own 12-Lead interpretation, just like we've acknowledged we own cardiac arrest care, and I don't think this algorithm adds to what we already have.