Friday, July 6, 2012

Distinguishing STEMI from LVH on the ECG

I've started saying really something annoying at work.

When a resident starts interpreting an ECG, listing off features that either suggest or weigh against active ischemia, they will invariably (and properly) mention the deviation in millimeters of ST segments. I will then sanctamoniously intone

"Amateurs measure millimeters, professionals interpret the whole ECG."

If you think I'm smug about ECGs, ask about my thoughts on Huey Lewis.
Despite the my pretentiousness, there is an important element of truth there. (Dr Smith makes this point without smugness, but that's no fun. For me.) If you slavishly follow the standard criteria for STEMI diagnosis on the ECG (with so-and-so number of millimeters in this lead, so many leads with such elevation, etc.), you're going to make a lot of mistakes. 

One example is LBBB, a very common etiology of STE on the ECG. Usually, of course, the LBBB is immediately identified, owing to its characteristic appearance, and discussion switches to old vs new, and a review of Sgarbossa vs Smith criteria (peer-reviewed video lecture!).

The Problem with Left Ventricular Hypertrophy
Left ventricular hypertrophy (LVH), however, may not be as obvious, and ED physicians may mistakenly activate the cath lab based on the degree of ST segment elevation. Now, mistakes happen, and we want to have a certain "over-triage" rate, just like surgeons (before the use of CT scanners) had to have a certain "false appendectomy" rate. 

LVH, however, is very common, especially in the ED patient population. In fact, LVH is found in about a quarter of the ECGs in the ED that show STE, and is the most common cause.

Brady 2001
And it looks like emergency physicians get fooled by LVH fairly often. In one study of cath lab activations, it was found that LVH predicted a "false-postive" result, with an odds ration of 3.1.

How do we keep from getting fooled by this harmless (in the short-term) mimic of STEMI? It isn't simple. Indeed, one recent article suggested that "novice interpreters" of ECGs avoid diagnosing a STEMI in any patient with deep R- or S-waves.

Don't do it!

Well, I hope you are aspiring to more than "novice" status. I also hope that I can provide you some guidance beyond my usual semi-mystical injunctions and platitudes. 

"Be the ECG, Danny"
The Article

Fortunately, there are non-novice ECG interpreters out there who want to share their secrets, like this group out of UC-Davis. The authors used  a database ("ACTIVATE-SF) of all the emergency physician-initiated cath lab activations over a period of 3 years in the San Francisco area. This gave a denominator of 411 activations.

Applying any of 3 different scoring systems (Cornell, Sokolow-Lyon, or aVR > 11mm), they found that 79 of these patients met ECG criteria for LVH. Now, some of these patients with LVH ended up having culprit arteries identified at angiography, but most didn't. 

They went further, and analyzed this select group of patients who had LVH and a positive cath. They found a few predictors on the ECG.

Certain locations of STE, the degree of STE, and the number of leads with STE all increased the odds of a true STEMI. So did the presence of Q-waves or reciprocal ST depression.

Now, LVH usually only gives you a pseudo-STEMI appearance in the anterior leads, so elevations in the inferior leads, for example, won't cause much confusion. You can see from the table above that STE in the inferior or lateral leads, or a true posterior STEMI, will be easy to distinguish. But what we really care about is trying to distinguish an anterior MI from LVH, and the data above confirms the difficulty; STE in V1-V3 was more often found to not have a culprit artery.

ST Elevation: Absolute and Relative
 Rather than just measuring the absolute amount of STE, however, the authors also analyzed the relative amount of STE, normalized against the difference of the preceding R- and S-waves. 
% STE = (Height T-P segment to J-point) / (Height of R-wave minus depth of S-wave)
Here is an illustration of applying this, using an example of LVH without STEMI from the paper:

True STEMIs, overall, had a higher percent STE (25%) versus "false" STEMI (9%).  More importantly, if the percent STE < 25%, STEMI was essentially ruled out. If the percent STE > 25%, a STEMI was predicted with a sensitivity of 77%, and 91% specificity. 

When they compared this rule, using relative STE, versus the conventional criteria for STEMI that use absolute STE, they found it to be more sensitive and far more specific (73% and 58%, respectively).

For example this patient has LVH, and STE in V1-V3:

But they also have a ratio of STE/(R - S) = 45% STE

Indicating that it's a true STEMI.

Contrast this ECG with this patient with LVH:

They have STE, just like the preceding patient, but a closer look at the anterior leads shows...

... a fairly insignificant STE/R-S ratio, suggesting no active ischemia.

Putting it all together...
For practical use, the authors summarized, in an algorithm, the practical use of their findings, when interpreting an ECG with ST segment elevations.

The Bottom Line

Even if you don't crunch the numbers, try to appreciate the pattern, the proportionality, of the elevations and deviations. You need to look beyond the numbers, feel the waveforms, you need to...

Ah heck, I did it again.


  1. hahaha your opening comments sound a whole lot like me when I coined the phrase, "millimeter criteria are for amateurs," while pushing for cath on particularly subtle STEMI case a couple of months ago. Well stated.

  2. Proportionality cuts both ways. The LVH ECG is interesting and a bit suspicious, IMHO. The tiny QRS in lead aVL has a disproportionately large T-wave and I can almost hallucinate a smidgeon of ST-elevation. Throw in the apparent reciprocal changes in leads III and aVF (discordant, yes, but not typical-looking for LVH) and it's enough for me to take a second look at lead V2-V4. S-wave is only 13 mm in lead V2. Very small in lead V4 (but I always take transition leads with a grain of salt). The biggest surprise is the lack of typical strain in leads V5 and V6. I think Garcia and Holtz mention this pattern (notched J-point, upright T-wave) but it's rare, at least in my experience. Is this Hurst's type 2 (or type B) LVH? I looked into this a while back and couldn't find any references.

    1. Oh, I'm with you how that would be consistent with a high-lateral STEMI. Heck, it looks like this example of a diagonal occlusion from Dr. Smith (

      Marriott would suggest that the lack of a strain pattern implies chamber dilation without a subsequent increase in myocardial mass. Can't really speak authoritatively in that area. As Dr. Smith has suggested, the answer probably lies in the clinical context (and echo!). See ( for that discussion, as well as the ECG I SHOULD have used as an example in my post.

  3. A few of the ECG's they gave in their article I've emailed the primary author, as I could not figure out what ST-Elevation they were measuring! Figure 2.C is a great example (not published on this site).

    1. Tom, Chris -

      Looking over the examples in the article, I wondered if I should instead have found my own examples, since the figures provided by the authors had me a bit turned around as well. The results cohere with my experience, however, so I don't doubt the conclusions, just the illustration.

      Since I emailed him as well, and gave him a link to the post, I'm hoping we get some insight from Dr. McCabe.

    2. Dr. McCabe replied to my email and stated that their algorithm only requires 1 lead of V1-V3 to have >=25% ST-E relative to |R-S|, but that to continue all of V1-V3 should have *some* ST-E. He also stated that they chose subtle/borderline ECG's to send home their point.

      This seems to clear up my confusion with their examples. Granted I still have trouble not seeing "LVH with Strain" on many of the ECG's.

      All in all I like their criteria!

  4. Great study... I am particularly tickled that they've nearly rederived Dr. Smith's proportionality modification to the Sgarbossa criteria (for LBBB). They use full R-S height instead of the depth of the S wave, and Dr. Smith eventually switched from 25% to 20% as his cutoff, but otherwise it's very familiar. If nothing else I would consider this a partial validation of the generalizability of the "proportional discordance" concept to LVH and other rhythms -- something many of us have found anecdotally true but hasn't had too much support in the literature.

    1. There's one more "25%" criteria in ECG interpretation we should know about.

      When faced with an ECG that shows diffuse ST elevation, and the differential is pericarditis vs early repolerization, check out the relative heights of the ST and T segments in V6. Pericarditis will have, almost always, ST/T > 25%.

      Ginzton 1982 (

    2. Interesting... new to me. Although I don't know how many times that's really been the vexing differential for my own part...

  5. Edited the post, and added an illustration, to make the explanation and calculation of % STE more explicit.

  6. I am confused by the first example. The criteria for LVH do not seem to be even close to being met. I do not get 35 mm, even if I add all of the S waves and R waves in V1, V2, V5, and V6. What am I missing?

    It would seem that using a lead with ST elevation and a large S wave would require an unusually large amount of STE to rule in. An example with a deep S wave might help.


    1. They used 3 sets of criteria for diagnosing STEMI, one of which is R-wave in aVL > 11 mm. The other two were the Sokolow-Lyon, and the Cornell. The ECG only had to satisfy one of the 3 sets of criteria.

  7. Just seeing this now - Your bottom line quoting Yoda tells all. It is more than the numbers .... Fine to use numbers as a starting point - but the better one gets as an interpreter, the less the dependence on strict numbers (in my opinion).

  8. Small mistype that gave me brief palpitations: (have I been doing this wrong!?!) LVH criteria is R wave > 11 mm in aVL), under "Article", although it's presented correctly in the table right below.

  9. This comment has been removed by the author.

  10. Very interesting, love the algorithm.
    i have a simple question: how did you calculate R wave in the first example, when the QRS is a negative ?