Friday, January 26, 2018

Our patients deserve better than the "STEMI criteria"

Written by Pendell Meyers, of the Stony Brook class of 2019, with edits by Steve Smith

Two Cases

Imagine that two patients present simultaneously to your Emergency Department with moderately concerning chest pain and the following ECGs (no priors available).

The first patient has this ECG:

The second patient has this ECG:

Do either of them, neither of them, or both, require emergent reperfusion therapy?

STEMI guidelines (and therefore most clinicians) provide you with a clear answer. Clear and yet perfectly wrong in these two cases.

Let's take them one at a time.

Case #1

A woman in her 50s presented with 2 weeks of fatigue and chest heaviness that started several hours prior to arrival. She stated that she had been diagnosed with influenza 2 weeks ago and had felt worse today with new chest heaviness, so she went to an urgent care facility where she was sent to the ED for an abnormal ECG.

Here is her initial ECG on arrival, complaining of active chest heaviness:
What do you think?

The ECG shows sinus bradycardia with deep wide Q waves and a large amount of ST elevation in the  inferior leads, with reciprocal depression and T-wave inversion in aVL and I. There are also Q waves in V4-V6 with ST elevation and relatively large T-waves. There is no ST depression in V2-V3, but there is a biphasic T-wave in V2.

I showed this ECG to Dr. Smith with no other information. He immediately replied: "normal variant or LVH." We knew this was not acute coronary occlusion, despite the features listed above. I thought the Q waves might be indicative of prior completed inferior MI, but I did not think the morphology looked consistent with even subacute MI. I felt that there was absolutely no evidence of acute coronary occlusion on this ECG.

Why we can say that it doesn't look like ACO despite these objective features is difficult to explain. 

The best answer is that you simply have to see thousands of ECGs and correlate which ones turn out to be ACO and which are false positives. When you do that, you will have seen this ECG many times. The first few times you will be fooled, and then you will learn why it's not ACO.

For those that want a less accurate but more concrete explanation, here are a few points to consider:

1) When there is ST elevation in inferior leads with reciprocal ST depression in aVL, inferior MI is all but diagnosed, unless this STE and STD is preceded by an abnormal QRS! (no LVH, no LBBB, no paced rhythm, no WPW, and no abnormal Q-waves)

Any abnormal depolarization causes abnormal repolarization. Here there is markedly abnormal depolarization with large and deep Q waves inferiorly. No matter what the cause of these Q waves, whether it be old MI, LVH, or a normal variant, they all cause abnormal repolarization. Usually this manifests as ST and T-wave deviation in the opposite direction of the largest/widest abnormality in the QRS ("appropriate discordance"). Such deep Q waves will frequently generate ST elevation and reciprocal changes in the opposite leads.

2) Another important feature that makes it NOT look like MI is the extreme concavity of the ST segments, such that the base of the T-wave is narrow, causing some peaking of the T-wave.

3) There is a small slurring of the end of the QRS complex in some leads, consistent with a J-wave. This is not a perfect rule, but it is very rarely found in true ACO.

The clinicians were alarmed by the ECG findings (as is recommended by all available guidelines), and activated the cath lab. Here are the cath images:

Normal left main, LAD, and LCX.

Normal RCA.

Her catheterization showed completely normal coronary arteries. She had three consecutive negative troponins. Echo was completely normal, no evidence of LVH or any wall motion abnormalities. Extended viral panel was negative.

This is always the most interesting part of the chart review for me, when the clinicians struggle to reconcile normal results with the intense mobilization of resources used and an ECG that they cannot bring themselves to consider a possible normal variant: The clinicians concluded that the patient has "viral myocarditis."

Smith comment: this is NOT myocarditis.  First, the ECG is a normal variant.  Second, myocarditis will always have a positive troponin.

I find it highly unlikely (but not impossible) that the patient has viral myocarditis causing her chest pain and ECG abnormalities without any troponin elevation, wall motion abnormalities, coronary artery abnormalities, or pericardial effusion. It is much more likely that this unfortunate woman simply has a very scary normal variant baseline ECG. Unless she presents to the same institution or carries a copy of her ECG for the rest of her life, this problem will recur. We have many similar cases on this blog.

Her post-cath ECG is unchanged:

This is more proof that is is a normal variant. Acutely abnormal ECGs, whether due to ischemia or myocarditis or takotsubo, will evolve. There is no evolution of the ST segments or T-waves in II, III, aVF, or aVL.

Case #2

A male in his 50s presented with episodic chest pain over the past few days, which recurred and became worse 1 hour prior to arrival. At the time of presentation his pain was 4/10 and decreasing.

Here is his initial ECG:
What is your interpretation?

I sent this ECG to Dr. Smith without any clinical information and he immediately replied "inferior MI."

Congratulations if you see the subtle but important abnormalities here. There is a very tiny amount of ST elevation in lead III which is proportionally significant given a small normal QRS complex. The T-waves are borderline large for their QRS complexes, and alone could not be called hyperacute. However, there is a tiny amount of ST depression and (more noticeably) T-wave inversion in lead aVL which is far out of proportion to the very small QRS complex. The findings in aVL are not appropriate for or explained by the normal QRS complex, and therefore confirm that the subtle findings in lead III are new and indicative of ACO.

There is also ST depression in V3-V6, maximal in V4. This should be assumed to indicate posterior wall involvement of ACO.

These findings were not appreciated. The patient received nitroglycerin and the pain resolved. He was planned to undergo coronary CT angiography for further risk stratification. First troponin was negative.

Repeat ECGs were appropriately obtained, while the patient was completely pain free:

These show (incomplete) resolution of prior abnormalities, confirming that they are dynamic findings and also confirming reperfusion. There is also a change in heart rate which could affect the ST segment and T-waves.  The vessel is open at the time of this ECG.

While waiting for CTCA, his second troponin T turned positive at 0.020 ng/mL. CTCA was cancelled. The third troponin returned at 0.030 ng/mL.

At this point the notes state that the patient had a recurrence of chest pain, was given one dose of sublingual nitroglycerin, followed by lightheadedness with sinus bradycardia in the 30s. Unfortunately there is no mention of an ECG being performed at this time, and no ECG on file. The patient was admitted to cardiology with plans for an urgent cath the next morning. No further troponins were measured.

The next morning, cardiac cath revealed the following:

 This was called a 99% proximal RCA thrombotic stenosis, however I don't see any distal flow. This appears to be 100% occlusion. Thrombectomy and stent was performed with the following angiographic result:

After stenting, the full course of the RCA is now visible.

Unfortunately no more troponins were ordered, so the peak troponin value is unavailable. Echo performed 8 hours after cath did not show any wall motion abnormalities but was suboptimal quality.

Both patients did well.

Great comments by Pendell Meyers:

These two selected cases demonstrate a critically important problem with the current paradigm of MI management. The patients who benefit from emergent cath are those with acute coronary occlusion or near occlusion with insufficient collateral circulation to a substantial portion of the myocardium - this is the anatomic substrate of the entity we are supposed to call "STEMI."

Unfortunately, the term "STEMI" itself and all the guidelines for the past 30 years have restricted our minds into thinking that acute coronary occlusion is diagnosed only by "STEMI criteria." As we have shown countless times on this blog, acute coronary occlusion (ACO) is a complex and dynamic process, and cannot be simplified to measuring millimeters of ST elevation. Numerous studies have documented the STEMI criteria missing at least 25% of ACO. Conversely, we all know there is an unacceptably high rate of false positives using the STEMI criteria, which distracts the clinician from a patient's actual dangerous pathology, causes premature closure, puts the patient at risk of unnecessary procedures and their complications, as well as unnecessary mobilization of scarce resources.

Expert ECG interpretation is almost certianly superior to STEMI criteria in every way. If we are to progress in the management of acute MI, we have no choice but to break the current paradigm. The term "STEMI" cognitively inspires you to think that only the ST segments matter, and that the ST segment deviation is the reliable marker of ACO. To combat this problem we have invented other terms such as "subtle STEMI," "STEMI-equivalent," or "semi-STEMI" in attempt to convey that there are other ECG findings that indicate the same pathology as classic true positive STEMI, but these have not produced widespread change in the knowledge or perception of optimal ACO management, except among the readership of this blog and other similar resources. We have tried too long to keep the term "STEMI" in the name to pay homage to the breakthroughs in the 1970s-1990s which pulled us into the reperfusion era. Eugene Braunwald and others have suggested a requiem for the term "unstable angina," yet we have shown repeatedly on this blog that unstable angina is alive and well. But perhaps the time has come for a requiem for the term "STEMI." While it is true that using STEMI criteria in the 1980s and 1990s was proven better than flipping a coin to decide who needs emergent reperfusion therapy, it is no longer our best option and is actually inhibiting us from discovering that we already have better options. Too long has it restricted our thinking and prevented further research from showing who actually benefits from emergent reperfusion. Experts routinely perform better, and hopefully one day neural networks or the like will make all this a moot point.

The next available step is to formally show that experts can recognize ACO with more accuracy than STEMI criteria and current practice. We are planning a retrospective study now, which will be designed as the first study in an arc of research aimed at evaluating whether expert ECG interpretation can cause better outcomes for patients by more accurately predicting who needs emergent reperfusion therapy. If retrospective studies can show that expert ECG interpretation may have led to faster reperfusion of ACO, then prospective studies may one day be justified in cases where there is otherwise clinical equipoise. If such an ECG were reviewed by an expert, prospectively at the moment of the decision to emergently reperfuse or not, I believe that we could show patient-important outcome improvements correlating with increased accuracy of ACO detection and emergent reperfusion.

Perhaps most importantly, please remember that ACO sometimes manifests with completely normal serial ECGs. The ECG is an accurate surrogate marker of ACO in the hands of experts, but can never be perfect. Some patients deserve immediate diagnostic cath even without ECG abnormalities.

See this lecture by Dr. Smith:
"The False STEMI-NonSTEMI Dichotomy"


  1. Superb presentation by Pendell Meyers! The feature that for me is most against the 1st ECG being an acute stemi is the diffuseness of the abnormal findings. Virtually all leads show either abnormal T wave peaking (the inferior leads with ST elevation) or T wave inversion — and, large Q waves are seen in no less than 6 leads. The width and fragmentation of the Q waves in leads III and aVF is beyond that seen with “septal q waves” — with unusual loss of R wave from V3-to-V4 and uncharacteristically deep Q waves in V5,V6. What surprised me most about this case is that the Echo was completely normal. I think use of the term, “normal” variant to describe these findings is potentially problematic. By exclusion, the follow-up in this case confirms that the ST-T wave findings constitute a repolarization variant. I find it hard to describe these Q waves as “normal”. What the insightful follow-up to this case proves is that an “MI Mimic” tracing such as this CAN be found in a patient with no underlying heart disease. THAT is an important lesson to be learned!

    Credit to Dr. Smith for his instantaneous right-on accurate diagnosis of acute RCA occlusion for ECG #2. For those not appreciating the subtle hyperacute inferior lead ST-T wave changes with near mirror-image opposite picture in lead aVL (to that seen in lead III) — the important point that should not be missed by clinicians on the front line is that no less than 6 leads show ST depression (leads I, aVL, V3-thru-V6) in this patient with “moderately concerning new chest pain” — and regardless of whatever “stemi criteria” one uses — ST depression of this degree should be enough to convey the need for timely cath. THANKS again to Drs. Meyers & Smith for this excellent blog post!

  2. I identied the case 2 by lead 2 and aVL changes by IRBBB and is its concordant changes ... But case 1 was offcourse STEMI for me if the patient was symptomatic.
    In my opinion the STEMI criteria is required to help us in courtrooms, otherwise doctors will be sued more times than ever because reaching to such an expertise level will be problematic for the majority for the lack of mentors/cases/time etc etc..
    So guidelines need to be there and for better patient management yes we need to be aware of what you have been tryin to teach us.. 🙂

    1. They are fine to help diagnose some MIs and avoid some false negatives. But they miss many. Better to have a false positive cath lab activation than a false negative!

  3. "...when the clinicians struggle to reconcile normal results with the intense mobilization of resources used and an ECG that they cannot bring themselves to consider a possible normal variant: The clinicians concluded that the patient has "viral myocarditis." "

    The name given to this phenomenon is "cognitive dissonance".

    Excellent post.

  4. Drs. Smith and Meyers,

    My suspicion is that the two of you are more expert at identifying acute coronary occlusion on the EKG than many of the cardiologists at your respective institutions. So what happens when you are presented with an EKG such as #2? Do you handle it in the same manner as an EKG meeting traditional STEMI criteria? And what is the conversation like if/when the cardiologist does not agree with this interpretation?

    1. Matt, I can only speak for my very short experience at my own institution. When we have a patient with clear ACO which manifests with obvious ST elevation (a classic, clear cut STEMI), we activate the cath lab and the cardiologists come down immediately to evaluate the patient. When we have anything less than that, but we're worried that the patient might still need emergent cath, we generally don't activate the cath lab outright but instead call the cardiology fellow or attending directly, to explain our concerns and have them look at the ECG. If the cardiologist doesn't agree, the conversation doesn't usually make a difference. They decide whether and when the patient goes to the lab. We try to get a stat echo done on these patients. Subtle findings such as in EKG #2 are generally not accepted at my institution (except by the emergency medicine providers who keep up with this blog and other similar resources) - I imagine this is the situation for at least 90% of all institutions worldwide.

      I'll let Dr. Smith explain the details of how the process works at his institution, but overall it sounds a little different. They have a formal "pathway B" for their patients with high suspicion and/or subtle findings, including immediate formal contrast echos, low threshold to take the patient for immediate diagnostic cath, and the benefit of having spent 30 years working with Dr. Smith. So you can imagine how they might be more receptive to subtle findings!

  5. Compleatly agree with Matt. MI is a much wider concept than STEMI. But for most common ED Docs and cardiologists, STEMI concept provides a framework which is handy and easily available. Nevertheless, I do agree bigger efforts can be made by most of us to widen our scope of this potentially mortal disease. But unfortunately, our efforts and energy have to focus on many other issues and most of us cannot embrace them all strong enough.

  6. Drs,
    When reading this post over morning coffee (daily routine for past 6 years), I’m reminded of a case posted here involving an 80 something year old female with chest pain and a paced rhythm. This ecg presented with concordant depression in the inferior leads an concordant elevation in aVR. It was not recognized by the intensivist and she eventually arrested expired.

    With regards to this post, would it still have been considered wrong considering occlusion within a paced rhythm (abnormal qrs) in the inferior leads even though there were hints of reciprocal changes between III and aVL? Does a paced rhythm/abnormal qrs exclude cath lab consideration?


DEAR READER: I have loved receiving your comments, but I am no longer able to moderate them. Since the vast majority are SPAM, I need to moderate them all. Therefore, comments will rarely be published any more. So Sorry.

Recommended Resources