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.

Friday, November 2, 2012

STEMI called from the field - or is it?

As I have remarked before, the computer interpretation of the ECG can only go so far. Occasionally, it mistakes a benign, common finding (e.g. early repolarization) for a STEMI. Other times, it misses an atypical form of a STEMI.

And other times, the computer interpretation has the potential to mislead the clinician in a dangerous direction...

The patient
A mid-thirties male with a history of hypertension had an acute onset of chest pain while at rest. It radiated to his back, and was constant. It was non-pleuritic, constant, and had no associated dyspnea. Interestingly, he had just taken a ever-the-counter "sexual enhancement" pill (such drugs occasionally contain potent - and illegally added - active ingredients).

EMS found him to be bradycardic, and profoundly hypertensive, with a SBP of 230 mm Hg. His ECG was obtained:

Based on the symptoms and the ECG, along with the computer interpretation, EMS alerted the ED that they were bringing in a cath-lab candidate.

ED evaluation
The ED physician met EMS at the door, and called for cath-lab activation immediately. The patient was clearly in severe pain (he would eventually receive a total of 65 mg of morphine in the ED), with waves of sweat rolling off his face and chest. In addition to the severe chest pain, he also described numbness and weakness in his left leg, despite moving it as he writhed in pain on the stretcher.

An ECG was obtained in the ED:

Also a chest x-ray:

The ED physician had a brief conversation with the interventional cardiologist, who asked that the patient be started on the anti-platelet drug prasugrel while he started speeding on his way to the hospital at warp factor 9.  He had viewed an image of the EMS ECG on his smartphone, and thought that a diagnosis of STEMI was likely as well.

Now, you might suspect this patient did not end up having a STEMI, or else why would I post the case?

Where, then, did the patient end up? Cath lab? ICU? "Other?"

Subsequent course
The patient's appearance was "sicker than the ECG," and suggested an alternate diagnosis. An echocardiogram suggested widening of the aortic root. An emergent CT angiogram of the chest was performed before the cath lab was ready.

The diagnosis of a type A aortic dissection, extending from the carotid arteries down to the left iliac, was made based on the CT.

Labetalol and nitroprusside were started to control his blood pressure, and the patient was taken to the OR by cardiothoracic surgery.

So, what about the ECG and the "STEMI?"
There is no prior ECG available for comparison, but the ST and T wave abnormalities found on the EMS tracing were stable during the initial hospital course. This suggests that there was no active cardiac ischemia.

Probably, the ST elevation in the anterior leads was likely a repolarization effect, secondary to left ventricular hypertrophy (LVH), a very common STEMI mimic. LVH is typically caused by chronic hypertension, a well-established risk factor for aortic dissection.

Can we use the ECG to diagnose aortic dissection?
Probably not. The diagnostic standard will remain the CT scan, given the excellent visualization.

But are there some ECG findings that could suggest aortic dissection? A recent study provides some answers.

A 2010 study looked at 159 patients with a diagnosis of dissection, and classified the ECGs they found in various ways. Most importantly, they looked at both acute and chronic changes.

Just as seen in our patient, LVH due to long-standing hypertension was fairly common, around 10%. We would expect this ECG finding to be chronic, however, and evident on prior ECGs. Curiously, this patient also had sinus bradycardia, which was seen in 11.3% of the study patients. It isn't clear what this is caused by.

What about the 8.2% of patients that had acute ST elevation? How do we avoid sending these patients to the cath lab, or giving them heparin or other anticoagulants?

First off, aortic dissection is rare, and ST elevation is overwhelmingly more likely to be due to a "true STEMI" than a dissection. Second, a dissection will typically have an abrupt onset, with "ripping" pain radiating through to the back. Since a dissection can propagate up the carotid or vertebral arteries, you can find stroke symptoms along with the chest pain. Similarly, as the dissection can extend into other arteries, there may be ischemia of the arms or legs, along with pulse deficits.

Third, dissection will infrequently cause ST segment elevation due to occlusion of a coronary artery as the dissecting layer spreads towards the aortic root.  The right coronary artery is usually taught as being the one most commonly affected, and the study supports this; over half of the patients with ST elevation had those elevations in the inferior region, supplied by the right coronary. Unfortunately, a number of other patients had left main coronary occlusions, and the ECG findings were much more variable with those.

The Bottom Line
It isn't always a straight-forward STEMI, even if the computer is telling you so!


  1. Notice how that initial beat from the first strip (in V1-V3) looks like it has Q waves, until you see the next beat with a decent R in it. If a true Q, this would have been a pretty typical LVA-looking strip, other than what you could argue to be inferior reciprocal changes. With the follow-up tracing it's less concerning.

    I'd call this a "good false positive," I think. Probably not reasonable or even appropriate to try and weed out these; better to put them into the STEMI flow but then maintain suspicion for something like dissection at the same time. Can't see how management could have been improved here except avoiding antiplatelets; TAD is a tough chameleon. Clinical presentation should probably put it on your radar, but if possible I'd be transporting somewhere with both PCI and surgical capability.

  2. I offer this case as a lesson in matching the patient to the EKG, and in reviewing the role of the computerized interpretation. If you think to yourself "What was the computer thinking?" then you probably should reconsider the diagnosis.

    Boil that all down to - paramedics (and physicians!) need to interpret ECGs, and not interpret interpretations!

  3. Marquette is usually better than this; I'm actually not sure what it was thinking, unless it really did see reciprocal changes. The quality could perhaps be cleaned up a bit -- but still. Tom B would probably be interested in this strip.

    1. aVF and V6? That was the thought, at first, when we called for the lab. And then those changes disappeared, even as the pain persisted.

    2. Yeah, I honestly would be suspicious, but for fairly nebulous reasons; that's why I'm surprised the algorithm felt the same way. Despite the fairly subtle signs, there's an anatomical localization, and I suspect that's what makes it feel different from more aberrant changes.

      In general, with a really clean tracing and no tachycardia or other severe weirdness, you could do much worse than to activate for every ACUTE MI called by the Marquette algorithm. I think that's one of the more plausible approaches for low-cost integration of prehospital STEMI systems.

    3. I'm dubious about outsourcing interpretation to R2-D2, and in general I believe that an educated EKG reader (medic or doc) has to be part of the equation. I don't know how many activations for BER or LVH a "low-cost" system could tolerate!

      I'll have to look around for a Marquette vs human comparison study, but I'm betting that Christopher W. knows far more than I on that theme...

    4. The nice thing is that Marquette is old enough it's been heavily studied. (And, of course, GE wants to promote this stuff, so it's nicely documented; if you don't have a copy of the accuracy/validation guide let me know, I can email it to you.) In most cases the *** ACUTE MI SUSPECTED message has been shown to be around 50-75% sensitive and 90%+ specific. (So not a replacement for a good medic, although a nice rule-in tool in his arsenal... but potentially a valuable addition for systems where activation would otherwise not be possible, such as BLS only or where cardiologists who don't trust EMS.) For comparisons, see particularly --

      "the positive predictive value of the computer- and physician-interpreted ECG was, respectively, 94% and 86% and the negative predictive value was 81% and 85%. . . . "The present algorithm is clearly adequate for first line screening of patients with chest pain by paramedics or in the emergency department. Its sensitivity is no worse than that of the emergency physician and its specificity is superior to the trained electrocardiographer. . . . Although more sensitive, the electrocardiographer had an overall incidence of a 5% false positive diagnosis, including a 22% incidence of false positive diagnoses in patients with isolated ST segment elevation. In contrast, the computer was nearly perfect at excluding patients without acute myocardial infarction, but did so at the expense of diminished sensitivity.”'

  4. And then there's, where ECG transmission for ED interpretation was found to be far more accurate than the prior system of computer-interpretation "screening." Frankly, the 90+% specificity of the algorithm seems suspect to me, as I see false-positives so often - it's a running theme of this blog!

    But all of this is sort of off the point. Really, the main point is that the computer interpretation played a role in steering the patient towards the cath lab, which could have been a fatal detour. The ECG, I would submit, is pretty weak for STEMI criteria, and I doubt we would have activated the cath lab in the absence of the computerized statement on the EMS ECG. Certainly, the ED ECG was not supportive of STEMI.

    So, while there exist good reasons to aggressively reduce D2B times, there are real and substantial risks. Premature diagnostic closure, encouraged by the LP-12, was a factor here, and it's worthwhile to review how to be aware of this in the future.

    Thanks for the comments Brandon - as usual, you've made me hit the books before I'm ready to reply!

  5. As I often tell people when I'm trying to teach them to be be thinking medics, "The 12 Lead ECG is confirmatory, not diagnostic".

    I think it was Marriott who said that the diagnosis of MI is based on H&P and not the electrocardiagram.

    Or was we say in ACLS, treat the patient, not the monitor.

  6. The KEY point to me about Optimizing Use of Computerized Interpretations is for Non-Expert Interpreters to first COVER UP what the Computer says. This FORCES them to make their own interpretation. After they commit themselves - THEN looking at what the computer says may be helpful. As I showed in my small study on Computerized Tracings - Potential Benefit from using Computerized Interpretations is DIFFERENT depending on User Experience. For expert electrocardiographers - I found using the computer to read ECGs for 30 providers in an out-patient clinic literally tripled my speed. For the nonexpert - the principal benefit is providing a legitimate 2nd opinion - but MUCH of that benefit is lost if the interpreter is biased by reading what the computer said BEFORE they look at the tracing themself.

    I realize all of this takes time - but NOT REALLY that much time - since EMS recognition of acute STEMI should be fairly quick once trained - and it feels REALLY GOOD to then get that "2nd opinion" (which may bring up points not initially seen, that occasion a 2nd look).

    Easy download of a user-friendly PDF I made up consisting of a series of computerized ECGs with "what the computer says" - and then my interpretations - PLUS - the chapter on this that I wrote for my 12-lead ECG book is found at: (Be sure to right-click to download to your desktop, as links in this 11MB PDF file will otherwise not work well.)

    1. Ken -

      The PDF on understanding the computerized interpretation is great! This is a new and needed approach, acknowledging the ubiquitous interpretation, but attentive to the misdirection it can lead to.

      I've always felt a little disingenuous when I've told residents or students to "ignore the computer." In truth, I use it much as you describe, as another tool, another perspective.

      Since computer interpretations aren't going anywhere soon, we should teach EKG interpretation with an eye to understanding how R2-D2 can both help AND hinder. I like this approach, and I'll see if I can incorporate this into my teaching.