Sunday, July 22, 2018

A pathognomonic ECG you should recognize instantly

Written by Pendell Meyers

Here's a pathognomonic ECG from a patient presenting with generalized weakness and malaise.

After so many examples of this on the blog, you should recognize it instantly with almost zero mental effort! What is it?

Hypokalemia. There is prolonged QT interval with characteristic ST depression followed by down-up T-waves in the precordial leads. The "up" portion of the T-wave is considered a U-wave. No signs of ischemia. Sinus rhythm.

Indeed, the potassium level was 2.3 mEq/L.

Serial troponins were negative. The patient was rehydrated and potassium was repleted.

To prove that these changes were new and associated with hypokalemia, here is the patient's baseline ECG on file from 1 year ago:

Normal ECG. All findings of hypokalemia absent at baseline.

Key Points:

You must be able to recognize such clear cases of hypokalemia instantly.

Comment by KEN GRAUER, MD (7/22/2018):
I agree completely with Dr. Meyers that on seeing this ECG and being told that the patient presented to the ED with “generalized weakness” — that your reflexive response (ie, that should be made in no more than 2-3 seconds) — should be, “What’s the serum K+?” That said — I’d add the following points:
  • What is the serum Mg++ level? I was taught that hypomagnesemia may result in virtually identical physiologic and ECG effects as hypokalemia. These 2 conditions often coexist — and some patients with hypokalemia are relatively resistant to correction of K+ until low Mg++ is also corrected. Since serum Mg++ is not always included in routine “electrolyte panels” — it is worth checking. As is the case for K+, most body Mg++ is intracellular — so extracellular (ie, serum) levels do not necessarily reflect accurate body (and intracardiac) levels of cation. As a result, the serum levels we get back from the lab tell only part of the story — and a “low normal level” (ie, a serum Mg++ ~1.7 mEq/L) may still reflect low body stores of this cation if the patient’s baseline level was initially >2.0 mEq/L. Clinical context and serial values over time may be needed to accurately gauge optimal cation replacement.
  • Could “something else” (ie, in addition to low K+) be going on in this patient? Other entities may contribute to nonspecific ST-T wave flattening with QT (or QU) prolongation. Could there have been ST elevation from “silent infarction” in this patient with new weakness — with the ST-T wave changes seen here being attenuated by coexistent hypokalemia? And while I fully acknowledge that this is NOT a likely scenario — if you see enough patients, a “1 in a 100 situation” will eventually occur on your watch if you fail to consider other possibilities. NOTE: U waves can be seen with bradycardia — and both medication use and CNS disorders may produce weakness, nonspecific ST-T wave abnormalities and QT prolongation. Clearly — Hypokalemia is the #1, #2 and #3 diagnosis in this case until proven otherwise. That said — other entities should still be considered until lab results return and careful clinical correlation is completed.
  • I LOVE this case — because we are provided with a comparison tracing. Thus the Bottom ECG (BLUE border) in Figure-1 was obtained 1 year earlier on this patient. Note that there was some ST elevation in the prior ECG in the inferior leads. This most probably reflects the benign variant of early repolarization — but I suspect it accounts for the fact that although inferior lead ST-T waves are flattened in the ED ECG (RED border in Figure-1) — they are not as depressed as I might have expected with the exceedingly low serum K+ level of 2.3 mEq/L. I find it helpful to repeat the ECG of a patient with a new electrolyte disorder after treatment — in this case, just to be sure that the T wave inversion in leads V3, V4 and other nonspecific ST-T wave abnormalities on the initial tracing are improving.
Spend another moment reviewing the prior ECG (BLUE border in Figure-1):
  • For educational purposes — I think it worthwhile to point out: iThat the T wave inversion in lead aVL of the prior tracing is not abnormal, given the relatively vertical frontal plane axis with resultant predominant negativity for the QRS in aVL (ie, the T wave axis often "follows" the QRS axis); and iiThat there is probably some misplacement of one or more chest leads — given P wave negativity with shallow T wave inversion in leads V1 and V2 (often suggests these leads have been positioned 1 or 2 interspaces too high) — with then rather abrupt transition from the tiny r wave in V2, to a much more positive R wave already by V3.
Figure-1: Comparison of the 2 ECGs in this case (See text).
Our THANKS to Dr. Meyers for this insightful case!

Friday, July 20, 2018

Bupropion Overdose Followed by Cardiac Arrest and, Later, ST Elevation. Is it STEMI?

A young woman presented with status seizures and apparent overdose of bupropion.  There was a question of cocaine use too (with later suspicion of possible ingestion or body stuffing).

She had status seizures for which she was intubated and medically treated (successfully) with propofol and benzos.

An ECG was recorded:
Sinus tach, with a slightly widened QRS (113 ms) and slightly long QT
There is a slightly abnormally large R-wave in aVR.
So there might be some sodium channel blockade here, which is expected with cocaine.
Bicarbonate was given.

This was recorded 8 hours later:
QRS = 148 ms and large R-wave in aVR (very dangerous)
This is typical of Na channel blockade.  
Bupropion and Cocaine are both powerful Na channel blockers.
Computerized QTc = 486 ms
Bazett correction = 546 ms
Fridericia correction = 528 ms

My measurement = 550 ms
Bazett = 618 ms
Fridericia = 598 ms

(Very long QT)

Shortly thereafter, the patient had a witnessed PEA arrest that resolved with epinephrine and bicarbonate.

The arrest was not due to an arrhythmia (not Torsades de Pointes).

Here is the post arrest ECG:
There is RBBB with QRS duration is 137 ms and there is a very large R-wave in aVR.

There are 2 bumps on the T-wave; one could be a U-wave.
Since a long QU is also dangerous, let's assume it is all T-wave.
Computerized QT = 420 ms, with QTc = 490 ms. 
I measure 480 ms.  Fridericia correction = 615 ms.  Very long. 

The exact etiology of the PEA arrest was uncertain but a because the Na channel blocking activity, sodium bicarb was given and the patient stabilized.  (Most arrest from wide QRS in Na channel blockade starts with ventricular tachycardia, and most with long QT starts with polymorphic VT (Torsades de Pointes).)

The next day, this was recorded at a time when electrolytes were normal:
There is ST elevation and large T-waves.
The computer read ***STEMI***

It does meet STEMI criteria of at least 1.5 mm in V2 and V3 in a woman.
Is this STEMI?
Computerized QT = 462 (Bazett correction = 556 ms; Fridericia correction = 525 ms)
Troponin returned at 4.5 ng/mL.

The tox consultant texted me to confirm her opinion that this is not STEMI.  Is it?

Is it STEMI?  No.  There is a bizarrely long QT interval.  My measurement = 520 ms (Bazett correction = 634 ms; Fridericia correction 591 ms).  

The morphology of the ST segment is just not right for STEMI.  

Moreover, the clinical context is not right for STEMI.  It is right for acute myocardial injury due to direct toxic effect or to type 2 myocardial infarction due to hypoperfusion during arrest or shock.

This is an ECG that one would commonly see in Takotsubo Stress Cardiomyopathy, whether or not there is actual echo evidence of it.  

No echo was done, but my advice was that this was a purely toxicologic ECG, with extremely long QT.

EKG day 3:
QRS 110
Computerized QT 415, QTc 441

Again, there is an extremely long QT that the computer did not detect.Large U-waves, K = 3.9

EKG day 4:

The patient eventually did well and recovered.

Learning Points:

1. Recognize the effects of sodium channel blockade: large R-wave in aVR and widened QRS, often with RBBB morphology.

2.  A widened QRS in Na channel blockade toxicity is very dangerous and often leads to cardiac arrest

3. The computer is terrible at measuring the QT interval.  You must measure it yourself.

4.  Not all acute ST elevation is due to acute coronary occlusion.

5.  A large bupropion overdose can be very dangerous

Comment by KEN GRAUER, MD (7/20/2018):
Superb case by Dr. Smith for illustrating sequential effects of Sodium-Channel Blockade with the important clinical implications to be aware of. I’d add the following:
  • Great example of how serial ECGs on this patient simulate a number of other ECG conditions. As per Dr. Smith — despite troponin elevation and ST elevation in leads V2 and V3 of the 4th ECG — both the shape of these ST segments and the clinical context argued strongly against this being a STEMI.
  • Similarly, despite the QRS widening, upright R wave in V1 and S waves in lateral leads in the first 3 ECGs — the shape of the QRS in lead V1 was not suggestive of RBBB due to a conduction defect. For a QRS complex as wide as it is in these tracings — one simply would not expect the small (less than 5mm tall) amount of R wave positivity and narrowness that we see in V1 if there was true RBBB. Instead, the clinical context + combination of marked right axis, large terminal R wave in aVR plus QRS prolongation with marked QTc widening suggests severe toxic overdose as the cause of these ECG abnormalities (Figure-1).
  • The KEY to optimal use of computerized interpretations lies with knowing HOW to use the computer. Computers analyze data. If data-IN is poor — then data-OUT will be poor. The ECGs in this case for which computer estimation of the QTc was grossly inaccurate, showed baseline artifact + indistinct borders for the end of the QT interval in most leads. If YOU are having trouble quickly discerning where the QT interval ends — then the computer will also have trouble doing so. In such cases — Do NOT expect computer estimation of the QTc to be at all accurate. In contrast — when artifact is minimal or absent, and the onset and offset of the QT interval is easy to ascertain on the ECG — it is much more likely that the computer WILL be accurate in its assessment of intervals. This does NOT mean that you should ever accept computer values without overreading them. But it does mean that no more than a brief glance may be needed for you to validate that the computer is accurate in its assessment of intervals when the quality of the tracing is good, and all limits are clearly definable. That said, in THIS case — it should be obvious from the start that the computer can not be trusted at all for its assessment of any interval.
  • To clarify the process of QTc estimation in this case — I’ve marked in Figure-1 the 2 leads in which I am comfortable with my notation of where I believe the QT interval ends. I measure the QT ~550-560ms — which at a heart rate of ~75/minute, corrects to a QTc of >600ms! We could also have used lead V4 to assess the QT in this tracing — but probably none of the other 9 leads. The indistinct delineation of the end of the QT interval explains how the computer estimate of 486ms for the QTc came to be so far off from what the QTc actually is.
Figure-1: 2nd ECG in this case — illustrating atypical RBBB morphology and indistinct borders for the end of the QTc in most leads. (See text).
Our THANKS to Dr. Smith for this rare insight into the serial ECG changes of severe Sodium-Channel Blockade!

Tuesday, July 17, 2018

Syncope in a 20-something woman

A 20-something was outside exerting herself.  She states that  it was hot outside and that she was probably dehydrated. At one point, she felt lightheaded and then can't remember anything until waking up in the ambulance.  Her friends saw her lose consciousness and fall on the ground.  She regained consciousness spontaneously before responders arrived. Fire department was on scene first, who noted a cyanotic color to the patient's face.  EMS arrived and also noted cyanotic color which improved en route to HCMC. She denies head pain, neck pain, back pain, abdominal pain or any pain at this time.  There was no nausea or vomiting.

In the ED her exam was normal.  All vital signs were normal, with a pulse of 65.

Here is her ECG:
Computer interpretation: borderline long QT interval
What do you think?

Physician assessment (who apparently took the computer read as truth): "ECG had borderline prolonged QT interval, but otherwise did not have signs of arrhythmia or RH strain (Pulmonary embolism) that would be underlying cause of syncope."

Should we believe the computer's assessment?  Never!  When you have ANY suspicion that the QT is prolonged on visualizing the ECG, you must measure it.  The computer is often wrong.

Here I measure the QT:

One should measure the longest QT of the 12 leads.
I did not do that, but just chose one that looked long
See the 2 black lines at the beginning of the QRS and end of the T-wave.
This measures about 580 ms.
The heart rate is barely above 60, so correction will not change it much.
So this is a dangerously long QT (any corrected QT above 480 - 500 ms is associated with a significant risk of Torsades de Pointes--TdP)

There is actually a second hump on many of the T-waves.  Are these large U-waves?  Perhaps, but either way, it is very abnormal and puts the patients at risk for TdP.

See this post on QT correction: 

QT Correction Formulas Compared to The Rule of Thumb ("Half the RR")

Final physician assessment: dehydration.  The patient was discharged.


Another Emergency physician was later reading the ECGs on the system and putting in final interpretations.  She immediately recognized long QT.  She called the patient back.  The patient was admitted.  

She was on no medications that would prolong the QT.  Her K was 3.4 but the ECG findings persisted after correction.  

Thus, she was diagnosed with long QT syndrome.  The electrophysiologist diagnosed it as "probable type 2" long QT (beyond the scope of this blog post and of my understanding).

Genetic testing was undertaken, results pending.

Nadolol 20 mg prescribed.  It is typical to prescribe beta blockers for congenital long QT, though that depends on the type of congenital long QT.

For more on syncope:  

Emergency Department Syncope Workup: After H and P, ECG is the Only Test Required for Every Patient.....

Comment by KEN GRAUER, MD (7/17/2018):
The KEY to optimal ECG interpretation is to routinely use a Systematic Approach. This case is a perfect example of how easy it is to miss a seemingly obvious finding if one fails to systematically assess each and every tracing encountered. This happened in this case. The computer did not detect QT prolongation. The initial treating physician believed the computer report — and failed to independently assess the QT interval. Perhaps because T waves were indistinct in most leads — it wasn’t until a second physician overread the tracing (after the patient had already been discharged …) that marked QTc prolongation was finally picked up.
  • The 6 key parameters to assess in interpretation of any 12-lead ECG include Rate – Rhythm – Intervals (PR-QRS-QT– Axis – Chamber Enlargement – and QRST Changes for Ischemia/Infarction. Sequentially assessing every ECG for these 6 parameters organizes your approach — speeds up your interpretation (because you are organized) — and avoids missing important findings.
  • As a quick estimate — the QTc is prolonged if the QT interval is clearly more than half the R-R interval. One selects a lead where the limits of the QT interval are clearly defined — and the interval looks to be longest. Ignoring the leads in which the ST-T wave is fairly flat and amorphous — one could select any of the lateral chest leads (we favor lead V4) — which clearly shows marked QT prolongation.
  • In view of this patient’s presentation (ie, exertional syncope), and the fact that mild hypokalemia was not the cause of her ECG abnormalities — the diagnosis of LQTS (Long QT Syndromewas made. Referral to EP Cardiology was appropriate given potential for arrhythmic sudden death if not treated.
As per Dr. Smith — specifics of the diagnosis of LQTS are complex, and beyond the scope of most non-cardiologists. That said, I’ll make the following points:
  • Congenital LQTS is not a common disorder. Yet it is an important entity to be aware of — because it is a potentially lethal disorder that may occur in otherwise healthy young individuals. Think of this possibility in the setting of unexplained syncope. These patients are susceptible to episodes of Torsades de Pointes that are often precipitated by adrenergic stimulation (as occurs with physical exertion or mental or emotional stress).
  • There are a number of forms of congenital LQTS, which are thought to be caused by different ion channel mutations. Clinical and ECG manifestations may differ among the various forms — with the common denominator being QT prolongation not due to other cause. That said, the amount of QT prolongation will not always be marked. In such cases, a high index of suspicion may be needed to make the diagnosis.
  • ECG features may suggest one of the more common types of LQTS. To illustrate this — I have excerpted the picture in Figure-1 from Michael Crawford’s Cardiology Text. My purpose in doing so is not to imply that the non-cardiologist needs to know what the ECG features of LQT1 vs LQT2 vs LQT3 are — but rather that awareness of an ususual QTc morphology may clue you in to the possibility that the patient in front of you may have a form of LQTS.
  • Take another look at the admission ECG for the 20-year old woman in this case (Figure-2, which I’ve placed just below Figure-1). The ECG in Figure-2 lacks the ST depression and usually distinct U wave humps of hypokalemia. Instead, the T wave is almost amorphous in the 6 limb leads — marked by a distinct (and unusual) notch in lead V3 — and a bizarrely-shaped ST-T wave in leads V4, V5, V6. Comparing this ECG with the examples in Figure-1 — I can see why the referring EP Cardiologist thought, “probable Type 2 LQTS”. 
Figure-1: Excerpted from Crawford MH: Current Diagnosis & Treatment Cardiology (4th Edition): McGraw Hill, New York, 2014.  (See text).
Figure-2: Admission ECG for the 20yo woman in this case. (See text).

Friday, July 13, 2018

A man in his 40s with chest pain

Case submitted and written by Alex Bracey, with edits by Pendell Meyers and Steve Smith

This ECG was tossed onto my desk on my first day of a new rotation at a community site. The technician was nowhere to be found by the time I turned to ask what the story is or where the patient is located.
Initial ECG at 1350 - are you concerned?

- There is 0.5 mm STE in aVL, no clear STE in lead I.

- There is ST depression in II, III, and aVF.

- Nicely demonstrated here, leads III and aVL are reciprocal: STD in III is reciprocal ST depression to STE in aVL. This is diagnostic of occlusion (OMI).

- There is also some slight STD in V4-V6. Out of context, STD in V4-V6 as well as II, III, and aVF, with slight STE in aVR may make you think of diffuse supply demand mismatch ischemia as can happen with any pathology involving diminished blood flow to the majority of the heart, such as left main stenosis or severe three vessel disease. However, the morphology and STE in aVL with the morphology of the reciprocal STD inferiorly sets this case apart and identifies OMI rather than just global supply/demand mismatch.
- STEMI criteria are not met.

No prior ECGs were available.

My co-resident and I went to find the patient in the main ED. He was a man in a his late 40s who was alert but looked as though he was trying to maintain a smile through some discomfort. He told me that he had been having chest pain for ~6 hours, described as a heaviness in his chest, that started just after signing his divorce papers. He was an active smoker with hypertension and had a closely related family member that had an MI at the age of 49.

We gave him aspirin and asked for a repeat ECG.

While awaiting the repeat ECG, it was suggested that cardiology be consulted emergently for intervention. A senior team member, while concerned about the ECG wanted to wait for the troponin to result before involving cardiology. At 14:35, ~45 minutes after in the initial ECG the patient, the troponin I resulted as 0.701 ng/mL (normal range 0.000 - 0.045 ng/mL).

At this point, even if the ECG is confusing to you, the diagnosis is made: chest pain with an elevated troponin in this context is acute myocardial infarction. No matter what the ECG shows, with continued chest pain the patient must go emergently to the cath lab. In this case, you have BOTH a diagnostic ECG and a diagnostic troponin.

A repeat ECG was obtained shortly after:
Repeat ECG @ 1447

Similar to prior. The artery is still occluded.

At this point, cardiology was consulted and I described the findings and relayed concern for acute coronary occlusion in the absence of a STEMI. The cath lab was not immediately activated, but cardiology staff was sent to evaluate the patient. The patient was admitted to cardiology and underwent cardiac catheterization ~2.5 hours after presentation and initial ECG.

Prior to the publishing of the cath note, we received a phone call from the interventional cardiology who commended us and informed us, with some surprise, that he had found thrombus in the LCx second obtuse marginal, which had successfully undergone PCI with one drug eluting stent.

On the actual cath report, no initial stenosis percentage or TIMI flow were reported; rather, only the post procedure TIMI III and residual stenosis of 0% are documented.

This is the subsequent ECG taken just after PCI:
ST changes are resolving.

One day later, the troponin I peaked at 48.2 ng/mL. He was discharged home the subsequent day with a diagnosis of "NSTEMI."


STEMI criteria will miss a significant number of acute coronary occlusions. While this criteria exists and is still recognized as the standard for diagnosing acute coronary occlusion, patients will continue to have delay to definitive management (e.g., PCI). As in this case, the diagnosis was made at hospital presentation, but he patient waited for 2 hours until intervention. Even then, the cardiologist was surprised to find an acute lesion. As we've said before, we must change the language to include those that will benefit from emergent intervention even though they do not manifest STEMI criteria (see this post: OMI Manifesto).

We were fortunate to have a fairly receptive interventionalist on this case; however, that will not always be the case. In such instances, serial ECGs become invaluable, as the patient may manifest worsening STE or even simply dynamic ECG changes which may be enough to convince an interventionist to emergently activate the cath lab. But notice that in this case there was no serial change; you cannot depend upon it!

Do not wait for the troponin to result before getting cardiology involved with a concerning ECG and story; the troponin plays a supporting and complimentary role in the diagnosis of OMI. Keep in mind that contemporary troponin assays will only start to manifest at the 4-6 hours mark, and many people with ACO will present well before that point. Had this patient’s troponin been undetectable, it would not have changed where this patient needed to go: the cath lab. Had it been negative, it may have delayed his time to PCI. Time is myocardium.

Comment by KEN GRAUER, MD (7/13/2018):
It is cases like this one that emphasize the tremendous importance and educational impact of Dr. Smith’s ECG Blog. This case should not have been delayed for several hours before being recognized for the OMI that it is. I’ll add the following points:
  • As per Drs. Bracey, Smith and Meyers  — dependence on strict STEMI criteria will miss a significant number of acute coronary occlusions. This is especially so in cases like this one, in which only lead aVL manifests ST elevation. It is important to appreciate that lead aVL is the “highest” lateral lead — and with acute occlusion of the 1st or 2nd obtuse marginal branch of the LCx — lead aVL may occasionally be the only lead to show ST elevation.
  • There is ample evidence on the rest of the initial ECG (done at 13:50) to support the conclusion that changes in lead aVL must be assumed acute until proven otherwise. As per Dr. Bracey, Smith and Meyers — this evidence includes reciprocal ST depression in each of the inferior leads, with an exact “mirror-image opposite picture” of the ST elevation in aVL seen by the shape of the ST-T depression in lead III. In addition, there is ST segment flattening with 0.5-1mm of ST depression in leads V4, V5 and V6.
  • The appearance of anterior leads V1, V2 and V3 is also worthy of mention. On initial inspection of the 1st ECG (done at 13:50) — there appears to be a QS complex in each of these 3 anterior leads. This raises suspicion of anteroseptal infarction at some point in time. The disproportionately tall (and fatter-than-expected-at-its-peak) T wave in lead V3 might raise suspicion of an acute event in the LAD distribution, if not for 2 points: ithe even more convincing ECG picture in the limb leads that suggests acute LCx Obtuse Marginal occlusion; andiithe likelihood of some chest electrode lead misplacement, given the prominent negative P wave component in V1, V2 — and, the unexpected decrease in S wave amplitude from V1-to-V2, that then unexpectedly increases again by lead V3. Sure enough, a definite initial positive deflection (r waveis seen in lead V3 of the 2nd ECG (done at 14:47) — which makes prior anteroseptal infarction less likely.
COMMENT: The presentation of this case reminds me of my experience as the provider charged with interpreting all ECGs done by the 35 practitioners at our ambulatory Family Medicine Center over the 30 years that I served in the Residency Program as full-time core faculty. In that role, I was often presented a series of tracings without any history. The challenge was determining which tracing(s) showed findings that mandated dropping whatever I was doing to immediately find out the history, because of concern for a potential acute condition. Credit to Dr. Alex Bracey for doing just that in this case! Had he not directly hunted down the patient in the ED — there is a very good chance that this patient’s OMI would not have been recognized in time.
  • MORALIt’s impossible to appropriately interpret ECGs without clinical correlation! The initial (13:50) ECG in this case might not necessarily be cause for alarm in the absence of symptoms — especially if a prior ECG was found showing similar changes. But on learning that the patient in this case had new-onset chest discomfort just a few hours earlier — acute OMI has to be assumed until proven otherwise, regardless of what serum troponin shows — and regardless of whether the 2nd ECG shows serial changes.

Recommended Resources