Wednesday, May 15, 2019

Potassium and Magnesium: how low is too low?

A 30-something woman with h/o HTN on losartan presented for vomiting, diarrhea, and cough for 2-3 days.

K returned at 2.4 mEq/L (3.5 - 5.0 is normal) and Mg returned at 0.7 mg/dL (1.6-2.6 is normal)

An ECG was recorded:
What do you think?

Here is her ECG a couple years prior with a K of 3.7 mEq/L:


The QT is long at about 470 ms, with a QTc that is longer (that number depends on the correction formula used).   The heart rate is 81 beats per minutes (RR interval = 0.74 sec, or 740 ms).   Our computer calculated the QTc at 505 ms.  I believe it uses the Hodges formula.

There are very minimal (normal) U-waves.

(The previous QT was normal.)

Other formulas:

The QTc-Bazett is 546 ms
The QTc-Fridericia is 519 ms
The QTc-Framingham is 510 ms.

How acutely dangerous is a QTc greater than 500 ms?

That is uncertain.  We will shortly be publishing a review of this and of QT correction methods in the Western Journal of Emergency Medicine, but I will not (yet) give details of that.

But we do know that hypokalemia with an abnormal ECG (long QT or prominent U-waves) is dangerous.  

We just published an abstract on hypokalemia, and it is being presented at SAEM:, abstract 253.

The Incidence of Serious Adverse Events After Emergency Department Treatment for Hypokalemia
Brian Driver1, Jacob Helmer, Ashley Nelsen, Joseph Pasquarella, Maros Cunderlik, Lauren R. Klein1, Rajesh Satpathy1, Michael A. Puskarich2, and Stephen W. Smith.  Hennepin County Medical Center/University of Minnesota

Conclusion: Adverse events in patients with hypoK were rare and not directly related to the K level.  

But just one case of death is enough to affect patient management.

Case continued:

The patient felt better after treatment, including K supplementation, and wanted to go home.  She was told that this was a very bad idea and that she could die.

Shortly thereafter, she became unresponsive and was found to be in ventricular fibrillation.

She was defibrillated.

All serial troponins were negative.

Learning Point:

Hypokalemia with associated ECG abnormalities is very dangerous.

Here is a previous case of ventricular fibrillation from hypokalemia:

Prehospital Ventricular Fibrillation in a Young Woman. What is the Diagnosis?

Here is a detailed post on hypokalemia:

Comment by KEN GRAUER, MD (5/15/2019):
I found this case extremely interesting — because had I not been told at the outset what serum K+ and Mg++ values were — I never would have guessed they were critically low from looking at the initial ECG.
  • For clarity — I have put both tracings together in Figure-1.
Figure-1: The 2 ECGs that are shown in this case (See text).
MThoughts on ECG #1:
  • There is a regular sinus rhythm at ~80/minute. The PR and QRS intervals are of normal duration — but the QTc is clearly prolonged. I estimated a corrected QTc ~510 ms (similar to values calculated by Dr. Smith).
  • Other descriptive analysis findings on this tracing included — a normal frontal plane axis (about +35 degrees) — no chamber enlargement — and peaked waves in multiple leads.
  • COMMENT — Had I not been aware of this patient’s severe electrolyte imbalance — my clinical impression of this tracing would have centered on 2 ECG findings: ithe prolonged QTcandiipeaked T waves in multiple leads. I would not have suspected critically low values of serum K+ and serum Mg++ — because these electrolyte abnormalities are most commonly associated with ST-T wave flattening (sometimes with ST depression) — U waves — and, a long QT (QU) interval — but not with the overly tall and peaked T waves that we see in ECG #1. If anything — I would have considered hyperkalemia from the T wave appearance in ECG #1 (albeit the base of peaked T waves with pure hyperkalemia tends to be narrower than that seen here). NOTE: Combined hyperkalemia + hypocalcemia (as is sometimes seen with renal failure) — may produce a picture of peaked T waves + a prolonged QTc.
Regarding the Prolonged QTc:
In a patient with a normal-duration QRS (ie, no BBB or IVCDand no sign of ischemia/infarction — the finding of a Prolonged QTc should suggest: iDrug Effect (from one or more of any number of medications that may prolong the QTc— iiElectrolyte Disturbance (low K+/low Mg++/low Ca++ — or some combination of these— and/oriiiCNS Catastrophe (ie, stroke, seizure, CNS bleed, trauma, coma, tumor — may all produce marked ST-T wave abnormalities, together with marked prolongation of the QT interval).
  • Clinical correlation is needed to discern between which one or more of the above common causes of QTc prolongation is operative in the case at hand. From the brief history given in this case — it sounds like the principal cause of the long QTc was the severe electrolyte disturbance.
MThoughts on ECG #2:
  • There is sinus tachycardia at ~110/minute. The PR and QRS intervals are both normal. The frontal plane axis is normal — and there is no chamber enlargement.
  • Assessment of the QT interval is more difficult when the heart rate is rapid (especially when >100/minute). Considering the rapid rate in ECG #2 — I thought the QTc here was borderline prolonged. That said — the QTc has clearly become longer in ECG #1 than it was in this baseline tracing.
  • T waves were prominent in the prior ECG (ie, in ECG #2). Especially in leads V3, V4 and V5 (and to a lesser extent in leads II and V6) — these T waves are tall, peaked if not pointed, symmetric in ascent and descent, and manifest a narrow base. This appearance could clearly be consistent with hyperkalemia.
COMMENT — It should be remembered that the overwhelming majority of body K+ (and body Mg++) is intracellular — so a single serum (extracellular) K+ value in the lower normal range = 3.7 mEq/L does not necessarily reflect “adequate” body stores of this cation. The serum Mg++ value of 0.7 mg/dL in ECG #1 was critically low — and we do not know serum Mg++ at the time ECG #2 was obtained — so it’s difficult to know how to compare relative ST-T wave changes in these 2 tracings on the basis of serum K+ values alone.
  • T wave peaking can be seen transiently in normal subjects during exercise (I saw numerous examples of this in many ETTs I did over the years in otherwise healthy young adults). Then again, this young adult woman at seemingly low risk of acute ischemic heart disease did develop VFib, and only did so after a period of electrolyte replacement …
  • Speculation  I would have loved to see a true baseline ECG on this patient obtained at a time when we knew body electrolyte stores were normal — and, when heart rate was normal. Only then would we know with certainty IF this patient’s baseline tracing had prominent T wave peaking — or — if perhaps both the long QTc and the prominent T waves with wide base in ECG #1 might be an acute response to something in addition to acute electrolyte disturbance (ie, acute ischemia …).
FINAL Thoughts: It proved lifesaving that the clinicians in this case were able to convince this patient not to go home after partial electrolyte replacement!
  • Severe hypokalemia and hypomagnesemia do not always manifest the ECG findings expected with these 2 electrolyte disorders. If anything — ECG #1 should prompt consideration of hyperkalemia rather than hypokalemia …
  • Recognition of the prolonged QTc in ECG #1 (which is clearly longer than on the prior ECG of this patient) — proved critical to appreciating high risk for adverse event in this patient!
Our THANKS to Dr. Smith for presenting this insightful case!


  1. Really would like to see what her baseline was after treatment... this is the downside of diuretics with flushing K+ out of the system. I noticed she had some fragmentation in III - maybe that played an impacted role or was a conduction abnormality directly in relation to hypokalemia. Interesting case though Steve!

  2. I can't understand it. The patient has hypokalemia and on the EKG she presents high and pointed T waves. I can not understand it. I would have been afraid to administer potassium with high and pointed T waves. I do not get it...

    1. I understand your confusion, but perhaps this will help: T-waves of hyperkalemia are much sharper and the base of the T-wave is narrow. These T-waves have a very wide base, which is why they do not end up in a narrow sharp point. So these do not look like hyperK and DO look like hypoK!

    2. @ Doinel — Please see My Comment (written above, at the bottom of the post).

  3. Steve and Ken...

    It is well-known that QTc prolongation in the setting of hypokalemia (not to mention hypomagnesemia) is a really BAD combination. Is it known whether this patient developed Torsade de Pointes prior to fibrillating?

    Although I thought the T waves were somewhat peaked and very symmetrical, the base was simply too wide for hyperK+. However, I would NEVER have guessed HYPOkalemia! This is a great reminder that hypokalemia ALSO kills.

    1. THANKS for your comment Jerry! You, me and Steve are ALL on the same line of the same page that the T waves in ECG #1 are wider-than-expected for pure hyperkalemia (ergo, the reason I've drawn an Eiffel Tower in my ECG "Pocket Brain" books for many years to emphasize the tall, peaked, pointed T waves + NARROW base [like the Eiffel Tower]). But not all cases of hyperkalemia are "pure" (ie, they are often mixed with other entities that may affect the ST-T wave + QTc, not to mention presence of underlying ST-T wave abnormalities in the "baseline" tracing before the hyperkalemia developed) — so from a clinical perspective, when I saw ECG #1 — I thought "Check to ensure serum K+ is not elevated", even though I also thought to myself that the T wave base was wider-than-expected for pure hyperkalemia). That said, would you agree that the T wave base in ECG #2 ( = the prior ECG on this patient) DOES manifest pointed and NARROW-based T waves in a number of the chest leads? BOTTOM LINE for ME when I looked at ECG #1 — despite the long QTc, I never would have guessed hypokalemia — and for me, that's what makes this such a GREAT LEARNING case! Thanks again (as always!) for your comment! — :)

    2. Ken...

      The T waves in leads V3 - V5 in the 2nd ECG could certainly pass muster as hypERkalemic rather than hypOkalemic and perhaps the 3rd T wave in Lead II as well.

      I DO agree with you that had I been managing this case, I would certainly have had an eye out for the K+ level but I definitely would not have begun pushing Ca++ based on those T waves (in the 1st ECG) unless the patient had a history, for instance, of renal failure and non-compliance with dialysis. T waves in THOSE cases could look like ANYTHING! However, in spite of the findings in THIS case, I hope physicians are no less timely in their use of calcium in cases of suspected hyperkalemia.

      This was definitely a learning case for everyone!

    3. I agree completely with your last comment Jerry! THANK YOU — :)


Recommended Resources