This post is by one of my residents and the new EMCrit Intern, Pendell Meyers. He has a knowledge and love for ECGs that far exceeds his years of medical training. He has worked extensively with Steve Smith on the ECG blog and will be doing a fellowship in Resuscitation after he graduates at Stony Brook.–Scott

Critical hyperkalemia is one of the quintessential EM emergencies: immediately life threatening, immediately recognizable, and immediately treatable. No other specialty sees such a high frequency of undifferentiated hyperkalemia, and therefore no one else needs to be better at recognizing hyperkalemia on the ECG. You check a fingerstick glucose on every altered patient because hypoglycemia is rapidly lethal, easily identifiable, and immediately correctable, right? Checking the ECG for hyperkalemia should be just as obvious of a priority for a wide variety of patient presentations ranging from weakness to unexplained bradycardia to cardiac arrest, but requires us to be expert at recognizing and treating it.

Skilled ECG interpretation is likely the most important factor to identify and prevent hyperkalemic deaths. Point of care potassium testing is not universally available and is not always trusted due to frequently erroneous and misleading results. The main lab is not fast enough when it really matters, as patients who decompensate usually do so before the lab results are back or within the next 30 minutes after results return.¹ Cardiologists are unlikely to help you recognize significant hyperkalemia on the ECG even in the few cases in which they are involved in real time. They aren’t trained for this and don’t see it nearly as much as we do. Even worse, they may be fooled (as we sometimes are) into thinking that the ECG changes are indicative of STEMI, taking the patient to the cath lab where they will have a delay in diagnosis and an unnecessary procedure if they survive long enough to get it.²

The classic teaching is the clear, reliable, do-not-pass-go, step-wise progression of ECG findings shown below.³

The common caveats that are always provided with these ECG findings include:

“Like all other processes and levels in the body, the faster the rise/onset, the more dramatic the effects; the slower the rise/onset the more able the body is to adapt to it.”

“Patients may have a normal ECG despite critically high potassium level.” ^4,5

“Patients may have hyperkalemic arrests with a preceding normal or near-normal ECG”

“Classic hyperkalemia findings are not sensitive enough to predict hyperkalemic adverse events.”

There certainly are recorded cases of hyperkalemic arrest with a preceding normal or near-normal ECG.⁶ On the other hand, there are also published case reports which are frequently misquoted as showing hyperkalemic arrest with preceding normal ECGs. For example, Dodge et al.⁷ is often referenced as a case of sudden hyperkalemic arrest with preceding normal ECG. In fact, the previously healthy young volunteer who died soon after receiving a single oral dose of more than 400 mEq oral potassium was noted to progress from normal to peaked T-waves to QRS widening, then short runs of VT before VF and death.

Until recently, the best literature we had on this topic was a retrospective chart review by Montague et al, which must be dissected and understood before we can put the newer literature in context:

Montague et al.⁸

Montague et al. retrospectively identified 90 patients with K>/= 6.0 mEq/L who had a coincident ECG within 1 h of laboratory draw and before any therapy for hyperkalemia. Sixteen of these (18%) met “strict” criteria for ECG changes and 47 (52%) showed some T-wave change (any of the criteria in table 1 below).

New QRS prolongation (QRS>120ms) was noted in only six ECGs at time of hyperkalemia. They found 14 patients with arrhythmias (N=12) or cardiac arrest (N=4), of which only one met “strict” criteria for ECG changes and fewer than half were noted to have new or resolving T-wave peaking or symmetry.

The biggest limitation of this study’s methodology is that T-wave peaking seems to have been the only ECG indicator studied as a predictor of adverse outcomes. They report that 6 patients had new QRS prolongation, but there is no mention of whether these patients had any of the recorded adverse outcomes. There is also no mention of how they defined QRS widening if the baseline ECG already had a wide QRS. They did not study any of the other known hyperkalemia ECG findings. Unfortunately they also did not define which rhythms were considered “arrhythmias”, and they do not mention which arrhythmias happened in those 12 cases, meaning these could have included anything from atrial fibrillation to asymptomatic bradycardia to ventricular tachycardia. Given the severe limitations of this study, the only reasonable conclusion to be drawn is that T-wave peaking alone is likely an insufficient predictor of the adverse events in this particular study.

Durfey et al.¹

Durfey et al. recently published a slightly larger and more methodologically rigorous retrospective chart review. They identified patients with K ? 6.5 mEq/L and excluded those with evidence of hemolysis, platelet count ? 500 x 10⁹/L, paced rhythm, and hyperkalemia treatment prior to ECG. They studied the following eight ECG findings as predictor variables:

(1) peaked T waves

(2) PR prolongation (>200ms or > previous if baseline PR prolongation)

(3) QRS prolongation (>110ms or > previous if baseline QRS prolongation)

(4) bradycardia (HR<50 bpm)

(5) 2nd or 3rd degree heart block

(6) junctional rhythm

(7) ventricular escape rhythm

(8) ventricular tachycardia

They defined adverse events as: symptomatic bradycardia requiring treatment, ventricular tachycardia, ventricular fibrillation, cardiopulmonary resuscitation (CPR), or death within six hours of lab measurement.

Of the 188 patients identified with K ? 6.5 mEq/L, 134 (71%) had at least one of the studied ECG findings. Of those with ECG findings, 57% had one one abnormality, and 43% had multiple, with the most frequent combination being QRS prolongation and peaked T-waves.

Twenty-eight adverse events (15%) occurred within six hours: symptomatic bradycardia (n=22), death (n=4), ventricular tachycardia (n=2) and CPR (n=2). The median time from the ECG to adverse event was 47 minutes. Adverse events occurred either prior to the laboratory notification of hyperkalemia (n=16, 57%) or shortly after (mean 36 minutes) the laboratory notification of hyperkalemia (n=12, 43%).

All 28 patients with an adverse event had at least one ECG abnormality, and 86% had more than one abnormality. The two most common abnormalities were QRS prolongation (n=22) and bradycardia < 50 bpm (n=17). Two patients had isolated bradycardia, one had isolated junctional rhythm, and one had isolated QRS prolongation, but none had isolated PR prolongation or isolated peaked T-waves. The findings with highest relative risk for adverse events were QRS prolongation (RR 4.74, 95% CI [2.01-11.15]), junctional rhythm (RR 7.46, 95% CI [5.28-11.13]), bradycardia (HR<50 bpm) (RR 12.29, 95%CI [6.69-22.57]), and ventricular escape rhythm (n=4, all 4 with an adverse event). No adverse event was recorded after calcium administration, and all symptomatic bradycardia or VT improved after treatment with calcium.

Notably, 75% of patients with adverse events did not have peaked T-waves, and there was no significant correlation between peaked T-waves and adverse outcomes (RR 0.77, 95% CI [0.35-1.70]). Among those with isolated peaked T-waves or isolated PR prolongation, there were no adverse events recorded.

The graph below demonstrates increasing sensitivity of ECG findings for both increasing levels of hyperkalemia and adverse outcomes. Note that even the cohort of patients with K = 6.5-6.9 had a short-term adverse event rate of nearly 10%.

Take a look at Table 2 with the performance details of each ECG finding:

Now pare it down to just the ECG findings and their relative risks:

Next, rearrange the order of listed findings from least to greatest relative risk:

Finally, compare the resulting table to the original list of classic ECG findings:

With some common sense and no statistical wizardry required, we can see that there is a correlation between the classic progression of ECG findings and increasing relative risks for adverse events in this study.

Problems with Durfey et al:

• Despite being the largest retrospective study to date, it is still a small study with a very small number of adverse outcomes.

Given the prevalence of dialysis related hyperkalemia alone, it should be possible to perform a larger retrospective or even prospective study.

• Ventricular tachycardia was an ECG “predictor” but also an adverse outcome.

We were concerned that VT was listed as both a predictive finding and an adverse outcome in this study. VT would likely be a strong predictor of further episodes of VT, but clearly the adverse outcome is already present and thus it cannot also be a predictor. Additionally, we wondered if they required two separate events of VT, one during the initial ECG and another at a later point in time. If only one event of VT was required to fulfill both initial predictor and adverse outcome, then the time from ECG to adverse event would be zero, and would falsely lower the overall average time from ECG “predictor” to adverse outcome measure in the entire cohort. Although we do not want to de-emphasize the importance of treating hyperkalemia before labs return, we wanted to understand the truth behind this data.

Dr. Durfey graciously responded to this question:

“Two separate events of VT were not required. If a patient had an initial ECG showing VT, then the initial ECG was recorded as VT and an adverse outcome was recorded as well.   

Only 2 episodes of VT occurred in the study group and both of these patients had VT on their initial ECG.  We agree that VT on an initial ECG cannot be considered a “predictor” for an adverse event of VT.  Accordingly, a relative risk for adverse event was not calculated.

We reported the median time from the ECG to the adverse event as 47 minutes. If the 2 VT patients are removed, the median time from the ECG to the adverse event was 58 minutes.”

• The most common ECG “predictor” of adverse events was bradycardia, and the most common adverse event was “symptomatic bradycardia”

We were similarly curious about the possible relationship of bradycardia and symptomatic bradycardia. The possible problems of this relationship in a retrospective chart review are complex. The hope is that these two events can be retrospectively identified in an “asymptomatic predictor leads to symptomatic predicted outcome” relationship, however this might not be the case clinically. If an ECG was ordered and shows bradycardia, it is possible that the patient may have had preexisting symptoms of symptomatic bradycardia prompting that ECG. Even if bradycardia was detected by routine vital signs, it then prompts immediate evaluation, repeat ECG, and questioning of symptoms, revealing symptomatic bradycardia and prompting immediate treatment.

The questions raised in our minds were:

Does this potential problem falsely inflate the perceived utility of ECG predictors, because some cases are not “predicting” the outcome but rather “discovering” it immediately?

Does this falsely decrease the recorded time from ECG to adverse outcome? There must be some or many of these 28 cases where an ECG showed bradycardia, then because of the ECG prompting evaluation the outcome of symptomatic bradycardia was immediately discovered (but was already present)?

Dr. Durfey’s response:

“This is a more difficult question and exposes the limitations of a retrospective chart review and of our definition of symptomatic bradycardia.

22 patients had adverse event of symptomatic bradycardia.  17 of the 22 patients with adverse event of symptomatic bradycardia were bradycardic on their initial ECG.

We made the decision that in order for the adverse outcome of symptomatic bradycardia to be clinically relevant, the bradycardia & symptoms should have been severe enough to warrant treatment (calcium, atropine, epinephrine, dopamine and/or pacing).  The time of the adverse outcome of symptomatic bradycardia was recorded as the time that the first treatment was administered.

In these 17 patients the median time from initial ECG to treatment of symptomatic bradycardia was 31 minutes.

You have pointed out the limitation of this definition.  While the median time to treatment from ECG was 31 minutes, we do not know how long the patient had severe symptoms prior to the treatment.  One would hope that the patient was treated promptly, but this is unable to be determined in a retrospective chart review.”

Learning points from Durfey et al:

• Durfey et al. confirms Montague et al.’s findings that T-wave changes alone don’t reliably predict adverse outcomes
• All adverse outcomes in this study were predicted by hyperkalemic ECG findings including QRS prolongation and abnormal bradycardic rhythms, but this is possibly limited by the concerns explained above
• No adverse event was recorded after calcium, and all patients with VT or bradyarrhythmias improved with calcium
• The increasing relative risks of peaked T-waves, PR prolongation, QRS prolongation, and worsening bradycardic rhythms illustrate the progression of hyperkalemic EKG findings prior to adverse outcomes

Evidence Meets Clinical Common Sense

• The ECG is not accurate for detecting minor hyperkalemia, but accuracy increases as hyperkalemia becomes more severe and more clinically relevant, to the point that most or all patients with adverse outcomes will have at least one ECG abnormality if the ECGs are ordered and an experienced interpreter carefully reviews the ECG for these findings. However, peaked T-waves alone have been consistently insufficient to predict adverse outcomes.
• Based on Durfey et al, it is possible that further research could provide effective risk-stratification tools for hyperkalemia based on ECG findings.
• Hyperkalemia can mimic almost anything else seen on ECG, especially ischemia. When hyperkalemia causes ST elevation, it most commonly manifests in the inferior or anterior leads and is present alongside other clear hyperkalemic abnormalities.
• In a perfect setting with constant monitoring and expert ECG analysis, it is likely that most patients with progressive, severe hyperkalemia would display the classic progression of ECG findings. But this is rarely the case in real practice due to various time and resource constraints, inexperience with subtle hyperkalemic findings, failure to order serial ECGs, laboratory error or delay with potassium levels, etc. The bottom line is that this progression happens quickly and is frequently unnoticed in a busy Emergency Department, such that in practice there will be patients who seem to arrest right after a seemingly normal ECG. Furthermore, there will likely always be the occasional rare case of hyperkalemic arrest soon after a truly normal ECG.
• Calcium is a (probably) life-saving but only temporizing intervention.

The most important hyperkalemic ECG abnormalities can be remembered by:

(Peaked T waves)

Broad (QRS widening)

Brady

Blocks (AV blocks)

Bizarre

Clinical Scenario Pearls:

• Regular really wide complex tachycardia – consider hyperK in addition to VT⁹
• Bradycardia or heart block, considering transvenous pacemaker – consider hyperkalemia, maybe even try empiric calcium first while you're setting up for the pacemaker
• ECG shows STEMI but also something weird or bizarre – consider hyperkalemia, consider diagnostic and therapeutic calcium
• Altered/found down/hypotensive/peri-arrest – consider hyperkalemia just like you would consider hypoglycemia, immediately diagnose/rule out/treat
• Large T-waves, considering hyperkalemia – make sure it’s not hyperacute T’s with acute coronary occlusion
• In hyperkalemia, the monitor may detect the large T-wave as another QRS complex, resulting in double counting the T-waves and falsely doubled heart rate (may prevent detection of bradycardia or suggest false tachycardia)

Treatment of Hyperkalemia

Therapy for hyperkalemia aims to prevent arrhythmias (calcium), to shift potassium into the cells (insulin, beta agonists such as albuterol or terbutaline,¹⁰ relatively basic solutions such as lactated ringer’s), and to excrete potassium from the body altogether (Na-K exchange resins, diuretics). Lactated ringer’s is the commonly available fluid of choice in the setting of hyperkalemia.¹¹ The details and nuances of each medication is beyond the scope of this post, but can be found in the supplementary resources below. Calcium is one of the most important life saving, but temporizing therapies, and deserves a brief review of its key points.

Calcium Gluconate:

Does not require hepatic metabolism¹²

No significant extravasation risk

Is not located in the code cart

Dose: 3 gm IV over 10 minutes (or slow IVP if needed)

Calcium Chloride:

Three times the amount of elemental calcium as the same mass of calcium gluconate

Terrible peripheral extravasation consequences

Located in the code cart, always immediately available

Dose: 1 gm slow IVP through a perfect, patent IV

Indications for Calcium (Gluconate or Chloride):

Hyperkalemia with normal ECG:

Controversial, with differing expert opinions and no high quality prospective data. Dr. Smith does not treat with calcium unless there are ECG findings. This is supported by Durfey et al., in which all patients with adverse outcomes had ECG findings. However, for this approach the clinician must be expert at recognizing subtle findings.

ECG changes presumed to be hyperkalemic, before laboratory confirmation:

Opinions range from “calcium only for widened QRS or worse” to “treat all ECGs with suspected hyperkalemia findings.” Dr. Smith recommends treatment with (at least) calcium for any ECG change with suspicion of hyperkalemia, even before laboratory confirmation.

Contraindications: probably none (the classic digoxin controversy is based on extremely poor evidence including several unsubstantial case reports from the 1920s); the serum calcium level is irrelevant

Onset of Action: rapid, somewhere between immediate to 5 minutes

Duration: supposedly 30-60 minutes, maybe closer to 20-30 minutes in real life

Repeat: Unknown, every ~3-5 minutes

Endpoint: Unknown, experts recommend titrating to normal/baseline QRS

Maximum Dose: Unknown, experts recommend no upper limit if the patient is in arrest or peri-code; some cases have documented needing more than 15 grams Ca Gluconate to maintain circulation until dialysis¹³

Monitoring: Obviously obtain repeat 12-lead ECGs as a guide, but usually the view on the cardiac monitor is an easily available method of tracking the morphologic changes. Beware the hyperkalemic T-waves being double-counted as QRS complexes, falsely doubling the heart rate on the monitor and potentially leading to R-on-T phenomenon in the case of attempted cardioversion.

More Resources:

• Dr. Smith’s ECG Blog Hyperkalemia Posts
• REBEL EM: Kayexalate
• REBEL EM: Hyperkalemia review
• First 10 in EM
• Core EM: Hyperkalemia
• EMCrit Hyperkalemia
• EMPharmMD: Hyperkalemia
• Essentials of Emergency Medicine: Amal Mattu on Regular Really Wide Complex Tachycardia
• Example of rapid reversal of hyperkalemic findings in real time

References:

1. Durfey N, Lehnhof B, Bergeson A, et al. Severe hyperkalemia: can the electrocardiogram risk stratify for short-term adverse events? West J Emerg Med. 2017; 18(5)963-971.
2. Smith S, Meyers H. You must recognize this pattern. Dr. Smith’s ECG Blog. Available at: http://hqmeded-ecg.blogspot.com/2013/11/you-must-recognize-this-pattern-even-if.html. Accessed November 18, 2017.
3. Hyperkalemia. Life in the Fast Lane. Available at: https://lifeinthefastlane.com/ecg-library/basics/hyperkalaemia/. Accessed November 18, 2017.
4. Khattak H, Khalid S, Manzoor K, et al. Recurrent life-threatening hyperkalemia without typical electrocardiographic changes. J Electrocardiol. 2014;47(1):95-7.
5. Szerlip H, Weiss J, Singer I. Profound hyperkalemia without electrocardiographic manifestations. Am J Kidney Dis. 1986;7(6):461-5.
6. Smith S. Are these peaked T waves the patient’s baseline T waves? Dr. Smith’s ECG Blog. Available at: http://hqmeded-ecg.blogspot.com/2014/07/are-these-peaked-t-waves-patients.html. Accessed November 18, 2017.
7. Dodge HT, Grant RP, Seavey PW, et al. The effect of induced hyperkalemia on the normal and abnormal electrocardiogram. Am Heart J. 1953 May;45(5):725-40.
8. Montague BT, Ouellette JR, Buller GK, et al. Retrospective review of the frequency of ECG changes in hyperkalemia. Clin J Am Soc Nephrol. 2008 Mar;3(2):324-30.
9. Mattu A. How do you avoid a clean kill with wide complex tachycardias? Available at: https://www.youtube.com/watch?v=UXh8PS9dtmo&feature=youtu.be. Accessed November 18, 2017.
10. Smith S. Terbutaline and albuterol for lowering of plasma potassium. Dr. Smith’s ECG Blog. Available at: http://hqmeded-ecg.blogspot.com/2013/12/terbutaline-and-albuterol-for-lowering.html. Accessed November 26, 2017.
11. Farkas J. Myth-busting: Lactated Ringers is safe in hyperkalemia, and is superior to normal saline. PulmCrit. Available at: https://emcrit.org/pulmcrit/myth-busting-lactated-ringers-is-safe-in-hyperkalemia-and-is-superior-to-ns/. Accessed November 18, 2017.
12. Martin TJ, Kang Y, Robertson KM, et al. Ionization and hemodynamic effects of calcium chloride and calcium gluconate in the absence of hepatic function. Anesthesiology. 1990 Jul;73(1);62-5.
13. Smith S. Weakness, prolonged PR interval, wide complex, ventricular tachycardia. Dr. Smith’s ECG Blog. Available at: http://hqmeded-ecg.blogspot.com/2011/02/weakness-prolonged-pr-interval-wide.html. Accesed November 18, 2017.