Is epinephrine harmful in cardiogenic shock?

We’re only going to focus on one facet of today’s article in our discussion, but Dr. Sean van Diepen wrote an excellent editorial that covers several other important points. [1] There is more to the paper than I’m going to dive into here, so check out his editorial if you have access around the paywall.


I’m hearing a lot of buzz concerning an article that was just published in the Journal of the American College of Cardiology. [2]


If you only read the abstract, you’d think epinephrine was hurting patients with cardiogenic shock.


I don’t want people to draw the wrong conclusions about epinephrine. I’m seeing posts on social media stating that epinephrine is harmful in cardiogenic shock, or that norepinephrine should be the definitive first-line pressor in all-comers.

You have to read the paper

As we see here, always read past the abstract in any article of even passing interest. It turns out the results of this trial aren’t as straightforward as its abstract suggests.


What did they do?

Let’s run through the Methods.

  • This was a multi-center randomized-controlled trial performed at 9 French ICUs
  • It recruited patients between 9/2011 and 8/2016
  • Patients were eligible if they met all the following criteria:
    • age over 18 years
    • pulmonary artery catheter inserted
    • systolic blood pressure (SBP) < 90 mmHg or mean arterial pressure (MAP) < 65 mmHg without a vasopressor, or need for a vasopressor to correct hypotension
    • cardiac index  < 2.2 without vasopressors
    • a pulmonary artery pressure > 15 mmHg
    • ejection fraction < 40% without inotropes
    • “at least one evidence of tissue hypoperfusion (e.g. skin mottling, oliguria, elevated lactate level, altered level of consciousness)” [This one is important!]
  • Patients were randomized and staff blinded in a reasonable fashion


The treatment protocol is a little confusing, but here’s my understanding of how it worked.

  1. A patient experiencing an acute MI presented in cardiogenic shock. They were probably started on a pressor before arriving in the ICU (e.g. in the field, emergency department, or cath lab).
  2. Once in the ICU, staff there identified the patient as a candidate for inclusion in the study and notified the pharmacy.
  3. Pharmacy then opened an envelope with a treatment assignment, and prepared either epinephrine or norepinephrine as directed. They labeled the unmarked syringe with the patient’s identification number and send it to the ICU.
  4. Physicians and nurses, blinded to the study drug, titrated it by 0.02 mcg/kg/min increments (“or higher in emergency cases”) to a target MAP of 65–70 mmHg.
  5. As the study drug was titrated up, the initial open-label pressor was titrated down and discontinued.
  6. If the patient didn’t respond to the study drug, the treating physician could stop it and switch to an open-label vasopressor.


They collected a bunch of variables, including vital signs, hemodynamic measurements and calculations, clinical scores, and a plethora of blood tests.

The baseline characteristics of the two study groups were similar across the board, except the norepi group ended up with a lot more males than females. The patients were randomized, so it’s probably just a spurious difference, but it’s a little reminder that randomization doesn’t produce perfectly equal groups.

Because of the strict inclusion criteria (i.e. they must have a pulmonary artery catheter), this wasn’t ever going to be a large trial. With small numbers, it would be under-powered to detect major clinical outcomes like a mortality benefit, so they chose the change in cardiac index as their primary outcome. They also examined a bunch of secondary endpoints; I’ll let you read those for yourself in the paper.



Over 5 years, at 9 centers, they ended up enrolling 27 patients in the epinephrine group and 30 patients in the norepinephrine group—a tiny sample. As you might expect with such small numbers, and given the collective experience of people using both medications (i.e. clinical equipoise), there weren’t very many significant differences in outcomes.

Their effects on cardiac index, the primary endpoint, were identical.

Among the secondary endpoints, the results were again alike. Patients required similar doses of both medications to achieve their goal blood pressures. They had similar changes in SBP, MAP, and several other hemodynamic measures. The only exception—epinephrine caused more tachycardia (as expected), along with a higher cardiac double product.

Looking at blood tests and biomarkers, both medications produced similar results across a wide range of endpoints. The only outlier was that epinephrine was associated with higher lactate levels and more metabolic acidosis in the first 24 hours—we will expand on this in a moment.

In terms of end-organ dysfunction, both drugs performed similarly.

There was no significant difference in mortality.

All this makes sense when you’re dealing with such a small sample; you won’t see many significant differences unless they’re glaringly obvious. What’s surprising, however, is that one of the safety endpoints showed marked divergence between the two medications in spite of the limited enrollment.

In fact, it led to the study being halted due to concerns for patient safety.


A shocking result

According to the authors and people monitoring the study, epinephrine caused a significant amount of refractory cardiogenic shock and acidosis.

“Refractory cardiogenic shock” wasn’t even defined as a safety endpoint at the start of the study. The imbalance between the two arms became so obvious to the folks monitoring the study, however, that they started looking while the trial was in progress. It led to them shutting the whole thing down.

What’s going on here? Well, let’s start by again looking at the authors’ definition of “tissue hypoperfusion” from the inclusion criteria section of the Methods:

at least one evidence of tissue hypoperfusion (e.g. skin mottling, oliguria, elevated lactate level, altered level of consciousness)

So, at the outset, the authors defined an elevated blood lactate level as evidence of shock. Keep that in mind.

Now, let’s look at their definition of “refractory cardiogenic shock”:

Refractory CS [cardiogenic shock] was defined as CS with major cardiac dysfunction assessed according to echocardiography, elevated lactate level, and acute deterioration of organ function (e.g., oliguria, liver failure) despite the use of >1 μg/kg/min of epinephrine/norepinephrine or dobutamine >10 μg/kg/min and/or intra-aortic balloon support and sustained hypotension (SAP <90 mm Hg or MAP <65 mm Hg) despite adequate intravascular volume.

That’s a lot of words. Let’s pare that down, using what we know from the secondary endpoints of the study.

  • These patients all had cardiogenic shock; it’s why they were enrolled.
  • They also had cardiac dysfunction on echo, as defined in the inclusion criteria, and a similar increase in left ventricular ejection fraction.
  • There seemed to be no significant difference in organ dysfunction between epi and norepi, so that’s unlikely to be the driver of these “refractory CS” cases.
  • Patients received comparable doses of epi and norepi.
  • Both groups required the same amount of inotropic support with dobutamine.
  • Both groups received intra-aortic balloon pumps at a similar rate.
  • Patients also had similar improvements in blood pressure in response to the pressor agents.

So, after eliminating all of those factors, what’s left?

Refractory CS [cardiogenic shock] was defined as CS with major cardiac dysfunction assessed according to echocardiography, elevated lactate level, and acute deterioration of organ function (e.g., oliguria, liver failure) despite the use of >1 μg/kg/min of epinephrine/norepinephrine or dobutamine >10 μg/kg/min and/or intra-aortic balloon support and sustained hypotension (SAP <90 mm Hg or MAP <65 mm Hg) despite adequate intravascular volume.

It seems likely that higher lactate levels drove the higher rates of “refractory cardiogenic shock” in the epinephrine arm. That’s borne out in the results, which show significantly higher arterial lactate levels—and later, lactate clearance—in the epinephrine group.

[The statistical methods of what I just did aren’t really legit, but I think it’s a workable shortcut given the limited data available, small patients numbers, and face-validity of my argument. You’re allowed to disagree.]

The authors also claim more acidosis in the epinephrine group (“p = 0.0004”), but they never show the pH values in the results, so we have to take their word on it. Maybe it’s in the online supplement (though it’s not referenced), but I haven’t been able to get access yet.

So, the epinephrine group had higher arterial lactate levels and more acidosis—does that really translate to “refractory cardiogenic shock”?


Probably not

If you’ve ever worked in an emergency department, you’ve seen a critically-elevated lactate flagged on a patient with difficulty breathing but no over signs of sepsis or shock. In hindsight, it comes to light that phlebotomy drew their labs during, or shortly after, a patient’s nebulizer treatment. Repeat levels return at a reasonable value and the initial lactate level gets dismissed as spurious.

Why do breathing treatments increase serum lactate levels?

Specifically, why does albuterol increase serum lactate levels?

It has to do with lactate production. Most of us were taught that lactate is a by-product of anaerobic respiration, so when a patient is in shock and tissues become hypoxic, they release lactate into the blood.

There’s a grain of truth there, but it’s probably not the source of the lactate we find in most shock patients.

There’s another way to make lactate: beta-adrenergic stimulation.

I won’t get into the physiology, and much of it is still being researched and debated, but it suffices to say that much of the lactate we see in shock states seems to result from catecholamines. They stimulate the release of lactate aerobically, through glycolysis. In fact, much of its production is driven by B2 adrenergic stimulation.

Albuterol is a B2-agonist. Epinephrine is a B2-agonist. Norepinephrine is not a significant B2-agonist.

For more on the role of catecholamines in lactate production, this is a nice primer. [3]

This theory is borne-out in research that shows albuterol and epinephrine cause increased lactate production, but norepinephrine does not. [4,5] It’s also seen in clinical practice, when patients on albuterol nebulizers have spuriously high lactates, or when ICU patients on epinephrine infusions do not clear their lactate despite improving clinically.

Lactate is not harmful—in fact, it seems to be an adaptive response that provides significant “fuel” to the heart and brain. [6] So why do the authors of this paper assume that the high lactate levels seen in the patients on epinephrine infusions are a sign of “refractory cardiogenic shock”?

I have no idea.

It seems to be a misapplication of the decades-old belief that lactate is a marker of tissue hypoxia. Complicating matters, their Discussion cites several papers which support the impression of lactate I describe above; those citations even share some authors with this trial! But this article then goes on to discuss lactate as a monitor of tissue perfusion, so I’m at a loss for where the authors actually stand.

What I do know, with reasonable certainty, is that lactate should not be the sole driver that labels patients with “refractory cardiogenic shock.” Unless the authors can provide data showing there were other factors at play, I have to throw out their argument that epinephrine caused any measurable harm in these patients with cardiogenic shock.


In summary

Half of the patients in this study were placed on an infusion that is known to increase serum lactate levels and lead to poor lactate clearance; however, that doesn’t mean lactate, or epinephrine, are evil.

The mechanism by which epinephrine increases serum lactate is not known to cause clinically-significant harm. In fact, lactate production may be an adaptive mechanism that aids the body in times of stress.

By every other clinically-relevant marker of efficacy, epinephrine performed as well as norepinephrine in patients with cardiogenic shock (with the exception of slightly more tachycardia). Epinephrine only caused harm in this study according to the authors’ arbitrary definition of “refractory cardiogenic shock”—a diagnosis that seemed to be driven by elevated lactates and no other measurable findings.


Should epinephrine by the default first-line pressor for cardiogenic shock? Probably not.

Do I know which pressor should be? No way.

Do I even have a preference? Absolutely not.

But I do know how to spot misleading studies, and I would hate to see epinephrine infusion maligned as harmful because of one fatally flawed trial.



  1. van Diepen, S. Norepinephrine as a First-Line Inopressor in Cardiogenic Shock: Oversimplification or Best Practice? JACC. 2018 Jul;72(2). doi:
  2. Levy B, et al.. Epinephrine Versus Norepinephrine for Cardiogenic Shock After Acute Myocardial Infarction. JACC. 2018 Jul;72(2). doi: .
  3. Garcia-Alvarez M, Marik P, Bellomo R. Sepsis-associated hyperlactatemia. Crit Care. 2014 Sep 9;18(5):503. doi: 10.1186/s13054-014-0503-3.
  4. Lewis LM, Ferguson I, House SL, Aubuchon K, Schneider J, Johnson K, Matsuda K. Albuterol administration is commonly associated with increases in serum lactate in patients with asthma treated for acute exacerbation of asthma. Chest. 2014 Jan;145(1):53-59. doi: 10.1378/chest.13-0930.
  5. Levy B, Perez P, Perny J, Thivilier C, Gerard A. Comparison of norepinephrine-dobutamine to epinephrine for hemodynamics, lactate metabolism, and organ function variables in cardiogenic shock. A prospective, randomized pilot study. Crit Care Med. 2011 Mar;39(3):450-5. doi: 10.1097/CCM.0b013e3181ffe0eb.
  6. Garcia-Alvarez M, Marik P, Bellomo R. Stress hyperlactataemia: present understanding and controversy. Lancet Diabetes Endocrinol. 2014 Apr;2(4):339-47. doi: 10.1016/S2213-8587(13)70154-2. Epub 2013 Nov 29.



  • Vince, it does appear that there were more deaths in the Epinephrine group. Is this statistically significant?

    • There’s a definite *trend* toward more deaths in the epi cohort, but nothing significant.

      I like this study for what it is—a super-intense look at the differences between epi and norepi in terms of hemodynamics and biomarker profiles—but that same strength also severely limited enrollment. With invasive PA monitoring as an inclusion requirement, it would have taken decades to get a large enough sample to measure any difference in clinical outcomes.

  • Michael Riediger says:

    Quote: “… espite the use of >1 mg/kg/min of epinephrine/norepinephrine …”

    You guys know the difference between mg and µg, do you?

    • Sorry about that; I copied the quote from the article and my browser automatically changed the “μ” to an “m”. Thanks for noticing, and thanks for pointing it out in the most passive-aggressive manner possible.

Leave a Reply

Your email address will not be published. Required fields are marked *