Without evidence of benefit, an intervention should not be presumed to be beneficial or safe.

- Rogue Medic

How to Study Epinephrine in Cardiac Arrest

No. This is not a post about a study that did everything the right way. I don’t even think there is any such study in progress. This is my description of what is necessary for a future study to be valid.

I have written a lot about epinephrine in cardiac arrest. So far there is no research to support the use of epinephrine, unless we believe that the surrogate endpoint of ROSC (Return Of Spontaneous Circulation) is important. Surrogate endpoints have a long history of leading people to make bad mistakes. I have already written about surrogate endpoints several times.[1]

Here is one explanation of the way surrogate endpoints are used:

From this unhappy rhythm – the beginning point.

We add a drug that we know does a great job of improving the surrogate endpoint and then we have this happy rhythm. The improved rhythm is the surrogate endpoint. It is a surrogate (substitute) for showing improved survival, because this does not require as many study participants.

The surrogate endpoint also led us to the rhythm below. The surrogate endpoint misled us.

Survival with good neurological function is the only meaningful end point. Once we had enough participants in a well controlled trial, we were able to find out that making the rhythm look better did not make the patient better.

The study drugs did a great job of making the rhythm look better. It looked like a great drug – A lifesaver! – until there was a study large enough to show a difference in survival.

When a study large enough to show a significant difference in survival was done, we could see that more than 3 times as many patients receiving the study drugs were dropping dead.

In looking at epinephrine for cardiac arrest, we need to forget about surrogate endpoints. We’ve studied that to death. We need to look at survival.

The study has to be large enough to show a difference in survival.

The study needs to control for confounding variables as much as possible.

The study needs to be randomized and double-blinded.

The study needs to track from before the first resuscitation drug is given to hospital admission, at the very minimum.

The study really should look at the difference in outcome between the study drug and the placebo after the first drug is given.

While the rate of patients resuscitated before medication should be equal in the placebo and study arms, we should probably recognize that these are not patients who will be affected by epinephrine. These may also be the patients least likely to present as needing aggressive post-resuscitation care. It is an assumption, but it does seem reasonable that, more responsive to initial interventions means less low-circulation/no-circulation time in cardiac arrest and, we might expect, less intensive post-resuscitation care.

What we want to know is the difference between the patients who do receive epinephrine and the patients who do not receive epinephrine, but would have received epinephrine if not in a study of epinephrine in cardiac arrest. These are the patients who do not receive epinephrine because of the study design, not because of responsiveness to earlier treatments. Nobody interacting with the medics should know which patient received which treatment.

We need some agreement on post-resuscitation care. I used to regularly present ACLS classes with this scenario of a patient just resuscitated. When someone would ask for vital signs, I would state that the blood pressure is 60 by palpation.

This presents them with an important treatment decision. This may be the most important treatment decision of this patient’s life.

How do we approach this patient?

1. A systolic pressure of less than 90. We must give fluids and dopamine.

2. We got a resuscitation. We’re going to Disney World. Maybe we’ll drop off the patient on the way.

3. The systolic pressure rose 60 points in the last minute. Maybe I should wait to see what happens in the next minute, and the minutes after the next minute, before making decisions about aggressive interventions that may discourage this patient from maintaining any pulse.

I think you will be able to guess what my approach is. I grew up with Disney movies on Sunday night. High Fives and high tail it to the park. 😉

We would need to have some sort of protocol to standardize post-resuscitation care, because I know a lot of people – doctors, nurses, and paramedics – who would already have dopamine running before the next blood pressure.

I think the research shows that therapeutic hypothermia works, so this would probably be a requirement for participation.

CPR should be standardized to whatever the next guidelines come up with before the beginning of the study. I hope that Dr. Ewy’s minimally interrupted, continuous compression CPR is what the AHA/ILCOR/ARC decide to adopt. (They did not. Ventilation is still viewed as important, even though there is not research to demonstrate that ventilations improve outcomes. We continue to mistakenly look at withholding an ineffective treatment as an intervention that needs to be supported by excellent evidence, rather than the actual treatment – ventilations, epinephrine, amiodarone, et cetera – as something that needs to be supported by excellent evidence.)

We need to look at subpopulations to see if epinephrine is effective in some groups, while not effective in other groups. This could explain why the overall effect is not impressive.

There needs to be a randomization of placebo and epinephrine syringes with many different identification markings, such as individual numbers with no pattern. The reason for the variety in the markings is to make it almost impossible for the medics to figure out which are active drugs and which are placebos. Those running the study should also be prevented from knowing which patients are receiving the active drug – epinephrine. Or some other method of randomization and blinding that can be shown to be effective at blinding EMS.

If this is done the right way, there should not be much doubt about what to do with epinephrine after the conclusion of the study. Either epinephrine is significantly better than placebo and belongs in resuscitation guidelines, or we only use epinephrine when specifically indicated – potentially reversible causes of cardiac arrest and for any subpopulation identified as having better survival after epinephrine.

I am sure there are things I have not included, but this is what is needed. Either we find out what is best for patients, or we continue to give a treatment that is based on tradition and the short-term ability to bring back a pulse.

Maybe I am wrong. Maybe this is already being done.


[1] C A S T and Narrative Fallacy
Rogue Medic

Some other surrogate endpoint posts:

New Series of Rants Second follow up

C A S T and Narrative Fallacy comment from Shaggy

EMS EdUCast – Journal Club 2: Episode 43

The association between emergency medical services staffing patterns and out-of-hospital cardiac arrest survival

Updated 12-06-10 at 13:49 and some wording changed for clarity. In other words, parts that didn’t make as much sense as I thought they did, when I originally wrote them.