Journal Club #1: Goto et al. Effects of prehospital epinephrine... Critical Care 2013

[Brandon O]

Hi folks!

There's been talk about starting a forum journal club. If you're not familiar, journal clubs are a longstanding practice in clinical and academic settings; odds are your local hospital or med school has a few. Basically, it's nothing more than a group that meets periodically to discuss recent research, usually focusing on a particular study that's relevant to their practice. You sit down and really dig into it, hashing out the good, the bad, and the ugly. It's a great way to stay current on the literature as well as practicing the skills needed to critically appraise all the nonsense that gets published.

So let's give it a try! We'll figure out the nuts and bolts as we go, but right now, I'm thinking that we can basically just post an article and go nuts on it. I have a list of questions to give some structure to the discussion; if you respond, it'd be great if you tried to answer at least one of the questions -- or try anyway! (Don't just say "I like it!" or "I hate it!" Let's get into why...)

So here's one to get us started...

Goto Y, Maeda T, Goto YN. Effects of prehospital epinephrine during out-of-hospital cardiac arrest with initial non-shockable rhythm: An observational cohort study. Critical Care. 2013 Sep 3;17(5):R188.

Questions:

1. What's the question we're looking at? Why do we care in EMS?
   
2. What type of study was this? If observational, was it retrospective or prospective?
   
3. What is the PRIMARY outcome? (Was there more than one?) What were the SECONDARY outcomes? (Were there more than three?) Do we care about these outcomes?
   
4. Any reason to suspect bias? Conflicts of interest among the authors? Who sponsored the study?
   
5. Who was the study population (what were the inclusion/exclusion criteria)? 1. Is it suitable to answer the question? 2. Is this population relevant for us?
   
6. What was the comparison/control? Was it appropriate? If randomized, what was the allocation method, was it concealed, and were the groups initially similar? If observational, are there potential variables that were not controlled, and what effect might they have?
   
7. Is there any reason why control patients might have been treated differently after enrollment, and was this controlled? If not, what effect might it have?
   
8. Were there any losses/failures after enrollment, and if so, were they analyzed using intention-to-treat?
   
9. Were patients blinded? Providers? Evaluators of the gold standard? Statistical analysts?
   
10. How reliable was the gold standard? How long was study follow-up, and was this adequate? Was the study stopped early?
    
11. What are the results? Are they statistically significant? Clinically meaningful? Are they the same primary/secondary outcomes initially described? Are they plausible?
    
12. What are the results in terms of NNT (for therapy) or LR (for diagnostics/risk)? What were the harms? Were benefits greater than harms?
    
13. What were the authors' conclusions, and are they supported by the results?

 

[Carlos Danger]

Very glad to see this.

I'll post later when I have more time

 

[Ewok Jerky]

good choice for a journal club starter

Rather than try to answer your questions in bullet point format, I will put my initial reaction to the article after a single read (not including the lengthy discussion section).

The question is, should we be giving epi prehospital to code Pts.

Study design is retrospective cohort. very large sample size, giving the study plenty of statistical power. includes what appears to be an equal distribution of the japanese population. no apparent bias (selection, migration, information, recall, etc.)

Ideally, we would answer this question with a randomized control study, but unfortunately that would be unethical (to withhold a treatment that is standard of care). so we are stuck looking back at code Pts that for some reason did not get epi and comparing their survival rates to peeps that DID get epi. now, im guessing that their are lots of confounding variables going on, which the authors tried to account for: EMS response time, bystander CPR, prehospital defib etc. interestingly, all these other interventions were found to be associated with increased survival in both groups (epi and non-epi).

results: In shockable rhythms, the non-epi group has better survival. my guess is that japanese EMS is shocking before administering epi, and Pts who have ROSC after a single shock are most likely to survive, and will NOT have been given epi, where peeps who require more CPR, more shocks, and eventually epi, are less likely to survive a priori. The other result is that Pts with NON-shockable rhythms who are given epi are associated with higher 1-month survival rates. if you look through the tables you will notice that PEA is about 4 times more survivable than asystole.

These results make sense to me. I already talked about the shockable rhythm Pts. It also makes sense that giving a vasoconstrictive agent to someone in PEA, thus increasing venous return and cardiac output, will increase odds of ROSC and 1-month survivabiltiy. Pts who present in asystole are dead, they might not be resuscitate-able no matter what intervention, so its not surprising that this group is used as a comparison to gauge the effect of epi.

now the real question, are my cardiac arrest patients similar to Japanese cardiac arrest Pts?

 

[Brandon O]

I guess I'll pick up my part of the load...

1. What is the PRIMARY outcome? (Was there more than one?) What were the SECONDARY outcomes? (Were there more than three?) Do we care about these outcomes?

This was actually unclear. They initially say that:

"Therefore, the first objective of the present study was to examine whether initial cardiac rhythm would be considered a key factor for predicting survival and favorable neurological outcomes at 1 month. The second objective was to determine whether prehospital epinephrine administration would improve 1-month survival in patients who had experienced OHCA with initial non-shockable rhythms."

But then they say...

"The primary study end point was survival at 1 month. The secondary end points were ROSC before arrival at the hospital and survival at 1 month with favorable neurological outcome (defined as a CPC of 1 or 2)"

So I'm going to read between the lines and say that the primary outcome was whether epi's effect on 1-month survival (with good neuro outcome) differed according to initial rhythm. Everything else was a secondary outcome (and they collected a zillion stats, so there are lots of those).

• What are the results in terms of NNT (for therapy) or LR (for diagnostics/risk)? What were the harms? Were benefits greater than harms?

NNT if you give epi for non-shockable rhythms:

You need to treat 4 to achieve one extra ROSC, 59 for one extra 1-month survival, and infinity (no effect) for 1-month survival with good neurological status.

NNH if you give epi for shockable rhythms:

You need to treat 21 to eliminate one chance at ROSC, 9 to eliminate one chance at 1-month survival, and 9 (yes, same figure) to eliminate one chance at 1-month survival with good neurological status.

 

[Ewok Jerky]

I guess I'll pick up my part of the load...

(and they collected a zillion stats, so there are lots of those).

based on the changing stated outcomes, and the sheer number of variables, do you think we are experiencing Type 1 error?

 

[Brandon O]

based on the changing stated outcomes, and the sheer number of variables, do you think we are experiencing Type 1 error?

Well, maybe. I don't think they're trying to pull a fast one, although I do think that there's a bit of a language barrier, and this (pre-print) paper could use some editing.

The most compelling argument against the results is that there has been quite a lot of research on epi, and this is just about the only study that's ever managed to show a positive effect on long-term survival. So why here but not before?

Maybe because it's big (there have been bigger studies but not many). Maybe because there's bias in this data-dredging, retrospective registry study, although the biggest confounder seems like the one you mentioned (patients receiving epi were probably sicker) and that should bias AWAY from a beneficial effect, not toward it.

But maybe also because the group that benefitted was fairly specific -- non-shockable rhythms with a reasonable short time-to-administration. The thing is, other than that interesting shockable-vs-not-shockable dichotomy, everything else here is very plausible and consistent with prior research -- including the fact that even in the optimal subgroup, epi still didn't improve long-term NEUROLOGICALLY INTACT survival.

The other tantalizing tidbit is that the effects are pretty large, both for epi in not-shockable and against epi in shockable. They're big enough that they look real -- and the dichotomized result would help explain why, when you pool everyone together, the positive effect disappears (it balances out). Thus the many studies suggesting no long-term effect (or even some harm) from epi.

So it's tempting to say that epi is bad for shockable rhythms, but for non-shockables can improve pretty legit long-term survival, although NOT survival with a good neurological outcome. And that conclusion is consistent enough with prior research that it's plausible.

Do we care about creating more surviving vegetables? I'll leave that for the roundtable.

 

[EpiEMS]

Interesting post! Glad to see a large n study on this.

The -- unsurprising -- result that epinephrine administration creates more ROSC in non-shockable rhythms but decreased neurological outcomes vs. no epinephrine administration seems to me to require a controlled study.

This is the key line in the paper, from my perspective: "Moreover, prehospital epinephrine administration was independently associated with deteriorated neurological outcomes at 1 month (adjusted OR, 0.71, 0.34, 0.21; 95% CI, 0.54–0.92, 0.28–0.42, 0.14–0.31; for epinephrine administration times < 9 min, 10–19 min, and > 20 min, respectively)"

Brandon's question is spot on. I don't want to create more "vegetables." That's not the goal of EMS.

Also, digging through the tables, there's a mention of advanced airway management. Now, they don't seem to distinguish between LMAs, ETTs, etc., but it's pretty intuitive that it doesn't help versus non-advanced airways, and may be harmful to ROSC (at p < 0.05). This data is at the bottom of Table 5 for non-shockable, and Table 4 for shockable, where the results are clear: advanced airways reduce ROSC substantially.

 

[Brandon O]

Also, digging through the tables, there's a mention of advanced airway management. Now, they don't seem to distinguish between LMAs, ETTs, etc., but it's pretty intuitive that it doesn't help versus non-advanced airways, and may be harmful to ROSC (at p < 0.05). This data is at the bottom of Table 5 for non-shockable, and Table 4 for shockable, where the results are clear: advanced airways reduce ROSC substantially.

I noticed that, but I tried not to get too excited, since the previously-raised issue applies here in spades -- those who were pulseless long enough to receive an advanced airway are going to be sicker patients than those who had ROSC prior to that.

Same problem applies to the observations that longer delays to epi = worse outcomes. Longer time to ANYTHING is going to mean sicker patients.

 

[Carlos Danger]

Ideally, we would answer this question with a randomized control study, but unfortunately that would be unethical (to withhold a treatment that is standard of care).

I know this is how research is approached, but I think we need to ask: Is it really unethical if the standard of care has never been conclusively shown to improve outcomes?

Personally, I have a much easier time questioning the ethics of exposing patients to the risks of a potent vasoactive pharmaceutical agent when we don't really have any reason to believe that it helps them in any meaningful way.

 

[EpiEMS]

I noticed that, but I tried not to get too excited, since the previously-raised issue applies here in spades -- those who were pulseless long enough to receive an advanced airway are going to be sicker patients than those who had ROSC prior to that.

Same problem applies to the observations that longer delays to epi = worse outcomes. Longer time to ANYTHING is going to mean sicker patients.

I would assume that you're right with that, I just don't have sufficiently granular data to confirm it. Along those lines, I don't think the study makes much mention of it, which suggests to me that they really didn't have much detail on a case-by-case level, at least, I don't think they had individual run reports, for example.

I know this is how research is approached, but I think we need to ask: Is it really unethical if the standard of care has never been conclusively shown to improve outcomes?

I would say no, too. Is there actually anybody with the cojones to run such a study under their own license, though?

 

[Brandon O]

I would assume that you're right with that, I just don't have sufficiently granular data to confirm it. Along those lines, I don't think the study makes much mention of it, which suggests to me that they really didn't have much detail on a case-by-case level, at least, I don't think they had individual run reports, for example.

I believe it was purely a data-dump from an Utstein-style registry (that is, standardized datafields much like your favorite ePCR uses). The Utstein template has become pretty widespread and universal now.

 

[mycrofft]

Paging Portland, OR......

 

[Christopher]

I noticed that, but I tried not to get too excited, since the previously-raised issue applies here in spades -- those who were pulseless long enough to receive an advanced airway are going to be sicker patients than those who had ROSC prior to that.

Same problem applies to the observations that longer delays to epi = worse outcomes. Longer time to ANYTHING is going to mean sicker patients.

It is worth mentioning how Japanese EMS handles cardiac arrest:

1. Run and gun / load and go
2. Largely BLS
3. Recent introduction of IV's + epi
4. Previously BLS airway, some LMA/CombiTube, now some ETI but that is rare.

 

[chaz90]

It is worth mentioning how Japanese EMS handles cardiac arrest:

1. Run and gun / load and go
2. Largely BLS
3. Recent introduction of IV's + epi
4. Previously BLS airway, some LMA/CombiTube, now some ETI but that is rare.

Based on this, I'd be very curious if they use mechanical CPR assist devices during transport. If anyone has a robot to do something for them, I'd think it would be the Japanese.

 

[Christopher]

Based on this, I'd be very curious if they use mechanical CPR assist devices during transport. If anyone has a robot to do something for them, I'd think it would be the Japanese.

Nope, they do not (or at least I've never seen any papers about wide spread use).

They do however use field ECMO in some areas. They call it ECLS.

 

[Christopher]

Based on this, I'd be very curious if they use mechanical CPR assist devices during transport. If anyone has a robot to do something for them, I'd think it would be the Japanese.

I'm a filthy liar, they apparently use it in some HEMS groups.

 

[EpiEMS]

It is worth mentioning how Japanese EMS handles cardiac arrest:

1. Run and gun / load and go
2. Largely BLS
3. Recent introduction of IV's + epi
4. Previously BLS airway, some LMA/CombiTube, now some ETI but that is rare.

Assuming short-ish transport times, that sounds like an OK (if not good) system for cardiac arrest, as far as the efficacy of interventions goes.

Nope, they do not (or at least I've never seen any papers about wide spread use).

They do however use field ECMO in some areas. They call it ECLS.

Whoa. I just did some (light) googling, and apparently they're doing it in EDs in the US...but in the field? Just woah

 

[Christopher]

Assuming short-ish transport times, that sounds like an OK (if not good) system for cardiac arrest, as far as the efficacy of interventions goes.

#1 on the list negates any benefit from #2-4.

Any system transporting cardiac arrest patients as a routine has protocolized low survival rates. It makes most research from these areas useless for services which perform CPR properly.

How do you judge a study on epi, good or bad, from a system which has a protocol to ensure their survival rates are 1/2 to 1/3 what they could be?

 

[EpiEMS]

#1 on the list negates any benefit from #2-4.

Any system transporting cardiac arrest patients as a routine has protocolized low survival rates. It makes most research from these areas useless for services which perform CPR properly.

How do you judge a study on epi, good or bad, from a system which has a protocol to ensure their survival rates are 1/2 to 1/3 what they could be?

I just reread my post...and you're totally right.

Transporting an arrest is quite silly. Work on scene, more or less, for all non-traumatic arrests, right?

 

[Brandon O]

Transporting an arrest is quite silly. Work on scene, more or less, for all non-traumatic arrests, right?

Barring reversible (and otherwise refractory) causes, that would be my vote.