The most recent episode of Emergency Medicine Cases Journal Jam takes a look at the evidence for epinephrine in cardiac arrest. (I suppose as I prepare for my move to New Zealand, I should probably get used to using the term adrenaline, but for now I will stick with the Canadian “epinephrine”.) These are the written notes to accompany that podcast.
Usually the summary goes at the end, but I have been told that some people don’t find the study details as exciting as I do, so I should just cut to the chase.
One might ask: why would I want to know about a lot of old observational research when I just spent so much time digesting the PARAMEDIC2 trial? Aside from the obvious answer that reading the medical literature is fun, understanding the literature before PARAMEDIC2 can help inform our interpretations of PARAMEDIC2. (If the past evidence strongly supported epinephrine, we would ask if PARAMEDIC2 warrants overturning the current standard of care. However, if the prior evidence was mostly negative, we might approach PARAMEDIC2 by asking: is this data strong enough to support using a thus far unproven medication?)
The data before PARAMEDIC2, although fundamentally flawed because it is almost all observational, is pretty easy to summarize, and is very consistent with the results we saw in PARAMEDIC2: using epinephrine results in higher rates of ROSC and more patients being admitted to hospital. It is unclear if more patients leave the hospital alive, but if they do, there is no increase in the truly important outcome of neurologically intact survival.
After PARAMEDIC2 there are still a lot of questions. Is there a better dose of epinephrine? Do we need to be giving it earlier? Should we be guiding our dose based on physiologic parameters such as the diastolic blood pressure? Are there other vasopressors we should be trying? These are all very reasonable questions. I expect epinephrine to have a continued role in cardiac arrest research going forward, but I think that while we wait for those results, the bulk of this data suggests that epinephrine is, on balance, causing more harm than good, and should be removed from standard cardiac arrest algorithms.
Holmberg M, Holmberg S, Herlitz J. Low chance of survival among patients requiring adrenaline (epinephrine) or intubation after out-of-hospital cardiac arrest in Sweden. Resuscitation. 2002; 54(1):37-45. PMID: 12104107
Quick summary: This is prospectively collected observational data in a large Swedish cardiac arrest registry. Between 1990 and 1995 they included 10,966 patients with out of hospital cardiac arrest, of whom 42% had been given epinephrine. Epinephrine was associated with increased 1 month mortality (6.3% vs 3.4%, p<0.001). This was true across all the subgroups.
Caveats: Groups were significantly unbalanced, with more VF arrests in the no epinephrine group, but more witnessed arrests and bystander CPR in the epi group. This is data from the early 1990s, so ACLS protocols were different. They don’t mention the dose of epinephrine used anywhere in the manuscript.
Hagihara A, Hasegawa M, Abe T, Nagata T, Wakata Y, Miyazaki S. Prehospital epinephrine use and survival among patients with out-of-hospital cardiac arrest. JAMA. 2012; 307(11):1161-8. PMID: 22436956
Quick summary: This is huge database of prospectively collected data from out of hospital cardiac arrests in Japan. They look at 417,188 adult cardiac arrest patients over a 4 year period. The groups were not similar in a number of key areas, such as bystander CPR and shockable rhythm. ROSC was more common the epinephrine group, both in the raw numbers (19% vs 6%) and the propensity matched groups (18% vs 11%). Survival at 1 month was slightly higher with epinephrine (5.4% vs 4.7%), however in the adjusted analyses the epinephrine group had a lower survival rate (5.1% vs 7.0%). Survival with good neurologic outcomes was lower with epinephrine in both the raw (1.4% vs 2.2%) and adjusted analysis (1.3% vs 3.1%).
Caveats: Epinephrine is used very rarely in this setting – only 15,000 patients were given epinephrine as compared to over 400,000 patients who were not. The big question, as it is for all the observational trials, is why these patients were given epinephrine, while others were not. What confounders might be driving the differences seen between the groups?
Nakahara S, Tomio J, Takahashi H, et al. Evaluation of pre-hospital administration of adrenaline (epinephrine) by emergency medical services for patients with out of hospital cardiac arrest in Japan: controlled propensity matched retrospective cohort study. BMJ (Clinical research ed.). 2013; 347:f6829. PMID: 24326886 [Free full text]
Quick summary: This is another study using the large Japanese out of hospital cardiac arrest database. The study years overlap a little bit with Hagihara 2012, so some of the patients will be the same. This study looked at patients between the ages of 15 and 94 who had bystander-witnessed out of hospital cardiac arrests. They used propensity matching to compare a group of patients who received epinephrine with a group of similar patients (in the variables they considered) who did not receive epinephrine. Among patients with VF/VTach arrests, epinephrine was associated with a higher rate of 1 month survival (17.0% v 13.4%; OR 1.34, 95% CI 1.12 to 1.60), but not a higher rate of neurologically intact survival (6.6% v 6.6%; OR 1.01, 95% CI 0.78 to 1.30). Among non-VF/VTach patients, epinephrine was associated with a higher rate of 1 month survival (4.0% v 2.4%; OR 1.72, 95% CI 1.45 to 2.04), and neurologically intact survival (0.7% v 0.4%; OR 1.57, 95% CI 1.04 to 2.37).
Caveats: Propensity matching makes a lot of assumptions, which can significantly impact the results. This is still observational data. Big numbers can make p values of small clinical differences seem really important. The database only contains information about pre-hospital care, so we don’t know how the patients were treated in hospital.
Goto Y, Maeda T, Goto Y. Effects of prehospital epinephrine during out-of-hospital cardiac arrest with initial non-shockable rhythm: an observational cohort study. Critical care. 2013; 17(5):R188. PMID: 24004456 [free full text]
Quick summary: This is another study that looks at the large Japanese cardiac arrest database with a similar timeframe, so we a probably talking about the same patients again (potential duplicate publication bias). They look at 209,577 out of hospital cardiac arrests, and divided them into shockable and non-shockable rhythm groups. In the shockable group, epinephrine was associated with worse outcomes across the board: ROSC 15% vs 28%, 1 month survival 19% vs 23%, and survival with good neurologic outcome 7% vs 5% (all p values <0.001). In the non-shockable group, epinephrine was associated with more ROSC (19% vs 4%) and 1 month survival (3% vs 2%), but there was no difference in survival with good neurologic outcome (0.6% vs 0.6%).
Dumas F, Bougouin W, Geri G, et al. Is epinephrine during cardiac arrest associated with worse outcomes in resuscitated patients? Journal of the American College of Cardiology. 2014; 64(22):2360-7. PMID: 25465423 [free full text]
Quick summary: This is an observational study based on prospectively collected data from a registry. They included 1556 cardiac arrest patients who had been admitted to hospital after achieving ROSC, 73% of whom received epinephrine. Among the epinephrine group, only 17% of patients had a good neurologic outcome (defined as a Cerebral Performance Category of 1 or 2). Among the no-epinephrine group, an amazing 63% had a good outcome.
Caveats: The groups were not well matched. The group that didn’t get epinephrine was more likely to have a shockable rhythm and a short resuscitation. I imagine the results are at least partly related to survivorship bias: healthier patients got ROSC early, and there was no time to give them epinephrine, whereas sicker patients had longer resuscitations, allowing for epinephrine to be given.
Sagisaka R, Tanaka H, Takyu H, Ueta H, Tanaka S. Effects of repeated epinephrine administration and administer timing on witnessed out-of-hospital cardiac arrest patients. The American journal of emergency medicine. 2017; 35(10):1462-1468. PMID: 28473275
Quick summary: This is a retrospective look at the large Japanese out of hospital cardiac arrest registry, with two goals: looking at the impact of the number of doses of epinephrine and timing of epinephrine on neurologic outcomes. There were 378,040 cardiac arrest patients during the 3 year study period, but because of fairly extensive exclusion criteria, they ended up with a sample of 11,876 patients. They divided the timing of epinephrine into early (0-20 minutes), intermediate (21-26 minutes), and late (27-60 minutes). Not surprisingly, the more epinephrine you got the worse you did, both in terms of overall survival and the primary outcome of neurological survival at 1 month. Good neurologic survival was 4.8% with 1mg of epinephrine, 2.4% with 2 doses, and 1.7% with 3 or more doses. Of course, that association almost certainly has nothing to do with the dose of epinephrine. Patients who get ROSC early get less epinephrine. Patients who never get ROSC get more epinephrine. (Survivorship bias). Similarly, overall survival and neurologically intact survival were worse with longer delays to epinephrine (5.3%, 2.5%, and 0.8% neurologically intact survival with the early, intermediate and late groups respectively), but this association likely has more to do with the patients and the timing of paramedic arrival than with the epinephrine.
Caveats: This is an extremely select group of patients, essentially ensuring selection bias has occurred. The time divisions seem a little crazy to me. 20 minutes does not seem like early administration of epinephrine. That being said, these groups are actually fairly evenly divided in terms of numbers of patients, giving a sense of epinephrine timing in a real world setting, and shining some light on the 21 minute average time to epinephrine in the PARAMEDIC 2 trial.
Ueta H, Tanaka H, Tanaka S, Sagisaka R, Takyu H. Quick epinephrine administration induces favorable neurological outcomes in out-of-hospital cardiac arrest patients. The American journal of emergency medicine. 2017; 35(5):676-680. PMID: 28087097
Quick summary: This is another retrospective look at the Japanese cardiac arrest registry (from the same group of authors as the last paper). Again, they look at a very select group of patients (13,326 out of 504,046 cardiac arrests during the study period). They were looking at early versus late administration of epinephrine, this time using a 10 minute cutoff. Early administration of epinephrine was associated with improved survival with good neurological outcome at 1 month (6.0% vs 2.4% if EMS arrived in less than 8 minutes and 4.2% vs 1.4% if EMS arrived between 8 and 16 minutes).
Caveats: There are a number of reasons to be wary of these results. This is retrospective observational data, with the potential for multiple hidden confounders. This is the same group of authors looking at the same database over essentially the same time period as the above study, just using a different time period. It makes you wonder how many different ways they decided to analyze this data, and whether we are just seeing a select few of the analyses (publication bias or data dredging). And again, there is massive selection bias, with only 2.5% of all cardiac arrests being included in the analysis, so even if if you believe the results (which is difficult when RCT data doesn’t support a neurologic benefit from epinephrine), they are definitely not generalizable to the general population.
In observational trials, the decision about whether or not to give epinephrine is not random. The clinicians on scene are making a clinical judgement. So the real question is: why would some patients get epinephrine while others don’t? I would imagine that younger patients, with shorter down times are more likely to get epinephrine, because we definitely don’t want to give up on those patients. On the other hand, older patients with multiple comorbidities might have a truncated resuscitation because their expectation of survival is so much lower at the outset. If that is true, the difference could be entirely driven by the expected survival of the patients before the epinephrine was given, rather than the epinephrine itself.
Despite all that, I think there are some clinical conclusions you can make here. Epinephrine is strongly associated with ROSC. However, the results seem to consistently suggest that epinephrine is not associated with neurologically intact survival. If you were trying to predict the results of PARAMEDIC2 based on the observational data, you would expect more patients to be admitted to the hospital, but your pretest probability of an improvement in neurologically intact survival would be very low.
Do any medications help in ACLS?
These trials take a slightly different approach. They compare giving ACLS medications to giving no medications at all. So although they aren’t specifically about epinephrine, they give us some information about epinephrine.
Quick summary: This is a prospective, before and after study, that looked at cardiac arrest outcomes as defibrillation and then ACLS medications were introduced to EMS crews in Ontario, Canada. They looked at 5638 adult out of hospital cardiac arrest patients and compared baseline BLS care with the care provided after ALS training, such that paramedics were trained in intubation, IV line placement and IV medication administration. During the ALS phase, 96% of patients received epinephrine, 87% received atropine, and 24% lidocaine. ROSC was more common after ALS was introduced (18% vs 13%, p<0.001). Admission to hospital was also more common (11% vs 15%). However, survival to hospital discharge was not changed (5% vs 5%, p=0.83). There were not statistical differences, but the numbers seem to favour better neurologic outcomes among survivors before ALS was introduced.
Ong ME, Tan EH, Ng FS, et al. Survival outcomes with the introduction of intravenous epinephrine in the management of out-of-hospital cardiac arrest. Annals of emergency medicine. 2007; 50(6):635-42. PMID: 17509730
Quick summary: This is another before and after study, similar to the OPALS study. In Singapore, they introduced the use of epinephrine for cardiac arrest to the EMS system in 2003, and study compares the year before the introduction to the year after. They included 1296 out of hospital cardiac arrests in patient 8 years or older. In the year after epinephrine was introduced, 301 of 681 patients (44%) received epi. There was no difference in survival to hospital discharge (1.0% vs 1.6%; OR 1.7, 95% CI 0.6-4.5). There was no change in ROSC (17.9% vs 15.7%; OR 0.9, 95% CI 0.6-1.2).
Caveats: Not all patients in the after group received epinephrine. The overall survival numbers here are very poor. Their protocol was only for a single dose of epinephrine before arrival in the ED. No other medications were used and intubation was not performed, so this is not looking at full ACLS.
Olasveengen TM, Sunde K, Brunborg C, Thowsen J, Steen PA, Wik L. Intravenous drug administration during out-of-hospital cardiac arrest: a randomized trial. JAMA. 2009; 302(20):2222-9. PMID: 19934423
Quick summary: This is a prospective, unblinded, randomized controlled trial performed in Norway. They randomized 851 adults with nontraumatic out of hospital cardiac arrest to either get intravenous ACLS medications or not get ACLS medications. ROSC and survival to admission was higher with ACLS drugs (32% vs 21%, p<0.001). However, 1 month survival was not statistically different (10.5% vs 9.2%, p=0.61). SImilarly, survival with good neurologic outcome was not improved (9.8% vs 8.1%, p=0.45).
Caveats: The study could not be blinded by design. Epinephrine was not the only drug studied, so it impossible to know what impact epinephrine specifically had on the results. There was some crossover: only 80% of the IV group actually received epinephrine, and 10% of the no IV group actually received epinephrine. Patients with asthma or anaphylaxis as the presumed cause of their arrest were excluded. The EMS system in Norway includes physicians in the field.
High dose epinephrine
Although I have never seen it used, there was a time when high dose epinephrine was considered for cardiac arrest. Although my primary focus is on whether epinephrine works at all, dose-response relationships are important to consider, so here are a few of the studies that look at epinephrine dosing:
Quick summary: A RCT of 650 patients in cardiac arrest, randomized to either high dose (7mg) or standard dose (1mg) epinephrine, both given every 5 minutes according to standard protocols. There was no difference in 1 hour survival (18% vs 23%, p=0.12, absolute difference 5%, 95% CI -1-12%), nor survival to hospital discharge (3% vs 5%, p=0.38, absolute difference 2%, 95% CI -2-5%). Among survivors, neurologic outcomes looked the same.
Brown CG, Martin DR, Pepe PE, et al. A comparison of standard-dose and high-dose epinephrine in cardiac arrest outside the hospital. The Multicenter High-Dose Epinephrine Study Group. The New England journal of medicine. 1992; 327(15):1051-5. BPMID: 1522841 [free full text]
Quick summary: This is a double blind RCT comparing standard dose epinephrine (0.02mg/kg) to high dose epinephrine (0.2mg/kg) in 1280 adult patients with nontraumatic out of hospital cardiac arrest. There were no differences between the groups in the rates of ROSC (30% vs 33%), survival to hospital admission (22% and 22%), survival to hospital discharge (4% and 5%), or neurologic outcomes in survivors.
Callaham M, Madsen CD, Barton CW, Saunders CE, Pointer J. A randomized clinical trial of high-dose epinephrine and norepinephrine vs standard-dose epinephrine in prehospital cardiac arrest. JAMA. 1992; 268(19):2667-72. PMID: 1433686
Quick summary: This is a RCT comparing standard dose epinephrine (1mg) to high dose epinephrine (15mg) or high dose norepinephrine (11mg) in 816 adult patients with nontraumatic out of hospital cardiac arrest. High dose epinephrine resulted in more ROSC (13% vs 8%), but there was no difference in survival to discharge (1.7% vs 1.2%). Cerebral performance scores were not statistically different, but the authors report a trend toward possible harm with high dose epinephrine (3.2 vs 2.3, p=0.1).
Perondi MB, Reis AG, Paiva EF, Nadkarni VM, Berg RA. A comparison of high-dose and standard-dose epinephrine in children with cardiac arrest. The New England journal of medicine. 2004; 350(17):1722-30. PMID: 15102998 [free full text]
Quick summary: This is a double blind study comparing high dose epinephrine (0.1mg/kg) and standard dose epinephrine (0.01mg/kg) in 68 pediatric patients in in-hospital cardiac arrests. ROSC was the same in both groups (59% vs 62%). Survival at 24 hours was worse in the high dose group (21% vs 3%, p=0.05). Survival to discharge wasn’t statistically different, but still looks worse (0% vs 4%, p=0.11).
High dose discussion
There is a reason that we don’t use high dose epinephrine. It doesn’t help, and it might hurt. The data hints that epinephrine is not benign, and should make us consider if we are still using too high a dose.
Jacobs IG, Finn JC, Jelinek GA, Oxer HF, Thompson PL. Effect of adrenaline on survival in out-of-hospital cardiac arrest: A randomised double-blind placebo-controlled trial. Resuscitation. 2011; 82(9):1138-43. PMID: 21745533
Quick summary: This is a double blind RCT from a single EMS service in Western Australia that was previously using no medications in cardiac arrest. They randomized 534 adult out of hospital cardiac arrest patients to epinephrine (1mg q3min) or placebo. Prehospital ROSC was more common in the epinephrine group (23.5% vs 8.4%; OR 3.4, 95% CI 2.0 – 5.6). More patients were also admitted to hospital (25.4% vs 13.0%). However, there was not a statistical difference in survival to hospital discharge (4.0% vs 1.9%; OR 2.2; 95% CI 0.7-6.3), but the study was underpowered. (Considering the results of PARAMEDIC2, this is likely a real difference). There were 2 bad neurologic outcomes, both in the epinephrine group.
Caveats: The big caveat is that this trial was supposed to be much bigger. Originally the power calculation called for 5,000 patients, but somehow politicians got involved, questioned the equipoise on the issue, and multiple other EMS agencies were forced to pull out. Selection bias is possible, as there were a large number of patients who were excluded because “resuscitation was not commenced” (2513 out of 4103 screened). This population seems to have better characteristics than other studies (46% shockable rhythm and 51% bystander CPR), possibly because of the exclusions. Individual paramedics had to volunteer for the trial, so only 40% of eligible patients were enrolled.
All of this, of course, bring us up to the PARAMEDIC-2 trial that was published this year. My full post on that paper can be found here.
The quick summary is that it was a multicenter, randomized, double-blind, placebo controlled trial looking at 8014 adult patients with out of hospital cardiac arrest for which advanced life support was provided by trial-trained paramedics. Patients were randomized to epinephrine 1mg q3min or placebo. The primary outcome was 30 day survival, and epinephrine showed a small but statistically significant benefit (3.2% vs 2.4%, p=0.018). However, there was no difference in survival with good neurological outcome.
Loomba RS, Nijhawan K, Aggarwal S, Arora RR. Increased return of spontaneous circulation at the expense of neurologic outcomes: Is prehospital epinephrine for out-of-hospital cardiac arrest really worth it? Journal of critical care. 2015; 30(6):1376-81. PMID: 26428074
Quick summary: This is a systematic review and meta-analysis including both observational trials and RCTs that compared epinephrine to no epinephrine in out of hospital cardiac arrest. They include 14 studies covering 655,853 patients (most of which were included above). Epinephrine increased ROSC (OR 2.84), but there was no difference in 1 month survival (OR 1.03), and neurologic outcomes were worse in the epinephrine group (OR 0.51).
Lin S, Callaway CW, Shah PS, et al. Adrenaline for out-of-hospital cardiac arrest resuscitation: a systematic review and meta-analysis of randomized controlled trials. Resuscitation. 2014; 85(6):732-40. PMID: 24642404
Quick summary: This is another systematic review, but looking at slightly different papers. It only includes RCTs, but also looks at papers comparing epinephrine to vasopressin, and papers comparing high and standard dose epinephrine. Compared to high dose epinephrine, standard dose resulted in less ROSC (RR 0.85; 95% CI 0.75–0.97, p = 0.02) and decreased survival to admission (RR 0.87, 95% CI 0.76–1.00, p = 0.049). However, there were no differences in survival to discharge (RR 1.04, 95% CI 0.76–1.42, p = 0.83) or good neurologic outcome (RR 1.20, 95% CI 0.74–1.96, p = 0.46). Six trials compared epinephrine alone to a combo of epinephrine and vasopressin, and there was no difference. A single trial compared epinephrine to vasopressin, and there was no statistical difference.
Like most topics in medicine, the evidence here is far from perfect, and probably leaves us with more questions than answers. It seems fairly certain that epinephrine increases the rate of ROSC, as well as the number of patients admitted to hospital. It also seems fairly clear that, at least in the dosing regimens we are currently using, epinephrine does not meaningfully increase survival with good neurologic function. There are probably a small number of extra patients alive when epinephrine is used, but many of these patients are left living lives most of us wouldn’t envy.
All of the above research focuses on out of hospital cardiac arrest. In hospital arrests are different, and so these results shouldn’t be extrapolated (although I am not aware of any data that would suggest that epinephrine will help in hospital arrests either). However you end up interpreting the results, remember that clinical judgment is required. Although my baseline is not using epinephrine, an arrest secondary to anaphylaxis or asthma would probably change my mind.
The 2 observational trials from 2017 hinted at an association between giving epinephrine early and better outcomes. There are many reasons to be skeptical of those results, given that the same authors performed multiple retrospective looks at a very select group of patients. I think it is a reasonable hypothesis that PARAMEDIC 2 failed because epinephrine was given too late, but based on the totality of this literature I think it is more likely that epinephrine simply doesn’t work. Exploring this hypothesis in the setting of a study seems reasonable, but I would not let these flawed observational studies guide clinical practice.
Personally, I don’t think this question should be left in the hands of doctors. We like like to think we are the best judges of what is right in medicine, but these aren’t really medical questions – these are questions about humanity.
That brings up one of my favourite parts of the PARAMEDIC2 trial – even though they hid it in the discussion. Before they even started this trial, they asked normal people – potential patients – what outcomes mattered to them. And the answers were clear: long term survival without brain damage is much more important than just survival alone.
So based on that, they really should have made neurologically intact survival their primary outcome, but I understand why they didn’t. This was already a huge trial, and it would have had to have been much bigger if they were powering it for neuro outcomes.
For me, the real value of going through all this data is that it lets me assess PARAMEDIC 2 from a Bayesian perspective. It lets me set my pretest probability. And I think that if you take all this data as a whole, before PARAMEDIC2 was published, there was very little reason to think that epinephrine was going to improve patient oriented outcomes (in other words, neurologically intact survival). We already knew that it was going to increase ROSC, and the best guess was that there was going to be a small increase in overall survival, but I think this data tells us that our pretest probability before PARAMEDIC2 was that epinephrine wasn’t helping patients.
In that context, PARAMEDIC 2 is clearly a negative trial. What we wanted to see was an improvement in survival with good neurologic function, and we didn’t see it. We can’t completely exclude the possibility of benefit, but PARAMEDIC2 shows no benefit.
So I think you have to consider how you would interpret PARAMEDIC 2 if you weren’t currently using epinephrine – because the previous data says we probably shouldn’t have been using it. From that perspective, I think it is pretty clear that this is not a good enough trial to convince us to start using epinephrine. If was a brand new drug, rather than something we are all familiar with, we wouldn’t be using it.
I would take it a step further, because we have to consider the harms. We have no convincing evidence of patient oriented benefit. We can’t exclude it, but if its there its tiny. On the other hand, there are clearly harms with epinephrine. There are the costs of running ALS EMS services. There are the costs of putting hundreds of thousands of extra patients into hospitals and ICUs. There are the harms to the patients that are put into the hospital only to die. (I don’t know any person who wants to spend time in an ICU when they are guaranteed to die). And there are the harms of neurologic disability.
I have already said that I don’t know how to balance those harms against the potential benefit. This is a value judgement. I don’t want to impose my values on others.
But currently, we have no evidence of patient oriented benefit and lots of evidence of harm, so my take is that epinephrine should be removed from ACLS. It shouldn’t be used routinely.
There is room for more research (should we be using a different dose or maybe focusing on a different subset of patients), and there is always room for clinical judgement. The profound shock that we call PEA is very difference from v fib, and may respond to some appropriate dose of epinephrine.
But epinephrine as a routine, in my mind, should be out.
I am concerned, after reviewing all this data, that we are harming our patients with the routine use of epinephrine in cardiac arrest. I would remove it from the algorithms, but recognize that there are still many questions, and that there may be individual patient circumstances in which a bolus of epinephrine might make sense.