Headline news in the New England Journal: ECPR doubles survival in out of hospital cardiac arrest! Does that mean everyone needs to start learning how to crash patients onto ECMO?
The question
Does prehospital ECPR (extracorporeal cardiopulmonary resuscitation) improve mortality in young patients with refractory out of hospital cardiac arrest?
The paper
Reis Miranda D, Moors X, Ali S, et al. Prehospital Extracorporeal Life Support in Refractory Out-of-Hospital Cardiac Arrest. N Engl J Med. 2026 Nov 7. doi: 10.1056/NEJMoa2602919. PMID: 43001740 NCT04620070
The methods
This is a prehospital observational trial from the Netherlands. (They refer to it as a ‘step wedge design’ but it is an unusual use of that terminology.)
Patients
Out of hospital cardiac arrest patients between the ages of 18 and 50 with:
- An initial shockable rhythm or PEA that was specifically thought to be due to pulmonary embolism
- Witnessed arrest or presence of signs of life (defined by gasping, pupil reactions, or movements)
- Refractory arrest (at least 20 minutes after HEMS dispatch)
- Readiness for needle puncture for ECMO within 45 minutes of HEMS dispatch
Patients were excluded if:
- They had end tidal CO2 less than 1.2 kPa during CPR
- An estimated transfer time from collapse to ECMO center of less than 30 minutes
- No clear ultrasound visualization of either the femoral artery or vein
They initially had exclusions that only applied after hospital arrival, but seeing as that is contradictory to their stated intent-to-treat analysis, they dropped these exclusions.
Exposure
They are looking at the role out of HEMS based ECPR. In the Netherlands, a HEMS team staffed by a consultant anesthesiologist or trauma surgeon and a nurse are available 24 hours a day 7 days a week, can be airborne within 2 minutes of notification, and basically the entire area of the Netherlands can be reached by helicopter within 30 minutes.
In the ‘after’ or ECPR phase, ECMO was available to the HEMS service and could be started after 20 minutes without ROSC.
Comparison
During the conventional or before period of this trial, HEMS had no ECMO on board, and care was based on standard advanced life support algorithms and physician judgement.
Outcome
The primary outcome was survival until hospital discharge.
The results
During the study period there were 3362 HEMS activations, and after exclusions 231 patients are included in the study. Most of the exclusions are expected, including 1722 where the HEMS team was called off before they even landed on scene. However, there are 124 patients excluded with a listed “unknown reason” which is actually more than the number of patients excluded for age reasons (101), and could represent an important source of bias.
Cannulation was only attempted in 68% of the ECPR group. 22% of this group had ROSC before ECPR could be started, 8% could not have their femoral vessels visualized on ultrasound, and 2% met all criteria but were left out anyway.
The primary outcome, survival until hospital discharge, was 40% with ECPR and 20% with conventional therapy (odds ratio 2.72, 95% CI 1.59-4.66). Survival with favourable neurologic outcomes at 6 months were also significantly better in the HEMS group (36% vs 13%).
My thoughts
ECPR is exciting. There is little doubt that it provides better circulation than conventional CPR. However, CPR by itself is not actually a therapy. It is just a bridge to therapy. ECMO is probably the same. Patients who die are likely to remain dead unless we can treat the underlying cause, no matter what technology we use to keep their blood oxygenated and circulating. That is why we still need science to prove the value of ECMO, and unfortunately this observational data is not what we need.
ECPR is associated with better outcomes here, but there are so many potential confounders in this observational data that the association is essentially impossible to interpret.
They refer to this trial as “stepped-wedge”, but I think that gives the wrong impression. I have only ever seen the term steeped-wedge used in the context of an RCT. I have searched fairly widely, and essentially all the references I can find to “stepped-wedge” are in the context of cluster randomized trials. They use a lot of other terminology, such as “intention to treat” and “assigned group”, that you normally only see in RCTs. Although the terminology may be technically correct, it gives the impression that this trial is a cluster RCT, which it is not. It is just a before and after observational trial using fancy terminology.
Before and after trials are always problematic. It is very difficult to tease apart the impact of the specific change being studied from wider systems changes. Cardiac arrest outcomes have been improving for some time, and so you would expect improvement in any before/after trial design. More importantly, when you introduce a brand new technology, teams come together for extra training, and a new optimism spreads around resuscitation. The improved outcomes might have nothing to do with the therapy and everything to do with systems changes that were occurring simultaneously, or changes in decisions driven by the broader changes.
Mortality seems like an objective outcome, but we obviously have a lot of say in when we decide to stop resuscitation. If you roll into an emergency department doing conventional CPR on a patient who has been down for 45 minutes, we are likely to terminate resuscitation pretty quickly. However, if you arrive with a patient on ECMO, that patient is going to be given every possible chance to survive. We can clearly see that this happened, given that 50% of the conventional group died in the emergency department as compared to only 13% of the ECPR group. In the long run, this might be precisely why ECMO is valuable – because it gives us enough time to get the patient to definitive therapy. But for now, it is a dramatic confounder in this observational data.
This Hawthorne effect could certainly skew the results, but the biggest and most important source of bias probably occurred at enrollment. Enrollment occurred on scene, and although the study has defined criteria, many are open to judgement. When that happens, you worry that one group (usually the one getting the fancy new therapy) will select healthier patients in general, biasing the results. Looking at their baseline table, I think it is very clear this trial compares apples and oranges.
A shockable rhythm is one of the strongest predictors of good outcomes in out of hospital cardiac arrest. It occurred in 89% of the ECPR group and only 75% of the conventional group. Perhaps the inverse is more important to emphasize. PEA has a dismal outcome, and occurred in 11% of the ECPR group as compared to 25% of the conventional group. There were signs of life on scene in 38% of the ECPR group as compared to 30% of the conventional group. ECPR was simply being applied to a different cohort of patients than were included in the conventional CPR group.
They didn’t randomize the order in which the HEMS bases implemented ECPR. That allows for the potential to start with your highest performing HEMS site, which would artificially improve outcomes in the ECPR group. Some information on baseline performance at each site would have been nice.
Perhaps the biggest indicator of selection bias is just the survival rate. 40% survival to hospital discharge is simply unheard of in out of hospital cardiac arrest. Some people will use that number to create hype around ECPR, but the most likely explanation is simple: the ECPR group only included a select group of patients with a higher likelihood of survival.
There are RCTs of ECPR. Suverein and colleagues (2023) randomized 160 patients to ECPR or standard therapy. The trial was statistically negative, with 20% of the ECPR group being alive with a favourable neurologic outcome at 30 days as compared to 16% of the conventional CPR group. Belohlavek and colleagues (2022) randomized 256 patients, was stopped early, and was also statistically insignificant (31% vs 22% survival with favourable neurologic outcomes at 180 days). These two RCTs are similar. Both were underpowered. Both hint at a real benefit from ECPR. But most importantly for our purposes, both had survival rates well less than 40%, which suggests to me significant selection bias in this current observational trial.
There is a lot more that could be said. For example, external generalizability might be questioned, given that the entire country can be reached within 30 minutes by air. (Not to mention the fact that huge areas of the world don’t have physician led prehospital services.) However, questions of generalizability only really matter if you believe the 20% absolute improvement in mortality was real. I think the significant bias in this observational data combined with negative (although promising) RCT data means you have to assume, for now, that this results here do not represent a true causal effect from ECPR.
The results are dramatic, but we need to be cautious in our interpretation. ECPR makes a lot of sense, but therapies with good theory fail all the time in medicine. Some people will take these results and declare ECPR the new standard of care. Realistically, we just cannot know based on this data. Blinding of ECPR will be incredibly difficult, so my guess is that we never see a properly blinded trial, but randomization with strictly defined outcomes (where death isn’t simply left to the judgement of the clinician, but is judged identically in both groups) are necessary before we decide to implement this therapy widely.
Bottom line
Based on the prior RCTs, it seems like ECPR will be an important part of cardiac arrest care in careful selected patients in the future. However, despite the dramatic results, this observational data does not really change anything. We still need properly controlled RCTs to determine whether this technology is actually saving lives.
Did you find this information useful? There are many other First10EM critical appraisals, which can be found here. Also, consider subscribing to this blog:
Morgenstern, J. Prehospital ECPR: A 20% survival benefit?!?!, First10EM, November 16, 2026. Available at:
https://doi.org/10.51684/FIRS.146761
For regular readers, I have obviously dramatically decreased the frequency of new posts. To some extent, that is simply because there have been fewer papers published that I have felt compelled to comment on. However, to a larger extent, it is just life. I don’t have nearly as much time for this hobby as I did years ago. For the foreseeable future, First10EM will simply have less content. That will likely change again in the future, so please stay subscribed if you remain interested in the content. All the best – Justin.
Other FOAMed
Evidence based medicine is easy
Evidence based medicine resources
References
Belohlavek J, Smalcova J, Rob D, Franek O, Smid O, Pokorna M, Horák J, Mrazek V, Kovarnik T, Zemanek D, Kral A, Havranek S, Kavalkova P, Kompelentova L, Tomková H, Mejstrik A, Valasek J, Peran D, Pekara J, Rulisek J, Balik M, Huptych M, Jarkovsky J, Malik J, Valerianova A, Mlejnsky F, Kolouch P, Havrankova P, Romportl D, Komarek A, Linhart A; Prague OHCA Study Group. Effect of Intra-arrest Transport, Extracorporeal Cardiopulmonary Resuscitation, and Immediate Invasive Assessment and Treatment on Functional Neurologic Outcome in Refractory Out-of-Hospital Cardiac Arrest: A Randomized Clinical Trial. JAMA. 2022 Feb 22;327(8):737-747. doi: 10.1001/jama.2022.1025. PMID: 35191923
Reis Miranda D, Moors X, Ali S, et al. Prehospital Extracorporeal Life Support in Refractory Out-of-Hospital Cardiac Arrest. N Engl J Med. 2026 Nov 7. doi: 10.1056/NEJMoa2602919. PMID: 43001740
Suverein MM, Delnoij TSR, Lorusso R, Brandon Bravo Bruinsma GJ, Otterspoor L, Elzo Kraemer CV, Vlaar APJ, van der Heijden JJ, Scholten E, den Uil C, Jansen T, van den Bogaard B, Kuijpers M, Lam KY, Montero Cabezas JM, Driessen AHG, Rittersma SZH, Heijnen BG, Dos Reis Miranda D, Bleeker G, de Metz J, Hermanides RS, Lopez Matta J, Eberl S, Donker DW, van Thiel RJ, Akin S, van Meer O, Henriques J, Bokhoven KC, Mandigers L, Bunge JJH, Bol ME, Winkens B, Essers B, Weerwind PW, Maessen JG, van de Poll MCG. Early Extracorporeal CPR for Refractory Out-of-Hospital Cardiac Arrest. N Engl J Med. 2023 Jan 26;388(4):299-309. doi: 10.1056/NEJMoa2204511. PMID: 36720132

