CPR has progressed from adventurous reanimators pulling drowned bodies from rivers to the now well-oiled ILCOR ALS algorithms of the twenty-first century. Even though the hesitation to start CPR has almost completely dissipated, CPR yields a meagre 50% of normal cardiac output at best. In fact, one may quote to the inquisitive examiner, that CPR provides varying cardiac outputs of between 10% to 50% in poor dogs (Fitzgerald et al., 1981; Silver et al., 1981) , pigs, rats and other such animals that were subject to human-induced electrical VF. We have yet to be able to do this (and document it) to humans, fortunately. If the cardiac output is so feeble, it makes sense that one would instead try to restore full circulatory support using ECMO during CPR. E-CPR is now done widely across the world with increased success and promising outcomes (Abrams et al., 2022; Alfalasi et al., 2022) . For the seminal guideline I refer the reader to the ELSO paper collaborated by ELSO and our friends at The Alfred (Richardson et al., 2021) .
Free full cardiovascular support, oxygenation the brain, perfusing coronary vessels,
why not? Well, restoring full flow on an arrested patient with no possibility of coming back is not ideal. The indications for E-CPR encompass factors that are favourable for an adequate neurological outcome for the patient. Although varying from centre to centre, the ELSO indications are the best to memorise in the event of memory lapses:
As with other uses of ECMO, the contraindications include established multi organ failure or known terminal co-morbidities such as metastatic malignancy. VA ECMO doesn’t work well with blood escaping through an incompetent aortic valve, so significant aortic valve problems are generally contraindicated.
PEA is worth discussing, with PEA showing better outcomes in general for OOHCA and a higher chance to ROSC. It is probably true, given that asystole mostly equates to death. Whether the PEA group should have E-CPR is another thing, though. The likelihood of coronary ischaemia as the cause for PEA is lower, and many other nasty non-E-CPR candidate causes such as intracranial haemorrhage, can present with PEA. Interestingly, PEA is included in the Prague OHCA (assessed to be probable cardiac cause), Bernard studies, but not in the ARREST and 2CHEER studies (Belohlavek et al., 2022; Bernard et al., 2022) . In the end, if you suspect a cardiac cause, that makes the patient more likely to benefit from E-CPR.
Being not without consequence, it is advised to wait and see if ROSC is achieved by standard ALS algorithm techniques before initiating ECMO. There is a suggestion that the probability of ROSC diminishes greatly after 20 minutes, and the majority of neurologically favourable survivors have less than 35 minutes of CPR (Gravesteijn et al., 2020; Perkins et al., 2018; Reynolds et al., 2016) . Greater than that and less than around 60 minutes, where outcomes start to diminish with E-CPR, is probably the ideal window of opportunity for one to get those pipes in.
E-CPR is done with VA ECMO. The artery can be a fiend to cannulate in the presence of large circulating doses of adrenaline and some centres (like the team on those Louvre or Paris underground shots) always do a surgical or a hybrid cut down technique, even in the field. Locally in Australia, percutaneous cannulation seems to be the weapon of choice and in hand-to-hand combat, the technique that the
operator is most familiar with will probably be the best. Femoral arterial with femoral venous cannulation is the preferred configuration, as there will be someone or something doing CPR and an airway person of sorts at the top end. Wire and cannulae position should be confirmed on ultrasound and if available, TOE is helpful, as is fluoroscopy if you happen to be in the angiogram suite or it is magically available wherever you are, but they aren’t essential for establishing flow initially.
There is some debate in the ECMO communities about E-CPR dine in vs takeaway. Risks and benefits are in abundance for both options and probably depend on resources, staffing, skill mix, paramedic services as well as population and geographic factors such as hospital location, surrounding population density and
traffic structures. Certainly, having a 24-hour retrieval roster can be a significant strain to the institution but at the same time, minimising time to flow and the coolness of going out in a bond-like van, or with a parade of police like in Paris, cannot be understated. In times of recession, who knows what health funding is heading to, though. More important for an institution is probably maintenance of pathways, guidelines and the appropriate training, including simulation, to ensure high standards.
It’s quite hard to compare apples to apples when comparing conventional CPR with E-CPR. Crossover into an E-CPR is almost inevitable when performing a study and equipoise and objectivity is unachievable, with human beings and ethics, at least. The most recent Prague OHCA trial by Belohlavek et al. was rather interesting, randomising patients to a resuscitation basket of early coronary angiogram and intervention, and ECMO if the study criteria were met. The standard arm, however, had strong suggestions to the treating physicians for early coronary angiogram and intervention anyway. There was no statistically significant difference between the two groups in 180 day neurologically favourable survival (even if one argues that the absolute numbers were different) but it was very close. Recruitment also started in 2013 and perhaps we are even better at E-CPR now. This study had something quite striking though: the standard arm had 22% survival at 180 days! That is huge, compared to what we normally quote our junior doctors and medical students, which is around less than 10%, being generous, and large studies suggesting something like 3%. Bernard et al. from The Alfred recently did something akin to Prague OHCA for Melbourne, Australia, looking at patients coming into ED with CPR in progress. They had a very, unfortunately, similar result, with neurologically favourable just not meeting statistical significance in the E-CPR group. In Melbourne, however, there is only one centre that does E-CPR and the numbers in this study weren’t large (49 in the E-CPR group).
The slightly older ARREST trial (2020), (but more recent intervention, perhaps, with patients in 2019 and 2020, compared to Prague OHCA) by Yannopoulos et al. (Yannopoulos et al., 2020) randomised 30 patients to essentially E-CPR or standard resuscitation and found 43% survival to discharge in the E-CPR arm (small group etc.) while the 2CHEER trial in Sydney showed a 44% neurologically favourable survival in a cohort of 25 E-CPR patients (Dennis et al., 2020) . As more data of relevant recency comes out for E-CPR, it seems that things will only improve, as reflected by meta-analyses.