The evidence base for current advanced life support practice

There is a surprising lack of sensible evidence behind the routine process of cardiopulmonary resuscitation. Question 19 from the first paper of 2007 was the only question to ever ask anything about this, and it was mainly testing whether the candidates had reviewed the recently changed ALS guidelines.  Those guidelines had undergone multiple changes since, without attracting any further questions, suggesting that a detailed knowledge of this material is no longer prioritised by CICM. 

In any case, what evidence could they possible ask about. It would be absurd to expect high-quality randomised controlled trials in this area. For instance, how would one randomise people to the non-CPR group? Thus much of what we do is a combination of physiology, instinct, and animal studies.

Evidence for CPR, and specifically early CPR

Cardiac compressions have been around forever, but until relatively recently   most of these resuscitative efforts have taken the form of thoracotomy and internal cardiac massage. Even with this dramatic step, haste was advised, as it was clear to everybody that the longer you remain without cardiac output, the worse your outcome. Working on this premise, "there should be no hesitation in opening the chest when the heart beat has ceased", commanded various published authorities like Clemetson (1959).

Still, there was clearly hesitation. With contemporary resuscitation techniques Cole et al (1958) reported survival of only 22% (33 patients from 150), where of the dead, the majority had a long delay between the onset of arrest and the beginning of treatment (i.e opening the chest). Kouwenhoven et al (1960), arguing that all delay could be abolished by the abandonment of the thoracotomy, applied their new external-only technique to twenty patients and demonstrated a 70% rate of "permanent survival". All you need is two hands, they famously asserted. This was CPR in its modern form, albeit with a compression rate of 60 per minute. 

Obviously these days this is not one of those things where you could confidently apply for ethics approval of a clinical trial, and so the strongest evidence we have that CPR in general is beneficial comes from the observational evidence that when CPR is for whatever reason withheld for a period, the outcome is worse. A famous Swedish study by Hasselqvist-Ax et al (2015, n = 30,381) found that, where bystanders attempted CPR, 30-day survival was 10.5%, whereas it was only 4% when the bystanders stupidly stood around and did nothing. To be more granular, the survival dropped from 15.6% (where CPR was started within 3 minutes) to 0.9% (where it was delayed by more than 14 minutes), giving a 1.4% drop in survival chance with every passing minute of bystander hesitation.

Evidence for the precordial thump

Of all the things you can do for a patient in cardiac arrest, the precordial thump is probably the fastest (short of starting CPR). The idea is theoretically sound: on handling the heart intraoperatively, minor mechanical stimuli cause a depolarisation (“There is usually a single response: one flick with a finger or needle results in one mechanical premature contraction” according to Koster, 2009). The promise of immediately cardioverting a malignant arrhythmia makes the thump seem like a highly attractive, if somewhat confronting, instrument of the first responder: you need no equipment other than the ulnar edge of your strong young fist, and it lends a certain bravado to the rescue effort, framing the situation in a more heroic light. Unfortunately, as a routine part of practice, it can’t be recommended with a straight face. It rarely works, and you usually only see it working in VT - Volkmann (1990) were able to get something like 77% success rate with this arrhythmia. In order to achieve this formidable response, you’d need to use it within twenty seconds of arrest. It will also be useless in VF or PEA. In short, the setting for a successful thump is really the CCU, where a fully monitored patient goes into totally unambiguous pulseless VT just as you are examining them.

Evidence for early defibrillation

There also seem to be a substantial improvement in outcome with early defibrillation. In an observational study of 553,426 cardiac arrest patients Goto et al (2016)  found that each minute of delay in the arrival of paramedics resulted in a 10.7% decrease in the chances of a good neurological outcome, and this can be used as a surrogate for the effect size of timely defibrillation, as the defibrillator is the most important piece of equipment on that ambulance. Multiple authors have demonstrated the importance of early defibrillation, such as:

• Weaver et al (1984), in out of hospital arrest, 62% survival with early defibrillation by first responders and only 27% if purely basic life support was provided
• Chan et al (2008), for in-hospital cardiac arrest with a shockable rhythm, found that delayed defibrillation (by more than 2 minutes) resulted in progressively worse outcome, dropping from 40% survival (defibrillated within 1 minute) to less than 15% (defibrillated after 6 minutes, or longer).
• Kang et al (2019), also for in-hospital cardiac arrest, but in a much larger sample, detected an inflection point - for patients who had received defibrillation within the first 3 minutes, the likelihood of good neurological outcome was much the same, and only a delay of more than 3 minutes was associated with a progressively greater risk of brain injury.

In short, for defibrillation, it seems sooner is better (wherever defibrillation is appropriate). However, the current guidelines recommend for a period of compressions prior to defibrillation, and there is some solid rationale for this.

Evidence for some compressions prior to defibrillation

From the aforementioned data, defibrillation looks so important that some had come to the conclusion that immediate defibrillation is the most important priority, and indeed there was some early signal that this is beneficial. Spearpoint et al (2000), among in-hospital VT/VF arrest,  tried defibrillation prior to any other measures, and were impressed with the results (increased survival was observed) but this study was small and these results could not be replicated by subsequent investigators. Moreover, it was found that there is a nonzero time period of wheeling the arrest trolley up to the patient and attaching the pads, during which time the patient remains without blood flow unless CPR is in progress. The rationale of immediately commencing compressions while waiting for somebody to get the defibrillator also has a sound physiological basis as it is known that a hypoxic acidaemic myocardium is resistant to defibrillation , and so it is sensible to push some oxygenated blood into it first, so that the shock might be more successful. Kerber & Sarnat (1979) and Edelson et al (2006) all found that long pre-shock pauses and poor pre-shock BLS resulted in a lower chance that the defibrillation will be successful.

Evidence for defibrillation energy

In bygone days the energy dialled up for defibrillating the fibrillating ventricle was determined according to the patient’s weight, on the basis of a linear relationship observed from animal experiments, such as Geddes et al (1974). That one was an especially spectacular work, where the investigators zapped an entire zoo of creatures (“50 dogs, 9 sheep, 7 horses, 3 goats, 2 ponies, 1 calf, and 1 camel”) and determined that the energy required for successful defibrillation likely increases in proportion to the mass of the myocardium. This seemed to make logical sense to everybody, and manufacturers flooded the market with monstrous defibrillator devices, capable to blasting the patient with up to 800 Joules of energy. This practice was ultimately buried by an avalanche of contrary data, such as Gascho et al (1979), who found much lower energies effective in up to 98% of cases. These days, most machines max out at 360J. ILCOR recommend you escalate to this setting in adults if the first shock of 200J has not worked, as there is little evidence of harm from using this energy, so their argument is why would you use a lower energy (also it makes ALS algorithms easier to teach). In children, weight-based recommendations persist, and 4J/kg is the recommended dose.

Evidence for the rate of compressions

We have only very indirect evidence to support the current recommendations for CPR compression rate. We suspect it would be around 100, because any greater rate than this causes resuscitator fatigue (its hard work!) whereas a lower rate is associated with a poorer outcome. A higher rate would also likely result in poor diastolic filling. There's not much data here to help decide. We rely on surrogate measures, as for example in the study by Kern et al (1992), where the investigators used end-tidal CO₂ to determine the effect on the cardiac output, and found a slight but statistically significant improvement with a rate of 120, as opposed to 80 (15 vs 13 mmHg). The best data for anything patient-centred would probably have to be Idris et al (2015), who retrospectively analysed compression rate data from 10,371 defibrillator recordings, and found that survival was greatest in patients who had compressions at a rate between 100 and 120.

Evidence for ratio of breaths to compressions

We don't know what the ideal respiratory rate should be during a cardiac arrest, but we suspect it is lower than normal (because the arrested patient is producing less CO₂, given that large swathes of tissue are using anaerobic metabolism). The ILCOR people have settled on a 30:2 ratio irrespective of the number of rescuers. This ratio was arrived at not necessarily by science, but by convenience (in order to simplify teaching and skills retention), but it also happens to be supported by at least theoretical data. For example,  Babbs & Kern (2002), modelling cardiac output mathematically, concluded that 30:2 was the optimum ratio to maintain cardiac output for classically trained rescuers, and 60:2 for lay persons.

That prompts the question, do we even need to give breaths? It has been said that (at least initially) the gas mixing which occurs in the chest during cardiac compressions is sufficient for the exchange of gas during an arrest, and that any extra breaths on top of this gas mixing is unnecessary. Moreover, interruptions in CPR are obviously undesirable, and the single lay rescuer will not be able to coordinate quick breaths with CPR very easily. It's also easier to teach, which potentially increases the rates of bystander participation, resulting in more lives saved. On the other hand, one must admit that the patients who have arrested due to hypoxia would probably benefit from some additional respiratory support.

Tidal volume generated by CPR is not zero, but it’s also not much. Deakin et al (2007) reported tidal volumes in the range of 30-60ml, i.e. much smaller than the anatomical dead space. However, one needs to note that about a hundred of these little volumes will be mixed over the course of a single minute, leading to a greater total “minute volume” of perhaps 3-6L. That might sound almost normal; but we need to remember that each individual volume is so small that alveolar gas may remain almost entirely undisturbed. The idea that this is not enough to result in gas exchange was also supported in a more recent study by Vanwulpen et al (2021), who hooked up a flow sensor to the ETT of freshly intubated out-of-hospital cardiac arrest patients for the first 30 compressions, and discovered that in the real world of the prehospital setting the median tidal volume generated by compressions was only 20ml.

Observational data (eg. Ogawa et al, 2011) also cautiously weighs in favour of conventional CPR techniques, especially in young patients and those whose cardiac arrest is of a noncardiac cause. A more specific study, looking at victims of drowning, could not detect any outcome differences (Fukuda et al, 2019). Overall, compression-only CPR has never overtaken conventional CPR, and is viewed as a viable alternative only because it is so much better than no CPR.

Intubation during CPR

Intubation is by no means mandatory during CPR  in a cardiac arrest, unless some sort of airway disaster was the actual cause of the cardiac arrest  - and even in that case, either simple airway manoeuvres will open the airway, or nothing will (thinking of some kind of massive bulky aspiration or asphyxiation by a foreign body). The only real advantage is the ability to deliver breaths asynchronously with the  In the event that one takes it upon themselves to instrument the airway while CPR is ongoing, the Australian ARC recommends the attempt should take no more than 30 seconds, or 5 seconds in NZ. 

On what do we base these recommendations? Observational studies are difficulty to interpret because the patient may have, or may have not, been intubated for completely explainable reasons, which would then go on to affect their outcome. Perhaps they were doing badly, and intubation was felt to be inappropriate. Or perhaps they were doing so well that it was felt unnecessary. Newell et al (2018) tried to review the existing data and came up with only a couple of randomised controlled trials (eg. the CAAM trial from 2018), all of which produced inconclusive results. Intubation is obviously not uniformly benign - apart from the added time of interrupted CPR, it is not without complications. For a disturbing example, 10% of the intubated CAAM patients had an oesophageal intubation.

Evidence for the depth of compressions

We don't know what the depth of compressions should be, or whether the depth correlates well with ventricular chamber compression. However, the deeper you go the more likely you are to have a successful defibrillation.  

Evidence for CPR following defibrillation

We know that after the restoration of electrical activity, cardiac output is still poor. CPR should continue for 2 minutes after each shock attempt.

Evidence for specific medications during cardiac arrest

We think there might be some benefit to amiodarone in shock-refractory VF; or rather, it seems slightly better than placebo. In the group of arrested patients being CPRed by ambulance officers, the use of amiodarone for "refractory" shock-resistant VF resulted in a slight (10% ARR) improvement in their chances of surviving the ambulance ride. Thats right, no talk of survival to discharge whatsoever.