Mechanical CPR (mCPR) is widely used as a tool in cardiac arrest management. There are several good arguments to support its use of course, it does not get tired, it can reduce the number of people needed to run an arrest, it can be used to facilitate other procedures such as invasive access/ECPR and it is safer to delver when the patient is being transported (especially where the ergonomics of transport make manual CPR difficult. I’ve also heard it said that organisations may get fewer legal claims for bad backs and injury during transport through the use of mCPR.
In Virchester mCPR is considered to be an extra member of the team. We use a pit stop model of cardiac arrest management which requires a certain number of people. If mCPR (we use the LUCAS device) is available then one fewer person is required for the team. This certainly makes practical sense but does it lead to better outcomes?
The patient outcome data for mCPR is pretty weak. A Cochrane review of mCPR published in 2018 found little definitive evidence for benefit and concluded that mechanical and manual CPR were probably equivalent. Similarly a recent systematic review of 15 trials and thousands of patients found no benefit to mCPR. In the FOAMed world both ourselves and theSGEM have come to similar conclusions.
Despite the lack of evidence for improvements in patient outcomes, the practical advantages have led to the widespread use of mCPR in practice. However, there are exceptions. Ambulance Victoria which has an enviable record for OOHCA management and who are a leading research organisation in OOHCA don’t use mCPR and they have some of the highest success rates in the world.
What is the real world experience though? Can we learn anything from registry data on the impact of using mCPR in practice? This month there is an interesting paper in Resuscitation that compares outcomes for in hospital cardiac arrest patients as it is associated with the use of mCPR. The abstract is below, but as always please do read the article yourself and come to your own conclusions.
What kind of paper is this?
This is a retrospective observational study where routinely collected data is examined to look for trends and associations. These studies are convenient and often fairly quick to do as the data is already there, but they are also prone to signficant bias. The data may not be as ‘good’ as in prospectively collected trials and patients may receive the intervention of interest (in this case mCPR) as a result of factors that are unrecorded or hidden. Such effects are partially nullified in randomised controlled trials which is why they the preferred method of evaluating interventions. That said, observational studies can sometimes help us understand the real world impact of interventions once they hit clinical practice. It is also notable that much of the research into mCPR has been done in the pre-hospital setting, and yet it’s use in other settings has spread and many in-hospital resuscitation teams now deploy mCPR as part of their approach.
What did the authors do?
The trial examines routinely collected data from 153 hospitals in the United States. They were able to examine clinical data on in-hospital patients who suffered a cardiac arrest., including whether mCPR was used. They recognised that the patients were heterogenous and so tried to adjust for this using multivariate statistics and propensity matching. These techniques adjust the outcome found according to known confounding factors.
Tell me more about the patients.
Patients were included if they had an in-hospital arrest that lasted for more than 5 minutes, or longer than 120 minutes. Arrests in the emergency room were excluded. This left over 111,143 patients suitable for analysis of whom 2,232 received mCPR.
What did they find?
Of those who received mCPR, 264 patients (11.8%) survived until discharge compared to 18,351 patients (16.9%) in the manual CPR group.
The multi-variable analysis also showed worse outcomes for the mCPR group (OR 0.57 95% CI 0.46–0.70, p < 0.01). The propensity score analysis also showed worse outcomes for those in the mCPR group (OR 0.68 95% CI 0.44–0 0.92, p < 0.01)
The authors then did several sensitivity analysis that similarly showed statistically worse outcomes for mCPR.
So on the data available to the authors, whichever way they analysed it, the outcomes for mCPR are worse.
So should we stop using mCPR in-hospital.
Whilst the difference in outcomes are stark we need to be cautious about the findings as this is a retrospective cohort study. This may lead to significant bias. The authors looked a large range of potential variables and adjusted for these, but not all potential variables could be accounted for. I am always very cautious about transferring this data to practice, but that said, it is available and we can appraise it.
Notably mCPR is more likely to be deployed the longer an arrest continues. There will therefore be a bias towards its use in longer arrests which are by their very nature less likely to be successful. This bias is common in many observational studies in cardiac arrest as the longer an arrest continues, the more likely a more advanced intervention is to be deployed, and yet the chances of survival are falling. The time of mCPR deployment was not part of the modelling in this study, although they did do a sensitivity analysis across different arrest duration which again showed no benefit.
Arguably the authors tried a number of methods of data analysis to account for differences between the observed groups but all of these showed no benefit, and indeed harm.
If we put this data together with what we already know about mCPR in prehospital settings, then the likely benefits of mCPR in hospital seem to be low, and may even be harmful. We also need to consider that in hospital settings there is usually no requirement for movement during CPR, and there are almost always lots of people around to do manual CPR. The question then comes down to whether the quality of manual CPR is good enough, though arguably we also need to observe and critique mCPR too as that’s not always well positioned or effective. Slightly off topic, but when I have an art line placed, and am using the LUCAS prehospital (or in hospital) it is really noticeable that poor positioning makes big differences to the observed blood pressure/trace. One of the key outcomes of placing an intra/peri-arrest art line is optimising CPR whether that is with manual or mechanical CPR, but the lesson here is that mCPR is not always optimised, and may be more difficult to ‘judge’ it’s effectiveness by the team leader/operator.
The bottom line
The bottom line is that the evidence for mCPR in terms of patient benefit is poor. There are good reasons to use it in certain environments and when transporting patients. It is questionable whether those practical reasons exist for most in-hospital arrests.
References
- Mechanical chest compression machines for cardiac arrest. https://www.cochrane.org/CD007260/VASC_mechanical-chest-compression-machines-cardiac-arrest
- Effectiveness of Mechanical Chest Compression Devices over Manual Cardiopulmonary Resuscitation: A Systematic Review with Meta-analysis and Trial Sequential Analysis. https://pubmed.ncbi.nlm.nih.gov/35353993/
- SGEM#136: CPR – MAN OR MACHINE? https://thesgem.com/2015/11/sgem136-cpr-man-or-machine/
- Simon Carley, “JC: PARAMEDIC trial m-CPR at St.Emlyn’s,” in St.Emlyn’s, November 21, 2014, https://www.stemlynsblog.org/jc-paramedic-trial-m-cpr-st-emlyns/.
- Record survival rates for cardiac arrest in Victoria https://www.ambulance.vic.gov.au/record-survival-rates-for-cardiac-arrest-in-victoria/