Good quality, practice-changing prehospital research is difficult to come by for a number of reasons, not least the fact that the standalone specialty itself is in relative infancy. Issues of consent and ethics, combined with the very urgent nature of most of the prehospital care we are interested in undertaking research into, make it very tricky to design and conduct a robust study in the prehospital arena, leaving aside the fact that most of the outcomes of interest occur in hospitals which may be run by entirely different agencies who may (or may not) be cooperative but who will, in any case, need to buy into the aims and methods of the study.
So it’s been great to see two papers electronically published in the last couple of weeks in prominent journals, both of which were conducted in the prehospital environment and both of which address controversies in the care of patients in cardiac arrest.
Amiodarone, Lidocaine or Placebo?
First up, NEJM published this study [pubmed link] which looked at the efficacy of anti-arrhythmics (specifically amiodarone or lidocaine) in improving outcomes for patients with out-of-hospital cardiac arrest, presenting with shock-refractory VF or VT.
What is this paper about?
This was a randomised, double-blind trial of patients aged 18 or over with non-traumatic cardiac arrest where the rhythm was either shock-refractory VF/pulseless VT or recurrent VF/pulseless VT (recurring after initial successful defibrillation/cardioversion). In order to be included patients had to have received at least one shock (use of an AED was included in this definition) and have IO/IV access for administration of the trial drugs.
Patients were excluded if they received either amiodarone or lidocaine before they were enrolled. Like many pre-hospital studies, especially those around cardiac arrest, the trial was conducted with exception from informed consent.
The authors wanted to examine this population to ascertain whether administering amiodarone, lidocaine or placebo (as the control) affected the patients’ outcomes; they selected survival to hospital discharge as their primary outcome.
What did they do?
Patients who met the criteria above were randomised to receive one of the following:
- 150mg amiodarone
- 60mg lidocaine
- Normal saline
The drugs and placebo were identically packaged in unlabelled kits of three syringes each containing 3ml of colourless liquid. This is a pretty reasonable attempt at blinding the administering EMS providers (prehospital team/paramedic equivalents). The research team distributed the kits among 55 EMS agencies who participated in the study in a ratio of 1:1:1. Randomisation used stratified permuted blocks, a strategy utilised to try to iron out differences between the centres and services recruiting patients.
Once a patient had been recruited, their trial kit was opened and two syringes of medication (or placebo) were administered (so 300mg amiodarone, 120mg lidocaine or 6ml normal saline) unless the patient weighed less than 45kg in which case a single syringe was administered. If VF/pulseless VT persisted beyond the initial dose of anti-arrhythmic, further shocks were given and the third syringe administered.
Care beyond the hospital front door was not determined by the researching team although they did try to measure it. The outcome of interest, survival to hospital discharge, was determined from medical care records. This might cause us some problems; as clinicians we know that survival post-cardiac arrest is not simply about having a functioning brainstem. It is not unreasonable to argue that quality of life is a key component of determining a good outcome from cardiac arrest; the authors did look at this in their secondary outcomes (survival with favourable neurological status, defined as a modified Rankin score of 3 or less), but the study is not powered to detect a difference in this outcome, only the primary outcome.
That said, a power calculation was undertaken. To detect a difference of 6.3 percentage points in the rate of survival between groups with 90% power they needed to recruit a sample of 3000 subjects in the per protocol population, 1000 in each group. The authors set out to undertake both per protocol and intention to treat analyses which is good practice for trials like this where unforeseen practical consequences of adhering to the study protocol might cause a significant proportion of protocol violations and lead to a large number of results which cannot be analysed.
What did they find?
Firstly, they found that only a small proportion of patients with out-of-hospital cardiac arrest had shock-refractory or recurrent VF/pulseless VT – 18.6% of their total population. This is actually a bit higher than I would expect – it certainly doesn’t seem as though that many of the patients we see in the ED in cardiac arrest are in rhythms other than PEA or asystole.
Although they managed to recruit more than 3000 patients (3026 in fact), they were not evenly distributed among the three groups for the per protocol analysis so the study was underpowered to detect the predetermined difference between the groups. The authors tell us that the first dose of trial drugs was given a mean of 19.3+/- 7.4mins after the initial call to EMS and after a median of three shocks, which reflect the current ALS algorithm used in the UK (although lidocaine is not routinely used).
According to the per protocol analysis, survival to discharge was determined in 24.4% of the amiodarone group, 23.7% of the lidocaine group and 21.0% of the placebo group with an absolute risk difference between amiodarone and placebo of 3.2 percentage points but with a 95% confidence interval of -0.4 to 7.0. As the confidence interval crosses zero, we can’t conclude that the study provides statistical evidence of a significant difference between amiodarone and placebo (or lidocaine and placebo for that matter) in this context but this does not, of course, preclude a clinically significant difference.
But what we are really interested in is the patients who do survive – how many have a reasonable functional status? We can also see that the drugs did do something – patients who weren’t receiving placebo were statistically significantly less likely to require an extra dose and further defibrillation.
We can also see the positive influence of witnessed arrest in the subgroup analysis, whether witnessed by bystander or EMS provider, with significantly higher survival proportions with the active drugs than placebo in both subgroups.
What does this mean?
We can conclude that out-of-hospital cardiac arrest has a low survival rate, something which we already know (but it’s always helpful to have more evidence). Survival to discharge with a good neurological outcome is even more rare, not that I would particularly consider a modified Rankin score of 2 a good outcome (I concede that might just be me).
The authors conclude that neither amiodarone nor lidocaine resulted in a significantly higher rate of survival to hospital discharge or favourable neurological outcome than the rate with placebo for these patients, which is a fair conclusion from the data they have obtained. But does this mean we should abandon amiodarone altogether?
I would suggest probably not (as it’s always difficult to justify such a dramatic change in practice from a single paper, even with a large sample but especially if the study is underpowered). Let’s not forget that clinical significance and statistical significance are not the same thing – even though there is not a statistically significant difference in the groups we need to understand whether a clinical difference is represented. We need to find some common ground between generating more survivors (who have increased care needs and low quality of life) and meaningfully resuscitating people who can go on to have relatively normal lives. I don’t think the answer will be an easy one, nor one devoid of emotional undertones.
How Long Should We Keep Up CPR?
This paper from Circulation [pubmed link] is interesting. I don’t know much about how medicine is practiced in Japan and this paper does a good job of outlining some of the key components of prehospital care. What is particularly interesting is that EMS providers are not allowed to discontinue resuscitation unless ROSC occurs and must commence resuscitation attempts immediately unless the patient is “obviously moribund” (I am assuming this means “well dead”).
What is this paper about?
The authors were interested in understanding prehospital resuscitation duration for neurologically intact survival, dividing their out-of-hospital cardiac arrest patients into four groups according to whether the rhythm was shockable or non-shockable and whether bystander CPR was performed or not.
What did they do?
They undertook an analysis of data from a country-wide registry of out-of-hospital cardiac arrest, including all patients who were transported to hospital and excluding those <18 years of age, those with cardiac arrest occurring after EMS responder arrival, those where key components (bystander presence, witness of arrest, rhythm) could not be ascertained and those in whom a DNAR order was in place.
They determined their outcome of interest as cerebral performance category score at 30 days post discharge, with a score of 1 or 2 being interpreted as a favourable neurological outcome.
The authors then looked at the relationships between the category of the patient (according to their four groups) and the call receipt to ROSC time, including a multiple logistic regression analysis for independent predictors of favourable 30 day neurological outcome.
What did they find?
They had a huge number of patients in the study – 910,581 patients were transported to hospital with prehospital resuscitation attempts during the study period, with only 282,183 being included in their analysis. The majority of exclusions occurred because the cardiac arrest was not witnessed (529,092 were excluded for this reason).
The four groups were:
- Shockable rhythm plus bystander CPR 23,028 (8.2%)
- Shockable rhythm, no bystander CPR 19,383 (6.9%)
- Non-shockable rhythm plus bystander CPR 104,212 (36.9%)
- Non-shockable rhythm, no bystander CPR 135,560 (48.0%)
Favourable neurological outcome at 30 days was seen in the following proportions:
- Shockable rhythm plus bystander CPR 4,601 (20.0%)
- Shockable rhythm, no bystander CPR 2,556 (13.2%)
- Non-shockable rhythm plus bystander CPR 4,538 (4.4%)
- Non-shockable rhythm, no bystander CPR 1,533 (1.1%)
The authors also found that the best chance of having a good neurological outcome occurred if the patient achieved ROSC in the prehospital environment – 83.1% had prehospital ROSC, whereas only 0.7% of those not achieving prehospital ROSC had a favourable neurological outcome.
They also describe a relationship with duration of resuscitation: regression showed that the likelihood of favourable 30 day neurological status decreased for every minute that prehospital resuscitation attempts continued (R²=0.295-0.037, P<0.001). They have undertaken some exploration around what they are calling “sensitivity” of resuscitation time in predicting a good neurological outcome and conclude that to attain a sensitivity of 99% for good neurological outcome resuscitation must be continued for at least 40mins. This doesn’t quite sit right with me; I suspect what might be more appropriate here is to conclude that if resuscitation has been continued for 40mins or more there is a very, very low chance that the patient will be discharged from hospital with good neurological function, rather than mandate 40mins resuscitation for everyone.
What does it mean?
The authors’ conclusion – that resuscitation efforts should be continued for at least 40mins from call receipt (including 33mins of EMS responder resuscitation) in all adult patients with bystander-witnessed out-of-hospital cardiac arrest – is a little bit cheeky. This is an answer to a question they haven’t really asked and I’m not completely convinced that they have employed appropriate statistical methods in coming to this conclusion.
What we can say from their data is that:
- Survival is most likely in arrests with shockable rhythm and bystander CPR.
- The longer it takes to achieve ROSC, the less likely the patient will survive with a good neurological outcome
- If the patient hasn’t achieved ROSC by the time they arrive in the hospital, it is very unlikely they will do well
I don’t think any of the above points are going to be news to any practitioners. I don’t agree with the authors’ conclusions but I think there’s a way to reframe them which does align with my current practice; if resuscitation for out-of-hospital cardiac arrest has been undertaken for more than 40mins, we can conclude that it is unlikely the patient will leave hospital with good neurological function and recognise that it is probably time to stop.
In my practice I would not usually discontinue resuscitation attempts immediately on arrival to the ED anyway – let’s not forget that everything about medicine is harder when you are trying to undertake it outside the hospital (be it in the street, a helicopter or the back of an ambulance) and it can be very helpful for all concerned to run a few cycles of resuscitation attempts in the ED to address any potential reversible causes before discontinuing. It’s useful to model that the hospital care is often no different from the tremendous job our prehospital colleagues do (often in tougher circumstances). It might also allow family members some time to spend with their loved one while resuscitation is ongoing which can be very helpful for the grieving process (more on this in an upcoming post!).
What Can We Take From These Studies?
Let’s remember first and foremost that prehospital research is difficult – we should salute and thank these authors for working hard to contribute to our understanding of prehospital practice.
What both of these papers emphasise is the importance of doing the simple things well. The chain of survival, simple as it is, is still relevant to clinical practice. A great example of how to put this into practice comes from the NW Resus Group, a fantastic paramedic-led initiative which brings together clinicians of all disciplines (including police and fire services) to break down and work on the key skills needed to give our patients the best chance of survival. If you’re based in the NW of England, please try to get along to support them at one of their meetings.
Key BVM messages: correct position, optimise seal with 2 hand technique, single hand squeeze for 2 secs. pic.twitter.com/nHWxhOBrVu— NWRG (@nwresusgroup) March 23, 2016
We should also consider the benefits of bystander CPR as a reason to promote CPR training to the public – this great initiative in New South Wales does exactly that. And of course – if you, as a healthcare practitioner who regularly updates your BLS skills, are not registered with GoodSam, PLEASE DO IT NOW. (Thanks, you’re awesome)
P.S. If you were hoping for a post full of puns and references to loose seals, sorry!
Before you go please don’t forget to…