This is the first entry in a new series called The Trip Report: Turning Research Into Practice. Watch for it regularly.
Beck B, Bray JE, Cameron P, Straney L, Andrew E, Bernard S, Smith K. Resuscitation attempts and duration in traumatic out-of-hospital cardiac arrest. Resuscitation, 2017 Feb; 111: 14-21.
This month’s research topic is trauma. While there are a lot of aspects to trauma, we are going to discuss a trauma-related topic that's usually not discussed in depth: traumatic cardiac arrest. The authors of the above-mentioned study examined this topic, and what they found might have validated some things we thought we knew and added some important information to our collective knowledge base.
The authors, led by Ben Beck, PhD, of Monash University in Australia, wanted to understand what factors played a role in a paramedic’s decision to attempt resuscitation for these patients, who are often considered futile. Their study also looked at how long resuscitation was attempted for these patients. To do this, they used data from a cardiac arrest registry. Registry data can contain a wealth of information but, as we will see in a bit, comes with some limitations.
One really impressive thing about this study, and the registry, was how long they collected data. This study was able to use data from as far back as 2008. While registries of all kinds (trauma, stroke, STEMI) are pretty common now, how many of us were still using paper PCRs in 2008? So the authors were able to use data from 2008 to 2014. This was a huge advantage since, relatively speaking, traumatic cardiac arrest is not that common.
The authors chose to exclude children (less than 16), drowning victims, hangings and all arrests of cardiac etiology. They still were able to identify more than 2,000 cases. That is a pretty large study population. This is very important. Smaller sample sizes can lead to misleading conclusions.
The EMS system had a protocol for withholding and/or ceasing resuscitation. This is hugely important, because without this protocol this study would likely not have been possible. The medics had to use their judgment and guidance from the protocol to decide to attempt resuscitation or not, as well as to determine when they'd done all they could. This let the authors really drill down and answer their research questions.
In total the authors found traumatic cardiac arrests accounted for a little over 7% of all out-of-hospital cardiac arrests. Also, they validated some things we suspected: Their traumatic cardiac arrest patients were younger (about 44 years old) and overwhelmingly male (68%). Their traumatic arrests happened more often in public places, and the most common cause was MVCs. Almost 80% of the initial rhythms were asystole. A really interesting statistic was that over 50% of traumatic cardiac arrests were bystander-witnessed, but only 20% of victims received bystander CPR. This could be due to the high percentage of MVCs and the difficulty of gaining access to entrapped patients. It could also be a sign that we need to emphasize bystander CPR even more than we do now.
Of the 2,000-plus cases, paramedics attempted resuscitation on 660 (less than 30%). The authors employed some statistical tests to analyze these data: They used univariate and multivariate logistic regression to identify factors related to beginning resuscitation efforts. Basically what this means is they looked at each factor first (univariate) with blinders on. They didn’t consider any other factor or how factors related to each other. So for instance, they looked into initial rhythm but didn’t consider mechanism of injury. Then they looked at mechanism of injury but didn’t consider initial rhythm. Yes, this has limitations, but it is common practice to start. Next they looked at all the factors available in the registry data and how they related to each other when impacting the choice to begin resuscitation (multivariable logistic regression). This can give a better estimate of the importance of each factor because nothing happens in isolation.
In the multivariable logistic regression model, they found some things that, again, we might have expected. Resuscitation was more likely to be started if EMS witnessed the arrest, if bystander CPR was started, if the initial rhythm was VT/VF or PEA, and if the mechanism was not an MVC. Rescuers also were more likely to begin a resuscitation if the patient did not have a prolonged downtime and was in an urban environment. None of this is probably surprising to you, but again, it is one of the first times it has been studied and reported. If it is not published and peer-reviewed, it is not accepted as scientific fact.
One important note is that the regression model did not pass an important test examining how the model describes the data, meaning it may not have been completely accurate. Typically this would be a challenge when trying to publish this study; however, since this is not a widely studied topic, the authors probably got a little extra leeway in reporting this information.
When looking into the duration of resuscitation efforts, they found that about a third of attempts lasted less than 10 minutes. These may have been cases where the medics had doubts about saving the patient but thought they should at least try, but then quickly determined their effort was, in fact, futile.
Another important finding was that there was no change over time in the percentage of patients who died on scene. That means we are not getting any better at caring for traumatic cardiac arrest. Given that we don’t study traumatic cardiac arrest, this may not be surprising. Overall, 95 of these patients had ROSC, and 25 survived to hospital discharge. Yes, those are not huge numbers, but those 25 patients represented 3.8% of the population where EMS started a resuscitation. Given that overall out-of-hospital cardiac arrest survival in this region was reported at 3.3%, maybe attempting resuscitation on traumatic cardiac arrest patients is not futile after all.
Every study has limitations. This one was done in a non-U.S. system, which should normally raise caution in applying its findings to American models. However, while there are differences between EMS in the U.S. and Australia, the systems are largely comparable. (For an explanation of “EMS Down Under,” see www.emsworld.com/10323796.)
A more important limitation here has to do with using registry data. The authors were limited in what they could examine to the data collected in the registry. So while they examined some important factors, they were unable to examine the severity of injury (likely a very important factor!). Again, the fact that their multivariable logistic regression model did not “fit” their data makes interpretation of their results difficult.
Nevertheless, the authors should be commended for examining an important and not-often-studied topic. This is one of the few studies available on traumatic cardiac arrest and significantly adds to our knowledge on the topic. It would be great to see this study replicated in a U.S. EMS system using more factors that are likely important (severity of injury, distance to a trauma center, etc.).
Antonio R. Fernandez, PhD, NRP, FAHA, is the research director at the EMS Performance Improvement Center and assistant professor in the Department of Emergency Medicine at the University of North Carolina - Chapel Hill.