If you’re an evidenced-based medicine advocate, you could be forgiven for being a bit confused about how to manage an airway in out-of-hospital cardiac arrest (not that it’s any clearer in-hospital). The literature on this topic prior to this month was based mostly on retrospective studies that, while fine for hypothesis generation, didn’t shed much light on whether we should use endotracheal intubation (ETI) or a blind-insertion supraglottic airway (SGA). I summarized the state of that literature for EMS World here. I ended that article by mentioning two trials (PART and AIRWAYS-2) that were underway at the time that might answer this question. Well, they’re out now.
I’ve been reading and re-reading these studies since they published. I put my big nerd glasses on and dove into the supplements and data tables. I’d like to share some of my thoughts on them and try to make sense of how I’m going to incorporate their findings in my system. I’ll address PART, (Pragmatic Airway Resuscitation Trial), led by Dr. Henry Wang (@DrHenryWang)1in this post. I’ll address AIRWAYS-2, the second blockbuster study, in a separate post.
PART was a NHLBI-funded study conducted as part of the ROC group. It was a randomized, controlled trial with 27 different US EMS agencies of patients with all cause adult out-of-hospital cardiac arrest. 3,004 patients were randomized by agency in several month clusters to initial advanced airway management using either ETI (1,499 patients) or SGA with a King LT (1,505 patients). The primary outcome was 72-hour survival. They found that patients in the King group had better outcomes than those in the ETI group (18.3% vs 15.4%, adjusted difference 2.9%, 95% Confidence interval 0.2% - 5.6%). The main limitation I see with this study, as I attempt to generalized it to my practice, is the very low rates of ETI success. Only 51.6% had first pass ETI success vs 90.3% with the King.
This is a huge difference and, in my opinion, is likely a large part of the reason for the improved outcome with the King. My agency runs between 88 and 92% first pass ETI success in cardiac arrests (still not where we’d like it). I’d like to think we’re not the only agency like this. As much as I want to apply the results of large RCTs to my practice, I this this one difference is a deal-breaker for me and my agencies.
While the performance of ETI is better in my agency, I don’t think we should assume it is better in all agencies. In fact, I suspect this is actually a good reflection of actual performance at most agencies. The intent of the PART trial was to be pragmatic. In other words, reflective of actual practice rather than what is seen in systems with optimal performance. One of the reasons I think this trial reflects performance by the majority of systems is that our FPS with ETI was 44%
prior to beginning our intensive QI efforts, a level consistent with that seen in this study. One of the factors in our improvement was switching to Video Laryngoscopy. There were only 17 cases in this trial using VL vs 1,906 using DL. Perhaps if more agencies adopted VL their success rate would improve and, perhaps, survival.
While there were some agencies that were using CPR compression measurement devices, that data hasn’t yet been reviewed and wasn’t included in the paper. We don’t know the extent to which compressions were interrupted with either the ETI or King approach. Prior work has suggested that, at least with DL, compressions are frequently interrupted, something associated with worse outcomes. We may assume any interruptions, if present, were likely to have been more pronounced in those patients without successful ETI. I suspect that poor ETI performance may be a surrogate marker for poorer resuscitation practices overall. To be clear, there is no evidence for this, it is just my suspicion. Dr. Wang has indicated they will analyze this compression data in a subsequent project. I certainly look forward to reading that paper.
There were 27 agencies included in this study and they varied greatly in size. Although there were several large agencies (Milwaukee FD and MedStar both had over 120,000 annual runs), the median annual 911 volume was 7,100 (IQR 3,150, 23,450) per year. One of my suspicions that I’m looking for an opportunity to test is that agencies with fewer opportunities to intubate may perform worse. Again, no evidence that this occurred, but it does make me curious. This study does not report on the number of procedures performed by medics nor by agency. I would like to see a secondary analysis of this data that looks at the association between ETI success and outcome.
The time from EMS arrival to AAM insertion was shorter in the King group. This mightbe associated with improved survival. There can be two potential reasons for this. One is that some of the agencies involved allowed King insertion as a BLS skill and BLS providers arrived prior to ALS ones. In this case, it may actually be the system design that is impacting survival and not the AAM strategy. Alternatively, it is important to recognize that ETI is a more complex skill that is likely to take longer to perform and this may have been the difference.
There were 352 patients who had BVM ventilations only. Of these, 200 were in the ETI group and 152 in the King group. These patients were appropriately included in the primary intention-to-treat analysis; however, it does create a bit of a problem with interpretation. Of this group, more of the patients in the King group had characteristics associated with improved survival. It is possible that, by including them in the analysis, this moved the results in favor of the King. In fact, in a pre-defined ‘as-treated’ analysis without these patients, the improvement in survival went away (LT survival 16.0% vs 13.5% with ETI, Difference 2.5% (95% CI -0.2 to 5.2%).The characteristics of this BVM only group is found in eTable 4 in the supplement. Here is a summary of the larger differences:
One of the pre-defined secondary outcomes was the proportion of patients with aspiration pneumonitis. There was no significant difference in this outcome between the two AAM strategies (26.1% vs 22.3%, difference 3.7 (95%CI -2.1 to 9.6). This suggests that AAM with the King LT does not have higher aspiration risks compared with ETI, something that has often been suggested (including by me).
This is an incredible effort that goes a long way to helping us understand this problem. Dr. Wang and colleagues are to be commended for this outstanding effort. I can’t wait to see the additional hypothesis generating secondary analyses coming out of this data.
I think this is an accurate pragmaticlook at how mostEMS systems perform. In that, I think medical directors of systems who do not have robust QI systems with documented high ETI success rates without compression interruptionshould adopt an SGA-first strategy for cardiac arrest. If I were a medical director in a system that wasn’t able to achieve this QI oversight, whether because of lack of resources or too many paramedics, I would abandon an ETI-first strategy.
While I will continue to think about this, I am not planning on changing the ETI-first approach in my system based on this study. My mind is, however, open to change with further evidence or with a degeneration in our ETI performance.
I think future analyses of these, and other, data can attempt to deal with indication bias by excluding all patients with ROSC and awareness prior to the median time needed to begin AAM efforts.
I really, really want to see the association between ETI success and survival, both on an individual patient and agency level. I remain convinced this is a big part of the results in this paper.
So, those are my thoughts on this study. I’d love to hear your’s.
Jeff Jarvis, MD (@DrJeffJarvis)