You may have heard it said that the hardest thing about performing a resuscitative thoracotomy is the decision to do it.
You may also have heard it said that the above statement is nonsense; that if you find the decision hard, you probably don’t know the procedure well enough;
“the intention to do it should be very quick, and you should know the patients and the evidence, and have the confidence and competence to do this if it’s indicated.” John Hinds (1980-2015), Crack the Chest – Get Crucified
John also reminds us that first and foremost we should ask ourselves whether our intentions are honourable; is the patient salvageable?
Resuscitative thoracotomy is a dangerous procedure: there are sharps (both surgical and bony fragments) and blood – lots of blood. It’s a risky procedure for you as a clinician. It’s not a technically difficult procedure; there are plenty of FOAM resources which will help you mentally rehearse the steps required although these are best supplemented by a cadaveric course.
So… How can we determine which patients are appropriate for performing a resuscitative thoracotomy?
The recent paper – “an evidence-based approach to patient selection for emergency department thoracotomy: a practice management guideline from the Eastern Association for the Surgery of Trauma” attempts to answer the question exactly that question. The paper is open-access – click the link above or the graphic below to open the paper in a new tab or window.
What kind of study was this?
The Eastern Association for the Study of Trauma (EAST) group performed a systematic review of studies in which ED thoracotomy was performed with the intention of separating out the pre-procedure state of the patient and the resulting outcomes. The objective was to evaluate whether ED thoracotomy actually improves the outcome for the patient, when compared with resuscitation without thoracotomy, in a set of possible circumstances in which the patient might present to the ED.
Who or what was studied?
The papers included in the systematic review involved resuscitative thoracotomy occurring in the ED in the context of traumatic cardiac arrest. There were six defined populations according to whether signs of life were present or absent and whether the injury was penetrating (thoracic or extrathoracic), or blunt:
- Pulseless with signs of life after penetrating thoracic injury
- Pulseless without signs of life after penetrating thoracic injury
- Pulseless with signs of life after penetrating extrathoracic injury
- Pulseless without signs of life after penetrating extrathoracic injury
- Pulseless with signs of life after blunt injury
- Pulseless without signs of life after blunt injury
The authors were interested in ED thoracotomy as the intervention (prehospital and operating room thoracotomy were excluded) and compared with resuscitation that did not involve thoracotomy. They then looked at two key outcomes: hospital survival and neurologically intact hospital survival. What is interesting here is that the literature did not include direct comparison of outcomes (you’ll struggle to find randomised-controlled trials of ED thoracotomy) and so as a proxy measure the baseline risk of hospital survival for patients presenting pulseless to the ED was used, based on estimates by the subcommittee; effectively, an expert consensus on the likely survival of each PICO group had they received standard resuscitation (large bore IV access, blood and crystalloid infusion, thoracostomy tube placement and emergency surgery) without thoracotomy. How robust this method is for generating a proxy control group probability of survival is certainly debateable.
72 included studies (from 2152 identified) provided 10,238 patients. Table 1 tells us a little bit about those patients; just 871 of 10,238 “survived”, with 408 of the 6746 for whom neurological status was recorded surviving neurologically intact.
A little bit about systematic reviews
Systematic reviews are a little different from other published papers; they aim to combine the results of a number of studies on the same topic in an organised, pre-determined way, to collate the evidence and form broader conclusions. Systematic reviews are a significant undertaking (as you’ll know if you’ve written a CTR for the dreaded FCEM) and they can seem to be held in higher esteem than other published papers BUT their value is – naturally – dependent upon the quality of the study undertaken and it is as important to be able to appraise systematic reviews as it is other published papers: there’s a nice explanation of how to do that here.
It might be helpful to use a checklist, like this one over at BestBETs.org, when assessing the quality of systematic reviews. Systematic reviews should be… well, systematic: it should be clear from the methodology;
- What the population, interventions, comparison and outcomes of interest were (a PICO question)
- What sort of papers would be included, together with details about how they would be identified, and
- How they would be assessed for methodological quality
There should also be other information in the methods section about how interventions and outcomes were defined for inclusion and exclusion.
Systematic reviews will often include pre-specified assessments of methodology; in this paper, the authors used the Grading of Recommendations, Assessment, Development and Evaluation (GRADE) methodology to assess the strengths, weaknesses, bias and inconsistencies of each included paper.
What did they find?
The findings are divided into recommendations with relative risks of survival for each of the population categories.
Pulseless with signs of life after penetrating thoracic injury
Recommendation: EDT strongly recommended
RR of surviving hospitalisation 7.6 (182 of 853 patients surviving)
RR of surviving neurologically intact 4.7 (53 of 454 patients surviving neurologically intact)
Pulseless without signs of life after penetrating thoracic injury
Recommendation: EDT strongly favoured
RR of surviving hospitalisation 41.3 (76 of 920 patients surviving)
RR of surviving neurologically intact 19.5 (25 of 641 patients surviving neurologically intact)
Pulseless with signs of life after penetrating extrathoracic injury
Recommendation: EDT conditionally recommended (this group excluded isolated cranial injuries and the recommendation recognises that anatomical injury site is important and that not all are equal in terms of prognostication)
RR of surviving hospitalisation 9.2 (25 of 160 patients surviving)
RR of surviving neurologically intact 11.0 (14 of 85 patients surviving neurologically intact)
Pulseless without signs of life after penetrating extrathoracic injury
Recommendation: EDT conditionally recommended
RR of surviving hospitalisation 28.8 (4 of 139 patients surviving)
RR of surviving neurologically intact 55.7 (3 of 60 patients surviving neurologically intact)
Pulseless with signs of life after blunt injury
Recommendation: EDT conditionally recommended
RR of surviving hospitalisation 9.3 (21 of 454 patients surviving)
RR of surviving neurologically intact 7.8 (7 of 298 patients surviving neurologically intact)
Pulseless without signs of life after blunt injury
Recommendation: EDT conditionally not recommended (the subcommittee group were unable to agree on the strength of the unanimous recommendation against EDT)
RR of surviving hospitalisation 704 (7 of 995 patients surviving)
RR of surviving neurologically intact 202 (1 of 825 patients surviving neurologically intact)
A Sidestep into Relative Risk
For a quick refresher on relative risk, try here.
What we see in the latter population groups demonstrates what happens to relative risk when a rare and unlikely outcome (survival, or neurologically intact survival) is observed in the context of a reasonably rare exposure (EDT in the context of injuries for which EDT might not traditionally have been deemed appropriate).
Remember, relative risk is calculated by dividing the experimental event rate (EER) by the control event rate (CER). In this paper, the control event rate was the predicted survival rate as agreed by expert consensus for each population group. Mathematically, relative risk is expressed as:
Relative risk (RR) = EER / CER
Where both event rates are small (because the subject numbers are small), the absolute risk reduction (CER – EER) will also be small – the closer the CER and EER, the smaller the absolute difference between the groups – but the relative risk gives a larger figure because we have compensated for the low base rate by dividing by the CER. For this reason number needed to treat (1/ARR) is often used in place of relative risk measures as this provides an absolute treatment effect size and a number which is easier for us as clinicians to understand.
It’s not clear to my why the authors have chosen to use relative risk measures in the written manuscript of the paper. What I really want to know is how many patients I would need to perform EDT on in order to have one extra patient survive their traumatic cardiac arrest (ideally, neurologically intact). These numbers are in the paper but they are in the tables rather than in the main body. Relative risk is often reported because it makes better headlines: for example, “more than four times as likely to survive neurologically intact” sounds better than “an extra 92 survivors who are neurologically intact for every 1000 extra patients undergoing ED thoracotomy” (figure 1: EDT in patients presenting pulseless with signs of life after penetrating thoracic injury).
Compare this with figure 6 (EDT in patients presenting pulseless without signs of life after blunt injury) – the relative risk is 202 because the control event rate was almost 0 (the authors estimated it at 0.0006%), so we could tell our colleagues that performing EDT in these patients means they are 202 times more likely to survive neurologically intact! That’s great, right?
Well, it’s not the full story. These patients are unlikely to survive no matter what we do. That single survivor who was neurologically intact has pushed the relative risk right up but look at the risk differences with EDT (which looks like a number needed to treat to me) – we would need to perform EDT in 1000 extra patients with this presentation to have an extra one survivor who is neurologically intact (and we might add an extra seven who have residual neurological deficit). Is this benefit worth the risk to healthcare workers of blood-borne virus exposure? Is it worth the risk of adding those extra patients with resultant neurological deficit? Is survival the ultimate goal or should likely quality of life play a part in our decision-making? These are questions we need to ask ourselves in understanding the impact of the paper and they aren’t easy to answer.
So let’s get back to the paper.
How does this paper change clinical practice?
There are a few problems with this study. For a start, the control event rate for each population group has been agreed by expert consensus and I have concerns that this is not particularly robust.
The authors have identified that there is serious risk of bias in all of the papers they included in their analysis and serious risk of imprecision in the later groups (4 & 6). Patients in traumatic cardiac arrest are a pretty heterogenous cohort, data tends to be collected retrospectively and against-the-odds survival is more likely to be reported than another death. But are we likely to be able to collect better data? It’s tricky, because there are so many factors related to injury survival – many of which we probably don’t appreciate yet – and the chances of a randomised controlled trial into ED thoracotomy are… slim :-).
I’m not sure this paper changes my practice at all; if patients are pulseless with signs of life in the context of penetrating thoracic injury causing traumatic cardiac arrest then thoracotomy is potentially life-saving (we knew this already) and for other groups it is helpful to see in writing prognostication we already felt we knew; extra-thoracic penetrating injury carries a worse prognosis than thoracic injury, loss of signs of life is bad, blunt injury is bad.
The recommendations made by the authors are helpful in as much as we can see neurologically intact survival in all groups – this doesn’t justify cracking every chest (of course not!) but in the hands of experienced trauma team leaders there’s an extra piece of evidence to reflect on when making those decisions, which may or may not be difficult after all.
Before you go please don’t forget to…