By Tom Heaton
It's a sunny weekday morning and the surgeons are elbow deep in an emergency laparotomy. The anaesthetic jobs are done and the patient is stable, needing only the regular glances at the monitors as input from us. The emergency list has been going well and I have had the chance to further practice my intubation skills. As a fairly new trainee at this point, I seize this lull in the otherwise busy day to make the most of the direct one-to-one consultant teaching available to us.
“Can I ask a bit more about etomidate as an induction agent? I’ve not actually seen it used yet.”
Silence descends upon the operating theatre. A peal of thunder rumbles ominously in the distance (I thought it was sunny!). Even the surgeons look up from the table, eyes widening above their facemasks. Someone drops a scalpel.
“Can I ask a bit more about etomidate as an induction agent? I’ve not actually seen it used yet.”
Silence descends upon the operating theatre. A peal of thunder rumbles ominously in the distance (I thought it was sunny!). Even the surgeons look up from the table, eyes widening above their facemasks. Someone drops a scalpel.
I don’t think that the anaesthetic textbooks had quite prepared me for the reactions that I experienced when I brought up the topic of this much feared agent. Dramatic story telling aside, I have still yet to find a consultant who uses etomidate any more, with a good selection of descriptions given to it (‘Etomideath’ is probably the most common nickname, whilst ‘Semen of the Devil’ remains my favourite). Yes I had heard a little bit about this whole adrenal suppression thing, but I’d also read about its favourable cardiovascular profile, making it a useful agent for inducing the sick, hypotensive patient. However, I had clearly not got a clear grasp of the balance of these pros and cons. Indeed I believe I am fairly lucky to have even cast eyes upon a vial of the stuff (locked tightly away in the deepest darkest theatre vault) and yet I do hear stories of its continuing use. As such, all this really peaked my interest in finding out a bit more about the evidence behind etomidate’s fall from grace.
When looking for a starting point on this topic, the meta-analysis of etomidate as an induction agent in septic patients seemed a fine place to start. Published in ‘Critical Care Medicine’ way back in 2012, this review aimed to pull together all the data on etomidate and try and definitively answer the question that remained about its use as a one off dose induction agent:
Etomidate Meta-Analysis: Critical Care Medicine
Etomidate Meta-Analysis: Critical Care Medicine
Tell Me About The Paper
The meta-analysis focused primarily on looking at mortality associated with a single dose of etomidate for induction of adult patients with sepsis who needed intubating. They included a secondary analysis looking for evidence of adrenal suppression in these patients, and trying to establish a relationship between the two. The authors set themselves some pretty strict inclusion criteria to try and use only the best quality evidence, and indeed have been (as far as I can tell) pretty thorough from this point of view, going to great depths to analyse the quality of the data they were drawing upon. They came up with 10 articles for inclusion in the meta-analysis; 5 for looking at mortality and 7 for looking at adrenal suppression, altogether involving 1,623 patients.
What Did They Find?
With regards to mortality, the five studies they used contained 865 patients. The studies consisted of one prospective observational cohort study and four randomised control trials (RCTs) of varying quality. The results they found showed a statistically significant higher risk of death in the patients who received etomidate as an induction agent compared with a selection of other agents. The pooled risk ratio for these patients was 1.20 with 95% confidence intervals of 1.02-1.42.
With regards to adrenal suppression, the seven studies they used included 1,303 patients. Three of the studies were RCTs, three were retrospective studies and one was a prospective study. The results confirmed the relationship between etomidate and adrenal suppression in sepsis. The pooled risk ratio for these patients was 1.33, with 95% confidence intervals of 1.22 to 1.46.
On the surface of it then, some pretty compelling results about etomidate in sepsis. The relationship between etomidate and adrenal suppression was reconfirmed, with a demonstration of increased mortality too, something that the previous studies had failed to convincingly demonstrate. Let’s have a closer look at the paper and see what we can find, though I admit I am mainly limiting it to looking at the mortality side of things, given that this is the outcome that remains a bit contentious.
With regards to adrenal suppression, the seven studies they used included 1,303 patients. Three of the studies were RCTs, three were retrospective studies and one was a prospective study. The results confirmed the relationship between etomidate and adrenal suppression in sepsis. The pooled risk ratio for these patients was 1.33, with 95% confidence intervals of 1.22 to 1.46.
On the surface of it then, some pretty compelling results about etomidate in sepsis. The relationship between etomidate and adrenal suppression was reconfirmed, with a demonstration of increased mortality too, something that the previous studies had failed to convincingly demonstrate. Let’s have a closer look at the paper and see what we can find, though I admit I am mainly limiting it to looking at the mortality side of things, given that this is the outcome that remains a bit contentious.
Is It Any Good?
Let’s start with a look at the plus points. Firstly, I think the review meta-analysis nature lends it a good deal of weight, especially when you have a look at how it has been put together. The authors have been pretty rigorous with their design to try and bring together the highest quality data available to them; for example, limiting themselves to only prospective trials when looking at mortality to try and reduce bias. They have been explicit in their search design and exclusion criteria and you can clearly follow the path to the final trials that they have ended up with.
The authors then go further from this starting point and apply multiple tests to their source information to further appraise its quality. I hadn’t heard of the Jadad scale prior to reading this paper, never mind the modified Jadad scale, but it is apparently a method of scoring papers to attribute a numerical value to their quality (based on blinding, randomisation etc. 1 is rubbish, 7 is great). This was one method the authors used to appraise their starting papers, going on to evaluate study heterogeneity, potential publication bias, and look at how study quality might affect the results by doing sub-group analysis.
The authors then go further from this starting point and apply multiple tests to their source information to further appraise its quality. I hadn’t heard of the Jadad scale prior to reading this paper, never mind the modified Jadad scale, but it is apparently a method of scoring papers to attribute a numerical value to their quality (based on blinding, randomisation etc. 1 is rubbish, 7 is great). This was one method the authors used to appraise their starting papers, going on to evaluate study heterogeneity, potential publication bias, and look at how study quality might affect the results by doing sub-group analysis.
In a very positive sense I think that all this analysis allows you to have a clear grasp on what starting data was being used to calculate the final results and therefore draw some of your own conclusions. For example, one of the studies used in the meta-analysis for mortality was just a prospective observational cohort study with a modified Jadad score of 1. If we don’t include this paper and only look at the RCTs, we get only a slightly smaller group of patients (759) but similar results with a risk ratio of 1.26 and 95% confidence intervals of 1.06-1.50. The authors have done this subgroup analysis a few times to give us more information to base conclusions upon.
Whilst a lot of this detailed analysis works to increase your confidence that etomidate does kill people, some of it urges a bit more caution. The analysis for potential publication bias is a notable point when the authors use the ‘trim and fill’ method to compensate for potential publication bias. If we include all five of the trials on mortality then it looks much the same, but if we are only looking at the four RCTs, the results slip back towards non-significance with a pooled risk ratio of 1.10 and 95% confidence intervals of 0.97-1.22. Is it just a case that there is a bit of publication bias with all the boring papers showing etomidate is safe languishing in a draw somewhere?
Whilst a lot of this detailed analysis works to increase your confidence that etomidate does kill people, some of it urges a bit more caution. The analysis for potential publication bias is a notable point when the authors use the ‘trim and fill’ method to compensate for potential publication bias. If we include all five of the trials on mortality then it looks much the same, but if we are only looking at the four RCTs, the results slip back towards non-significance with a pooled risk ratio of 1.10 and 95% confidence intervals of 0.97-1.22. Is it just a case that there is a bit of publication bias with all the boring papers showing etomidate is safe languishing in a draw somewhere?
There is one more worrying point about the final results of this paper that comes from looking closely at the trials that have gone in to the meta-analysis. This is most clearly noted when you look at the forest plots for the mortality results. Only one of the papers used had, by itself, shown a statistically significant outcome with regards to etomidate causing death, the others showing no statistically significant difference. However, this paper (Cuthbertson et al. 2009) contributed 37.66% of the weight to the final meta-analysis outcome, and 44.60% if you are only looking at the RCTs. That is a lot of weight of the meta-analysis calculations resting on a single RCT. Closer inspection reveals that this isn’t really a true RCT of etomidate vs alternative induction agent, but actually a sub-study of the CORTICUS trial that was an RCT of Hydrocortisone vs placebo in septic patients. Hmmm… does that count as an RCT?
Final Thoughts
Where does this leave us? After looking through this particular meta-analysis I am less sure of the strength of the evidence between etomidate and death, primarily due to the weight that is put on one single, potentially less than ideal RCT (though the publication bias analysis also causes a pause for thought). That said, the article does flag up a worrying picture overall and from my (inexpert) reading about the subject, the link between etomidate and adrenal suppression (if not mortality) seems pretty conclusive now. The authors have done a good job of bringing all the data together and analysing it as best as possible and the link between etomidate and increased mortality in sepsis is certainly visible, even if not conclusively so.
I wanted to wrap up with a few thoughts that I have borrowed from some of the people I have been talking to about this. I think perhaps the most important question to ask is: do the known benefits outweigh the potential risks? The overwhelming answer I’ve got back is no. Ketamine seems to be shamelessly hogging the limelight for these sorts of patients at the moment, again for its favourable cardiovascular profile, but I'm repeatedly told by my consultants that even careful and appropriate use of the core induction agents (thio and propofol) can minimise the cardiovascular insult of induction. If we can use these, why do we need to take the risk of using etomidate?
I wanted to wrap up with a few thoughts that I have borrowed from some of the people I have been talking to about this. I think perhaps the most important question to ask is: do the known benefits outweigh the potential risks? The overwhelming answer I’ve got back is no. Ketamine seems to be shamelessly hogging the limelight for these sorts of patients at the moment, again for its favourable cardiovascular profile, but I'm repeatedly told by my consultants that even careful and appropriate use of the core induction agents (thio and propofol) can minimise the cardiovascular insult of induction. If we can use these, why do we need to take the risk of using etomidate?
That is it from me here. Thank you once again for reading, and I hope it has been relatively interesting. For a more expert appraisal on the subject there is quite a bit out there and to which I have included a handful of links below. As always, please let me know your opinions on the topic, especially if etomidate is something you see a bit in clinical practice or know of some other good resources about it. If I find time I might try and look at some of the other stuff out there on etomidate and I encourage you to do the same and post stuff on here. Until then, thanks for reading.
Tom Heaton
Tom Heaton
Links and Resources
Etomidate Overview - The LITFL page on etomidate
Intubation, Hypotension and Shock - The LITFL page on induction of the shocked patient.
Intubation, Hypotension and Shock - The LITFL page on induction of the shocked patient.
Etomidate: Unsafe for Intubation? - A more concise look at the paper by PulmCCM
Etomidate and RSI - A great little post by 'The Skeptics Guide To Emergency Medicine' looking at some of the problems with the meta-analysis
Initial image courtesy of Victor Habbick/FreeDigitalPhotos.net. Edited by Tom Heaton
RSS Feed
