Monday, 29 February 2016

Apneic Oxygenation During Intubation; Curbing the Enthusiasm.

The current enthusiasm for apnoeic oxygenation has been based more on physiologic plausibility & wishful thinking rather than any real science from good quality RCT’s.

One always needs to be a bit skeptical of physiologic plausibility. Of course it is an essential ingredient for assessing causality, but it is far from the only requirement. To be sure, I would imagine that one could think of a physiologic reason why we could fart ourselves to the moon.

Now we have the largest RCT investigating the efficacy of apnoeic oxygenation. A whole 150 patients.

Inclusion criteria were adults requiring intubation in a single medical ICU in Nashville, Tennessee. The intubator was a “pulmonary and critical care medicine fellow.” (This is probably about a PGY-4 doctor)

There were some exclusion criteria that probably meant that some of the sicker patients were not enrolled.

The study protocol was pragmatic and simulated “real world” ICU practice.  Therefore it governed only the provision of supplemental oxygen during apnoea. All other decisions about medications, devices, etc. were made by the clinical team.

Both groups could get pre-oxygenation. The patients randomized to apnoeic oxygenation got a high-flow nasal cannula set to 15L/min flow of 100% oxygen before induction and kept in place until the intubation was complete. The control group received no nasal cannula oxygen.

The primary outcome was the lowest pulse oximetry reading between induction and 2 minutes after successful intubation. Some might argue this is not the most patient oriented outcome but at least it is practical and easy to measure.

Results? There was no statistical difference between the two groups. The median lowest sat in the apnoeic oxygenation group was 92% vs. 90% (95% CI for the difference -1.6%-7.4%) in the control arm. Just for interest, they had only 2/3 success in first pass intubation rate.

This study was powered to have an 80% probability of detecting a 5% difference in the primary outcome if one truly existed in the population. All of the trends favoured the apnoeic oxygenation group and it is possible that the study was underpowered. However, one might question if such smaller differences are clinically meaningful. Probably not.

Limitations were several, but overall this was a well conducted study. Consider lack of blinding, single centre study, perhaps lack of intubator experience, issues of external validity, disease oriented outcome measure and some others.

So what are we to conclude? I don’t think this study is going to change anyone’s minds. True believers will still believe. But it certainly emphasizes the lack of good quality evidence to support the practice of apnoeic oxygenation. I personally don’t bother with it. But at the same time we should not have polarized views until better quality evidence provides us with some answers.


Semier MW, Janz DR, Lentz RJ, et al. Randomized Trial of Apneic Oxygenation during Endotracheal Intubation of the Critically Ill. Am J Respir Crit Care Med 2016;193:273-80.

A New Valsalva for SVT. Try this at home... or at an ED near you!

The REVERT trial was an elegant study with excellent methods. How often can we say that? It was genuinely a pleasure to review.

The authors from England randomized 433 patients with SVT to either a standard or modified Valsalva. Both groups were asked to generate a 40mm Hg pressure strain over 15 seconds. The standard group stopped here. But in the modified group the patients were repositioned to be supine and with a passive leg raise immediately after the Valsalva.

There is an excellent two minute video created by the authors that demonstrates the technique:

The primary outcome was proportion of patients in sinus rhythm at 1 minute after intervention.

In the end 17% of the standard Valsalva achieved sinus rhythm vs. 43% in the modified group. Wow!

These results are so good that I almost struggle to believe it is true. But it certainly can’t hurt to try. It would be nice to see if others are getting similar results.

Although this study used a bit of a fancy gizmo to measure 40mm of pressure, they make it clear that a standard 10ml syringe can be used. Just ask the patient blow hard enough just to move the plunger for 15 seconds. Then of course lay them supine and passively raise the legs.

If it works, whoopee! And tell them to try this at home.


Appelboam A, Reuben A, Mann C, et al. Postural modification to the standard Valsalva manoeuvre for emergency treatment of supraventricular tachycardias (REVERT): a randomised controlled trial. Lancet 2015 Oct 31;386(10005):1747-53. doi: 10.1016/S0140-6736(15)61485-4

Friday, 26 February 2016

Acupuncture for renal colic? How could such an awful paper get published?

The essential ingredients for conducting a randomized clinical trial are rather simple.

First you start with a research question that is hopefully feasible, interesting, novel, ethical and relevant. Then you come up with a hypothesis, null hypothesis and a primary outcome measure.

Next try to determine the smallest clinically meaningful difference that you think would be important to detect. Power your trial accordingly, get ethics approval and off you go!

Did this “RCT” do any of this?


Well, I guess they did do the “off you go” part.

This paper is so brilliantly awful that I will not discuss any of the results or conclusions as they are really irrelevant. However, I will focus on a few details.

This trial was conducted in three “clinics” in Turkey. Patients were evidently randomized (although there was zero mention of how they were randomized… not to be pedantic but this is important) to one of three arms; IM diclofenac, acupuncture, or intravenous acetaminophen (paracetamol). There was no primary outcome measure but they reported all kinds of measures over time on a VAS and VRS.

As mentioned previously, there was no power calculation performed but they managed to enroll about 40 patients in each arm. It doesn’t take a rocket scientist to conclude that they could only ever find very large differences with these small numbers (probably about 60mm on a VAS). In statistical speak, the possibility for type II error is huge.

Another question is ethics.

I do not believe that a Human Research Ethics Committee (HREC) in Australia would have approved the experimental design of this study. Before asking participants to take on some form of risk, pain or inconvenience, one needs to know if the study is designed to answer the research question. Otherwise it is a waste of time and exposes volunteers to unnecessary harm. This trial had no power calculation, no idea if it could answer any questions and as such could not be considered ethical.

Well done studies looking at acupuncture show that it is no better than placebo. (These are often conducted using sham acupuncture.) When we have an established treatment that is known to be effective, is it ethical to randomize a patient into a study arm that is placebo? For the methodology nerds, there was no clinical equipoise. This is mandatory before conducting an ethical randomized clinical trial.

There is so much more we could criticize about the methods, statistical analysis, bias’ and completely irrelevant discussion, but what’s the point.

Unfortunately a popular and generally well regarded educational program, EMRAP reviewed this paper and completely missed. It goes to show that even the best educational programs will get it wrong sometimes.

How did this paper get published? Not to be harsh, but this publication is not exactly a high impact journal. The process for peer review can be rather hit or miss in some of these fringe journals. In this case, it was a big miss.


Keynar M, Koyuncu F, Buldu I, et al. Comparison of the efficacy of diclofenac, acupuncture, and acetaminophen in the treatment of renal colic. Am J Emerg Med 2015;30:749-53.

Does an undetectable troponin upon presentation to the ED rule out MI?

Spoiler alert: Probably yes if they have had pain for a couple of hours.

There have been many consistent studies proving the high sensitive troponin assays become positive very early after myocardial infarction. Therefore one can presumably use it in a rapid rule out MI strategy. Unfortunately the international guidelines have yet to catch up with the science.

Along comes another study with a slightly different spin that shows one can exclude MI early using a high sensitive troponin assay showing an undetectable result. This was a single centre study done in an Australian ED looking at an “undetectable” troponin at presentation and whether this could exclude MI. The thought being this could hopefully result in early discharge without further observation.

Once again, this was not a troponin simply below the usual reference range of normal but it was below the detectable threshold of the assay. Of course one could reasonably be concerned this will crush the specificity of the test- we’ll get to this in a moment.

The primary outcome was a final diagnosis of MI in patients with undetectable troponin at presentation to the ED. (They also included some risk stratification with TIMI scores, but feel free to ignore this as it was not helpful).

Results? The undetectable troponin was 97% sensitive (95% CI 93-99%) out of the 1076 patients studied. There were four patients missed by this strategy. A post hoc analysis (a reasonable data dredge) showed that an undetectable troponin was 100% sensitive in those who had pain for at least 2 hours before presentation to the ED. Of course this makes sense and is consistent with all of the other literature on the subject.  

Interestingly, I was quite surprised to see 60% of the patients in this cohort had an undetectable troponin upon presentation. I would have thought this number would have been much worse… goes to show what I know.

In the end, this study really is just confirming what we already know about the new high sensitive troponin assays. They are very good at picking up MI’s early after onset of infarction. Almost all patients who rule-in will do so within a few hours of symptoms. The days of the six hour troponin are probably gone. Come on guidelines… catch up please!


Kelly A-M, Klim S. Does undetectable troponin I at presentation using a contemporary sensitive assay rule out myocardial infarction? A cohort study.  Emerg Med J 2014;0:1-4.