Often in the interpretation of medical literature we are forced into false dichotomies. Compelled to choose between two hypotheses, neither of which are representative of the data presented. Such is the case with a recent trial published in Intensive Care Medicine in which Guitton et al1 examine the efficacy of apneic oxygenation during RSI in critically ill adults admitted to the ICU. Like previous trials examining this question, the authors failed to identify a therapeutic benefit. This may very well be because there simply is not one, or it may be because we continue to ask the wrong question and expect in return to get the right answer.
The authors prospectively enrolled patients without severe hypoxia (defined as a P/F < 200), who required emergent intubation to either standard preoxygenation using a BVM or preoxygenation with a high-flow nasal cannula (HFNC) system.
Preoxygenation was performed over a 4-min period with either BVM or HFNC, during which the patients were placed in a head-up position. In the HFNC group, preoxygenation was performed using an AIRVO-2 HFNC system at 60 L/min and 100% FiO2. The control group received preoxygenation with a standard BVM set at 15 L/min oxygen flow, held in place to ensure appropriate mask seal. After preoxygenation, patients underwent a rapid sequence induction, the specifics of which were left to the discretion of the treating clinician. In patients randomized to the HFNC group, the high-flow cannula was left on throughout the intubation process in an attempt to achieve some degree of apneic oxygenation.
From April 2016 to June 2017 the authors enrolled 192 patients, 184 of which were included in their intention to treat analysis. Intubation was most often performed for neurologic reasons in comatose patients, 68% in HFNC and 75% in SMO patients. The authors reported no significant difference in their primary outcome, the median lowest SpO2 during intubation, 100% vs 99% (P=0.30), in the HFNC and BVM groups respectively. Results were similar when they examined the same results using the per protocol analysis.
Once again a well done RCT failed to identify a benefit of an apneic oxygenation strategy during RSI in critical ill patients in the ICU. Despite a strong physiological basis and early literature demonstrating apneic oxygenation prolongs the apneic period in patients in the OR2,3,4 ,why do we continue to fail to find a benefit with its use in critically ill patients? Apneic oxygenation may be ineffective in this subset of patients, or perhaps we are just asking the wrong question.
Like the previous trials examining the efficacy of apneic oxygenation, Guitton et al utilized the median lowest SpO2 during intubation as the primary outcome, estimating that HFNC would improve this value by 5% (from 88% in the control group to 93%). In order to ensure an 80% power to identify this difference, the authors planned to enroll 193 patients. We have discussed the issues with selecting such an outcome measure as a trial’s primary outcome in previous posts. In short, it assumes that oxygen saturation is an interval value (92% is better than 91% just as 83% is better than 82%). This is not the case. Because of the shape of the hemoglobin dissociation curve, a drop from 95% to 92% is not the same as one from 88% to 85% and different still from 85% to 82%. In addition, it is well known that the accuracy of a waveform pulse oximeter falters as the values drop below 90% and is essentially unusable below 80% 5,6. The idea that these values can be used as a continuous scale from least to most desirable is tenuous at best.
Not only is the median lowest SpO2 a poor outcome measure, but its selection as the primary outcome in turn limits the remainder of the trial’s secondary outcome measures. Because the authors powered their study to identify a difference in a continuous variable, which requires a much smaller sample size than if they utilized a dichotomous outcome (i.e. proportion of patients with a saturation <80%), the trial was severely underpowered to assess the other far more clinically relevant dichotomous outcomes. At least one severe complication occurred in 16% of the control group and only 6% of the HFNC group. Severe hypoxia (< 80%) occurred in 8% vs 2%, severe hypotension 9% vs 4%. Modest complications occurred in 7% vs 0 and esophageal intubation occurred in 6% vs 0. Some of these differences crossed the threshold to reach statistical significance, others did not and as such we are unable to differentiate them from errors in sampling. But if true, these findings would represent a clinically important benefit of apneic oxygenation.
Frequentist statistics do not allow us to view secondary endpoints as anything more than hypothesis generators, as multiple comparators are prone to type 1 error. But when study after study continue to identify the same beneficial trends, one has to suspect that the differences observed are representative of a true underlying benefit of apneic oxygenation. Some using HFNC, other a standard nasal cannula, but almost every RCT examining the use of apneic O2 in a critically ill patient population has failed to detect a difference in lowest median saturation. But almost all have demonstrated trends in a multitude of clinically important secondary outcomes but due to a small sample size are incapable of differentiating these observed difference from statistical chance. Jaber et al7 observed that severe procedural complications during intubation procedure occurred in 38% of the control group and 24% of the intervention group. Severe hypoxemia was observed in 21% of the control vs 4% of the intervention group. Miguel-Montanes et al8 reported a rate of desaturation < 80% in 14% of the controls, and only 2% of the apneic O2 group. Semler et al 9 observed the rate of desaturation <80% was 25% in the controls vs only 15.8% in the apneic O2 group. Vourc’h et al10 found the rate of severe complications occurred in 66.6% of the control and only 58.1% of the apneic O2 group. Only two trials failed to observe these differences in the rates of severe complications11,12.
The potential benefits of apneic oxygenation during RSI are yet to be clearly empirically elucidated. There are signals of benefit present, but because of the blunt measurement tools utilized, differentiating these findings from statistical noise is impossible. One thing is clear, if we continue to ask the wrong question, we will never get the correct answers.
- EM Nerd-The Case of the Partial Cohort - May 24, 2020
- EM Nerd: The Case of the Sour Remedy Continues - January 20, 2020
- EM Nerd-The Case of the Adjacent Contradictions - December 23, 2019
Nice post. I agree with the main points about patient centered vs surrogate outcomes. I have only two quibbles.
1.) I too was taught in the 90s to distrust SpO2 90 is less bad than 85–>80 depends on whether you assign primacy to PaO2 or CaO2
Interesting point, would you care to expand?
Also what was your second quibble?
I read a few of the articles. Won’t pretend I follow every point. Sounds like some of the authors are just better, faster getting the ET tube in and giving the pt adequate o2.
Sorry, but reference article 12 not very compelling with the spo2 chart. Maybe some of those tubes in that group were in the esophagus. 🙂