Despite being described by Sellick almost 60 years ago, there has never been any convincing evidence supporting cricoid pressure during rapid sequence intubation. (Sellick 1961) Based on physiologic reasoning, it is frequently described as being “standard of care”. However, when studied, cricoid pressure doesn’t actually seem to decrease aspiration. (Ellis 2007; Neilipovitz 2007; Fenton 2009). Worse, cricoid pressure has been shown to worsen laryngeal view during intubation, and even completely obstruct the airway. (Allman 1995; Palmer 2000; Levitan 2006; Oh 2013) The result has been a relatively classic debate in medicine between a historical “standard” and science. Unfortunately, the available science has been relatively weak, until now…
Birenbaum A, Hajage D, Roche S, et al. Effect of Cricoid Pressure Compared With a Sham Procedure in the Rapid Sequence Induction of Anesthesia: The IRIS Randomized Clinical Trial. JAMA surgery. 2018; PMID:30347104
The IRIS study was a multi-center, randomized, double-blind, non-inferiority trial.
Patients: Adult patients undergoing general anesthesia requiring RSI because of a “full stomach” (<6 hours fasting) or at least 1 risk factor for aspiration.
- Exclusion criteria: refusal of patient to participate, younger than 18 years, pregnancy, inclusion in another randomized trial, lack of national health care insurance, contraindication for the use of the cricoid pressure or succinylcholine, pneumonia or pulmonary contusion, upper respiratory tract abnormalities, consciousness disorders,and patients requiring an alternative technique to laryngoscopy.
Intervention: Sham cricoid pressure.
Comparison: Cricoid pressure.
- All participating clinicians underwent a special training session to reinforce the correct way to apply cricoid pressure.
- Blinding was done with an opaque sheet, although I imagine unblinding could happen if people aren’t great actors (or don’t understand how important blinding is to valid studies).
Outcome: The primary outcome was the incidence of visualized pulmonary aspiration.
They recruited 3472 patients.
For the primary outcome of pulmonary aspiration, there was no difference between the two groups (0.6% with cricoid pressure and 0.5% without, RR 0.9, 95% CI 0.33-2.38). However, because of the wide confidence intervals, non-inferiority was not demonstrated.
Rates of suspected pneumonia with 24 hours (0.9% vs 0.6%), aspiration pneumonia (0.2% vs 0.2%), and severe pneumonia (0.1% vs 0.1%) were also unchanged.
The secondary outcomes looking at difficult intubation were all worse in the cricoid pressure group. Intubation took longer in the cricoid pressure group. There were more Cormack Lehane grade 3 and 4 (bad) views in the cricoid pressure group (10% vs 5%, p <0.01). The cricoid pressure group also had more patients that required more than 2 intubation attempts (3% vs 2%, p=0.05).
The big issue with this study is the non-inferiority design. I think it is important, whenever you read a trial, to ask yourself: what question do I want answered here? Too often the study asks and answers a completely different question. Regarding cricoid pressure, there is no evidence that it helps, but plenty of evidence that it interferes with intubation. (Ellis 2007; Neilipovitz 2007; Allman 1995; Palmer 2000; Levitan 2006; Fenton 2009) The question I want answered is: does cricoid pressure help patients (is there a net benefit, considering the known harms)? The awkward question asked by this study is: does sham cricoid pressure result in a rate of pulmonary aspiration that, when considering the confidence intervals, is no more than 50% higher than pulmonary aspiration when using cricoid pressure? I don’t think that is the right question.
Non-inferiority trials make a large assumption in their methodology. They choose one of the two treatments as the “standard”, which biases the calculations in favour of that standard. (Instead of starting with the null hypothesis we are used to, non-inferiority trials actually start by assuming that one intervention is inferior to the standard). This assumption is fine if the intervention being studied has clear evidence of benefit, such as warfarin in atrial fibrillation, and we want to compare a new medication that may not be any better in hard outcomes, but might have other advantages. However, there is no reason to believe that cricoid pressure is beneficial, so it doesn’t make sense to ask if another intervention is at least as good as it. I would have designed this as a superiority trial. In other words, I think you need to prove that cricoid pressure provides benefit in order to justify its use. (And that clearly didn’t happen here.)
If I haven’t bored you enough about non-inferiority trials yet, it is worth quickly reviewing how the math works. First, you have to make a clinical decision: how much worse than the standard therapy could the new therapy be and still be acceptable? In this trial, they decided that a 50% (relative) increase in pulmonary aspiration would be an acceptable difference, because the base rate is so low. (This equated to to a 1.4% absolute increase.) After you run the trial, you look at the confidence intervals, and if the confidence intervals cross that predetermined “inferiority” threshold, then you conclude that the intervention is “not non-inferior” because it statistically includes the possibility of being that much worse.
Hence, non-inferiority trials really hinge on confidence intervals. This means that the size of the trial is incredibly important. Smaller sample sizes result in larger confidence intervals, while larger sample sizes result in smaller confidence intervals. The rate of the primary outcome is also important, because the confidence intervals are based on the number of events, not the number of people in the trial. Therefore, rarer outcomes will have larger confidence intervals. (Aspiration occurred less often here than anticipated, meaning that their original power calculation were flawed.) The upshot is that non-inferiority trials require massive numbers if they are studying rare events (like pulmonary aspiration during airway management), or they risk false negatives, simply because trials are underpowered to meet strict non-inferiority criteria chosen.
The 95% confidence intervals in this trial are wide, leading to the authors conclusion that “we failed to demonstrate the noninferiority (δ = 50%) of a sham procedure as compared with the cricoid pressure in preventing pulmonary aspiration.” However, if you actually look at the 95% confidence intervals, you will notice that the range actually favours the no cricoid pressure group. If you ran this trial hundreds of times, you would expect the majority to demonstrate that the no cricoid pressure group did better, and relatively few to end with a result worse than their predefined inferiority cutoff. It doesn’t make a lot of sense to look at those numbers and decide to use cricoid pressure (and I am not even talking about the reported harms yet).
This leads to my biggest advice when analyzing non-inferiority trials: just look at the raw numbers. When I look at the raw numbers in this trial, it is pretty clear these two groups had equivalent outcomes. The confidence intervals are large, but balanced. It is possible one of these interventions is superior, but based on this data, it is just as likely to be the no cricoid pressure group. (Realistically, this data tells us pretty definitively that you are going to see similar rates of pulmonary aspiration whether or not you use cricoid pressure).
Although there doesn’t seem to be any benefit to using cricoid pressure, this study reinforces our concerns about harms. Cricoid pressure clearly interferes with intubation. Thus, we are left with a net negative, which should provoke practice change for anyone still using cricoid pressure routinely.
Does this definitely prove that cricoid pressure is dead? No. That isn’t how science works. This trial only looked at adult patients undergoing general anesthesia. Although they had risk factors for aspiration, they were actually at very low risk for aspiration. (It is a reasonable question whether NPO status is even a risk factor.) There may be a higher risk population in which cricoid pressure could work. (For example, pregnant patients, who were specifically excluded here). However, there has never been any good science supporting that position. For now, this RCT is the best science we have, and it is a pretty convincing demonstration that cricoid pressure is not just unnecessary, but actually probably harmful.
Despite the confusing non-inferiority conclusion, this study clearly demonstrates that cricoid pressure causes harm by interfering with intubation without providing any benefit. Until further data becomes available, the practice should be abandoned.
Allman KG. The effect of cricoid pressure application on airway patency. Journal of clinical anesthesia. 1995; 7(3):197-9. PMID:7669308
Ellis DY, Harris T, Zideman D. Cricoid pressure in emergency department rapid sequence tracheal intubations: a risk-benefit analysis. Annals of emergency medicine. 2007; 50(6):653-65. PMID:17681642
Fenton PM, Reynolds F. Life-saving or ineffective? An observational study of the use of cricoid pressure and maternal outcome in an African setting. International journal of obstetric anesthesia. 2009; 18(2):106-10. PMID:19144507
Levitan RM, Kinkle WC, Levin WJ, Everett WW. Laryngeal view during laryngoscopy: a randomized trial comparing cricoid pressure, backward-upward-rightward pressure, and bimanual laryngoscopy. Annals of emergency medicine. 2006; 47(6):548-55. PMID:16713784
Neilipovitz DT, Crosby ET. No evidence for decreased incidence of aspiration after rapid sequence induction. Canadian journal of anaesthesia = Journal canadien d’anesthesie. 2007; 54(9):748-64. PMID:17766743
Oh J, Lim T, Chee Y, et al. Videographic analysis of glottic view with increasing cricoid pressure force. Annals of emergency medicine. 2013; 61(4):407-13. PMID:23306455
Palmer JH, Ball DR. The effect of cricoid pressure on the cricoid cartilage and vocal cords: an endoscopic study in anaesthetised patients. Anaesthesia. 2000; 55(3):263-8. PMID:10671846
SELLICK BA. Cricoid pressure to control regurgitation of stomach contents during induction of anaesthesia. Lancet (London, England). 1961; 2(7199):404-6. PMID:13749923
Morgenstern, J. Cricoid pressure is dead (Birenbaum 2018), First10EM, December 10, 2018. Available at: