Bayesian Analysis of the Pragmatic Airway Resuscitation Trial

Abstract

Study objective: Intubation and laryngeal tube insertion are common airway management strategies in out-of-hospital cardiac arrest. Bayesian analysis offers an alternate statistical approach to assess the results of a trial. We use Bayesian analysis to compare the effectiveness of initial laryngeal tube versus initial intubation strategies on outcomes after out-of-hospital cardiac arrest in the Pragmatic Airway Resuscitation Trial. Methods: We performed a post hoc Bayesian analysis of the Pragmatic Airway Resuscitation Trial. We defined prior distributions representing neutral or skeptical estimates of laryngeal tube benefit. Using Bayesian log binomial models, we fit models for 72-hour survival, hospital survival, and hospital survival with favorable neurologic status. We estimated the posterior probability (the probability of observing an effect difference between treatment groups) of the benefit of laryngeal tube over intubation on outof-hospital cardiac arrest outcomes. Results: The parent trial enrolled 3,004 patients (1,505 laryngeal tube, 1,499 intubation). Under a neutral prior distribution (relative risk 1.0), laryngeal tube was better than intubation (72-hour survival risk difference 1.8% [95% credible interval {CrI}-0.9% to 4.5%], posterior probability 91%; hospital survival 1.4% [95% CrI-0.4% to 3.4%], posterior probability 93%; and hospital survival with favorable neurologic status 0.7% [95% CrI-0.5% to 2.1%], posterior probability 86%). Under a skeptical prior distribution (relative risk 0.83 to 0.92), laryngeal tube was also better than intubation (72-hour survival risk difference 1.7% [95% CrI-0.9% to 4.3%], posterior probability 89%; hospital survival 1.3% [95% CrI-0.5% to 3.3%], posterior probability 91%; and hospital survival with favorable neurologic status 0.6% [95% CrI-0.5% to 2.0%], posterior probability 82%). Conclusion: Under various prior assumptions, post hoc Bayesian analysis of the Pragmatic Airway Resuscitation Trial confirmed better out-of-hospital cardiac arrest outcomes with a strategy of initial laryngeal tube than initial intubation. [Ann Emerg Med.

Bayesian analysis of amiodarone or lidocaine versus placebo for out-of-hospital cardiac arrest

“…Likewise, the ANDROMEDA-SHOCK trial of 424 patients with severe sepsis were unable to show benefit between two targeted resuscitation strategies,18 but a Bayesian reassessment estimated a posterior probability of superiority of the peripheral-perfusion targeted strategy greater than 90% for all priors 19. The Pragmatic Airway Resuscitation Trial found that prehospital laryngeal tubes resulted in better OHCA outcomes than endotracheal intubation,20 and a post-hoc Bayesian analysis strongly affirmed the results of the parent trial 21. As demonstrated by these analyses, a Bayesian approach may better inform clinicians about intervention efficacy by combining existing evidence with new evidence to calculate an overall estimate of benefit, while also supporting future studies by estimating residual uncertainty in the evidence base for a given clinical question.…”

Randomized controlled trials in resuscitation

“…Re-analysis of trial results using Bayesian statistics is becoming common in resuscitation and critical care; however to fully utilize the Bayesian methods, this should be part of the pre-planned study analysis. 49 , 50 …”

Advanced airway management in out-of-hospital cardiac arrest – to intubate or not to intubate: a narrative review of the existing literature

“…In the Bayesian analysis of the pragmatic airway resuscitation trial, 72-hour survival and hospital survival were compared to estimate the benefit of laryngeal tube over tracheal intubation on OHCA outcomes [ 45 ]. Both under a neutral and sceptical prior distribution, better outcomes were observed with laryngeal tube than with tracheal intubation.…”