Emergency Medical Services
Background: Endotracheal intubation (ETI) and laryngeal tube (LT) insertion are commonly used to manage the airway during out-of-hospital cardiac arrest (OHCA). In the cluster randomized Pragmatic Airway Resuscitation Trial (PART) an initial-LT strategy showed improved outcomes over initial-ETI in adult OHCA. Bayesian analyses may aid interpretation of trial results by incorporating knowledge from prior studies. We used Bayesian analysis to estimate the probability that initial-LT is superior to initial-ETI for OHCA.
Methods: We performed a post-hoc Bayesian analysis of data from PART. Using meta-analysis of prior studies, we defined prior distributions reflecting a) neutral or b) skeptical estimates of LT benefit over ETI. We used Bayesian log binomial models with randomization cluster as a random effect. We fit separate models for a) 72-hour survival, b) survival to hospital discharge, and c) survival with favorable neurologic status (Modified Rankin Scale≤3), estimating risk differences, 95% credible intervals (CrI) and posterior probabilities (PostPr) of LT benefit over ETI. We conducted the analyses using neutral and skeptical prior distributions
Results: The parent trial enrolled 3,004 patients (1,505 LT, 1,499 ETI). Outcomes were: 72-hour survival LT 18.3% vs. ETI 15.4%, survival to hospital discharge 10.8% vs. 8.1%, survival with favorable neurologic status 7.1% vs 5.0%. Under a neutral prior distribution, LT had better outcomes; risk difference 72-hour survival 1.8% (95% CrI: -0.9%, 4.5%), PostPr 90.5%; survival to hospital discharge 1.4% (-0.4%, 3.4%), PostPr 93.5%; survival with favorable neurologic status 0.7% (-0.5%, 2.1%), PostPr 88.2%. Under a skeptical prior distribution (LT worse than ETI, RR 0.82-0.91), LT was again better; risk difference 72-hour survival 1.7% (95% CrI: -0.9%, 4.3%), PostPr 89.5%; survival to hospital discharge 1.3% (-0.5%, 3.3%), PostPr 92.2%; survival with favorable neurologic status 0.6% (-0.5%, 2.0%), PostPr 83.9%.
Conclusion: Under varying prior assumptions, post-hoc Bayesian analysis of PART data indicated high probabilities that adult OHCA outcomes are better with initial-LT than initial-ETI.