Perioperative dexmedetomidine for the prevention of postoperative delirium after cardiac surgery: a systematic review, Bayesian meta-analysis, and Bayesian re-analysis of the DECADE trial

Abstract

Dexmedetomidine is seen as a promising agent for the prevention of postoperative delirium after cardiac surgery, but the largest study (DECADE) paradoxically suggested an increased risk of delirium. Studies were selected using inclusion/exclusion criteria after conducting online database searches for randomised controlled trials. The primary outcome was the incidence of postoperative delirium with a minimum clinically important difference (MCID), defined as an odds ratio >1.20 or <0.84. Publication bias was quantified using Bayesian model averaging. A random-effects meta-analysis with weakly informative priors was performed. Bayesian re-analysis of DECADE using several different priors including a prior based on this meta-analysis (excluding DECADE) was also performed. We identified 12 eligible randomised controlled trials (3539 participants). The overall pooled effect showed a mean benefit from dexmedetomidine in delirium prevention (odds ratio 0.67 [95% credible interval 0.45, 0.92]), but any definitive evidence of benefit disappeared after accounting for publication bias (odds ratio 1.15 [95% credible interval 0.93, 2.51]). Bayesian re-analysis of the DECADE trial under a vague prior showed a 1.5% posterior probability of any benefit and only a 0.1% probability of an MCID for benefit. Combining DECADE with the unadjusted meta-analysis-derived prior increased the probability of an MCID for benefit to 17.8%, which decreased to 0.2% using the meta-analysis-derived prior adjusted for publication bias. Pooled evidence suggests dexmedetomidine is associated with reduced incidence of postoperative delirium; however, this is highly sensitive to the possibility of publication bias. Bayesian re-analysis of the recently published DECADE trial showed the effect of dexmedetomidine differs markedly depending on the weight given to previous studies. Hence, any true benefit from the use of dexmedetomidine over standard care in this population cannot be definitively characterised at present. CRD42023401623 (PROSPERO).