Effect of peri‐operative pharmacological interventions on postoperative delirium in patients having cardiac surgery: a systematic review and Bayesian network meta‐analysis

Summary Introduction Postoperative delirium is a common complication following cardiac surgery. Despite its known impact on patient outcomes, effective preventative strategies remain elusive. We aimed to perform a comprehensive Bayesian network meta‐analysis of randomised controlled trials assessing the effect of pharmacological interventions on the incidence of postoperative delirium. Methods Databases were searched from inception to September 2024. Our search was updated in January 2025. Eligible studies included randomised controlled trials reporting the incidence of postoperative delirium in patients having cardiac surgery treated with pharmacological interventions. Bayesian models were used to estimate risk ratios (RR) and mean differences with 95%CrI through Markov chain Monte Carlo. Interventions were ranked using the surface under the cumulative ranking curve. Sensitivity analyses and grading of recommendations, assessment, development and evaluation assessment were conducted to evaluate the robustness and certainty of evidence. Results Seventy‐nine randomised controlled trials comprising 24,827 patients were included, with 29 pharmacological interventions compared. Dexmedetomidine combined with melatonin was the most effective intervention, reducing the incidence of postoperative delirium compared with placebo (risk ratio 0.31, 95%CrI 0.13–0.69; low certainty). Dexmedetomidine with melatonin also significantly decreased intensive care unit stay (2.4 days, 95%CrI ‐3.50–1.10) and hospital stay (1.32 days, 95%CrI ‐2.09 to ‐0.57). Other interventions, including ketamine and risperidone, showed potential efficacy but with low or very low certainty of evidence. Discussion We identified dexmedetomidine with melatonin as the most effective pharmacological strategy for preventing postoperative delirium following cardiac surgery. Whilst these findings highlight potential benefits, the low certainty of evidence underscores the need for more high‐quality primary evidence.