Perioperative sleep interventions for postoperative delirium prevention: a Bayesian network meta-analysis and systematic review

Perioperative sleep disorders significantly increase the risk of postoperative delirium (POD). However, there is no consensus on the preventive effects of specific perioperative sleep interventions on POD. This study aimed to assess the efficacy of different perioperative sleep interventions for preventing POD and to determine their relative rankings. This study systematically analyzed randomized clinical trials (RCTs) that investigated the impact of perioperative sleep interventions on POD in adult patients. Data were sourced from PubMed, the Cochrane Library, EMBASE, and Web of Science, spanning from inception to December 2025. A Bayesian network meta-analysis (NMA) was performed using the BUGSnet package in R to evaluate and rank the effectiveness of different interventions in reducing the incidence of POD. The optimal model for analysis was selected by comparing the deviance information criterion. A total of 20 RCTs were included. This NMA suggested that dexmedetomidine (log OR -0.69; 95% CrI, -1.33 to -0.04), melatonin combined with dexmedetomidine (log OR -1.87; 95% CrI, -3.69 to -0.07), and multi-component non-pharmacological sleep interventions (log OR -1.69; 95% CrI, -3.04 to -0.49) significantly reduced the incidence of POD. Melatonin combined with dexmedetomidine and the multi-component non-pharmacological sleep interventions yielded the highest SUCRA values. This NMA suggests that multi-component non-pharmacological interventions and melatonin combined with dexmedetomidine had the highest ranking probabilities for reducing POD, while not all sleep interventions were found beneficial. However, rankings only reflect probability rather than definitive superiority, and the estimate for the combination regimen rests on a single trial with wide uncertainty and was rated only moderate certainty, so it should be interpreted cautiously. This Bayesian network meta-Analysis goes beyond simple efficacy assessment by offering a quantitative hierarchy of perioperative sleep interventions for preventing POD. Perioperative multi-component non-pharmacological interventions and melatonin combined with dexmedetomidine had the highest ranking probabilities for reducing POD. Perioperative sleep interventions did not significantly reduce the duration of POD, ICU or hospital stay.