Effect of Remimazolam on Postoperative Delirium in Surgical ICU Patients: A Single‐Center Prospective Cohort Study

Aim Postoperative delirium is a clinically important complication among patients admitted to a surgical intensive care unit (ICU). Remimazolam is increasingly used for perioperative and ICU sedation, but its association with postoperative delirium after elective surgery remains uncertain. This study evaluated the association between remimazolam‐based postoperative sedation and postoperative delirium in surgical ICU patients. Design A single‐center prospective cohort study. Methods Adult patients admitted to a surgical ICU after elective surgery between June 2024 and May 2025 were categorized according to their early postoperative sedation strategy. The remimazolam group was compared with a nonremimazolam sedation strategy group, which included patients receiving dexmedetomidine, midazolam, propofol/ciprofol, or no continuous sedative infusion. The primary outcome was postoperative delirium within the first 5 postoperative days or during the ICU stay, whichever was shorter. Delirium was assessed using the Confusion Assessment Method for the ICU and/or the Intensive Care Delirium Screening Checklist when patients were sufficiently arousable. Stabilized inverse probability of treatment weighting was used to reduce measured baseline imbalance. Weighted logistic regression was used for delirium, and weighted median regression was used for skewed continuous outcomes. Results Of 307 patients assessed for eligibility, 208 were enrolled, and 204 were included in the final analysis, of whom 125 received remimazolam and 79 received nonremimazolam sedation strategies. Postoperative delirium occurred in 15 patients (12.0%) in the remimazolam group and 7 patients (8.9%) in the comparator group. After inverse probability weighting, remimazolam was not statistically significantly associated with postoperative delirium (adjusted odds ratio 1.69, 95% confidence interval 0.61–4.69; p = 0.312). Mechanical ventilation duration was longer in the remimazolam group (median difference 6.00 h, 95% confidence interval 2.00–8.00; p < 0.001). Conclusion Remimazolam was not statistically significantly associated with postoperative delirium compared with nonremimazolam sedation strategies. However, the low event rate, wide confidence interval, and heterogeneous comparator group mean that clinically important differences cannot be excluded. The longer ventilation duration observed in the remimazolam group should be interpreted cautiously because residual confounding may remain. Trial Registration: Chinese Clinical Trial Registry: ChiCTR2400085877