Sevoflurane Versus Propofol for Postoperative Sedation After Cardiac Surgery: A Systematic Review of Randomized...

Optimal postoperative sedation after cardiac surgery remains debated. Sevoflurane delivered via anesthetic conserving devices has emerged as an alternative to intravenous (IV) propofol, but its comparative effectiveness and safety in the intensive care unit (ICU) remain uncertain. This systematic review and meta-analysis evaluated the efficacy and safety of sevoflurane versus propofol for postoperative ICU sedation after cardiac surgery. We conducted the review in accordance with Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines and the Cochrane Handbook. Medline, Web of Science, Scopus, and CENTRAL were searched without date restrictions. Randomized controlled trials (RCTs) comparing inhaled sevoflurane with IV propofol in mechanically ventilated adult patients admitted to the ICU after cardiac surgery were included. Primary outcomes were time to extubation, ICU length of stay (LOS), and hospital LOS. Secondary outcomes included acute renal failure (ARF), atrial fibrillation, mortality, and postoperative nausea and vomiting (PONV). Eight RCTs involving 688 patients were included. Four studies reporting time to extubation showed no significant difference between groups overall (mean difference (MD) -62.67, 95% confidence interval (CI) -157.23 to 31.90; p = 0.2977), with high heterogeneity (I² = 91.7%). Sensitivity analysis excluding one study showed significantly shorter extubation time with sevoflurane (MD -72.51, 95% CI -135.83 to -9.20; p = 0.0388). No significant differences were observed for ICU LOS (MD -0.29, 95% CI -4.10 to 3.51) or hospital LOS (MD -0.40, 95% CI -2.67 to 1.63). Secondary outcomes, including mortality, ARF, atrial fibrillation, and PONV, were also comparable between groups. Sevoflurane appears comparable to propofol for postoperative sedation after cardiac surgery and may reduce time to extubation in selected analyses; however, the evidence remains limited by heterogeneity and possible publication bias.