Perioperative intravenous dexamethasone on postoperative delirium and delayed neurocognitive recovery: a systematic review and meta-analysis

Introduction Perioperative neurocognitive disorders include postoperative delirium (POD), delayed neurocognitive recovery (dNCR), and postoperative neurocognitive disorder. Although neuroinflammation may contribute to these conditions, randomised trials of perioperative dexamethasone have reported inconsistent findings. We updated the evidence on intravenous dexamethasone for POD and dNCR, as well as recovery and safety outcomes. Methods We searched PubMed, Embase, Web of Science, Ovid MEDLINE, and the Cochrane Library through September 4, 2026. Randomised trials comparing perioperative intravenous dexamethasone with placebo or usual care in adults undergoing surgery were eligible. The co-primary outcomes were POD and trial-defined postoperative cognitive decline mapped to the dNCR time window according to the 2018 nomenclature framework. Risk of bias and certainty of evidence were assessed for the co-primary outcomes using RoB 2 and GRADE. Results Eight reports representing seven independent trials were included. Four independent trials involving 1,069 unique participants contributed to the POD analysis (RR 0.62, 95% CI 0.37–1.03; I 2 = 51%). Four independent trials comprising six comparisons and 1,533 unique participants contributed to the dNCR synthesis (RR 0.95, 95% CI 0.65–1.41; I 2 = 64%). Certainty of evidence was very low for both co-primary outcomes. Exploratory estimates suggested shorter time to extubation and hospital and intensive care unit stays, based on two small cardiac-surgery trials with substantial methodological limitations. Infection and hyperglycaemia estimates were imprecise, whereas continuous glucose measures indicated greater perioperative glucose exposure with dexamethasone. Conclusions A possible reduction in POD remains uncertain, while no clear effect was identified for dNCR. Certainty was very low for both outcomes. Recovery findings remain exploratory and are influenced by a trial with substantial post-randomisation losses and high risk of bias for POD. These findings should be considered alongside increased perioperative glucose concentrations. Systematic Review Registration https://www.crd.york.ac.uk/PROSPERO/view/CRD420261381836 , PROSPERO CRD420261381836.