Introduction

Postoperative delirium (POD) is a frequent and serious complication after cardiac surgery. The glucose-to-albumin ratio (GAR) integrates metabolic and nutritional status, but its predictive value for POD in intensive care unit (ICU) patients remains uncertain.

Methods

We conducted a retrospective cohort study using data from the MIMIC-IV (v3.1) database. A total of 4,307 adult ICU patients admitted within 24 h after cardiac surgery were included. GAR was calculated by dividing serum glucose (mg/dL) by albumin (g/dL). The primary outcome was POD, assessed using the Confusion Assessment Method (CAM). GAR was evaluated as a continuous variable, by quartiles, and by a dichotomized cutoff determined from the receiver operating characteristic (ROC) curve. Logistic regression, restricted cubic splines, and mediation analyses were applied. To reduce confounding, propensity score matching was performed, and the main logistic regression analysis was repeated in the matched cohort.

Results

Of 4,307 patients, 658 (15.3%) developed POD and 120 (2.8%) died within 90 days. Higher GAR was independently associated with POD (continuous odds ratio [OR] 1.01, 95% confidence interval [CI] 1.00–1.01, p = 0.027; highest vs. lowest quartile OR 1.54, 95% CI 1.16–2.05, p = 0.003). The optimal GAR cutoff was 35 (AUROC 0.654). Patients with GAR ≥ 35 had higher delirium risk in both the full cohort (adjusted OR 1.37, 95% CI 1.13–1.67, p = 0.001) and the matched cohort (OR 1.31, 95% CI 1.06–1.61, p = 0.012). Mediation analysis suggested that prolonged mechanical ventilation and ICU stay partly explained the GAR–POD relationship.

Conclusion

GAR is an independent predictor of postoperative delirium in ICU patients undergoing cardiac surgery. It may serve as a practical biomarker for early delirium risk stratification. Prospective studies are warranted to validate these findings and to further establish the clinical utility of GAR in perioperative care.