AimStudy results on blood glucose and the risk of delirium in patients receiving cardiac surgery are inconsistent, and there is also a gap in how to manage blood glucose after coronary artery bypass grafting (CABG). This study focused on patients with diabetes mellitus (DM) undergoing CABG and explored the associations of different blood glucose-related indexes and blood glucose change trajectory with postoperative delirium (POD), with the aim of providing some information for the management of blood glucose in this population.MethodsData of patients with DM undergoing CABG were extracted from the Medical Information Mart for Intensive Care (MIMIC)-IV database in this retrospective cohort study. The blood glucose-related indexes included baseline blood glucose, mean blood glucose (MBG), mean absolute glucose (MAG), mean amplitude of glycemic excursions (MAGE), glycemic lability index (GLI), and largest amplitude of glycemic excursions (LAGE). The MBG trajectory was classified using the latent growth mixture modeling (LGMM) method. Univariate and multivariate logistic regression analyses were utilized to screen covariates and explore the associations of blood glucose-related indexes and MBG trajectory with POD. These relationships were also assessed in subgroups of age, gender, race, estimated glomerular filtration rate (eGFR), international normalized ratio (INR), sepsis, mechanical ventilation use, and vasopressor use. In addition, the potential interaction effect between blood glucose and hepatorenal function on POD was investigated. The evaluation indexes were odds ratios (ORs), relative excess risk due to interaction (RERI), attributable proportion of interaction (AP), and 95% confidence intervals (CIs).ResultsAmong the eligible 1,951 patients, 180 had POD. After adjusting for covariates, higher levels of MBG (OR = 3.703, 95% CI: 1.743–7.870), MAG >0.77 mmol/L/h (OR = 1.754, 95% CI: 1.235–2.490), and GLI >2.6 (mmol/L)2/h/per se (OR = 1.458, 95% CI: 1.033–2.058) were associated with higher odds of POD. The positive associations of MBG, MAG, and GLI with POD were observed in patients aged <65 years old, male patients, White patients, those with eGFR <60 and INR <1.5, patients with sepsis, and those who received mechanical ventilation and vasopressors (all p < 0.05). Patients with class 3 (OR = 3.465, 95% CI: 1.122–10.696) and class 4 (OR = 3.864, 95% CI: 2.083–7.170) MBG trajectory seemed to have higher odds of POD, compared to those with a class 1 MBG trajectory. Moreover, MAG (RERI = 0.71, 95% CI: 0.14–1.27, AP = 0.71, 95% CI: 0.12–1.19) and GLI (RERI = 0.78, 95% CI: 0.19–1.39, AP = 0.69, 95% CI: 0.16–1.12) both had a potential synergistic effect with INR on POD.ConclusionFocusing on levels of MBG, MAG, GLI, and MBG trajectory may be more beneficial to assess the potential risk of POD than the blood glucose level upon ICU admission in patients with DM undergoing CABG.