Association between group-based trajectory modeling of blood glucose and acute delirium incidence in critically ill patients: a retrospective cohort study based on the MIMIC-IV database.

Delirium is a frequent complication in critically ill patients and is associated with adverse outcomes such as long-term cognitive impairment and increased mortality. The relationship between the dynamic changes in blood glucose and the onset of acute delirium remains unclear. This study aims to explore the effect of 24-h blood glucose trajectory on acute delirium in patients via latent category trajectory modeling, and additionally investigate its association with in-hospital mortality in this population. This retrospective cohort study examined patients in the intensive care unit (ICU) using the MIMIC-IV database. Changes in the trajectories of blood glucose within 24 h after admission to the ICU were categorized using latent category trajectory modeling. The outcome examined was the occurrence of acute delirium during ICU hospitalization, with the secondary outcome being in-hospital mortality. The study included 21,940 critically ill patients, of which 2,633 developed acute delirium during ICU hospitalization. The blood glucose trajectories within 24 h were classified into four categories using the LCMM model. After fully adjusting for various confounders, Classes 4 and 2 were associated with a higher risk of acute delirium compared with Class 1, and the respective ORs (95% CIs) were 1.34 (1.08-1.64) and 1.18 (1.04-1.35). For the secondary outcome, a similar trend was observed between Class and in-hospital mortality: OR (95% CI) was 1.62 (1.29-2.02) for Class 4 and 1.45 (1.25-1.67) for Class 2. The 24-h blood glucose trajectory is significantly associated with both the risk of acute delirium and in-hospital mortality in critically ill patients. Focusing on levels of blood glucose trajectory may be beneficial to assess the potential risk of acute delirium and in-hospital mortality.