Association of glycemic variability with delirium and mortality in critically ill patients with community-acquired pneumonia: a retrospective analysis

Delirium frequently complicates the course of critically ill patients with community-acquired pneumonia (CAP). Nursing screening tools such as the CAM-ICU remain subjective and fail to capture underlying metabolic disturbances. Glycemic variability (GV) which can be calculated from routine glucose measurements collected by nursing staff, has been linked to delirium and poor outcomes in general ICU populations but has not been fully explored in CAP patients. This study evaluates whether GV is associated with delirium onset and one-year mortality in ICU-admitted CAP patients, with the goal of informing targeted, evidence-based nursing interventions. This retrospective cohort study analyzed 6,167 ICU patients with CAP from the MIMIC-IV database. GV was measured using the coefficient of variation of blood glucose values and divided into tertiles. The primary outcome was delirium, identified by at least one positive CAM-ICU assessment. Secondary outcomes included 1-year mortality and ICU length of stay. Multivariable logistic regression was used to evaluate the association between GV and delirium, while Cox proportional hazards regression was used to evaluate the association between GV and 1-year mortality. Kaplan–Meier curves and restricted cubic spline analyses were used to examine the associations between GV and clinical outcomes, and receiver operating characteristic (ROC) curve analysis was performed to evaluate the discriminatory performance of multivariable models incorporating GV and prespecified covariates. A sensitivity analysis using a fixed 72-hour exposure window was performed to assess the robustness of the primary findings. Among 6,167 critically ill patients with CAP, 54.3% developed delirium, the overall 1-year mortality rate was 45.2%, and the median ICU length of stay was 4.30 days (IQR 2.37–9.08). After full adjustment, each unit increase in log-GV was independently associated with higher odds of delirium (OR, 1.328; 95% CI, 1.177–1.499) and a higher hazard of 1-year mortality (HR, 1.081; 95% CI, 1.006–1.162). The highest GV tertile (GV3) exhibited a delirium incidence of 58.40% and significantly greater mortality compared with GV1 ( P < 0.001). Kaplan–Meier analysis confirmed lower survival in GV3 (log-rank P < 0.001). Restricted cubic splines showed a nonlinear positive association with delirium (P nonlinearity = 0.002) and a linear relationship with mortality (P nonlinearity = 0.65). ROC curves yielded AUCs of 0.82 for delirium and 0.74 for 1-year mortality. GV was significantly associated with delirium and 1-year mortality in critically ill patients with CAP. GV may serve as a potential risk marker for identifying patients at higher risk of adverse outcomes. Further prospective studies are warranted to determine whether targeted assessment or modification of GV can improve clinical outcomes.