Background: This study aimed to identify distinct trajectories of serum osmolality within the first 72 h for patients with sepsis-associated encephalopathy (SAE) in the MIMIC-IV and eICU-CRD databases and assess their impact on mortality and adverse clinical outcomes.
Methods: In this retrospective cohort study, patients with SAE from the MIMIC-IV database were included. Group-based trajectory modeling (GBTM) was used to categorize distinct patterns of serum osmolality changes over 72 h in ICU patients. Differences in survival across the trajectory groups were compared using Kaplan-Meier (K-M) survival curves.
Results: A total of 11,376 patients with SAE were included in the analysis, with a median age of 65.6 ± 16.5 years. The in-hospital mortality rate at 30 days was 12.8%. Based on model-defined criteria, three distinct osmolality trajectory groups were identified: Group 1 (59.6%), Group 2 (36.4%), and Group 3 (4.0%). Kaplan-Meier survival analysis indicated that patients with relatively lower serum osmolality within the normal range (Group 1) had a lower 30-day mortality rate compared to those in the other groups (Group 2 and 3). Subgroup analysis demonstrated significant interactions (P < 0.05) between osmolality trajectories and covariates such as the Sequential Organ Failure Assessment (SOFA), vasopressor use and renal replacement therapy (RRT).
Conclusion: Identifying distinct serum osmolality trajectories may help recognize SAE patient subgroups with varying risks of adverse outcomes, providing clinically meaningful stratification.
Keywords: Dynamic trajectory; MIMIC-IV database; Osmolality; Sepsis-associated encephalopathy.
© 2025. The Author(s).