Group-based trajectory modeling of anion gap and mortality in patients with sepsis: a retrospective analysis of the MIMIC-IV database.

BACKGROUND

The purpose of this study was to identify distinct trajectories of the anion gap (AG) of patients with sepsis within the first 48 h following intensive care unit (ICU) admission and to explore the relationship between these trajectories and all-cause mortality.

METHODS

This study was carried out involving patients with sepsis from the Medical Information Mart for Intensive Care IV (MIMIC-IV) database. Group-based trajectory modeling (GBTM) was utilized to identify the distinct trajectory groups for the AG values. The primary outcome was 30-day mortality, and the secondary outcomes were 90-day and 1-year mortality. Both univariable and multivariable Cox proportional hazards regression models were performed to explore the relationship between different AG longitudinal trajectories and mortality. Stratified analyses were performed to investigate the stability of the relationship between AG trajectories and the primary outcome.

RESULTS

A total of 6960 patients with sepsis were included for trajectory grouping. Four distinct AG trajectories based on the model fitting standard were identified: group 1 (11.19%), group 2 (52.87%), group 3 (29.86%), and group 4 (6.08%). Using trajectory group 1 as the reference, after adjusting for all potential confounders, group 2, group 3, and group 4 still had 1.32 (95% confidence interval [CI] 1.07-1.63), 1.67 (95% CI 1.33-2.09), and 1.87 (95% CI 1.40-2.51) times the risk of 30-day mortality, respectively. Similar results were also found for 90-day mortality and 1-year mortality. The associations were directionally similar across all subgroups.

CONCLUSIONS

Trajectory groups with higher AG levels were associated with an increased risk for all-cause mortality. Identifying distinct AG trajectories may help identify patient subgroups with varying risks of mortality, providing valuable implications for both research and clinical practice.