Abstract
The effects of diabetes and glucose on the outcomes of patients with sepsis are somewhat conflicting. This retrospective study enrolled 1214 consecutive patients with sepsis, including a subpopulation of 148 patients with immune profiles. The septic patients were stratified according to their Diabetes mellitus (DM) status or peak glucose level (three-group tool; P1: ≤140 mg/dL, P2: 141-220 mg/dL, P3: >220 mg/dL) on day 1. Although the DM group had a lower hazard ratio (HR) for 90-day mortality compared to non-DM patients, the adjusted HRs were insignificant. The modified sequential organ failure assessment-glucose (mSOFA-g) score can predict 90-day survival in patients with and without diabetes (β = 1.098, p < 0.001; β = 1.202, p < 0.001). The goodness of fit of the mSOFA-g score was 5% higher than the SOFA score of the subgroup without diabetes. The SOFA score and human leukocyte antigen-D-related (HLA-DR) expression were comparable between the groups. The P3 group had lower HLA-DR expression on days 1 and 3 and a higher 90-day mortality. The three-group tool was useful for predicting 90-day mortality in patients with separate Kaplan-Meier survival curves and mortality HRs in the construction and validation cohorts. The peak glucose level, instead of diabetes status, can be used as an easy adjunctive tool for mortality risk stratification in critically ill septic patients.