Prognostic Value of 72-h Reassessment Using SOFA Score, C-Reactive Protein, and Lactate in a 72-h Landmark Cohort of...

Background and Objectives: Sepsis caused by Gram-negative bloodstream infections is associated with substantial mortality in intensive care units (ICUs). Admission measurements reflect initial disease severity but may not adequately capture subsequent clinical evolution or treatment response. This study evaluated whether a model based on 72 h reassessment demonstrated greater discrimination than a baseline model for mortality occurring between the 72 h landmark and day 28. Materials and Methods: This retrospective, single-center cohort study included adult ICU patients with Sepsis-3-defined sepsis and confirmed Gram-negative bloodstream infection admitted between January 2018 and November 2023. The primary analysis was conducted as a 72 h landmark cohort analysis with a fixed-horizon binary outcome and was restricted to patients who were alive and under observation at 72 h and had complete clinical and laboratory data. Separate multivariable logistic regression models were constructed using baseline and 72 h variables. Model discrimination was assessed using the area under the receiver operating characteristic curve (AUC), and the AUCs were compared using DeLong's test for correlated ROC curves. Internal validation of the final 72 h model was performed using 1000 bootstrap resamples. Results: A total of 422 patients were included, of whom 134 (31.8%) died within 28 days. Baseline septic shock (adjusted odds ratio [OR] = 3.994; 95% confidence interval [CI]: 2.296-6.949; p < 0.001), higher 72 h SOFA score (adjusted OR = 1.603 per point; 95% CI: 1.424-1.805; p < 0.001), higher 72 h C-reactive protein level (adjusted OR = 1.005 per mg/L; 95% CI: 1.001-1.008; p = 0.005), and higher 72 h lactate level (adjusted OR = 1.189 per mmol/L; 95% CI: 1.007-1.403; p = 0.041) were independently associated with mortality occurring between the 72 h landmark and day 28. The 72 h model demonstrated greater discrimination than the baseline model (AUC: 0.867 vs. 0.799; AUC difference = 0.068; 95% CI: 0.027-0.109; p = 0.0012). The optimism-corrected AUC after bootstrap validation was 0.861. Conclusions: Among patients with Gram-negative sepsis who were alive at the 72 h landmark, a model incorporating baseline septic shock, 72 h SOFA score, CRP, and lactate demonstrated greater discrimination than the baseline model for subsequent mortality through day 28. These findings support the clinical relevance of 72 h reassessment for risk stratification, although external validation is required before routine clinical implementation.