Performance of sequential organ failure assessment 1 versus 2 for sepsis identification in patients with suspected infection: a multicenter retrospective cohort study

The revised Sequential Organ Failure Assessment (SOFA-2) score was recently developed to update definitions of acute organ dysfunction in intensive care units (ICU). This study aimed to evaluate performance between SOFA-1 and SOFA-2 definition for sepsis identification in patients with suspected infection. We analyzed 24,510 patients with suspected infection in Chinese multicenter cohort and validated in the MIMIC-IV database. Patients were divided into four categories based on concordance between SOFA versions: SOFA-1(-)/SOFA-2(-), SOFA-1(+)/SOFA-2(+), SOFA-1(-)/SOFA-2(+), and SOFA-1(+)/SOFA-2(-). Kaplan-Meier analyses and cox proportional hazards models assessed the associations with in-hospital mortality. Most patients met sepsis criteria under both SOFA versions (18,179/24,510; 74.2%95% confidential interval (CI), 73.6%-74.7%), while 9.9% (95% CI, 9.53%-10.2%) were classified as SOFA-1(+)/SOFA-2(-), 5.5% (95% CI, 5.2%-5.8%) as SOFA-1(-)/SOFA-2(+), and 10.2% (95% CI, 9.8%-10.6%) as SOFA-1(-)/SOFA-2(-). Patients in SOFA-1(-)/SOFA-2(+) group were younger, more often male, higher rates of pneumonia, gastrointestinal infection, and neurologic infection. Superior discrimination for in-hospital mortality was observed in SOFA-2 area under the receiver operating characteristic curve (AUROC) 0.746; 95% confidence interval (CI) 0.737–0.756 compared with SOFA-1 (0.679; 95% CI 0.668–0.689). In-hospital mortality was markedly higher in SOFA-2 (+) groups 2.9% in SOFA-1(-)/SOFA-2(-), 3.3% in SOFA-1(+)/SOFA-2(-), 10.6% in SOFA-1(-)/SOFA-2(+), and 14.3% in SOFA-1(+)/SOFA-2(+); p < 0.001. Compared with SOFA-1(-)/SOFA-2(-), the hazard ratios for in-hospital mortality were 0.99 (95% CI, 0.72–1.36) for SOFA-1(+)/SOFA-2(-), 3.52 (95% CI, 2.66–4.67) for SOFA-1(-)/SOFA-2(+), and 4.44 (95% CI, 3.52–5.60) for SOFA-1(+)/SOFA-2(+). These patterns were consistent in sensitivity analyses. SOFA-2-based sepsis identified a clinically meaningful higher-risk subgroup, with consistent results across sensitivity analyses. Nevertheless, cohort-specific differences emphasize the need for cautious interpretation and further validation in diverse clinical settings.