External validation of SOFA 2 score: a retrospective Chinese cohort study

The revised Sequential Organ Failure Assessment (SOFA) 2 score was developed to improve early identification of acute organ dysfunction, which restricted from incomplete data in resource-limited settings. To compare the performance of original (SOFA-1) and updated SOFA-2 scores in these settings.

The retrospective multicenter analysis included 118,542 adult patients admitted to intensive care units (ICUs) between 2014 and 2024. SOFA scores were calculated using clinical data from the first seven ICU days. The primary outcome was ICU mortality. Multiple approaches were applied to address missing data. We assessed the discriminative performance of total SOFA and organ-specific scores using the area under the receiver operating characteristic curve (AUROC).

A total of 5,286 patients (4.45%) died in the ICU, and 6,173 patients (5.2%) died during hospitalization. Across all cohorts, higher SOFA-2 subscores (0-4) and total scores were associated with lower frequency and higher ICU mortality. SOFA-2 demonstrated significantly better discrimination for ICU mortality than SOFA-1 [AUROC of 80.9% (95% CI, 80.3%-81.5%) vs. 76.7% (95% CI, 76.0%-77.4%)]. Each 1-point increase in SOFA-2 was associated with increasing odds of ICU mortality (OR 1.33, 95% CI 1.33-1.35). Between-cohort variance was lower for SOFA-2 (0.07) than for SOFA-1 (0.23), indicating greater robustness across these cohorts. Longitudinal analyses showed widening divergence in SOFA-2 trajectories between survivors and non-survivors over the first 7 days.

In this external validation, SOFA-2 demonstrated improved discrimination for ICU mortality and more consistent score distribution than SOFA-1. These findings support the broader applicability of SOFA-2 across diverse healthcare settings, including those with substantial resource constraints.