Augmenting SOFA-2 with immune markers to capture dysregulated host response in patients with suspected infection

The updated SOFA-2 score excludes the immune system due to specificity concerns, creating a conceptual gap given that sepsis is defined by a dysregulated host response. We hypothesized that augmenting SOFA-2 with widely available immune markers-white blood cell counts, lymphocyte counts, and the neutrophil-to-lymphocyte ratio (NLR)-would improve mortality prediction in patients with suspected infection.

We analyzed 75,203 adult patients from the MIMIC-IV, MIMIC-III, and eICU databases. We constructed an immune-augmented SOFA-2 model using provisional consensus criteria for cell counts and data-driven thresholds for NLR derived via generalized additive models. Discrimination was evaluated using random-effects meta-analysis of the area under the receiver operating characteristic (AUROC) curve and net reclassification improvement (NRI).

The immune-augmented model significantly outperformed standard SOFA-2 across all cohorts. In the pooled analysis, the immune-augmented SOFA-2 yielded a higher AUROC (0.750 [95% CI, 0.715-0.785]) compared to standard SOFA-2 (0.740 [95% CI, 0.712-0.768]; p=0.003). Additionally, incorporating immune markers improved clinical risk stratification, resulting in a significant pooled total NRI of 0.046 (95% CI, 0.014-0.078; p=0.005).

Excluding the immune system limits the ability of the SOFA-2 score to capture the dysregulated host response essential to severe infection. Integrating immune markers significantly enhances prognostic accuracy and risk reclassification for patients with suspected infection.