SOFA-2 vs. SOFA and early warning scores for short-term mortality in older adults with sepsis in the emergency department

SOFA-2 was developed to update the SOFA score for contemporary clinical practice, but its performance in older adult emergency department (ED) patients with sepsis is uncertain. We evaluated SOFA-2 for short-term mortality prediction and compared it with SOFA and commonly used early warning scores. This single-center retrospective cohort study included 330 hospitalized ED patients aged ≥ 65 years with sepsis. SOFA-2, SOFA, NEWS, NEWS-L, and MEWS were calculated from the initial ED assessment. The outcomes were 1-day and 7-day in-hospital mortality. Discrimination was assessed using the area under the receiver operating characteristic curve (AUC), with pairwise comparisons by DeLong’s test and Holm–Bonferroni correction. Threshold-based sensitivity and specificity were also evaluated. Mortality occurred in 43 patients (13.0%) within 1 day and 141 (42.7%) within 7 days. SOFA-2 had an AUC of 0.83 (95% CI 0.77–0.89) for 1-day mortality and 0.76 (95% CI 0.70–0.81) for 7-day mortality, compared with 0.82 (95% CI 0.76–0.89) and 0.75 (95% CI 0.70–0.80), respectively, for SOFA; these differences were not significant (adjusted p = 0.526 and p = 0.756). SOFA-2 also did not differ significantly from NEWS-L for either outcome. It outperformed MEWS for both outcomes and NEWS for 7-day mortality, whereas the difference from NEWS for 1-day mortality was not significant after correction. At the intermediate-or-high-risk threshold, SOFA-2 had lower sensitivity than SOFA for 1-day (86.0% vs. 100.0%) and 7-day mortality (67.4% vs. 92.9%), but higher specificity (59.2% vs. 20.6% and 68.8% vs. 25.9%, respectively). SOFA-2 showed good discrimination for short-term mortality but did not improve prognostic performance over the original SOFA score in this cohort. Its lower sensitivity at the intermediate-or-high-risk threshold is a direct function of the non-validated cutoffs applied and should not be interpreted as evidence of reduced clinical utility. Larger multicenter studies are needed for external validation.