Mortality Discrimination of SOFA-2 versus SOFA-1 in Older Patients with Sepsis: An Age- Stratified Single-Centre Cohort Study

Abstract Objective To compare the mortality discrimination of SOFA-2 and SOFA-1 across the ageing spectrum in older patients with sepsis, with particular attention to the oldest-old (aged ≥ 85 years), for ICU, 28-day, and 6-month mortality. Methods We conducted a retrospective, single-centre cohort study of consecutive patients aged ≥ 60 years with sepsis admitted to the ICU of the General Hospital of Southern Theater Command between 2020 and 2025. SOFA-1 and SOFA-2 were calculated at ICU admission. Discrimination for ICU, 28-day, and 6-month mortality was assessed using the area under the curve (AUC). Score changes between SOFA-1 and SOFA-2 were examined, and analyses were stratified by age (60–84 and ≥ 85 years). Within each age stratum, paired AUCs were compared using DeLong's test. To formally assess heterogeneity by age, differences in the change in AUC (ΔAUC = AUC_SOFA-2 − AUC_SOFA-1) between age strata were estimated; 95% confidence intervals were additionally evaluated using 5,000 stratified bootstrap resamples. Results Among 972 older patients with sepsis, 635 (65.3%) were aged ≥ 85 years. SOFA-2 assigned a different total score from SOFA-1 in 65.3% of patients, with scores decreasing in 45.7% and increasing in 19.6%. Overall, SOFA-2 showed higher discrimination than SOFA-1 for ICU mortality (AUC 0.699 vs. 0.667), 28-day mortality (0.719 vs. 0.695), and 6-month mortality (0.686 vs. 0.671; all P < 0.05). Among patients aged ≥ 85 years, SOFA-2 showed higher discrimination than SOFA-1 for ICU (0.714 vs. 0.676), 28-day (0.726 vs. 0.696), and 6-month mortality (0.689 vs. 0.668; all P < 0.05). In patients aged 60–84 years, higher discrimination was observed only for ICU mortality (0.710 vs. 0.684; P < 0.05), but not for 28-day (0.709 vs. 0.697) or 6-month mortality (0.692 vs. 0.685; both P > 0.05). Improvements in AUC were numerically greater in patients aged ≥ 85 years, but formal heterogeneity testing showed no statistically significant between-age-group differences in ΔAUC for ICU mortality (difference, 0.013; 95% CI, − 0.012 to 0.038; P = 0.308), 28-day mortality (0.018; 95% CI, − 0.005 to 0.041; P = 0.128), or 6-month mortality (0.014; 95% CI, − 0.008 to 0.036; P = 0.214). Conclusions SOFA-2 modestly improved mortality discrimination over SOFA-1 in older patients with sepsis. The improvement was statistically significant across all three mortality endpoints among patients aged ≥ 85 years, whereas only ICU mortality showed a significant improvement among those aged 60–84 years. Although improvements were numerically greater in the oldest-old, formal heterogeneity testing did not confirm a statistically significant difference in ΔAUC between age groups. These findings suggest, but do not establish, potential age-related heterogeneity in the prognostic performance of SOFA-2.