Early Mortality Stratification with Serum Albumin and the Sequential Organ Failure Assessment Score at Emergency Department Admission in Septic Shock Patients

Life (Basel). 2024 Oct 2;14(10):1257. doi: 10.3390/life14101257.

Abstract

Background: Early risk stratification is crucial due to septic patients' heterogeneity. Serum albumin level may reflect the severity of sepsis and host status. This study aimed to evaluate the prognostic ability of the initial sequential organ failure assessment (SOFA) score alone and combined with serum albumin levels for predicting 28-day mortality in patients with septic shock. Methods: We conducted an observational study using a prospective, multicenter registry of septic shock patients between October 2015 and May 2022 from 12 emergency departments in the Korean Shock Society and the results were validated by examining those from the septic shock cohort in Asan Medical Center. The primary outcome was 28-day mortality. The area under the receiver operating characteristic (ROC) curve was used to compare the predictive values of SOFA score alone and SOFA score combined with serum albumin level. Results: Among 5805 septic shock patients, 1529 (26.3%) died within 28 days. Mortality increased stepwise with decreasing serum albumin levels (13.6% in albumin ≥3.5, 20.7% in 3.5-3.0, 29.7% in 3.0-2.5, 44.0% in 2.5-2.0, 56.4% in <2.0). The albumin SOFA score was calculated by adding the initial SOFA score to the 4 points assigned for albumin levels. ROC analysis for predicting 28-day mortality showed that the area under the curve for the albumin SOFA score was 0.71 (95% CI 0.70-0.73), which was significantly higher than that of the initial SOFA score alone (0.68, 95% CI: 0.67-0.69). Conclusions: The combination of the initial SOFA score with albumin can improve prognostic accuracy for patients with septic shock, suggesting the albumin SOFA score may be used as an early mortality stratification tool.

Keywords: SOFA score; albumin; prognosis; risk stratification; septic shock.

Grants and funding

This research was supported by a research grant from Medical AI Co., Ltd.