Sepsis, a life-threatening condition, remains a leading cause of death in ICUs worldwide. To assess the severity of organ dysfunction in critically ill patients, clinicians routinely use the Sequential Organ Failure Assessment (SOFA) score, a cornerstone of modern critical care and a central component of the Sepsis-3 definition. An acute rise of two or more SOFA points signals clinically significant organ failure and helps guide diagnosis, treatment decisions, and prognostication.
Recently, SOFA has been updated to a revised version, known as SOFA-2. The update was designed to better reflect contemporary ICU practice, incorporating changes in organ-support technologies, commonly used vasoactive drugs, and revised thresholds aligned with current mortality risk. While SOFA-2 was evaluated in broad ICU populations, one key question remained unanswered: does SOFA-2 perform as expected, specifically in patients with sepsis?
That question matters because SOFA is deeply embedded in sepsis care and research. If the revised score behaved differently in sepsis patients, it could alter how organ dysfunction is identified or how patients are classified under Sepsis-3 criteria. Addressing this gap, investigators from First affiliated hospital, Sun Yat-sen University conducted the first validation of SOFA-2 in a well-defined sepsis cohort and explored whether incorporating immune markers could further improve its performance.
The study was a secondary analysis of the TESTS trial, a large, multicenter randomized controlled trial conducted in China. The trial enrolled adult ICU patients diagnosed with sepsis according to Sepsis-3 criteria. Using a clinical trial dataset is a notable strength, as sepsis diagnoses in trials are typically more consistent and carefully adjudicated than in retrospective studies relying on electronic health records.
The analysis included 1,089 sepsis patients with a median age of 64.5 years, and over 31% were female. ICU mortality was 9.2%. Researchers recalculated both the original SOFA (SOFA-1) and the updated SOFA-2 at the time of randomization and compared score distributions and their ability to discriminate between patients who survived and those who died in the ICU.
Overall, SOFA-2 scores were slightly lower than SOFA-1 scores, with a median of 6 versus 7. Differences were mainly driven by changes in respiratory, cardiovascular, and liver components, while kidney scores tended to be higher under SOFA-2. Despite these shifts, the two versions were closely aligned at clinically important thresholds. Only 2.2% of patients met the traditional SOFA-1 threshold of two or more points but fell below that threshold using SOFA-2. “ This suggests that, in practice, switching from SOFA-1 to SOFA-2 would rarely change whether a sepsis patient is identified as having clinically meaningful organ dysfunction,” the authors said.
The two scores also showed nearly identical ability to discriminate between ICU survivors and non-survivors. The area under the receiver operating characteristic curve was 0.646 for SOFA-2 and 0.641 for SOFA-1, with similar results for 28-day and 90-day mortality. While SOFA was never intended to be a standalone mortality prediction tool, mortality is closely linked to the severity of organ failure. Comparable performance, therefore, suggests that SOFA-2 preserves the clinical signal clinicians rely on while reflecting modern ICU care.
The investigators also examined whether adding immune markers could improve SOFA-2’s predictive performance. Although immune dysfunction is central to sepsis, an immune domain was excluded from SOFA-2 because commonly available markers, such as white blood cell or lymphocyte counts, were considered insufficient proxies. Using a machine-learning approach, the researchers tested five immune-related indicators available in the dataset—including white blood cell count, lymphocyte count, monocyte HLA-DR, neutrophil-to-lymphocyte ratio, and regulatory T-cell percentage—across multiple combinations. None meaningfully improved predictive performance.
Rather than a failure, the authors interpret this as reflecting the biological complexity of sepsis. Immune dysregulation is dynamic, heterogeneous, and difficult to capture with single baseline measurements, limiting the usefulness of simple immune markers in a one-number score.
“Together, these findings provide the first sepsis-specific validation of SOFA-2, supporting its use in sepsis care and research. The study also underscores the challenges of integrating immune biomarkers into bedside scoring systems,” the authors conclude.
About the Authors
Qingui Chen is an Associate Research Fellow and PhD advisor in the Department of Critical Care Medicine at the First Affiliated Hospital of Sun Yat-sen University, and a recipient of Sun Yat-sen University’s Hundred Talents Program. His research interests focus on clinical and translational critical care research, with an emphasis on integrating big data analytics and state-of-the-art clinical research methodologies.
Jianfeng Wu , Professor and PhD Supervisor, Department of Critical Care Medicine, The First Affiliated Hospital of Sun Yat-sen University. He serves as a Council Member and Deputy Secretary-General of the Critical Care Medicine Branch of the Chinese Medical Association. As corresponding author or first author, he has published original research articles in leading international journals, including BMJ, Signal Transduction and Targeted Therapy, Military Medical Research, Critical Care Medicine, Critical Care, and Annals of Intensive Care. He has received multiple prestigious awards, including the First Prize of the Guangdong Provincial Science and Technology Progress Award, the First Prize of the Military Science and Technology Progress Award, and the Third Prize of the Chinese Medical Science and Technology Award.
Journal of Intensive Medicine
Observational study
Validation of SOFA-2 score in sepsis and exploration of its extension with additional immune markers
8-Jan-2026
JW is an Editorial Board Member for Journal of Intensive Medicine and was not involved in the editorial review or the decision to publish this article. All authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.