Development and Validation of the Sequential Organ Failure Assessment (SOFA)-2 Score

Otavio T. Ranzani, MD, MSc, PhD^1,2,3; Mervyn Singer, MD⁴; Jorge I. F. Salluh, MD, PhD⁵et al

2025년 10월 29일

JAMA Published Online: October 29, 2025

DOI: 10.1001/jama.2025.20516

Abstract

Importance  Acute dysfunction of vital organs is the hallmark of critical illness. The Sequential Organ Failure Assessment (SOFA) score, the most widely adopted approach to describe organ dysfunction, has not been updated in 30 years and therefore may not appropriately capture current clinical practice and outcomes.

Objectives   To inform the data-driven component of an updated score (SOFA-2) in varied geographical and resource settings (stages 6-8) after expert input via a modified Delphi process (stages 1-5).

Design, Setting, and Participants   A federated analysis was performed on data collected from adult patients admitted to 1319 intensive care units (ICUs) in 9 countries (Australia, Austria, Brazil, France, Italy, Japan, Nepal, New Zealand, United States) between 2014 and 2023. Four representative multicenter cohorts containing data from 2 098 356 patients were used for data-driven score development and internal validation. External validation was performed on 6 cohorts containing data from 1 241 114 patients.

Main Outcomes and Measures     Content validity for organ dysfunction identified through the modified Delphi process should be reflected by predictive validity using the area under the receiver operating characteristic (AUROC) curve of the score measured on the first ICU day (higher scores indicate worse organ dysfunction).

Results   Of 3.34 million patient encounters, 270 108 (8.1%) died in the ICU (range, 4.5% to 20.5% across the 10 cohorts). SOFA-2 modified the 6 organ systems of the original SOFA score (brain, respiratory, cardiovascular, liver, kidney, hemostasis), including new variables and revised thresholds that better describe the organ dysfunction distribution from 0 to 4 points and their associated mortality (SOFA-2 AUROC, 0.79; 95% CI, 0.76-0.81; SOFA-1 AUROC, 0.77; 95% CI, 0.74-0.81). Evaluation of sequential SOFA-2 data from ICU day 1 to day 7 maintained its predictive validity. Insufficient data and lack of content validity precluded incorporation of gastrointestinal and immune dysfunction scores into SOFA-2.

Conclusions and Relevance   The SOFA-2 score, updated to include contemporary organ support treatments and new score thresholds, describes organ dysfunction in a large, geographically and socioeconomically diverse population of critically ill adults.

Figure 1.  SOFA-2 Update Stages and Methods

Stages 1-8 were aligned with Sequential Organ Failure Assessment (SOFA) principles and with 6 score utility domains: stages 1-5 and 7, reliability; content; construct; criterion validity: and clarity measurement, burden, and timeliness. Stages 6 and 8: construct and predictive validity.

Further details on rationale and Delphi methodology are provided in a companion article.¹¹

AUROC indicates area under the receiver operating characteristic curve; CART, classification and regression tree; GAM, generalized additive model; ICU, intensive care unit.

Key Points

Question   Does an updated Sequential Organ Failure Assessment (SOFA)-2 score describe organ dysfunction in critically ill patients and its association with intensive care unit (ICU) mortality?

Findings   The SOFA-2 score was developed and validated in 10 international multicenter cohorts of 3.3 million adult ICU patients. SOFA-2 includes the original 6 organ systems with a total score ranging from 0 to 24 (higher scores indicate worse organ dysfunction). Possible inclusion of immune and gastrointestinal systems was investigated but not added. The updated score now incorporates commonly used drugs and mechanical organ supports that were rarely or not used when the original version was published in 1996. Some thresholds were modified to improve predictive validity against ICU mortality.

Meaning   The SOFA-2 score, updated to include contemporary organ support treatments and new score thresholds, describes organ dysfunction, supported by good predictive validity, in a large, geographically and socioeconomically diverse population of critically ill adults.

요약

·         중요성   급성 장기 기능 부전은 중증 질환의 특징적인 징후이다. 가장 널리 사용되는 장기 기능 부전 평가 방법인 연속 장기 부전 평가(Sequential Organ Failure Assessment, SOFA) 점수는 

30년 동안 개정되지 않았으며, 따라서 현재의 임상 실무와 결과를 적절히 반영하지 못할 수 있다.

·         목적  전문가 의견을 수정된 델파이(modified Delphi) 과정을 통해 수집한 후(1–5단계), 다양한 지역적·자원 환경(6–8단계)에서 데이터 기반 구성요소를 포함한 업데이트된 점수(SOFA-2)를 개발하기 위한 근거를 마련하는 것이다.

·         연구 설계, 장소, 대상자   이 연구는 연합형(federated) 분석으로 수행되었으며, 2014년 부터2023년 사이에 호주, 오스트리아, 브라질, 프랑스, 이탈리아, 일본, 네팔, 뉴질랜드, 미국 등 9개국의 1319개 중환자실(ICU)에 입원한 성인 환자로부터 수집된 데이터를 이용하였다.

데이터 기반 점수 개발과 내부 검증을 위해 총 2,098,356명의 환자 데이터를 포함하는 4개의 다기관코호트가 사용되었으며, 외부 검증은 1,241,114명의 환자 데이터를 포함한 6개의 코호트를 이용하여 수행되었다.

·         주요 결과 및 측정치  수정된 델파이 과정을 통해 식별된 장기 기능 부전의 내용 타당성(content validity)은 중환자실 입원 첫날 측정된 점수의 수용자조작특성곡선 아래 면적(AUROC)을 이용하여평가된 예측 타당성(predictive validity)으로 반영되어야 한다. (점수가 높을수록 장기 기능 부전이 심함을 의미한다.)

·         결과  총 334만 건의 환자 입원 사례 중 270,108명(8.1%)이 중환자실에서 사망하였다 .

(10개 코호트에서 사망률 범위: 4.5%–20.5%)

SOFA-2는 기존 SOFA 점수의 6개 장기 시스템(뇌, 호흡기, 심혈관, 간, 신장, 혈액응고)을 수정하여,

새로운 변수와 개정된 임계값(thresholds)을 포함함으로써 0점에서 4점까지의 장기 기능 부전 분포와 이에 따른 사망률을 보다 정확하게 기술하였다.

• SOFA-2 AUROC: 0.79 (95% CI, 0.76–0.81)

• SOFA-1 AUROC: 0.77 (95% CI, 0.74–0.81)

  
  

중환자실 입원 1일부터 7일까지의 연속적인 SOFA-2 점수 평가에서도 예측 타당성이 유지되었다.

그러나 소화기계(gastrointestinal)와 면역(immune) 기능 부전 점수는 데이터 부족 및 내용 타당성 부족으로 인해 SOFA-2에 포함되지 않았다.

·         결론 및 임상적 의의   SOFA-2 점수는 현대의 장기 치료(organ support) 방식과 새로운 점수 

임계값을 반영하도록 개정되어, 전 세계적으로 지리적 및 사회경제적으로 다양한 중증 성인 환자 집단에서 장기 기능 부전을 기술하는 도구로서 유효성을 입증하였다.