Use of the SOFA score to assess the incidence of organ dysfunction/failure in intensive care units: results of a multicenter, prospective study. Working group on "sepsis-related problems" of the European Society of Intensive Care Medicine

Abstract

Objective: To evaluate the use of the Sequential Organ Failure Assessment (SOFA) score in assessing the incidence and severity of organ dysfunction in critically ill patients.

Design: Prospective, multicenter study.

Setting: Forty intensive care units (ICUs) in 16 countries.

Patients: Patients admitted to the ICU in May 1995 (n = 1,449), excluding patients who underwent uncomplicated elective surgery with an ICU length of stay <48 hrs.

Interventions: None.

Measurements and main results: The main outcome measures included incidence of dysfunction/failure of different organs and the relationship of this dysfunction with outcome. In this cohort of patients, the median length of ICU stay was 5 days, and the ICU mortality rate was 22%. Multiple organ dysfunction and high SOFA scores for any individual organ were associated with increased mortality. The presence of infection on admission (28.7% of patients) was associated with higher SOFA scores for each organ. The evaluation of a subgroup of 544 patients who stayed in the ICU for at least 1 wk showed that survivors and nonsurvivors followed a different course. This subgroup had greater respiratory, cardiovascular, and neurologic scores than the other patients. In this subgroup, the total SOFA score increased in 44% of the nonsurvivors but in only 20% of the survivors (p < .001). Conversely, the total SOFA score decreased in 33% of the survivors compared with 21% of the nonsurvivors (p < .001).

Conclusions: The SOFA score is a simple, but effective method to describe organ dysfunction/failure in critically ill patients. Regular, repeated scoring enables patient condition and disease development to be monitored and better understood. The SOFA score may enable comparison between patients that would benefit clinical trials.