Performance of the quick SOFA in very old ICU patients admitted with sepsis

Abstract

BACKGROUND: The number of very elderly ICU patients (abbreviated to VOPs; ≥80 years) with sepsis increases. Sepsis was redefined in 2016 (sepsis 3.0) using the quick SOFA (qSOFA) score. Since then, multiple studies have validated qSOFA for prognostication in different patient categories, but the prognostic value in VOPs with sepsis is still unknown. METHODS: Retrospective cohort study including patients admitted to Dutch ICUs with sepsis, in the period 2012 to 2016, evaluating the outcome and the performance of qSOFA, an extended qSOFA model, SOFA, SAPS II and APACHE IV for hospital mortality. RESULTS: 5969 patients were included, of which 935 VOPs. Crude hospital mortality rates were 19%, 28% and 39% for patients aged 18-65, 65-80 and ≥80 years respectively. Discriminative performance of qSOFA for in-hospital mortality in VOPs was poor (AUC 0.596) and lower than that of SOFA, APACHE IV and SAPS II (0.704, 0.722 and 0.780 respectively). A qSOFA model extended with several other characteristics (AUC 0.643) was non-inferior to the full SOFA, but still inferior to APACHE IV and SAPS II, for all age groups. The Hosmer-Lemeshow goodness-of-fit test showed non-significant p-values for all models. Accuracy for both qSOFA and the extended qSOFA was lower compared to APACHE IV and SAPS II (Brier scores 0.227, 0.223, 0.184 and 0.183 respectively). CONCLUSION: The qSOFA showed worse discriminative performance to predict mortality than SOFA, APACHE IV and SAPS II in both VOPs and younger patients admitted with sepsis.