Prognostic Accuracy of Quick SOFA is different according to the severity of illness in infectious patients

Background: Sepsis-3 proposed the quick Sequential Organ Failure Assessment (qSOFA) to identify sepsis patients likely to have poor outcome. The clinical utility of qSOFA still remains controversial because its predictive accuracy for mortality is quite different across the validation studies. We hypothesized that one of the major causes for these controversial findings was the heterogeneity in severity across the studies, and evaluated the association between severity of illness and the prognostic accuracy of qSOFA. Materials and methods: This was a post hoc analysis of a prospective nationwide cohort of consecutive adult patients with sepsis in 59 intensive care units in Japan. Regression trees analysis for survival was used to classify patients according to severity of illness as determined by SOFA score on registration. We conducted receiver operating characteristic (ROC) analyses and evaluated the differences in the area under the ROC curve (AUROC). As a subgroup analysis, we conducted the above evaluations in emergency department (ED) and non-ED patients separately. Results: We included 1114 patients fulfilling the criteria and classified them into three subsets according to severity. The AUROC for mortality was significantly different according to the severity of illness (p = 0.007), with the highest AUROC being in the low-severity subset (patients with SOFA score ≤ 7). Interestingly, our subgroup analysis revealed that a significant difference in the AUROC of qSOFA was observed only in ED patients. Conclusion: This study suggested that lower severity of illness was associated with the relatively higher prognostic accuracy of qSOFA, especially in ED patients.