Systemic inflammatory response syndrome criteria and the prediction of hospital mortality in critically ill patients: a retrospective cohort study

OBJECTIVE This study intended to determine whether the systemic inflammatory response syndrome criteria can predict hospital mortality in a Brazilian cohort of critically ill patients. METHODS We performed a retrospective cohort study at a private tertiary hospital in São Paulo (SP), Brazil. We extracted information from the adult intensive care unit database (Sistema EpimedTM). We compared the SAPS 3 and the systemic inflammatory response syndrome model as dichotomous (≥ 2 criteria: systemic inflammatory response syndrome -positive versus 0 - 1 criterion: systemic inflammatory response syndrome -negative) and ordinal variables from 0 to 4 (according to the number of systemic inflammatory response syndrome criteria met) in the prediction of hospital mortality at intensive care unit admission. Model discrimination was compared using the area under the receiver operating characteristics (AUROC) curve. RESULTS From January to December 2012, we studied 932 patients (60.4% were systemic inflammatory response syndrome -positive). systemic inflammatory response syndrome -positive patients were more critically ill than systemic inflammatory response syndrome -negative patients and had higher hospital mortality (16.9% versus 8.1%, p < 0.001). In the adjusted analysis, being systemic inflammatory response syndrome -positive independently increased the risk of death by 82% (odds ratio 1.82; 95% confidence interval [CI] 1.12 - 2.96, p = 0.016). However, the AUROC curve for the SAPS 3 model was higher (0.81, 95%CI 0.78 - 0.85) compared to the systemic inflammatory response syndrome model with the systemic inflammatory response syndrome criteria as a dichotomous variable (0.60, 95%CI 0.55 - 0.65) and as an ordinal variable (0.62, 95%CI 0.57 - 0.68; p < 0.001) for hospital mortality. CONCLUSION Although systemic inflammatory response syndrome is associated with hospital mortality, the systemic inflammatory response syndrome criteria show low accuracy in the prediction of mortality compared with the SAPS 3.