Comparison of the RIFLE, AKIN and KDIGO criteria to predict mortality in critically ill patients

OBJECTIVE: Acute kidney injury is a common complication in critically ill patients, and the RIFLE, AKIN and KDIGO criteria are used to classify these patients. The present study's aim was to compare these criteria as predictors of mortality in critically ill patients. METHODS: Prospective cohort study using medical records as the source of data. All patients admitted to the intensive care unit were included. The exclusion criteria were hospitalization for less than 24 hours and death. Patients were followed until discharge or death. Student's t test, chi-squared analysis, a multivariate logistic regression and ROC curves were used for the data analysis. RESULTS: The mean patient age was 64 years old, and the majority of patients were women of African descent. According to RIFLE, the mortality rates were 17.74%, 22.58%, 24.19% and 35.48% for patients without acute kidney injury (AKI) in stages of Risk, Injury and Failure, respectively. For AKIN, the mortality rates were 17.74%, 29.03%, 12.90% and 40.32% for patients without AKI and at stage I, stage II and stage III, respectively. For KDIGO 2012, the mortality rates were 17.74%, 29.03%, 11.29% and 41.94% for patients without AKI and at stage I, stage II and stage III, respectively. All three classification systems showed similar ROC curves for mortality. CONCLUSION: The RIFLE, AKIN and KDIGO criteria were good tools for predicting mortality in critically ill patients with no significant difference between them.