Objectives: There is an intense debate on whether the RIFLE (R-renal risk, I-injury, F-failure, L-loss of kidney function, E-end-stage renal disease) classification or its recent modification, the Acute Kidney Injury Network definition and classification system should be used to standardize research on acute kidney injury. In this study we compared these classifications with regard to (1) the detection of acute kidney injury, (2) their agreement according to the grading of acute kidney injury across classes, and (3) their prognostic value.
Methods: We prospectively enrolled 282 cardiac surgery patients undergoing cardiopulmonary bypass and assigned a RIFLE and Acute Kidney Injury Network class to each patient. The incidence of acute kidney injury and in-hospital mortality across classes was compared by using the chi(2) test, and their prognostic value was compared by using the area under the curve receiver-operating characteristic for in-hospital mortality.
Results: According to the RIFLE (45.8%) or Acute Kidney Injury Network (44.7%) classification, a similar proportion of patients had acute kidney injury. There was large agreement between classifications according to patients graded as having nonacute kidney injury; however, there was some disagreement across classes for staging the severity of acute kidney injury. The area under the curve for in-hospital mortality was similar for all classifications: 0.91 for the RIFLE classification (95% confidence interval, 0.82-0.99) and 0.94 for the Acute Kidney Injury Network classification (95% confidence interval, 0.81-0.97; P = .6 for area under the curve comparison).
Conclusions: In patients undergoing cardiac surgery, modifications of the RIFLE classification for acute kidney injury do not materially improve the clinical usefulness of the definition. Other factors, such as the applicability of the acute kidney injury definition and classification system to be applied, need to be considered.