Objective: To evaluate the sensitivity/accuracy of 2 different acute kidney injury (AKI) diagnosis/classification criteria, the RIFLE (risk, injury, failure, loss of kidney function, end-stage kidney disease) and the acute kidney injury network (AKIN), for patients in intensive care unit (ICU).
Methods: Clinical data were collected from all adult patients admitted to the Department of Intensive Medicine in Guangdong General Hospital between October 2009 and July 2010, and AKI cases were identified/classified using RIFLE and AKIN criteria separately, for statistical evaluation of their diagnostic sensitivity, and accuracy in hospital mortality prediction.
Results: In all 524 patients evaluated, AKI were identified by RIFLE criteria in 95 of them, while by AKIN, 135. The AKI incidence by RIFLE (18.1%), and AKIN (25.8%) were significantly different (P < 0.05). Meanwhile, AKI incidence was found independent from the mortality, either by RIFLE or AKIN (both P < 0.001). In all patients, the area under the receiver operator characteristic curve (ROC curve), the index for hospital mortality prediction, was 0.7293 for RIFLE [with 95% confidence interval (95%CI) ranging from 0.6005 to 0.8581, P < 0.001], and for AKIN, 0.7777 (95%CI: 0.6664 - 0.8890, P < 0.001). No significant difference was found between the total hospital mortality by the two criteria (37.9% vs. 34.1%, P > 0.05).
Conclusion: Although AKIN criteria has higher sensitivity in AKI diagnosis, it is not different from the RIFLE criteria in predicting hospital mortality in critically ill patients.