Relationship between patients' outcomes and the changes in serum creatinine and urine output and RIFLE classification in a large critical care cohort database.

We report the stepwise application of the RIFLE classification in 155,624 admissions in the UK Intensive Care National Audit & Research Centre Case Mix Programme database. The assumptions required to define RIFLE and their relationship with renal replacement therapy (RRT) and ICU mortality were assessed. Previous reports had not explored the method of estimating baseline creatinine, the position of class boundaries, or interactions between urine volume (AKI-U) and the peak/estimated baseline creatinine (AKI-Cr) within 24 h of ICU admission. The risk of developing AKI strongly depended on the assumed GFR increasing from 36 to 58% across the recommended range. AKI-U was often seen without AKI-Cr, and moderate oliguria (under 850 ml/24 h) was a stronger predictor of mortality than any degree of AKI-Cr partly because mortality fell when peak/estimated baseline creatinine ratios exceed fourfold. Mild oliguria (850-1500 ml/24 h) was common (38,928 admissions, 26%) and had a similar association with mortality (relative risk 1.6, 95% CI: 1.5-1.6) as did AKI-Cr defined Failure (risk ratio 1.5, 95% CI: 1.5-1.6). However, AKI-Cr was a strong predictor for RRT, which was used in 17,802 (11%) of admissions. Nearly half (48%) of the Failure patients never received RRT; nonetheless, most (66%) survived critical care. Thus, although the RIFLE classification may be attempted in large population cohorts, there is significant heterogeneity of both renal and, in particular, vital outcomes within each class.