Sequential evaluation of prognostic models in the early diagnosis of acute kidney injury in the intensive care unit.

General and specific severity scores for patients with acute kidney injury have significant limitations due in part to the diversity of methods that have been used. Here we prospectively validated five general (APACHE II, SAPS II, SOFA, LODS, and OSF) and three specific (SHARF, Liaño, and Mehta) scoring systems in 366 critically ill patients who developed acute kidney injury in the intensive care unit. Sequential scores in each system were determined on the day that acute kidney injury was diagnosed, on the day when acute kidney injury-specific score criteria were achieved, and on the day of initial nephrology consultation. Acute kidney injury, defined as an increase of 50% or more in the baseline serum creatinine, was mainly due to sepsis, and had an incidence of 19% and an overall 68% mortality. A progressive improvement in score performance was found. On the day of initial nephrology consultation, most scores showed a good performance and two indices (SAPS II and SHARF) achieved an area under the receiver operating characteristic curve above 0.80. Calibration was good on all three defining days, except for OSF when score criteria were achieved, and Mehta at the time of nephrology consultation. Our study shows that early and sequential evaluation is a better approach for prognostic scoring in critically ill patients who develop acute kidney injury.