External Validation Of Three Clinical Scores Predicting Acute Renal Failure After Cardiac Surgery

External Validation of Three Clinical Scores Predicting Acute Renal Failure after Cardiac SurgeryObjectiveTo assess the prognostic accuracy for acute renal failure(ARF) after cardiac surgery of 3 different predictive models in Chinese patients underwent cardiac surgery.MethodsBetween June,2009 and November,2010.504 patients with complete clinical data undergoing cardiac surgery in our institution were prospectively scored for postoperative renal replacement therapy(RRT) and AKI using Cleveland ARF Score, Acute kidney injury prediction following elective cardiac surgery (AKICS) and Simplified Renal Index (SRI)。Calibration Was estimated using the Hosmer—Lemeshow goodness-of-fit test. Discrimination was evaluated using receiver operating characteristic (ROC) curves and area under a ROC curve(AUC).ResultFollow-up was able to be completed in 504 prospectively scored patients. The overall incidence of postoperative RRT was 3.17%(16/504), among which the mortality was 37.5%, and the incidence of AKI was 5.36%(27/504). Discrimination for the prediction of RRT and AKI was good for SRI scoring models measured using areas under the receiver operating characteristic curve (AUROCs):0.759 (95% CI,0.643-0.874) for RRT and 0.773 (95% CI,0.677-0.868) for AKI. Unlike the outcome reported by other centers, SRI score showed better discrimination compared with the Cleveland ARF score and AKICS.ConclusionThe SRI scoring system offers the best discriminative value to predict postoperative RRT and was most suitable to the Chinese patients undergoing cardiac surgery. It also can be used for prediction of the composite end point of AKI, which enables broader application to patients at risk of postoperative kidney dysfunction.