Renji acute kidney injury score is a useful tool to predict acute kidney injury after cardiac surgery

Objective To validate the effect of Renji acute kidney injury score (RAKIS) on predicting patients with acute kidney injury (AKI) after cardiac surgeries, and make comparison with Cleveland score, simplified renal index (SRI) and acute kidney injury following cardiac surgery (AKICS). Methods Patients undergoing open heart surgery from 2008/01/01 to 2010/10/31 in Renji hospital were enrolled, and their scores of those four scoring models were calculated. AKI patients were diagnosed by KDIGO, and those scores of AKI patients and non-AKI patients were compared. Receiver operating characteristic (ROC) curve and area under curve (AUC) were used to decide the predictive values of those models. Results A total of 1126 patients were chosen in this cohort, with the average age of (58.43±14.88) years (rang from 18 to 88). The male to female ratio was 1.47∶1. And 355(31.5%) patients were developed AKI. AKI stage Ⅰ, Ⅱ and Ⅲ were 65.4%, 23.7% and 11.0% respectively. RAKIS was significantly higher in AKI patients than in non-AKI patients (17.5 vs 9.0, P<0.001). The AUCs of RAKIS to predict AKI, AKI Ⅱ-Ⅲ stages, renal replacement therapy (RRT) and in-hospital death were 0.818, 0.819, 0.800 and 0.784 respectively. The AUCs of Cleveland score and SRI were 0.659 to 0.710, lower than those of RAKIS and AKICS. AKICS had lower value for predicting AKI and AKI Ⅱ-Ⅲ stages (AUC 0.766 and 0.793), but good value in predicting RRT and in-hospital death after surgery (AUC 0.804 and 0.835) as compared with RAKIS. Conclusions RAKIS is valid and accurate in the discrimination of KDIGO defined AKI patients, while for predicting the composite end point, AKICS may be more useful. Key words: Renal insufficiency, acute; Cardiac surgical procedures; Forecasting