Prediction of the severity of acute kidney injury after on-pump cardiac surgery

On-pump cardiac surgery is associated with a high risk of acute kidney injury (AKI), which can substantially affect risk of mortality and morbidity depending on its severity. Current methods are limited in predicting AKI severity. This study aimed to develop and validate a model to predict AKI severity after on-pump cardiac surgery.Observational retrospective cohort study.Two tertiary general hospitals in China.The sample in this study came from two hospitals: 6919 patients who underwent on-pump cardiac surgery between January 1, 2011 and June 30, 2017 at West China Hospital of Sichuan University, and 1575 patients who underwent on-pump cardiac surgery between September 1, 2013 and June 30, 2017 at the Second Affiliated Hospital of Zhejiang University.Data from West China Hospital was used to develop an individualized prediction for AKI severity. The model was internally validated, and a cohort from the Second Affiliated Hospital of Zhejiang University was used for external validation.AKI incidence was 14.7% in the development cohort and 42.3% in the external validation cohort. Ten predictors of AKI severity were identified: age; sex; preoperative levels of serum creatinine, serum cystatin C, and blood urea nitrogen; preoperative red blood cell count; hypertension; cardiopulmonary bypass time; operation time; and red blood cell transfusion. The model incorporating these variables showed a concordance index of 0.730 (95% CI 0.713, 0.747) for predicting stage I AKI, 0.772 (95% CI 0.738, 0.806) for predicting stage II, and 0.770 (95% CI 0.712, 0.828) for predicting stage III in the development cohort. The corresponding indices for the external validation cohort were 0.676 (95% CI 0.650, 0.703), 0.730 (95% CI 0.691, 0.769), and 0.795 (95% CI 0.737, 0.852).The prediction model incorporating 10 predictors may be useful for predicting severity of AKI after on-pump cardiac surgery.