Validation of Four Prediction Scores for Cardiac Surgery-Associated Acute Kidney Injury in Chinese Patients

Abstract

ObjectiveTo assess the clinical value of four models for the prediction of cardiac surgery-associated acute kidney injury (CSA-AKI) and severe AKI which renal replacement therapy was needed (RRT-AKI) in Chinese patients.Methods1587 patients who underwent cardiac surgery in the department of cardiac surgery in the Zhongshan Hospital, Fudan University, between January 2013 and December 2013 were enrolled in this research. Evaluating the predicting value for cardiac surgery-associated AKI (AKICS score) and RRT-AKI (Cleveland score, SRI and Mehta score) by Hosmer-Lemeshow goodness-of-fit test for the calibration and area under receiver operating characteristic curve (AUROC) for the discrimination.ResultsBased on 2012 KDIGO (Kidney Disease: Improving Global Outcomes) AKI definition, the incidence of AKI and RRT-AKI was 37.4% (594/1587) and 1.1% (18/1587), respectively. The mortality of AKI and RRT-AKI was 6.1% (36/594) and 66.7% (12/18), respectively, while the total mortality was 2.8% (44/1587). The discrimination (AUROC=0.610) for the prediction of CSA-AKI of AKICS was low, while the calibration (x2=7.55, P=0.109) was fair. For the prediction of RRT-AKI, the discrimination of Cleveland score (AUROC=0.684), Mehta score (AUROC=0.708) and SRI (AUROC=0.622) were not good; while the calibration of them were fair (Cleveland score x2=1.918, P=0.166; Mehta score x2=9.209, P=0.238; SRI x2=2.976, P=0.271).ConclusionIn our single-center study, based upon valve surgery dominant and less diabetes mellitus patients, according to KDIGO AKI definition, the predictive value of the four models, combining discrimination and calibration, for respective primary event, were not convincible.

Pre-, Intra-, and Post-Operative Factors for Kidney Injury of Patients Underwent Cardiac Surgery: A Retrospective Cohort Study

“…The incidence of AKI in the analysis was in concordance with the findings of other studies conducted on Asian populations [15,23]. However, the incidence reported in the current finding was higher than reported in the other studies conducted using the AKIN criteria [24] and the RIFLE criteria [25]. The difference in the incidence reported might be attributed to several reasons, including varying definitions used to classify AKI, the timing of AKI evaluation, urine output versus SCr criteria, and population characteristics.…”

Acute Kidney Injury after Cardiac Surgery in Patients Without Chronic Kidney Disease

“…Among individuals undergoing cardiac surgery, mortality has been reported to be as high as 8% and postoperative AKI can increase the mortality rate to over 60% [ 7 ] . The occurrence of AKI in patients undergoing cardiac surgery raises the mortality rate from 0.4-4.4% to 1.3-22.3%; when those same patients require dialysis, rates range from 25% to 88.9%, making severe postoperative AKI an independent risk factor for mortality that results in an 8-fold increase in the risk of death [ 8 , 9 ] . Therefore, cardiac surgery AKI is associated with serious complications as well as with prolonged intensive care unit (ICU) stays and with a worse quality of life.…”

Prediction Scores for Acute Kidney Injury following Adult Cardiac Surgery

“…However, due to differences in race, religious belief, living habits and medical level, foreign prediction models cannot be effectively applied to other individual, local patient cohorts. Previous models have demonstrated poor discrimination and calibration in non-local populations [19][20][21][22]; hence, it is necessary to build prediction models for AKI for individual populations after cardiac surgery. For the Asian population, Jiang et al developed a dynamic predictive score for any-stage AKI after cardiac surgery.…”

“…Third, only two published scores (the Cleveland clinic score and the Mehta score) have undergone independent external validation by investigators. However, neither has demonstrated adequate discrimination and calibration in non-North American patient populations [19][20][21][22]. Fourth, the sample sizes of previous prediction models were relatively limited.…”