Diagnostic criteria for acute kidney injury based upon serum creatinine reference change in the diagnosis of acute kidney injury and prediction of progression of chronic kidney disease and risk of allcause mortality in patients with chronic kidney disease

Objective To evaluate a new criterion based upon the creatinine reference change (cROCK)for diagnosing acute-on-chronic kidney injury(ACKI)and predicting renal outcomes and allcause mortality of patients with stage 3 chronic kidney disease(CKD). Methods From January 1,2016 to June 30,2018,clinical data were collected from 90,864 patients admitted into Xuzhou Central Hospital,including 52 647 males with an average age of(60. 75±18. 99)(3-102)years and 38 217 females with an average age of(58. 36±18. 54)(3-104)years old. The cROCK and KDIGO criteria were employed for diagnosing ACKI. Follow-ups continued for 3 years from the time of ACKI diagnosis. Renal composite endpoint events and all-cause mortality were recorded as the endpoints. Results A total of 790 adult patients with stage 3 CKD were included. More ACKI patients were detected by cROCK than KDIGO criteria(68. 1% vs 59. 7%,P=0. 001). Kappa value of KDIGO and cROCK criteria for diagnosing ACKI was 0. 640,indicating a relatively strong consistency(P<0. 001). According to whether a diagnosis of ACKI fulfilled the KDIGO and/or cROCK criteria,790 patients were divided into four groups of KDIGO(-)cROCK(-),KDIGO(-)cROCK(+),KDIGO(+)cROCK(-)and KDIGO (+)cROCK(+). The results indicated that patients had earlier and higher risk of renal composite endpoint events and all-cause mortality in KDIGO(+)cROCK(+)group than those in KDIGO(+) cROCK(-) group (P<0. 01). Patients had earlier and higher risk renal outcomes in KDIGO (-) cROCK(+)group than those in KDIGO(-)cROCK(-)group,as well as higher risk of all-cause mortality(P<0. 01). COX regression analysis revealed that KDIGO(-)cROCK(+)group had worse renal prognosis(adjusted HR=2. 547,95% CI:1. 928-3. 365,P<0. 01)and worse survival outcomes(adjusted HR=2. 199,95% CI:1. 263-3. 829,P<0. 01)than KDIGO(-)cROCK(-)group. Similarly, KDIGO(+)cROCK(+)group had worse renal prognosis(adjusted HR=1. 995,95% CI:1. 322-3. 011,P<0. 01)and worse survival outcomes(adjusted HR=3. 136,95% CI:1. 716-5. 732,P<0. 01) than KDIGO(+)cROCK(-)group. According to the receiver operating characteristic(ROC)curve, KDIGO & cROCK had higher capabilities of predicting renal outcomes(AUC=0. 844 for KDIGO & cROCK;AUC=0. 803 for KDIGO;AUC=0. 842 for cROCK,P<0. 001) and all-cause mortality (AUC=0. 887 for KDIGO & cROCK;AUC=0. 865 for cROCK;AUC=0. 880 for KDIGO,P<0. 001) than KDIGO/cROCK alone. Conclusion As compared with the KDIGO criteria,cROCK is more sensitive in detecting ACKI. And combining cROCK with KDIGO criteria in diagnosing ACKI is beneficial for risk stratification of stage 3 CKD patients and it further improves timeliness of identifying poor prognosis. Differences exist in predicting adverse outcomes between two criteria. And cROCK predicts renal outcomes better than KDIGO and KDIGO forecasts all-cause mortality better than cROCK. Also cROCK plus KDIGO criteria may hint at poor outcomes better than cROCK/KDIGO alone.