Improving the performance of the Cleveland Clinic Score for predicting acute kidney injury after cardiac surgery: a prospective multicenter cohort study

Abstract

Background: Cardiac surgery-associated acute kidney injury (CSA-AKI) is associated with high short- and long-term mortality rates. The prediction of CSA-AKI is crucial for early detection and treatment. Current predictive models may be improved by potentially useful preoperative and intraoperative information.

Methods: This multicenter prospective cohort study recruited 261 consecutive patients at high risk for developing CSA-AKI, based on a Cleveland Clinical Score (CCS) of ≥4 points from July to December 2017 in 14 hospitals in Spain and the UK. Postoperative AKI occurred in 145 (55.5%) patients. The receiver operating characteristics curve (AUC) of a base model including only the CCS was compared with models including additional preoperative and intraoperative variables such as the estimated glomerular filtration rate (eGFR) instead of plasmatic creatinine, intraoperative urine output, baseline hemoglobin, nadir hemoglobin, and glycosylated hemoglobin (HbA<inf>1c</inf>) instead of diabetes mellitus. The performance of each model for AKI was compared.

Results: The CCS alone gave an AUC of 0.67 (95% CI, 0.56-0.78) for postoperative AKI. None of the single variables added to the base model CCS improve discrimination. The AUC for postoperative AKI was improved when baseline hemoglobin, eGFR instead of plasmatic creatinine, HbA<inf>1c</inf>, and nadir hemoglobin were added to the CCS (AUC=0.77; 95% CI, 0.67-0.87; P=0.02).

Conclusions: The addition of baseline hemoglobin, eGFR, HbA<inf>1c</inf>, and nadir intraoperative hemoglobin may be useful for improving the discrimination of the clinical predictive risk scores for AKI.