A risk score predicting unplanned renal replacement therapy after coronary catheterization.

Background Iodinated contrast medium exposure is associated with acute kidney injury (AKI). Patients with severe contrast-induced AKI (CI-AKI) may need renal replacement therapy (RRT). Prediction models exist for CI-AKI, but few need RRT. We aimed to establish a preprocedural score model to stratify patients at risk of unplanned postprocedural RRT following invasive coronary angiography (ICA) or percutaneous coronary intervention (PCI). Methods and results Between January 2010 and December 2015, a series of 3,469 patients were randomly divided into two cohorts at a 2 : 1 ratio for model development and validation, respectively. A total of 36 patients (1.0%) needed unplanned postprocedural RRT following ICA and/or PCI. Multivariable logistic regression was used to build the risk model. C-statistic and Hosmer-Lemeshow tests were used to evaluate the performance of the model. Five preprocedural variables - independently associated with unplanned postprocedural RRT - were identified as factors of the risk score model with different scores: age > 75 years (1), serum creatinine level ≥ 1.5 mg/dL (1), diabetes mellitus (1), hypotension (2), and acute myocardial infarction (2). The risk score model was demonstrated with high discrimination (C-statistic = 0.872) and goodness of fit (χ2 = 3.769, p = 0.438). Furthermore, the model allowed a hierarchical classification of low, intermediate, and high risk, within which the observed unplanned RRT rates were ~ 0.4, 3.0, and 20.0%, respectively. Conclusion Using preprocedural variables, we developed and validated a risk model for unplanned postprocedural RRT following ICA and/or PCI.