It is unknown which definition of contrast-induced acute kidney injury (CI-AKI) in the setting of percutaneous coronary interventions is best associated with inpatient mortality and whether this association is stable across patients with various preprocedural serum creatinine (SCr) values.

We applied logistic regression models to multiple CI-AKI definitions used by the Kidney Disease Improving Global Outcomes guidelines and previously published studies to examine the impact of preprocedural SCr on a candidate definition's correlation with the adverse outcome of inpatient mortality. We used likelihood ratio tests to examine candidate definitions and identify those where association with inpatient mortality remained constant regardless of preprocedural SCr. These definitions were assessed for specificity, sensitivity, and positive and negative predictive values to identify an optimal definition.

Our study cohort included 119,554 patients who underwent percutaneous coronary intervention in Michigan between 2010 and 2014. Most commonly used definitions were not associated with inpatient mortality in a constant fashion across various preprocedural SCr values. Of the 266 candidate definitions examined, 16 definition's association with inpatient mortality was not significantly altered by preprocedural SCr. Contrast-induced acute kidney injury defined as an absolute increase of SCr ≥0.3 mg/dL and a relative SCr increase ≥50% was selected as the optimal candidate using Perkins and Shisterman decision theoretic optimality criteria and was highly predictive of and specific for inpatient mortality.

We identified the optimal definition for CI-AKI to be an absolute increase in SCr ≥0.3 mg/dL and a relative SCr increase ≥50%. Further work is needed to validate this definition in independent studies and to establish its utility for clinical trials and quality improvement efforts.