A simplified bedside model for prediction of acute kidney injury after percutaneous coronary intervention: CRASH...

Acute kidney injury (AKI) following percutaneous coronary intervention (PCI) is associated with poor outcomes. The National Cardiovascular Data Registry (NCDR) AKI prediction model includes a relatively high number of complex clinical variables. We propose a simplified bedside score including 5 pre-procedural variables from the NCDR model as an effective tool for AKI prediction after PCI.

The Mayo Clinic Rochester cardiac catheterization laboratory registry, which follows NCDR criteria, was analyzed between Q2 2022 and Q1 2023. Inclusion/exclusion criteria, definitions, and outcome data were based on the NCDR model. Logistic regression models were built to predict AKI, including a simplified integer score comprised of 5 pre-procedural variables: Cardiac arrest, Reduced glomerular filtration rate < 30 ml/min/1.73 m2, Anemia (Hb <10 mg/dl), Shock, and Heart failure history. Model calibration and discrimination were tested.

We identified 840 patients with a mean age of 70 (IQR 61-79), of which 593 (70.6 %) were men. AKI developed in 107 (12.7 %) patients. There was no significant difference between the groups in gender, vascular access, PCI indication, or contrast volume used (median: 150; IQR: 110-200 ml in the non-AKI group and median: 170 IQR: 123-213 ml in the AKI group; p = 0.14). The CRASH score demonstrated a c-statistic of 0.77 (0.60-0.88) and calibration intercept of 0.31 (-0.07-0.69) and slope of 1.03 (0.66-1.39) in the validation data, similar to the NCDR score.

A simplified integer score using 5 pre-procedural variables was an effective tool for risk prediction of post-PCI AKI, with performance similar to the more complex NCDR model.