Comparative Performance of Prediction Models for Contrast-Associated Acute Kidney Injury After Percutaneous Coronary...

Area of Science:

• Nephrology
• Cardiology
• Medical Informatics

Background:

• Contrast-associated acute kidney injury (CA-AKI) is a significant complication following percutaneous coronary intervention (PCI).
• Accurate risk stratification is essential for targeted prevention strategies and quality improvement in PCI procedures.
• Evaluating existing risk prediction models is a foundational step for improving patient outcomes.

Purpose of the Study:

• To validate and compare the performance of established risk prediction models for CA-AKI.
• To assess the effectiveness of model recalibration in enhancing risk stratification accuracy.
• To identify the best-performing models for stratifying CA-AKI risk in patients undergoing PCI.

Main Methods:

• Validation of seven CA-AKI and three dialysis-requiring AKI risk models using two CA-AKI definitions (KDIGO and historical).
• Performance assessment based on discrimination (C-statistics), calibration, and net reclassification index.
• Comparison of model performance before and after recalibration in a cohort of 7888 patients undergoing PCI.

Main Results:

• CA-AKI incidence was 4.2% (KDIGO) and 7.3% (historical definition); dialysis-requiring AKI was 0.6%.
• The two most recent CA-AKI models demonstrated superior discrimination (C-statistics 0.75-0.76) compared to older models (0.61-0.68).
• Model recalibration significantly improved risk stratification for some models, though calibration varied.

Conclusions:

• More recent CA-AKI prediction models offer better discrimination than older ones.
• Model recalibration is a critical step to enhance prediction accuracy, especially for guiding clinical management.
• External validation of recalibrated models is recommended to confirm improved accuracy in diverse patient populations.