Risk prediction models for contrast induced nephropathy: systematic review

Area of Science:

• Nephrology
• Radiology
• Medical Informatics

Background:

• Contrast-induced nephropathy (CIN) is a significant risk following radiocontrast procedures.
• Accurate prediction models are crucial for identifying at-risk patients and implementing preventive measures.

Purpose of the Study:

• To systematically review and characterize validated prediction models for CIN.
• To assess the performance and clinical utility of existing CIN prediction models.

Main Methods:

• Systematic review of Medline, Embase, and CINAHL databases (inception to 2015).
• Inclusion of studies with derivation and validation cohorts for prediction models.
• Extraction of data on patient/procedural characteristics, model performance, and clinical usefulness.

Main Results:

• 16 studies identified 12 distinct CIN prediction models.
• Significant heterogeneity observed across studies.
• Internal validation (C-statistic 0.61-0.95) and external validation (0.57-0.86) showed variable discrimination.
• Top models incorporated chronic kidney disease, age, diabetes, heart failure, and hypotension.

Conclusions:

• Current clinical CIN prediction models have limited accuracy and applicability, mainly for coronary angiography.
• Need for development of models that enhance patient-centered decision-making.
• Further research required to improve CIN prevention strategies.