Clinical prediction models for acute kidney injury

Area of Science:

• Nephrology
• Critical Care Medicine
• Biostatistics

Background:

• Acute kidney injury (AKI) is a significant complication in intensive care units (ICUs).
• Clinical prediction models aim to identify high-risk patients for early intervention.
• Heterogeneity in model development and validation poses challenges.

Purpose of the Study:

• To systematically review existing prediction models for AKI in adult ICUs.
• To assess the development, validation, and impact of these models.

Main Methods:

• Systematic literature search of PubMed for studies on AKI prediction models in ICUs.
• Inclusion criteria focused on model development, validation, or impact analysis.
• Screening of 583 articles, with 5 meeting final inclusion criteria.

Main Results:

• Models varied in complexity (5-19 predictors) and included sepsis, age, and creatinine.
• Internal validation showed good predictive performance (AUC 0.78-0.88).
• Only two models underwent external validation; definitions of AKI were inconsistent.

Conclusions:

• AKI prediction models can guide preventive strategies in ICUs.
• Lack of external validation and standardized definitions limits current clinical utility.
• Further research is required to validate models and assess their real-world impact.