Review assesses prediction models for acute kidney injury after cardiac surgery

A systematic review and meta-analysis published in Critical Care assessed clinical prediction models for cardiac surgery-associated acute kidney injury (CSA-AKI) in adults undergoing surgery with cardiopulmonary bypass. Researchers searched four databases, used independent data extraction, and pooled predictive performance with random-effects models where feasible. The supplied report does not specify the number of included studies or patients.

The pooled incidence of CSA-AKI was 26.3% (95% CI, 26.0–26.4), while 2.2% (95% CI, 2.1–2.2) required renal replacement therapy. Among validated models, Thakar 2005 had the most extensive validation and the best reported performance for predicting renal replacement therapy: pooled area under the curve 0.80 (95% CI, 0.77–0.82). At a threshold of at least 3 points, sensitivity was 0.87 and specificity was 0.49. Calibration analyses indicated risk underestimation. Mehta 2006 and Wijeysundera 2007 also showed good discrimination.

Prediction of milder CSA-AKI was less reliable. Newer models, including Demirjian 2022, showed promising discrimination but had limited validation evidence. These findings support risk stratification, although low specificity and underestimated absolute risk warrant caution when interpreting individual predictions. The review assessed predictive performance rather than demonstrating improved outcomes from model-guided care. Funding came from the Government of Navarre, Spain; the authors declared no competing interests.