Development and validation of a prediction model for acute kidney injury following cardiopulmonary bypass surgery

Acute renal impairment following cardiopulmonary bypass (CPB) is a serious perioperative complication with sustained occurrence. Proactive risk assessment and preventive strategies are crucial for enhancing surgical outcomes.

This prospective observational cohort study was conducted at a provincial tertiary hospital in Zhejiang Province, China, and involved 211 patients with heart disease who underwent CPB. Least absolute shrinkage and selection operator (LASSO) regression and best-subsets regression (BSR) were used to identify predictive variables for post-CPB acute kidney injury (AKI). A nomogram was constructed based on the selected predictors, and the model performance was evaluated using receiver operating characteristic curves, calibration plots, decision curve analysis (DCA), and internal validation via bootstrap resampling.

Among the 211 patients, 20.4% (43/211) developed AKI. LASSO and BSR identified body mass index, intrarenal venous flow pattern, ventilation time, and CPB time as significant predictors. These four prognostic factors were integrated into a web-based nomogram. The model yielded an area under the curve of 0.825 (95% confidence interval: 0.748-0.902). DCA demonstrated clinical utility across threshold probabilities ranging from 2 to 98%, confirming robust predictive accuracy.

We developed and validated a nomogram-based model for predicting post-CPB AKI and demonstrated its strong discriminative performance and clinical applicability. This tool may assist clinicians in risk stratification and early intervention.