Development and validation of a prediction model for acute kidney injury following cardiac valve surgery.

Abstract

Background: Acute kidney injury (AKI) often accompanies cardiac valve surgery, and worsens patient outcome. The aim of our study is to identify preoperative and intraoperative independent risk factors for AKI in patients undergoing cardiac valve surgery. Using these factors, we developed a risk prediction model for AKI after cardiac valve surgery and conducted external validation.

Methods: Our retrospective study recruited 497 adult patients undergoing cardiac valve surgery as a derivation cohort between February and August 2023. Patient demographics, including medical history and perioperative clinical information, were acquired, and patients were classified into one of two cohorts, AKI and non-AKI, according to the Kidney Disease: Improving Global Outcomes (KDIGO) guidelines. Using binary logistic stepwise regression analysis, we identified independent AKI risk factors after cardiac valve surgery. Lastly, we constructed a nomogram and conducted external validation in a validation cohort comprising 200 patients. The performance of the nomogram was evaluated based on the area under the receiver operating characteristic curve (AUC), calibration curves and decision curve analysis (DCA).

Results: In the derivation cohort, 172 developed AKI (34.6%). Relative to non-AKI patients, the AKI patients exhibited elevated postoperative complication incidences and worse outcome. Based on multivariate analysis, advanced age (OR: 1.855; p = 0.011), preoperative hypertension (OR: 1.91; p = 0.017), coronary heart disease (OR: 6.773; p < 0.001), preoperative albumin (OR: 0.924; p = 0.015), D-Dimer (OR: 1.001; p = 0.038), plasma creatinine (OR: 1.025; p = 0.001), cardiopulmonary bypass (CPB) duration (OR: 1.011; p = 0.001), repeat CPB (OR: 6.195; p = 0.010), intraoperative red blood cell transfusion (OR: 2.560; p < 0.001), urine volume (OR: 0.406 p < 0.001) and vasoactive-inotropic score (OR: 1.135; p = 0.009) were independent risk factors for AKI. The AUC of the nomogram in the derivation and validation cohorts were 0.814 (95%CI: 0.775-0.854) and 0.798 (95%CI: 0.726-0.871), respectively. Furthermore, the calibration curve revealed that the predicted outcome was in agreement with the actual observations. Finally, the DCA curves showed that the nomogram had a good clinical applicability value.

Conclusion: Several perioperative factors modulate AKI development following cardiac valve surgery, resulting in poor patient prognosis. The proposed AKI predictive model is both sensitive and precise, and can assist in high-risk patient screening in the clinics.