Oxygen delivery in pediatric cardiac surgery and its association with acute kidney injury using machine learning

Abstract

Objective: Acute kidney injury (AKI) after pediatric cardiac surgery with cardiopulmonary bypass (CPB) is a frequently reported complication. In this study we aimed to determine the oxygen delivery indexed to body surface area (Do₂i) threshold associated with postoperative AKI in pediatric patients during CPB, and whether it remains clinically important in the context of other known independent risk factors.

Methods: A single-institution, retrospective study, encompassing 396 pediatric patients, who underwent heart surgery between April 2019 and April 2021 was undertaken. Time spent below Do₂i thresholds were compared to determine the critical value for all stages of AKI occurring within 48 hours of surgery. Do₂i threshold was then included in a classification analysis with known risk factors including nephrotoxic drug usage, surgical complexity, intraoperative data, comorbidities and ventricular function data, and vasoactive inotrope requirement to determine Do₂i predictive importance.

Results: Logistic regression models showed cumulative time spent below a Do₂i value of 350 mL/min/m² was associated with AKI. Random forest models, incorporating established risk factors, showed Do₂i threshold still maintained predictive importance. Patients who developed post-CPB AKI were younger, had longer CPB and ischemic times, and required higher inotrope support postsurgery.

Conclusions: The present data support previous findings that Do₂i during CPB is an independent risk factor for AKI development in pediatric patients. Furthermore, the data support previous suggestions of a higher threshold value in children compared with that in adults and indicate that adjustments in Do₂i management might reduce incidence of postoperative AKI in the pediatric cardiac surgery population.

Keywords: acute kidney injury; cardiopulmonary bypass; goal-directed perfusion; machine learning; oxygen delivery.