Risk Models for Monitoring Postoperative Complication Rates After Paediatric Cardiac Surgery

Objectives
As postoperative mortality for paediatric cardiac surgery is very low, we aimed to develop methods for monitoring of postoperative complication rates, given their impact upon children’s health and well-being.

Methods
We used national registry data to develop and evaluate a suite of risk adjustment models for the outcomes of 6 defined postoperative complications, designed for use in complication monitoring for quality assurance.

Results
There were 23 423 30-day postoperative episodes in children under the age of 18 years undergoing cardiac surgery between 2015 and 2021 in England and Wales, with 361 (1.5%) deaths <30 days. Two hundred fifty-seven (1.9%) of 13 556 postoperative episodes in infants (<1 year) involved necrotizing enterocolitis; 158 (1.3%) of 12 408 postoperative episodes between 2018 and 2021 involved prolonged pleural effusion; and among the full sample of postoperative episodes, there were 526 (2.2%) acute neurological events, 446 (1.9%) extracorporeal life supports, 740 (3.6%) renal replacement therapies, and 1006 (4.3%) unplanned reinterventions within 30 days of surgery. The risk adjustment models were developed using clinical factors first defined for mortality monitoring. The models for prolonged pleural effusion, extracorporeal life support, and renal replacement performed very well with area under the curve (AUC) statistics >0.85. The performance of the models for necrotizing enterocolitis, acute neurological event, and unplanned reintervention was less good (AUC statistics 0.74-0.79).

Conclusions
Although complications are more complex outcome measures than mortality, national registry data can be used to capture them and to evaluate methods for risk adjustment of these outcomes. These methods may enable future risk-adjusted monitoring of complication metrics for quality assurance.