J Thorac Cardiovasc Surg. 2026 Oct 27:S0022-5223(26)01398-X. doi: 10.1016/j.jtcvs.2026.10.019. Online ahead of print.
ABSTRACT
OBJECTIVE: Adults with congenital heart disease (ACHD) undergoing cardiac surgery represent a growing population in whom conventional surgical risk models perform suboptimally. We externally validated a previously described ACHD-specific preoperative risk model and evaluated whether performance could be improved through recalibration/model updating.
METHODS: We conducted a prospective, multicenter observational study of patients ≥16 years undergoing congenital cardiac surgery at four centers (March 2020-April 2023). The primary composite adverse outcome (CAO) comprised in-hospital death, prolonged ventilation >7 days, or new-onset dialysis. We evaluated discrimination and calibration of the original preoperative model (including MELD-XI, cognitive impairment, body-mass index [BMI], prior chest wall incisions, and complex anatomy) and then performed bootstrap-based model updating (200 resamples) to derive refined models with and without the Adult Congenital Heart Surgery (ACHS) procedural mortality score.
RESULTS: Among 538 patients, CAO occurred in 4.3% (n=23). In external validation (n=383 complete cases), the original model preserved discrimination (AUC=0.729, 95% confidence interval: 0.602-0.855) but systematically overestimated risk, indicating need for recalibration. The refined model retained BMI, prior chest incisions, cognitive impairment, age, serum albumin level, and ventricular systolic dysfunction. Discrimination was good without procedural complexity (AUC=0.799; 0.697-0.900) and improved with inclusion of the ACHS score (AUC=0.857; 0.779-0.935).
CONCLUSIONS: In this prospective multicenter external validation, an ACHD-specific preoperative risk model maintained discrimination but required recalibration. Incorporating procedural complexity enhanced discrimination, supporting a hybrid patient- and procedure-based approach to guide perioperative planning.
PMID:42946203 | DOI:10.1016/j.jtcvs.2026.10.019