Surgical risk scores for congenital heart surgery are useful for long-term risk prediction

Andrew Dailey-Schwartz1, Krisy Kuo2, Yanxu Yang2

  • 1Division of Cardiology, Emory University School of Medicine, Atlanta, GA, USA.

Cardiology in the Young
|January 17, 2025
PubMed

Insights

Existing scoring systems accurately predict early mortality after pediatric heart surgery but show diminishing accuracy for long-term survival. New tools are needed to better predict long-term outcomes for congenital heart disease patients.

Area of Science:

  • Pediatric Cardiology
  • Cardiac Surgery Outcomes
  • Biostatistics

Background:

  • The Society of Thoracic Surgeons-European Association for Cardiothoracic Surgery (STAT) and Risk Adjusted Classification for Congenital Heart Surgery (RACHS) scoring systems are established for predicting early mortality in pediatric cardiac surgery.
  • However, their effectiveness in predicting long-term mortality following these procedures remains unexamined.

Purpose of the Study:

  • To evaluate the ability of STAT, STAT 2020, RACHS-1, and RACHS-2 scoring systems to predict both early and long-term mortality in pediatric patients undergoing congenital heart surgery.

Main Methods:

  • A retrospective cohort study utilized data from the Pediatric Cardiac Care Consortium (1982-2011).
  • STAT, STAT 2020, RACHS-1, and RACHS-2 scores were applied to predict early mortality and long-term survival.
  • Long-term outcomes were determined by linking patient data with the National Death Index through 2021.

Main Results:

  • All scoring systems demonstrated good performance in predicting early postoperative mortality (c-statistics ranging from 0.7668 to 0.7872).
  • For long-term mortality prediction, discriminative power decreased, with c-statistics ranging from 0.6741 to 0.7156.
  • STAT 2020 and RACHS-2 showed the highest predictive ability for long-term mortality.

Conclusions:

  • Current risk-adjusted scoring systems for congenital heart surgery maintain adequate but reduced accuracy for predicting long-term mortality.
  • Further development of predictive tools is necessary to improve the assessment of long-term survival in this patient population.