Development and Validation of Risk Models for Outcomes of Isolated Coronary Artery Bypass Grafting ― A Nationwide Analysis in Japan ―
Abstract
BACKGROUND: Reliable preoperative risk assessment is essential in cardiac surgery. In Japan, coronary artery bypass grafting (CABG) risk models have been developed from the Japan Cardiovascular Surgery Database - Adult section (JCVSD-A), with JapanSCORE I and II released previously. To reflect contemporary practice, we developed and validated JapanSCORE III, an updated risk model for isolated CABG, using JCVSD-A. METHODS AND RESULTS: We identified patients undergoing isolated CABG between 2018 and 2023 in the JCVSD-A. The dataset was temporally split into development (2018-2022) and validation (2023) cohorts. A least absolute shrinkage and selection operator model was developed to predict operative mortality, incorporating imaging-based predictors and clinically relevant factors. Secondary outcome models were constructed for stroke, reoperation, prolonged ventilation, renal failure, deep sternal wound infection (DSWI), gastrointestinal complications, and intensive care unit stay >7 days. Model performance was assessed by areas under the curve (AUC) and calibration plots. Among 70,180 patients (mean age 69.3 years; 20% women), operative mortality was 2.7%. The incidence of stroke, reoperation, prolonged ventilation, renal failure, DSWI, gastrointestinal complications, and prolonged ICU stay was 1.8%, 1.5%, 3.7%, 2.0%, 1.1%, 1.3%, and 13.2%, respectively. Operative mortality and renal failure models demonstrated good discrimination (AUC >0.8) and calibration in validation. CONCLUSIONS: Operative outcomes remained stable despite increasing predicted risk. The JapanSCORE III models showed good predictive performance, providing valid tools for preoperative risk stratification.
Related
- Risk factors for post-PCI restenosis in patients with coronary heart disease and the development and validation of a nomogram prediction model
- A Parsimonious Machine Learning Model for Predicting Invasive Coronary Angiography Referral After Coronary Computed Tomography Angiography: Development and Multicentre External Validation
- A machine learning model for predicting ischemic and bleeding risk after percutaneous coronary intervention: Development and external validation
- Development and Validation of a Prediction Model for Acute Coronary Occlusion in Patients With Non-ST-Segment Elevation Myocardial Infarction.
- Development and validation of machine learning models for preoperative prediction of perioperative outcomes in patients with thymic epithelial tumors
BibTeX
@article{hayashi2026development,
title = {{Development and Validation of Risk Models for Outcomes of Isolated Coronary Artery Bypass Grafting ― A Nationwide Analysis in Japan ―}},
author = {Sachiko Hayashi and Hiraku Kumamaru and Shiori Nishimura and Aya Saito and Hideyuki Shimizu and Noboru Motomura},
journal = {Circulation Journal},
year = {2026},
month = sep,
doi = {10.1253/circj.cj-26-0160},
url = {https://www.jstage.jst.go.jp/article/circj/advpub/0/advpub_CJ-26-0160/_article},
note = {CC BY-NC-ND},
}
RIS
TY - JOUR
TI - Development and Validation of Risk Models for Outcomes of Isolated Coronary Artery Bypass Grafting ― A Nationwide Analysis in Japan ―
AU - Sachiko Hayashi
AU - Hiraku Kumamaru
AU - Shiori Nishimura
AU - Aya Saito
AU - Hideyuki Shimizu
AU - Noboru Motomura
PY - 2026
DA - 2026/09/02
JO - Circulation Journal
DO - 10.1253/circj.cj-26-0160
UR - https://www.jstage.jst.go.jp/article/circj/advpub/0/advpub_CJ-26-0160/_article
AB - BACKGROUND: Reliable preoperative risk assessment is essential in cardiac surgery. In Japan, coronary artery bypass grafting (CABG) risk models have been developed from the Japan Cardiovascular Surgery Database - Adult section (JCVSD-A), with JapanSCORE I and II released previously. To reflect contemporary practice, we developed and validated JapanSCORE III, an updated risk model for isolated CABG, using JCVSD-A. METHODS AND RESULTS: We identified patients undergoing isolated CABG between 2018 and 2023 in the JCVSD-A. The dataset was temporally split into development (2018-2022) and validation (2023) cohorts. A least absolute shrinkage and selection operator model was developed to predict operative mortality, incorporating imaging-based predictors and clinically relevant factors. Secondary outcome models were constructed for stroke, reoperation, prolonged ventilation, renal failure, deep sternal wound infection (DSWI), gastrointestinal complications, and intensive care unit stay >7 days. Model performance was assessed by areas under the curve (AUC) and calibration plots. Among 70,180 patients (mean age 69.3 years; 20% women), operative mortality was 2.7%. The incidence of stroke, reoperation, prolonged ventilation, renal failure, DSWI, gastrointestinal complications, and prolonged ICU stay was 1.8%, 1.5%, 3.7%, 2.0%, 1.1%, 1.3%, and 13.2%, respectively. Operative mortality and renal failure models demonstrated good discrimination (AUC >0.8) and calibration in validation. CONCLUSIONS: Operative outcomes remained stable despite increasing predicted risk. The JapanSCORE III models showed good predictive performance, providing valid tools for preoperative risk stratification.
SN - 1347-4820
SN - 1346-9843
ER -