Performance of the GRACE and the ACTION risk model in the prediction of in-hospital mortality: external validation, model revision and updating in the Thai Percutaneous Coronary Intervention Registry
Issued Date
2025-05-21
Resource Type
ISSN
2398595X
eISSN
20533624
Scopus ID
2-s2.0-105006785972
Journal Title
Open Heart
Volume
12
Issue
1
Rights Holder(s)
SCOPUS
Bibliographic Citation
Open Heart Vol.12 No.1 (2025)
Suggested Citation
Kiatchoosakun S., Chamnarnphol N., Wongwipaporn C., Pussadhamma B., Roongsangmanoon W., Siriyotha S., Thakkinstian A., Sansanayudh N. Performance of the GRACE and the ACTION risk model in the prediction of in-hospital mortality: external validation, model revision and updating in the Thai Percutaneous Coronary Intervention Registry. Open Heart Vol.12 No.1 (2025). doi:10.1136/openhrt-2024-003027 Retrieved from: https://repository.li.mahidol.ac.th/handle/123456789/110482
Title
Performance of the GRACE and the ACTION risk model in the prediction of in-hospital mortality: external validation, model revision and updating in the Thai Percutaneous Coronary Intervention Registry
Corresponding Author(s)
Other Contributor(s)
Abstract
Background External validation is crucial before implementing a risk score model in clinical practice. This study examined the performance of Global Registry of Acute Coronary Events (GRACE) and Acute Coronary Treatment and Intervention Outcomes Network (ACTION) Registry-Get With The Guidelines (GWTG) (ACTION Registry-GWTG) Risk Score (AR-G RS) using the Thai Percutaneous Coronary Intervention Registry (TPCIR). Methods Included in this study were 11 455 patients with acute coronary syndrome (ACS) undergoing percutaneous coronary intervention (PCI) between November 2015 and May 2018. GRACE and AR-G RS models were externally validated, revised and updated using discrimination (C-statistic score) and calibration (Hosmer-Lemeshow (HL) indexes). Clinical predictors were selected stepwise from the multivariate analysis to evaluate the performance of each risk score in the revised and updated model. Results In-hospital mortality was 4.4%. GRACE and AR-G RS demonstrated good discrimination for in-hospital mortality (C-statistics 0.8957 and 0.8823, respectively) with optimal calibration (HL, p=0.036 and 0.006, respectively) and penalty rates of 0.005 and 0.006, respectively. The updated model significantly improved the discrimination performance compared with the original GRACE and AR-G RS models, with a C-statistic of 0.9118 and a penalty of 0.006. Conclusion GRACE and AR-G RS maintained a good performance in TPCIR. Based on routine PCI practice, we demonstrated that the updated model could improve the accuracy of GRACE and AR-G RS in predicting in-hospital mortality among patients with ACS who underwent PCI.