Validation of the Global Registry of Acute Coronary Event (GRACE) risk score for in-hospital mortality in patients with acute coronary syndrome in Canada
Received 10 March 2009; accepted 2 June 2009.
Background
The Global Registry of Acute Coronary Event (GRACE) risk score was developed in a large multinational registry to predict in-hospital mortality across the broad spectrum of acute coronary syndromes (ACS). Because of the substantial regional variation and temporal changes in patient characteristics and management patterns, we sought to validate this risk score in a contemporary Canadian population with ACS.
Methods
The main GRACE and GRACE2 registries are prospective, multicenter, observational studies of patients with ACS (June 1999 to December 2007). For each patient, we calculated the GRACE risk score and evaluated its discrimination and calibration by the c statistic and the Hosmer-Lemeshow goodness-of-fit test, respectively. To assess the impact of temporal changes in management on the GRACE risk score performance, we evaluated its discrimination and calibration after stratifying the study population into prespecified subgroups according to enrollment period, type of ACS, and whether the patient underwent coronary angiography or revascularization during index hospitalization.
Results
A total of 12,242 Canadian patients with ACS were included; the median GRACE risk score was 127 (25th and 75th percentiles were 103 and 157, respectively). Overall, the GRACE risk score demonstrated excellent discrimination (c statistic 0.84, 95% CI 0.82-0.86, P < .001) for in-hospital mortality. Similar results were seen in all the subgroups (all c statistics ≥0.8). However, calibration was suboptimal overall (Hosmer-Lemeshow P = .06) and in various subgroups.
Conclusions
GRACE risk score is a valid and powerful predictor of adverse outcomes across the wide range of Canadian patients with ACS. Its excellent discrimination is maintained despite advances in management over time and is evident in all patient subgroups. However, the predicted probability of in-hospital mortality may require recalibration in the specific health care setting and with advancements in treatment.
aTerrence Donnelly Heart Centre, Division of Cardiology, St Michael's Hospital, University of Toronto, Toronto, Ontario, Canada
bThe Canadian Heart Research Centre, Toronto, Ontario, Canada
iMontreal Heart Institute, Montreal, Quebec, Canada
jUniversity of Michigan Health System, Ann Arbor, MI
Reprint requests: Andrew T. Yan, MD, is to be contacted at Division of Cardiology, St Michael's Hospital, 30 Bond Street, Room 6-030 Queen, Toronto, Ontario, Canada M5B 1W8. Shaun G. Goodman, Division of Cardiology, St Michael's Hospital, 30 Bond Street, Room 6-034 Queen, Toronto, Ontario, Canada M5B 1W8.
ABSTRACT