Background:

GRACE score (G) estimates in-hospital mortality risk (IHM) in patients with acute coronary syndromes (ACS). Given the variability of patients´ characteristics, health systems, socioeconomic environment and changes in treatment; external validation should be performed. The TRIPOD statement was published to standardize validation studies´ methodology. The objective of this study is to asses if G predicts IHM with precision in patients with acute coronary syndrome at Juan A. Fernández Hospital (validation cohort -V-).

Methods:

Retrospective validation study. Consecutive patients admitted between 2001 and 2016 were included. Discrimination was evaluated using the area under the ROC curve (AUC) and calibration was evaluated using de Hosmer-Lemeshow test, calibration-in-the-large, calibration slope, validation graph and calibration belt. Subgroup analysis by type of ACS was performed.

Results:

A total of 2104 patients were analyzed. Myocardial infarction with ST elevation (40.4% and 32%) and IHM (5.56% and 4.6%) was more prevalent than G estimation. Calibration was poor, risks were underestimated between probabilities of 3 and 13%. The calibration slope was adequate, therefore the overall effect and magnitude of predictors on IHM was similar to G. The AUC was 0.86. The model performed with underestimation of risk between probabilities of 5% and 23% in patients with ACS and ST elevation, the discrimination was good. Calibration and discrimination performed well in patients with non ST elevation ACS.

Conclusions:

Despite the lack of statistical validity of the model in the total population, the variables were predictors of IHM in the validation cohort. Therefore, G score is clinically relevant, but should be recalibrated in our population.

        · resumo em Espanhol     · texto em Espanhol     · Inglês ( pdf ) | Espanhol ( pdf )