Ethnicity-dependent performance of the Global Registry of Acute Coronary Events risk score for prediction of non-ST-segment elevation myocardial infarction in-hospital mortality: nationwide cohort study

Abstract

Aims The Global Registry of Acute Coronary Events (GRACE) score was developed to evaluate risk in patients with the acute coronary syndrome with or without ST-segment elevation. Little is known about its performance at predicting in-hospital mortality for ethnic minority patients. Methods and results We identified 326 160 admissions with non-ST-segment elevation myocardial infarction (NSTEMI) in the Myocardial Infarction National Audit Project (MINAP), 2010–17, including White (n = 299 184) and ethnic minorities (excluding White minorities) (n = 26 976). We calculated the GRACE score for in-hospital mortality and assessed ethnic group baseline characteristics by low, intermediate and high risk. The performance of the GRACE risk score was estimated by discrimination [area under the receiver operating characteristic curve (AUC)] and calibration (calibration plots). Ethnic minorities presented younger and had increased prevalence of cardiometabolic risk factors in all GRACE risk groups. The GRACE risk score for White [AUC 0.87, 95% confidence interval (CI) 0.86–0.87] and ethnic minority (AUC 0.87, 95% CI 0.86–0.88) patients had good discrimination. However, whilst the GRACE risk model was well calibrated in White patients (expected to observed (E : O) in-hospital death rate ratio 0.99; slope 1.00), it overestimated risk in ethnic minority patients (E : O ratio 1.29; slope: 0.94). Conclusion The GRACE risk score provided good discrimination overall for in-hospital mortality, but was not well calibrated and overestimated risk for ethnic minorities with NSTEMI. Key question Does the performance of the Global Registry of Acute Coronary Events (GRACE) (v2.0) score in predicting in-hospital mortality for non-ST-segment elevation myocardial infarction (NSTEMI) differ by ethnicity? Key finding The GRACE risk score provided good discrimination overall for in-hospital mortality but was not well calibrated and overestimated risk for ethnic minority patients with NSTEMI. Take-home message Ethnicity or race should be considered during the development of risk scoring systems. Existing systems can be recalibrated in the population they serve to better address risk.

Diagnostic ability using fatty liver and metabolic markers for metabolic‐associated fatty liver disease stratified by metabolic/glycemic abnormalities

“…With this command, calibration was plotted in 10 groups across the risk spectrum, reporting the following components: (i) calibration in the large (CITL) index, describing the difference between the average predicted probabilities and the observed event frequencies, in which the ideal value should equal zero; (ii) the calibration slope, in which the ideal value should equal one; and (iii) the expected probability vs observed frequency (E:O) ratio, in which the ideal value should equal one. This method has been used in several studies describing calibration 36–38 …”

“…This method has been used in several studies describing calibration. [36][37][38] Assessment of model performance: discriminative ability and calibration Based on the best cutoff value of each predictor according to the Youden index, 39 we calculated the sensitivity, specificity, positive and negative likelihood ratios, positive and negative predictive values, diagnostic odds ratio, and kappa coefficient, which evaluates the agreement of categorization compared with the ultrasound-proven MAFLD. We also performed sexstratified analysis.…”

The Prognostic Value of the Age–D-Dimer–Albumin Score in Patients with Acute Myocardial Infarction Undergoing Percutaneous Coronary Intervention

“…6 However, there is a risk of overestimation of the in-hospital mortality when the GRACE score was used in a non-white population with non-ST-segment elevation MI. 3 Additionally, its impact on subsequent patient management remains unclear. The CREDO-Kyoto and the PARIS thrombotic risk scores can be used to assess the longterm thrombotic risk after PCI in patients with CAD.…”

Sex-specific evaluation and redevelopment of the GRACE score in non-ST-segment elevation acute coronary syndromes in populations from the UK and Switzerland: a multinational analysis with external cohort validation

“…Given that the XGBoost learning algorithm operates on single datasets, we used a single imputed dataset, generated as described, for training, testing, and external validation of the GRACE 3.0 score. Sensitivity analyses using complete cases were done to explore a potential effect of the imputation on the results (appendix pp [17][18][19][20][21]23). Internal validation of the GRACE 2.0 score evaluation in complete cases was done in each cohort by using 300 bootstrap samples (appendix p 2).…”

“…[11][12][13][14] Importantly, female patients are less likely to receive early invasive management. 15 Despite the growing awareness of its differing discriminative performance in specific patient groups, 16,17 including female patients with ST-segment elevation myocardial infarction, 18,19 the effect of sex differences on the performance of the GRACE score in patients with NSTE-ACS remains unclear. 14 Emerging evidence on a distinct profile of baseline risk factors in women and men with NSTE-ACS, the unequal strength of association of individual GRACE components with adverse outcomes, 12,[20][21][22] and the hitherto uniform handling in the GRACE score suggest sex differences in score performance, which might promote structural inequities in the treatment of patients with NSTE-ACS.…”