A47JACC April 1, 2014

Volume 63, Issue 12

Acute Coronary Syndromes

admiSSion hyperglyCemia improveS the graCe riSk SCore For prediCtion oF in-hoSpital mortality: inSightS From the euro heart Survey aCS iii

Poster ContributionsHall CSaturday, March 29, 2014, 10:00 a.m.-10:45 a.m.

Session Title: Acute Coronary Syndromes: NSTEMIAbstract Category: 1. Acute Coronary Syndromes: ClinicalPresentation Number: 1117-224

Authors: Francois Schiele, Hector Bueno, Matthias Hochadel, Marco Tubaro, Nicolas Meneveau, Wojciech Wojakowski, Marek Gierlotka, Jean-Pierre Bassand, Keith Fox, Anselm Gitt, University Hospital Jean Minjoz, Besancon, France

background: Stress hyperglycemia is associated with higher mortality. Using data from Euro Heart Survey Acute Coronary Syndromes (ACS) III, we determined the incremental prognostic value of adding admission hyperglycemia (AH) on top of the GRACE risk score.

methods: AH was defined as plasma glucose >140 mg/dL, irrespective of diabetic status. Prediction models for mortality were compared (AH alone vs GRACE score plus AH) by changes in the C-index. Discriminatory capacity of the 2 models was compared using the Integrated Discrimination Improvement (IDI) and Hosmer-Lemeshow (HL) tests. We report appropriateness of reclassification using the Net Reclassification Index (NRI). Reclassification was considered as appropriate when the “new” risk prediction was more accurate than GRACE alone. Similar tests were repeated (1) in selected subgroups (diabetics, elderly, STEMI, women, patients with GRACE score >140), and (2) with different thresholds for AH definition (from 120 to 200 mg/dL).

results: Among 21872 patients included in the Euro Heart Survey-ACS III, GRACE risk score and AH were available in 16896 (73%), in-hospital mortality was 4.3%. AH was observed in 37% of the patients. Improvement in C-statistic was significant when AH was added to GRACE score (0.8405 vs 0.8312, p<0.0001), relative IDI was also significant (p=0.03). NRI showed that AH added to GRACE allowed correct reclassification of 31% of those who died and 30% of survivors (p<0.0001). Incremental prognostic value was more pronounced in high risk subgroups (elderly, STEMI, GRACE score >140), but not in diabetics. Changes in threshold value to define AH did not yield to better improvement in risk assessment.

Conclusions: AH is associated with higher mortality in patients with acute coronary syndromes. Added to the GRACE score, AH allows better prognostication of in-hospital death.