BACKGROUND: Mean blood glucose (BG) during acute myocardial infarction (AMI) is an important predictor of inpatient mortality but does not capture glucose variability (GV), which has been shown to be independently associated with mortality in critically ill patients. Whether GV is associated with in-hospital mortality during AMI, after accounting for mean BG, is unknown. METHODS AND RESULTS: We analyzed 18 563 consecutive patients with AMI with ≥3 BGs in the first 48 hours admitted to 61 US hospitals from 2000 to 2008. Five different GV metrics were compared for their ability to predict in-hospital mortality (range, standard deviation, mean amplitude of glycemic excursions, mean absolute glucose change, and average daily risk range) using hierarchical logistic regression models that sequentially adjusted for mean BG, hypoglycemia (<70 mg/dL), and multiple patient characteristics. In unadjusted analyses, range and average daily risk range had the highest C-indices (0.620 for range, 0.635 for average daily risk range; both P<0.0001). Greater GV was associated with higher mortality for all metrics (eg, mortality was 3.8%, 5.5%, 7.1%, and 11.3% for increasing quartiles of range, P<0.0001); however, the association between GV and mortality for each metric was no longer significant after multivariable adjustment. In contrast, mean BG remained an important predictor of survival (P<0.001, all models). CONCLUSIONS: Although greater GV is associated with increased risk of in-hospital mortality in patients with AMI in unadjusted analyses, GV is no longer independently predictive after controlling for multiple patient factors, including mean BG. These findings suggest that GV does not provide additional prognostic value above and beyond already recognized risk factors for mortality during AMI.
BACKGROUND: Mean blood glucose (BG) during acute myocardial infarction (AMI) is an important predictor of inpatient mortality but does not capture glucose variability (GV), which has been shown to be independently associated with mortality in critically illpatients. Whether GV is associated with in-hospital mortality during AMI, after accounting for mean BG, is unknown. METHODS AND RESULTS: We analyzed 18 563 consecutive patients with AMI with ≥3 BGs in the first 48 hours admitted to 61 US hospitals from 2000 to 2008. Five different GV metrics were compared for their ability to predict in-hospital mortality (range, standard deviation, mean amplitude of glycemic excursions, mean absolute glucose change, and average daily risk range) using hierarchical logistic regression models that sequentially adjusted for mean BG, hypoglycemia (<70 mg/dL), and multiple patient characteristics. In unadjusted analyses, range and average daily risk range had the highest C-indices (0.620 for range, 0.635 for average daily risk range; both P<0.0001). Greater GV was associated with higher mortality for all metrics (eg, mortality was 3.8%, 5.5%, 7.1%, and 11.3% for increasing quartiles of range, P<0.0001); however, the association between GV and mortality for each metric was no longer significant after multivariable adjustment. In contrast, mean BG remained an important predictor of survival (P<0.001, all models). CONCLUSIONS: Although greater GV is associated with increased risk of in-hospital mortality in patients with AMI in unadjusted analyses, GV is no longer independently predictive after controlling for multiple patient factors, including mean BG. These findings suggest that GV does not provide additional prognostic value above and beyond already recognized risk factors for mortality during AMI.
Authors: Tara Lagu; Penelope S Pekow; Meng-Shiou Shieh; Mihaela Stefan; Quinn R Pack; Mohammad Amin Kashef; Auras R Atreya; Gregory Valania; Mara T Slawsky; Peter K Lindenauer Journal: Circ Heart Fail Date: 2016-08 Impact factor: 8.790