Abstract 3583: Glucometrics in Patients Hospitalized With Acute Myocardial Infraction: Defining the Optimal Measure of Risk
Background: Elevated blood glucose (BG) on admission is a known risk factor for higher mortality in patients (pts) with AMI. Whether metrics that incorporate multiple BG assessments during AMI hospitalization are better predictors of mortality than admission BG alone is unknown.
Methods: We evaluated 16,871 AMI pts hospitalized from 01/00–12/05, using Cerner Corporation’s Health Facts® database from 40 hospitals, which contains demographics, clinical and comprehensive laboratory data. Using logistic regression models and C-indices, 3 metrics of overall BG control (mean BG, time-averaged glucose (TAG), hyperglycemia index (time-average of hyperglycemic values, HGI)), each evaluated over 3 time windows (first 24 hours (hr), first 48 hr, entire hospitalization) were compared with admission BG for their ability to discriminate those who did and did not survive AMI hospitalization.
Results: 29% of the pts had diabetes. The median number of BG measurements per pt per hospitalization was 4 (IQR 2–8). All of the average BG metrics were better predictors of in-hospital mortality than admission BG (Table⇓). There was a gradual, significant increase in the ability of models to predict mortality as the time window increased from admission to 24 hr, 48 hr and to the entire hospitalization. Statistically significant, but small differences in the C-indices of mean BG, TAG and HGI were observed.
Conclusions: Measures of average BG control during AMI hospitalization are better predictors of mortality than admission glucose alone. At each time period, mean hospitalization BG appears to be the most practical metric of hyperglycemia-associated risk in AMI pts.