|Saturday, February 22|
|PS3 Poster Session III & Continental Breakfast||
Sat, Feb 22, 7:30 AM - 9:00 AM
Testing Differences in Glucose Profiles Using AUC and Mixed Models (302801)*Robbie Beyl, Pennington Biomedical Research Center
Jeffrey H. Burton, Pennington Biomedical Research Center
William D. Johnson, Pennington Biomedical Research Center
Keywords: AUC, Mixed model, OGTT, Glucose profiles
An oral glucose tolerance test (OGTT) measures glucose at several time points following ingestion of a glucose solution. Assessing the shape of the glucose curve is primarily done via area under the curve (AUC) analysis. The goal of this analysis is to determine if treatment groups have different glucose profiles. Another possible method that could capture differences in glucose profiles is a linear mixed model. OGTT data are simulated for treatment groups under the assumption that means change over time, but are equal between the treatment groups. P-values are computed for AUC analysis using the trapezoidal rule and adjusted for baseline. The p-values from the linear mixed model are based on testing a treatment by time effect while also adjusting for baseline. These p-values are compared to each other to determine how they correspond. Particular focus is on data resulting in a high p-value from one method and a low p-value from another method. P-values based on AUC can be vastly different from p-values based on the linear model. These differences generally occur when the glucose profiles cross. Thus both methods should be considered when modeling glucose profiles.