Regression Methodology for Comparing Longitudinal Rates of Change

• Activity Number: 482
• Type: Contributed
• Date/Time: Wednesday, August 7, 2013 : 8:30 AM to 10:20 AM
• Sponsor: Biometrics Section
• Author(s): Matthew Bryan*+
• Keywords: Longitudinal Data ; Rates of Change ; Regression Methods

Abstract

Longitudinal data can compare groups across an outcome at the mean level and the rate level. Powerful methods for longitudinal data, such as the linear model, have been developed for estimating mean level differences. In addition, the linear model can estimate rate level difference through the inclusion of appropriate interaction terms. However, when differentiating rates of change under the presence of non-linear trend in time, the standard linear models approach is insufficient. Other methods have been developed for modeling rates of change that allow a general time trajectory, but these methods have not addressed comparing rates across covariate defined groups. Thus, we propose regression methods for differentiating longitudinal rates of change. Our proposed method offers a parsimonious comparison of rates of change relative to a generally specified time trajectory. The regression model demonstrates increased power to detect group difference in the rate of change relative to a standard linear models approach. The proposed approach to comparing longitudinal rates of change is illustrated using a study of growth among infants exposed to HIV infection.