Abstract:
|
Latent class trajectory analysis (LCTA) has become an increasingly popular method to help better understand longitudinal risk factor trajectory patterns. LCTA categorizes individuals based on their patterns of change and allows for the discovery of patterns that may not be captured using a priori methods, such as certain weight fluctuations. While LCTA can be used to examine patterns in risk factors across age or over a pre-defined calendar time period, it is essential to select an appropriate time scale to best capture the period of change of interest. For example, unless each individual subject is followed both over the same calendar time and at the same ages, one could potentially get very different results depending on the choice of time scale. To investigate this issue, we use a simulation study to evaluate the types of inferences one could make using a variety of different approaches to handle time. We also simulate and evaluate the impact of staggered recruitment, age range restrictions and cohort effects. Based on our findings, we put forward guidelines and suggestions for choice of appropriate approaches to handle time in studies using LCTA.
|
ASA Meetings Department
732 North Washington Street, Alexandria, VA 22314
(703) 684-1221 • meetings@amstat.org
Copyright © American Statistical Association.