Using SAS for the Longitudinal Analysis of Difference Scores.
page Paper AA-09-2015. (2015)Dades longitudinals; SAS; Canvi.Abstract
Background: When longitudinal data has no missing baseline values, analysis of difference scores is one method of normalizing the error terms, even if the original outcome variable is non-normal. Adjusting for the baseline value as a covariate enables estimation of difference scores, with adjustment for the starting value. Objective and Methods: Derive the linear mixed model (LMM) for difference scores, which will include terms for time, treatment group, interaction between time and treatment, and baseline value. Demonstrate how to use the SAS data step to prepare a dataset for longitudinal analysis of difference scores. Present a SAS macro that uses Proc Mixed for analysis of difference scores, with adjustment for the baseline values of treatment groups. Derive the formulas for contrasts between change scores between treatment groups, adjusted for baseline. Show how to convert the contrast equations to SAS Estimate statements. Further, explain how the between-group contrasts can be adjusted for multiple comparisons. The example data will be from a diabetes study with three treatment groups with time points at baseline, 6-months, 12-months, and 18-months. Results. Examples will be presented that show the trajectory of an outcome over time between treatment groups, in table and graphic format. These will include the treatment group improving significantly, in comparison to the control group, and of the treatment group staying the same, while the control group worsened over time. Conclusion. Outcome analysis, based on a LMM on difference scores with baseline adjustment, is an effective analysis technique for longitudinal data.