Biostatistics Journal Club: Change Scores and Baseline Adjustment Revisited

The goal of many longitudinal studies is to determine factors (e.g., interventions or exposures) that influence change in an outcome over time in both randomized and observational settings. In the simplest design, with two groups measured at two time points, there are two approaches that are widely used. The first approach is based on analyses of “change scores” (defined as the change in an outcome between two time points). The second approach is based on analyses that make a “baseline adjustment” (i.e., analysis of the outcome measurement at the second time point while adjusting for the outcome measurement at the first time point). In the literature, there is much debate and conflicting recommendations about these two approaches. In this presentation, led by Garrett Fitzmaurice, ScD, Harvard T.H. Chan School of Public Health, the two approaches will be reviewed and some practical recommendations concerning their use will be made.

Speaker:

Garrett Fitzmaurice, ScD, Harvard T.H. Chan School of Public Health