Analysis of Covariance Using Pretreatment Baselines

What is it about?

In the analysis of clinical trials with a numerical measure as an endpoint, a pretreatment baseline of the measure is commonly included as a covariate in an analysis of covariance (ANCOVA). But the pretreatment baseline has random variation, just like the post-treatment measure. This paper shows that under certain conditions, the ANCOVA is still valid in this situation.

Featured Image

Why is it important?

Using a pretreatment baseline can increase the statistical power of a clinical trial and the accuracy of estimates based on the data. ANCOVA typically assumes the covariates are fixed, not random variables. Therefore it's important to understand whether applying ANCOVA techniques to clinical trials with pretreatment baselines is valid.