Regression analysis with difference values in a longitudinal study

Hello, I need some help with an analysis for my master thesis.

I'm doing a longitudinal study with three different times of measurement. It's about the quality of a talk and how this effects the development of psychological distress between two times of measurement. There are no between-subjects-factors like different interventions or something. I thought about doing a regression analysis with the quality of the talk as a predictor and the difference of t1(distress) and t2 (distress) as criterion. I know that difference values lead to a loss of information (for example the loss of the t1(distress)-data). That's why I thought I could probably add the t1(distress) as covariate. Would this be possible? Is there another possibility to analysize this which is better? (Maybe something like repeated measures anova without between-subjects-factor and instead the quality of the talk as covariate?)

Thank you very much for your help! (and sorry for English mistakes)