I had to go back and review this thread, given the age, and I see that Andrew replied back in 2024, albeit I do not see an explicit reference to his excellent 2001 paper on the topic:
It may be included in some of the other reference links in the prior replies however.
For the overall mean treatment effect as as primary endpoint, you can use the baseline mean measure as the covariate value for the model estimate and then derive percent change from that result, as has been noted.
You can also use other input values for baseline to assess the magnitude and perhaps even the direction of the treatment effect, over the range of observed baseline values. There is a good chance that doing so will also yield an obvious regression to the mean effect.