Reliability in health and disease (ICC,CoV) - Separtely or combined

I am currently writing an article. However, I have a concern regarding my statistics part. I calculated the ICC (Intraclass correlation) and CoV (Coefficient of Variation) over the group of healthy subjects and over the diseased subjects (measurement within two weeks) and they both are good to excellent. Or is it more the overall ICC en CoV, does it add something? I eventually want to use validate the measurement for diagnostic purposes.

If you have the answer, do you also have the reference for me where I can learn an understand which decision is the best to present in the paper.

Thanks, Martijn

EDIT: Example: If I add all the data together, I will defenitely end up with a heteroscedastic dataset. This, however, is unfair while this might not happen if I analyse them apart. Still, do I need to describe both groups seperate and together?