Hmm yeah, thanks for pointing that out. As corroborate can also mean confirm, I've said 'support' instead. While I agree retrospective studies are often hypothesis-generating, I also think they can provide additional data to better understand cause-and-effect relationships.