I don't think they're substituting diseases. As I read it, they don't differentiate down when the sample size is small, so you don't get Diabetes I or II for small samples. You just get diabetes.

It makes some sense to me, that if there are less than some number, say 1000 reponses, you can only drill so far into the data to protect privacy.

I'm with you on research and getting more data, but not at the expense of privacy. I already know people that have not hired woman because they mentioned wanting to have kids, or someone was concerned they would get pregnant. Can you imagine someone that would implement similar policies because someone was potentially ill? Or carried a gene for a disease?

We need to be sure that we are respecting individual rights when we consolidate and use data.