There is some degree of mathematical guarantee that one can provide, if one is only releasing statistics and not individual records. The technique is called "di

There is some degree of mathematical guarantee that one can provide, if one is only releasing statistics and not individual records. The technique is called "differential privacy" and it was invented in large part because of a few high-profile data de-anonymization stunts, one of which was in the health industry. At the very least, this allows one to quantify the tradeoff of security and specificity in what is released.