Absolutely, and the observations suggested that hormone treatment was possibly even beneficial. That's why it's completely true that correlation != causation. You can try to "control" for differences in the observed groups, and epidemiologists try this all the time, but unless they already understand everything involved, it's impossible to get this completely right. And sometimes the things they try to control for are themselves based on other epidemiological studies.

This is the main reason we are always hearing in the press about how a "study shows that X is bad for us" then a "study shows that X is good for us". The majority of the time when there are such conflicting studies, it is the result of epidemiology or poorly controlled studies.