However, case-control studies are retrospective and subject to bias. As in many cases in epidemiology, prospective cohort studies, which are much more expensive and difficult to do, fail to show as strong an association. You might ask, since at Harvard we have very large on going cohort studies, why this has not been looked at in the 100,000 women in the Nurses Health Study. You will notice that these cohorts have all been done in Asian populations. At least until recently, in the American population of the NHS, only 1-2% eat any soy which is too few to look at this question. We are however collecting more information on soy intake and maybe as it becomes more popular we can examine it in the future.