There isn't much selection against diseases that affect us after our children have grown, and what little there is (grandmother care effects, say) was likely swamped in most of human history by other causes of mortality.
It is reasonable to consider such hypotheses, but until you have some evidence it isn't very convincing. It's on about the same low level of intellectual rigor as assuming that "progress" is all wonderful. Neither good-old-days nor progress-is-always-good are valid modes of logical inference.
Take, for instance, the farming-based diet that our ancestors have eaten for millenia. In fact, there is evidence that it was worse than what went before. But there is also evidence that modern diets are healthier (in terms of robust development to biological potential). It isn't all one way or the other.
Here's a reference to a survey article:
You definitely have to take a very expansive view.
> Just because people in the medieval times and before did it is no reason to presume that it was good for health.
It is reasonable to assume that if your ancestors for a sufficiently large number of generations ate some diet it means that natural selection has ensured that diet is reasonably optimized for your health.