One way to think about this is to first determine what type of study would be actually be needed to prove conclusively, for example, that one egg per day is a n

One way to think about this is to first determine what type of study would be actually be needed to prove conclusively, for example, that one egg per day is a net negative or net positive and go from there. I'm assuming that cost / complexity / time associated with conducting such a study would be prohibitive. If so, we can simply conclude that we will never be able to tell one way or the other (with current technology at least) and stop worrying about it. This way, we ensure that our meta-analysis firepower (whatever/whoever that actually is), can be directed at studies that actually have a potential to uncover some real knowledge.