This is the problem with all cohort studies. Since there aren't strong controls, you can't determine causality.

Fundamentally, this is why we know get consistently conflicting advice about nutrition: we take cohort studies as law instead of using them to form hypotheses that are tested with well controlled experiments.