Tbf that’s almost always going to be the case with nutritional epidemiology because it’s too hard to run an RCT for long enough to measure long term outcomes. If they can be paired with rcts looking at biomarkers, causative explanations and well controlled for interfering factors they can offer some good evidence.