The journal Nature Medicine published a major study about a cohort of over 105,000 people followed for 30 years. This is that researchers found.
Correlation isn't causation. But that's still interesting.
Yea, we'd need to filter out the behavioural angles to clearly identify causal relationships.
Do people who live longer just happen to start eating differently etc.
For example, the better I feel, the better I eat. It's not the bad eating that makes me feel bad - the better eating is a consequence of feeling better.
I can tell exactly when my diet is going to tank, when my chronic conditions flare. Diet has zero effect on them.