Hmm… I thought I’d read the paper, but it turned out I got confused by an abstract with multiple sections. I thought it was very short. My mistake

So I can’t say whether the authors addressed whether the low carbohydrate group was selecting for people already known to be at risk.

This finding reminds me of the studies that found people who drank a little alcohol lived longer than those who drank no alcohol. Further investigation finds that the no alcohol group included reformed alcoholics, who has already done enough damage to their systems to shorten their life expectancy, and this extra group was enough to skew the figures.

So I think we need to ask: are there reasons to think taking <40% of calories as carbs is selecting for a group with shorter life expectancy? Maybe - anorexia would be one, although I’ve no idea of its prevalence among Japanese men.

The paper makes no mention of considering this sort of thing. edit: correction because I can’t read the paper.