However, we excluded participants with a history of cancer, heart disease, or stroke at baseline...
Sounds like multiple regression, which is used to make insignificant statistics sound more plausible, by reducing the effect of a handful of confounding variables that the researchers happened to think of. That leaves behind any significant correlations plus the infinite number of confounding variables they didn't think of.
As to why one would think nuts have protective effects at all -- I'm not a nutritionist, but from what I gather nuts do seem to have some desirable nutritional properties (high on unsaturated fats, protein and minerals), so that assertion doesn't seem that far-fetched to me.
nuts do seem to have some desirable nutritional properties (high on unsaturated fats, protein and minerals)
The problem is that none of those other components of food have been shown to prevent death, either, except in weak studies like this one. It's all very circular.
Indeed, clinical trials have shown that nut consumption has beneficial effects on some intermediate markers of chronic diseases, such as high cholesterol levels,3 oxidation,6,7 endothelial dysfunction,13 hyperglycemia,6,10 and insulin resistance.11,12
The above quotes from the NEJM article imply more than simple correlation, but the conclusion of that article stops short of stating that, specifically, "Nonetheless, epidemiologic observations establish associations, not causality, and not all findings from observational studies have been confirmed in controlled, randomized clinical trials."
Source: http://www.nejm.org/doi/full/10.1056/NEJMoa1307352#t=article