Hacker News Reader

Comment by jey | Hacker News Reader

Has a causal relationship been demonstrated from nuts to health, or is this only a correlation? Couldn't it be that health-conscious people are healthier on average, and they usually choose to snack on nuts instead of snacking on something unhealthy?

The line "individuals who ate more nuts were leaner, less likely to smoke, and more likely to exercise, use multivitamin supplements, consume more fruits and vegetables, and drink more alcohol" further reinforces my belief that there's just a confounder here, e.g. maybe being affluent and educated explains all of those. (Yes, including alcohol consumption: http://blogs.discovermagazine.com/gnxp/2010/05/people-of-cla...)

View on HN

Reverse causality is another possible explanation for our findings, because people with chronic disease and poor health might abstain from nut consumption. However, we excluded participants with a history of cancer, heart disease, or stroke at baseline, and we suspended further updating of all dietary variables when participants reported a diagnosis of stroke, heart disease, angina, or cancer. Moreover, the results remained significant when we excluded the first 2 years of follow-up and added a 2-year lag period between nut-intake assessment and each follow-up period.

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

This study sounds like a waste of time. How you could ever follow up and prove that nuts have all those protective effects is beyond me -- as is why you would even think they do in the first place.

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.

I wouldn't be so categorical. Regarding assessing causality, with an appropriate choice of instruments (food prices, tax on specific food categories) one could use instrumental variable regression to disambiguate the relationship. I'm not sure if this is standard practice in this field but a quick search seems to indicate this has been done before.

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.

Observational studies like this could conceivably be useful under different circumstances. I think the chances of distinguishing meaningful from meaningless correlations would probably be higher if the variable being studied wasn't as ubiquitous as eating nuts and the causes of death weren't ordinary diseases of aging. For instance, an observational study could help determine the risk of disease from living in a specific area that was contaminated with a specific toxic substance. In order to make a very convincing case that the toxin had caused the diseases, you'd want to support it with stronger kinds of studies, that, say, demonstrated that that chemical was capable of causing the diseases that were observed.

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.

Well, the correlation at least gives a great probability that nut consumption does not affect the health negatively!

Reply on news.ycombinator.com