Hacker News Reader

Comment by orange_soda | Hacker News Reader

>red meat: 0.17% over 10 years, 0.41% over 20 years and 0.62% over 30 years

Isn't this essentially nothing? Am I misinterpreting these results?

Welcome to the world of nutrition science, where the methodology is shoddy, the evidence is poor, and the recommendations are ridiculous.

> ... completed a food questionnaire and then were followed up for around 20 years

FFQ are useless. I barely remember what I ate 2 days ago.

> Isn't this essentially nothing?

Essentially nothing: for an individual? Yes. For a population? No.

You or I may not care about a 0.62% increased CVD rate for ourselves over 30 years, but that does translate to 6,200 extra possibly-preventable cases of CVD per million people. That's about 76,000 diagnoses annually in the US alone.

How strong is this effect relative to other effects though? With small effects, you get an increased risk of error in the analysis as there could be confounding variables not fully accounted for, reporting errors, etc.

How would this compare, for example, to not smoking, drinking less alcohol, wearing sunscreen, safe sex, not exercising enough, etc.

(Picked things that I think we have a high standard of evidence for and fairly large effect, but that people often neglect due to wanting to do the harmful activity, as with red meat)

For years low salt recommendations were justified based on lessened harm "on a population level" even if individual gains were small. Now it seems that these recommendations were wrong, according to Cochrane reviews.

Smoking increases lung cancer risk 15x-30x, comes at a steep tax-inflated price and people still won't give it up.

I basically read this as "it's ok to eat meat/sausages/whatever".

This way of interpreting the data is misleading. For population, it still could mean nothing. People are not living in an isolated environment just eating red meat or processed food and do nothing else. At the population level there are way more factors need to be considered. You can't even imagine how many things in the society will be impacted if people stop eating red meat or the processed food. CVD will be the least of concern at the population level.

So, yes, this tiny effect size means nothing.

Moreover, this tiny effect size could be much smaller than noises in the data.

It could mean nothing, or it could not. But a small effect size alone doesn’t tell you this, and you are correct that there are many factors — that is exactly the point of this study, to tease apart which other factors are causative.

Of course, this study says nothing about what happens if you try to change behavior, that seems like an entirely unrelated point.

Taken together, I’m not sure how you can be so confident that a small effect size therefore means nothing.

This is assuming this 0.62% actually shows a TRUE effect.

If it is noise/invalid/a mistake, it means nothing; the actual effect might actually be a DECREASE of CVD as far as we know.

And it is important to also look at all cause mortality; such studies have often shown a decrease in CVD but the same final number of deaths...

You should really include that this was under the assumption of only 2 servings a week, where a serving is about 3.5 ounces. I wonder how the rates change when you assume 2 servings per meal, which seems more in line with peoples actual diets.

7 ounces of steak per meal is average? That's an expensive diet.

Reply on news.ycombinator.com