Hacker News Reader

Comment by TaupeRanger | Hacker News Reader

This is one of those terrible studies that looks at every disease under the sun and inevitably something comes back with a positive association (even though it's probably just statistical noise) and they publish that single result. Take it with a huge grain of salt.

Looking at their other study using this same data, they found an INCREASE in major cardiovascular events among people with lower-than-average BMI with Vit D supplementation which was statistically significant: https://academic.oup.com/ajcn/article/115/5/1300/6496028

In other words: if an HN reader clicks on this thread and has the question "should I take Vitamin D?", the answer is: no one knows, and this study likely can't inform your decision.

View on HN

You write like they have some agenda to promote vitamin D3.

Instead, they publish both negative and positive results related to things they were interested in from the start. They find how things are.

The Finnish Vitamin D Trial (FIND) is very large randomized, double-blind, placebo-controlled, 5-year supplementation study of the benefits and risks of vitamin D in the primary prevention of cardiovascular (CVD) and cancer among 18000 men 60 years or older and women 65 years or older. https://classic.clinicaltrials.gov/ct2/show/NCT01463813

They don't find how things "are", they do the typical useless epidemiological study which I described and find spurious connections in statistical noise. This is why nutrition epidemiology is largely a failed science - it can't actually tell us anything useful: https://metarationality.com/nutrition

It seems to me that you just commented on the title.

If you actually read the study, you find out they did not find strong connection.

It’s not a failed science. It tells us useful things all the time including countless things that you take for granted. Pointing out industry interests and better evidence superseding worse evidence doesn't change that.

In my experience, harping on epidemiology always seems to boil down to inconsistent epistemological standards that favor the foods someone wants to eat.

Most commonly, epidemiology is shit because it gives us strong causal inferences for trans and saturated fats being bad for us BUT we want to eat meat and dairy so nvm epi studies suck except for the ones I take for granted.

As someone with the same general opinion, how do you know what kind of food/diet/vitamins are right for you?

Self experimentation, but default to "Lindy" nutrition. Look at things that have stood the test of time and allowed humans to flourish rather than die off. Focus on traditional diets, fast (don't eat 3 meals with snacks between - humans did not evolve to have that kind of caloric input, look at the obesity epidemic).

> This is why nutrition epidemiology is largely a failed science - it can't actually tell us anything useful

Hmmm.... I haven't suffered from scurvy, pellagra, rickets, or beriberi lately. You're saying that's just a coincidence?

They definitely have an incentive to publish results people will care about, as any other researcher that is part of an academic environment.

Researchers in the field care about results that show what is the real effect of D3 vitamin could be, and that's what these guys are doing.

Small, but not very big effect.

Which is the problem with the medical research.

They are figuring out if it is better to cut your toenails 2 times a week or 3 times a week, while obesity is killing millions.

Green jellybeans and all that: https://xkcd.com/882/

If you don't prepublish what you are specifically looking for, you can eventually find something that will match.

Reply on news.ycombinator.com