Hacker News Reader

Comment by shawncampbell | Hacker News Reader

> This population-based prospective cohort study included 495 077 women and men (mean (SD) age, 56.6 (8.1) years) from the UK Biobank study. Participants were recruited from 2006 to 2010 and were followed up through 2018. We evaluated all-cause mortality and mortality due to cardiovascular disease (CVD), cancer, respiratory and digestive disease. HRs and 95% CIs for all-cause and cause-specific mortality were calculated using Cox proportional hazards models with adjustment for potential confounding variables.

> Regular glucosamine supplementation was associated with lower mortality due to all causes, cancer, CVD, respiratory and digestive diseases.

View on HN

I am guessing they didn't control for lifestyle, etc. Someone who takes glucosamine either works out heavily and supplements (and probably watches their diet), or had joint surgery, and falls into the first group to some degree.

Edit: Yea, reading the study, I don't see anything about controlling for fitness level. Running and caring about one's diet alone would account for this effect without glucosamine.

So it's just another crappy observational study, which is about all we ever get on human nutrition. This is low quality on the evidence-based medicine scale.

> Yea, reading the study, I don't see anything about controlling for fitness level.

Doesn't table 2 and 3 show that they examined a bunch of relevant covariates including physical activity, smoking, alcohol use, diabetes, hypertension, cholesterol, statin and aspirin use? It's at least "something" at the very least and I'm not sure how you missed it if you were actually reading the study because it's very hard to miss.

"because it's very hard to miss". Haha rub it in, fair enough.

> Edit: Yea, reading the study, I don't see anything about controlling for fitness level.

They specifically control for the following lifestyle factors: "smoking status, alcohol consumption, physical activity, body mass index (BMI) and vegetable and fruit consumption."

(Ctrl+f for "covariates")

In the conclusion, they acknowledge: "Although we had carefully adjusted for potential confounding lifestyle-related factors in our analyses, we could not exclude the possibility that the results were confounded by unmeasured lifestyle-related factors."

But if someone "works out heavily and supplements," that should show up in self reported physical activity, and indirectly in BMI.

I missed that. Thanks a lot for the correction.

I don't see "other supplements" in that list; surely they had to control for that as well?

Awesome!, thanks for the tip `(Ctrl+f for "covariates")`

But if someone "works out heavily and supplements," that should show up in self reported physical activity, and indirectly in BMI.

Well, not to nitpick, but people with high muscle mass can easily end up with an "obese" BMI even if they have very low body fat. And it also doesn't take very many hours of working out per week to get jacked if you're doing heavy weightlifting and supplementing.

No idea if such people are common enough to skew the results of this study significantly, but it still makes me wonder a bit.

Sure, some (not a huge fraction, I'm guessing single digit percentage) of very physically active people are poorly modeled by BMI but it shouldn't matter if they are explictly handling these covariates.

It's more an issue if there is something they haven't accounted for, as they point out.

Even body builders will have a hard time of getting to obese levels of BMI most of them are in the “overweight” range usually around 26-27.

And no people like this aren’t common at all.

Reply on news.ycombinator.com