3 ms·

Wow, anyone reading the source paper in detail (or trying to take away conclusions) has to know that the stats are really, really rough. The authors are almost

Wow, anyone reading the source paper in detail (or trying to take away conclusions) has to know that the stats are really, really rough. The authors are almost just doing internet desk research -- counting the number of Covid deaths/cases and just plotting that against the average Vitamin D level in the population of each Indian state in general. Based on a couple thousand healthy patients of the general population. (BTW, I think it's really important that everyone learn to see through the math/symbols/technical jargon when reading such research papers to understand the populations really being studied. All those r and p numbers are meaningless if the underlying patient data are flawed.) That is really rough. I could not imagine this would be credible if you substituted for example (just to think about it locally with a population you might know), "New York" for "Maharastra" state in the plot and asked if you believed the strength of finding that average Vitamin D level in New York state correlates with Covid death rate. How about for Florida -- would that be believable too? And also, the noticeable points making up the perceived trend are so few, and the points are not even weighted for population size (?)! Much more work in detail -- at the patient level with Vitamin D measurement -- would have to be done to be credible. I would not act on this information in the present form.

NightMKoder 6y ago
It doesn’t seem like they controlled for any other variable at all - even time. From the paper: “Various literature databases like PubMed and Google Scholar were searched to find the mean of 25-hydroxyvitamin D [25(OH)D] levels in different states and union territories of India”. I would have to assume not all those studies were done at the same time. This would maybe be a great elevator pitch to get some resources. Not sure if that’s the intent. Relatedly, have there been proper double-blind studies for vitamin d injections & COVID outcomes? I would guess there’s not a lot of interest given other substances that have better “method of action” hypotheses. Or maybe it’s just lack of funding (hence we see an article like this every month or so)?
- sieabahlpark 6y ago
  > It doesn’t seem like they controlled for any other variable at all - even time. Yet this post is still popular on HN. Makes you think just how many other headlines are just so completely wrong hidden behind terrible studies. 100% of scientists agree with who guarantees their paychecks
- lolc 6y ago
  > Relatedly, have there been proper double-blind studies for vitamin d injections & COVID outcomes? I think there was a study that saw improved clinical outcomes when a metabolized version of vitamin D was administered. Don't know where to find it though.
  - speakeron 6y ago
    It was this one from the Reina Sofia University Hospital in Spain (high dose of Calcifediol significantly reduced need for ICU treatment): https://www.sciencedirect.com/science/article/pii/S0960076020302764 https://www.sciencedirect.com/science/article/pii/S096007602...

NightMKoder 6y ago

It doesn’t seem like they controlled for any other variable at all - even time. From the paper: “Various literature databases like PubMed and Google Scholar were searched to find the mean of 25-hydroxyvitamin D [25(OH)D] levels in different states and union territories of India”. I would have to assume not all those studies were done at the same time. This would maybe be a great elevator pitch to get some resources. Not sure if that’s the intent. Relatedly, have there been proper double-blind studies for vitamin d injections & COVID outcomes? I would guess there’s not a lot of interest given other substances that have better “method of action” hypotheses. Or maybe it’s just lack of funding (hence we see an article like this every month or so)?

sieabahlpark 6y ago
> It doesn’t seem like they controlled for any other variable at all - even time. Yet this post is still popular on HN. Makes you think just how many other headlines are just so completely wrong hidden behind terrible studies. 100% of scientists agree with who guarantees their paychecks
lolc 6y ago
> Relatedly, have there been proper double-blind studies for vitamin d injections & COVID outcomes? I think there was a study that saw improved clinical outcomes when a metabolized version of vitamin D was administered. Don't know where to find it though.
- speakeron 6y ago
  It was this one from the Reina Sofia University Hospital in Spain (high dose of Calcifediol significantly reduced need for ICU treatment): https://www.sciencedirect.com/science/article/pii/S0960076020302764 https://www.sciencedirect.com/science/article/pii/S096007602...