Most health articles ask you to act on their opinions. I am specifically asking you not to act on mine. In a moment, I’ll tell you whether or not I think Vitamin D prevents or treats coronavirus. But I’ll give you a free spoiler: I am less than 100% certain of what I’m about to say. So if you want to take Vitamin D, take it. If it does prevent or cure coronavirus, great. If not, the worst that will happen is you’ll have slightly better bone health. I can’t stress how much I don’t want to be those people who said they couldn’t prove face masks helped so you must not use face masks. Just ignore everything I’m saying, do a quick cost-benefit calculation, and take Vitamin D. That having been said:

Lots of people think Vitamin D treats coronavirus, and some of them have good evidence. For example, infection rate from coronavirus seems latitude dependent; in general, the further north an area, the worse it’s been hit. Northern areas get less sunlight, and sunlight helps produce Vitamin D, so whenever you see a disease that’s worse at high latitudes, Vitamin D should be on your short list of potential causes.

Also - in the US, COVID seemed to remit with the summer and worsen over the winter. It’s hard to distinguish this from general exponential growth and from the effect of playing ping-pong with gradually loosening/ tightening lockdowns, but the US spike this winter was pretty dramatic. Most Northern Hemisphere countries show such a pattern, most equatorial countries don’t, and some Southern Hemisphere countries arguably show the opposite. Whenever you see a disease that’s better in summer and worse in winter, Vitamin D is one of the possible culprits.

And also, black people get COVID 1.4x more than white people, and die of it 3x more often. There are lots of potential social causes for health disparities between black and white people. But among potential biological causes, one of the most important is that black people have much less Vitamin D at temperate latitudes - their dark skin blocks the sunlight that would usually help them produce it. This is another pattern that sometimes means Vitamin D could be involved (ask me about schizophrenia rates sometime!), although there are obviously lots of other things that could be going on here.

Also also, we know a lot of immune cells have Vitamin D receptors, and Vitamin D seems to modulate the immune system in some important way. A giant meta-analysis in 2017 found that Vitamin D modestly decreased rates of flus and colds, some of the coronavirus’ closest relatives.

So lots of people did studies, and some of them were pretty suggestive. For example, a couple of papers like Radujkovic et al found that coronavirus patients with low Vitamin D levels had worse outcomes (eg were more likely to need ICU care or to die) than patients with higher levels. A couple more papers like Annweiler et al found that patients who regularly took Vitamin D supplements for the year before they got coronavirus did better. A team from Quest Diagnostics, led by Harvey Kaufman, looked through 190,000 patients who they had tested for both coronavirus and vitamin D levels, and found that the coronavirus patients had notably lower Vitamin D levels than controls. Another team under Eugene Merzon in Israel looked through their databases and found the same in a 7807 patient sample.

Surely that’s enough evidence to reach a firm conclusion, right?

I’m not completely convinced.

In the US, more blacks than whites have gotten coronavirus. But this could be because of Vitamin D, or because of wealth/education disparities. One easy way to check this is to look at Asians. Despite their superficially lighter skin, they have Vitamin D deficiencies almost as bad as blacks. But on average they’re better-off and better-educated than whites. If blacks’ problem was Vitamin D, we would expect Asians to do worse than whites; if it was wealth and education, we’d expect them to do better. In fact, it’s a mix. They get coronavirus only about half as often as whites, but they do worse once they get it. This suggests that maybe blacks are are getting the virus more because of wealth/education disparities, but doing worse once they get it because of something that might be Vitamin D? So maybe Vitamin D is involved in severity, but not overall infection risk? We’ll come back to this.

If blacks get coronavirus more often because of socioeconomic reasons, and also have lower Vitamin D, anybody looking at coronavirus infection rate without adjusting for race is suspect. Merzon and the Israelis didn’t control for race (and Israel is quite racially diverse). Kaufman and the Quest team say they adjusted for race, but if you look at their paper, they didn’t have access to race data for any participants, so instead they looked at what zip code they were in, coded it as majority-black or majority-white or whatever, and adjusted for that. I live in a majority black zip code, so apparently I’m black now. And my lived experience as a person of color, which I hear is more trustworthy than any scientific study, tells me this is a big enough loophole to invalidate the entire paper. When white people in majority black zip codes have enough money/education to avoid the coronavirus more effectively than real black people, and also have higher Vitamin D because of their lighter skin color, the paper’s metholodogy is going to mistake this for Vitamin D preventing coronavirus.

Two other teams, Hastie and Raisi-Estabragh, independently analyzed data from UK Biobank’s 350,000 participants. They actually knew the race of all their participants instead of guessing (plus Biobank is 95% white anyway), and neither of them found any effect of Vitamin D status on risk of getting COVID. That makes me think the people who did find it were just failing to adjust for race and probably other things effectively.

(partisans of Merzon and Kaufman point out that the Biobank Vitamin D data was ten years old by the time they got the coronavirus data, so possibly out of date. But Vitamin D levels are usually stable over long periods, and their sample is more than big enough that this shouldn’t sink the study).

What about all the people who find that lower Vitamin D upon hospitalization for coronavirus corresponds to worse outcomes? Many diseases deplete Vitamin D, including coronavirus itself. So low Vitamin D levels could mean you have more comorbidities, could mean you’re sick and frail so you’re staying inside and avoiding sunlight, could mean you’re black and have whatever other risk factors black people have (including much more hypertension) - or it could just mean you have a worse case of coronavirus. All of these can make your outcomes worse. The people who took Vitamin D supplements consistently for a year before being hospitalized did better because people who can do anything consistently for a year will do better - this selects for high-conscientiousness people who care a lot about their health and have good relationships with their doctors.

Take away these confounders, and it’s another story. The same study (Annweiler et al) which found daily supplementation helped found that being given lots of Vitamin D just after entering the hospital didn’t seem to help. That’s because that’s more of a randomized trial type thing compared to just selecting for conscientious people. In fact, this team finds that although taking Vitamin D supplements consistently seems to decrease coronavirus risk, this is independent of how much Vitamin D you have, which seems more like what would happen with a confounder than with a real effect.

Are there real randomized controlled trials? Yes - three of them. One from Spain (n = 76) tried randomizing hospital patients to get or not get Vitamin D; the patients in the experimental group did much much much better (25x lower odds of having to go to the ICU!) than the control group. Another from India (n = 40) tested asymptomatic people with mild cases; everyone stayed mild but the patients treated with Vitamin D were three times more likely to clear the infection quickly. The last, from Brazil (n = 240) was a large multicenter RCT that tested whether Vitamin D affected the outcomes of hospital patients. It found no effect whatsoever, not even a hint of a trend.

I could take or leave the Indian trial; nobody is worrying very much about seroconversion in mild cases. The Spanish and Brazilian ones are a pretty jarring contrast. In the Spanish one, the Vitamin D treated patients did 25x better; in the Brazilian one, there was no benefit at all.

I side with Brazil. It’s bigger, more professionally-done, and has fewer minor statistical flaws that really shouldn’t matter this much but make me nervous. Also, in the past I have learned to side with negative RCTs rather than astoundingly massively positive ones when the two conflict. There were some early astoundingly massively positive RCTs for hydroxychloroquine, and then the bigger and more-professionally-done ones that came later showed no effect.

(yes, I’ve read the arguments that the bigger ones started too late, and HCQ works better for early prevention. But a lot of the small early ones that showed astounding results also started late. Sometimes small early studies just suck.)

To what degree should we be trying to come up with clever ways to reconcile the two studies? For example, the Spanish study used a version of Vitamin D called calcifediol; the Brazilian one used a slightly different version called cholecalciferol. Calcifediol becomes active more quickly than cholecalciferol, enough so that if I were in the business of defending the Spanish study I might argue that it takes a few weeks for cholecalciferol to work, so giving it to someone who will already be dead or recovered by then is meaningless. I don’t really want to be in the business of pretending to know things about the biochemistry of Vitamin D, so I am going to assume that whichever doctors ran these studies did not completely bungle them and use forms of Vitamin D that could not possibly have worked in the time period involved. It would make me more comfortable if someone who did understand Vitamin D biochemistry would confirm this, but I’m not going to hang the whole argument on it.

One last study worth looking at: Butler-Laporte et al from Montreal. They use Mendelian randomization, a high-tech method that tries to get experiment-quality evidence from observational data by looking at genes directly. The idea is: you can’t just measure whether people with low Vitamin D get COVID more, because that could be confounded by all sorts of things like whether sick people are more likely to stay indoors and get less sunlight or a thousand other things like that. So instead, you measure whether people with the genes for low Vitamin D get COVID more. We assume that people with the genes for low Vitamin D in fact have low Vitamin D. And this isn’t confounded by anything; we know their low Vitamin D is genetic. So if these people get COVID more, we can be pretty sure that their COVID is caused by Vitamin D.

But they don’t. People with genes for low Vitamin D get COVID at the same rate as everyone else. And their COVID becomes severe at the same rate as everyone else. The authors tried to limit their study to white people, so it shouldn’t be confounded by genes related to race. This is a pretty hard study to get around. If I really wanted to get around it, I could argue that Vitamin D variation within the normal range doesn’t matter, but getting a massive bolus at a hospital does, for some reason. But that throws out a lot of seemingly convincing arguments like the evidence from latitude.

So my impression is that the best-quality observational study (UK Biobank studies that adjusted for race correctly), the best-quality RCT (the Brazilian study), and an additional really neat high-tech method all converge on no effect of Vitamin D on coronavirus incidence or severity.

The loose ends that bother me the most are the seasonal pattern, the latitude data, plus the increased risk of hospitalization and death in Asians. I don’t have a great explanation for those. One possibility is that sunlight does help prevent coronavirus, it just isn’t Vitamin D mediated. I suspect that the “anything involving sunlight is Vitamin D” assumption a lot of epidemiologists have isn’t going to hold up very well - this seems especially true for cancer, where sunlight matters a lot but study after study has shown Vitamin D doesn’t help at all. It may also be true for schizophrenia, although I’m going off one really pathetic study there and it could very well turn out to be Vitamin D after all. Some people are doing a little bit of work to clear up what the sunlight-related-Vitamin-D-independent pathways might be; I think nitric oxide has come up a few times.

This would explain the failed RCT and Mendelian randomization. But if it were true, we would expect to see more of a correlation in observational data - the people with more nitric oxide (or whatever) would be the people who get the most sunlight would be the people who have the most Vitamin D. So I don’t think this really saves anything.

There’s also the argument on priors - if Vitamin D helps with other infections, shouldn’t it also help with this one? I don’t know. SACN reviews other evidence that Vitamin D only helps prevent infections in children and not adults, and at low doses and not high, which is enough slicing and dicing that it makes me skeptical of the whole thing. I don’t know if the other-infections data is on any stronger foundation than the COVID data.

I’ve waited until now to bring in the argument from authority. NICE, the UK health system’s guideline-making authority, says that “there [is] little evidence for using vitamin D supplements to prevent or treat COVID‑19”, which is a terrible framing (give me time to write a post on this) but gets the point across well enough. UpToDate, the private company that produces semi-canonical evidence aggregation for US doctors, says that “there is no clear evidence that vitamin D supplementation reduces the risk or severity of COVID-19” (sorry, you won’t be able to read that link without a subscription). Stuart Ritchie, who literally wrote the book on how to tell good science from bad, says he’s unsure, but in a way that sounds a bit skeptical. This seems like a pretty common position.

Here are my beliefs after doing this research:

Does Vitamin D significantly decrease the risk of getting COVID?: 25% chance this is true. The Biobank and Mendelian randomization studies are strong arguments against this; the latitude, seasonal, and racial differences are only weak evidence in favor.

Does Vitamin D use at a hospital significantly improve your chances?: 25% chance this is true. I trust the large Brazilian study more than the smaller Spanish one, but aside from size and a general bias towards skepticism I can’t justify this very well.

Do the benefits of taking a Vitamin D supplement at a normal dose equal or outweigh the costs for most people?: 75% chance this is true. The risks are pretty low, and it will probably bring you closer to rather than further from a natural range if you’re a modern indoor worker (side effects are few; the most serious is probably kidney stones, so don’t take it if you have any tendency towards that). And maybe some day, after countless false leads and stupid red herrings, one of the claims people make about this substance will actually pan out. Who knows?