Skip to content

Randomized but unblinded experiment on vitamin D as a coronavirus treatment. Let’s talk about what comes next. (Hint: it involves multilevel models.)

Under the heading, “Here we go again,” Dale Lehman writes:

If you want to blog on the continuing theme – try this (it’s from Marginal Revolution, the citation):

Vitamin D Can Likely End the COVID-19 Pandemic

What is striking is the analysis by the Rootclaim group – repeated reliance on p values as the be all and end all of analysis. They even have the statement “It is important to understand that once a p-value has been obtained, the sample size no longer matters.” I only wish the raw data was available, because I am unable to do my own analysis. The randomization was far from perfect – there are numerous substantial differences between the two groups. They do note this, but their defense is that they performed a logistic regression, including (some of) those different factors, and the p values indicate that the results are robust. But which additional factors? Hard to tell. And, once again, the data are not available (as far as I can tell). The Rootclaim analysis does include a sort of “decision analysis” of whether or not to use vitamin D, which is a good thing. But it seems to state far too much certainty in the estimates for my taste.

Also, it appears that they did a Bayesian analysis (I hadn’t read that far yet), so you might want to look at that.

Here’s their key finding:

Of 50 patients treated with calcifediol, one required admission to the ICU (2%), while of 26 untreated patients, 13 required admission (50%).

Indeed, if this difference is real (and not explained by pretreatment differences between the groups, or by admission-to-the-ICU decisions that were not based on medical conditions) then this is a huge effect, and statistical significance wouldn’t come into play. A graph of outcome vs. pre-treatment risk factors would help here, especially if you could get some sort of continuous outcome, not just ICU-or-not. would be useful here. The p-value is kind of irrelevant, but there’s no harm in calculating it either.

The thing that confuses me is that the title of the published paper says this is “a pilot randomized clinical study.” The purpose of a pilot study is to demonstrate the feasibility of an experiment, not to estimate the treatment effect. In this case, they had a controlled experiment, so why do they call it “a pilot randomized clinical study” and not just “a randomized clinical study”? Apparently the assignment was not double-blind, but then I’d think you’d call it “an unblinded randomized clinical study.” I don’t get why they called it a pilot.

I certainly don’t want to suggest that the results of an experiment should be discarded just because it was unblinded (or, for that matter, unrandomized, were that the case). Perfect studies are rate, and generally we just need to consider possible flaws, adjust as well as we can, replicate replicate replicate, and then study variation in the treatment effect.

That’s right, I’m talking about multilevel modeling again. You don’t think this treatment has the same effect on all people, right? It will help some people more than others, might even hurt some people. And given that the effect varies across people, it will vary across groups as well, so you’ll expect to see different effects in different experimental settings.

This doesn’t directly affect the analysis of the study described above, but it gives us a sense of what might come next.


  1. I think a lot of the time the purpose of a pilot study is to get enough preliminary data that a funding agency will get excited enough that they will find your full size study. If you happen to get unlucky enough to find a really large effect then you might not get the funding. After all the pilot was enough (yes this is tongue in cheek sarcasm)

  2. Clyde Schechter says:

    “I think a lot of the time the purpose of a pilot study is to get enough preliminary data that a funding agency will get excited enough that they will find your full size study. ”

    This is definitely true. And I have personally been in the position where a pilot study, short term, single site, and non-randomized in this case, proved surprisingly successful. The reviewers of our subsequent application to do a more extensive, multi-site, randomized study explicitly cited our pilot as having settled the issue and made the proposal moot. See Adelman J, Aschner J, Schechter C, Angert R, Weiss J, Rai A, Berger M, Reissman S, Parakkattu V, Chacko B, Racine A, Southern W. Use of temporary names for newborns and associated risks. Pediatrics 2015 136(2):327-33.

  3. Matt Skaggs says:

    From the background section of the actual paper:

    “Severely sick patients require hospital admission and about 20% of hospitalized patients will developed Acute Respiratory Distress Syndrome (ARDS) and require intensive care unit (ICU) treatment [3]”

    The study population was patients who were getting admitted to the hospital for COVID. In the study, 50% of the no-vitamin-D group needed to be transferred to the ICU. I would think that 20% and 50% are kind of not the same. A more conservative estimate of the effect would account for this difference.

    Also from the paper:

    “A statistically significant difference was identified for the variable hypertension (26 had a history of hypertension of which 11 (42 %) received Calcifediol and 15 (58 %) not…”

    I have warning lights flashing on this. Those numbers, 42% and 58%, are the percentage of the total with hypertension in each group. Those numbers mean nothing at all from a statistical standpoint, so why mention them? The numbers that matter are 15/26 with hypertension in the control group (still 58% because the no-treatment group was 26 and the total number with hypertension was also 26) and 11/50 (22%). That is a huge difference that seems to have been intentionally avoided in the paper, which…gives me pause.

  4. Esteban says:

    Despite the “pilot” moniker, given the large treatment effect that appears to have been observed for a hard endpoint, and even considering the relatively low number of subjects, one must ask if sufficient clinical equipoise remains to ethically justify conducting the larger trial?

    • This is a good point. Giving a large dose of Vit D is near cost free and near impossible to cause harm. Vit D deficiency is also surprisingly common in many parts of the country. It’s probably the case that we should just give them as a precursor to getting a COVID test.

      • confused says:

        yeah exactly. This isn’t like hydroxychloroquine, it’s just a vitamin.

        How plausible is vitamin D overdose? If it’s not, maybe CDC etc. should recommend everybody take vitamin D supplements.

        • Navigator says:

          It’s all fine, under the assumption your body can synthesize and process the supplement and reap the same benefit as it does when ingested with food with other micronutrients.

          A lot of adults in the developed world are vit. D deficient, but it is hardly for the lack of it in the food (most food is intentionally fortified).

          Getting in enough and being able to use it are two different things.

          The other issue is why mega-doses (of either C or D) would work better than just the recommended daily amount?

        • Vitamin D overdose is more or less a theory rather than a fact. It’s a fat-soluble vitamin, so there’s a theory where it could accumulate like crazy, but I’ve seen studies where they give something like 600,000 IU intramuscular injections without harm. Typical vitamin D pills are 2000 IU which is 0.05mg, this study gave 0.532mg = about 10 of these little 2000 IU pills. They’re easy to swallow, you can buy them in 220 tablets for $5 or something. Giving 10 oral vit D pills is virtually cost-free. You could shake them out into little salsa cups and hand them to people right before taking their swabs at testing stations… Total cost would be like an extra $1 on top of a test that currently costs $100+.

          You might not want to take 10 pills daily for weeks on end, but 10 pills one time initial dose, plus say 2 pills a day for a week would be totally fine.

          • Anoneuoid says:


            Vitamin D is a membrane antioxidant: thus Vitamin D3 (cholecalciferol) and its active metabolite 1,25-dihydroxycholecalciferol and also Vitamin D2 (ergocalciferol) and 7-dehydrocholesterol (pro-Vitamin D3) all inhibited iron-dependent liposomal lipid peroxidation.


            Compared with several well-known antioxidants, vitamin D3 may be one of the most powerful antioxidants in biological organisms as shown in the present study.


            Vitamin D a sa hormone never made sense to me at all because why wouldn’t the receptor just evolve to have an easier to synthesize ligand? And why should a hormone get depleted in viral illness?

            If its one of the most potent lipid soluble antioxidants it makes sense why the unique structure is required and it gets depleted in illness.

          • JDK says:

            “In the 1940s, massive doses of vitamin D (200,000–300,000 IU/day) were considered an effective treatment strategy for chronic illnesses as diverse as tuberculosis and rheumatoid arthritis. Because hypercalcemia was observed in some patients thus treated, individual doctors discontinued the massive doses and the symptoms of VDT disappeared after a few months” – Vitamin D Toxicity–A Clinical Perspective.
            It is rare but possible, I think noted supplement guru Gary Null gave himself Vit D poisoning from a poorly mixed dose of one of his supplements. I am skeptical of the above result having seen other clinical studies noting no benefit of Vit D in treating ARDS.
            It is interesting to look at the history of treatments and to see how persistent some of these zombie ideas are.

            • Anoneuoid says:

              It is interesting to look at the history of treatments and to see how persistent some of these zombie ideas are.

              The way medical research is done they would never discover that water puts out fires and paint stops rust. I am 100% serious about that. I have been wanting to build a bunch of small windmills exposed to various humidity and wind conditions then light them on fire. That would be followed at random times by random amounts of water (or not) as the treatment and checking how many are still functional by the next day.

              You can’t just give arbitrary doses of something at arbitrary times and expect it to work even if it is the most effective intervention.

            • Sure, I think it’s clear that 200k to 300k IU/day for any extended period is way too much. But the US RDA is 400 for an adult male (based on looking at the back of a multi-vit). And the recommended daily maximum for routine use is 4000 IU for normal people. The idea that you’d somehow hurt people by giving 20k IU once on diagnosis followed by the recommended daily maximum of 4k daily for 2 weeks post diagnosis seems unlikely. (and this study only uses the 1-time 20k dose)

              The wikipedia article suggests: “Vitamin D toxicity is rare.[29] It is caused by supplementing with high doses of vitamin D rather than sunlight. The threshold for vitamin D toxicity has not been established; however, according to some research, the tolerable upper intake level (UL) is 4,000 IU/day for ages 9–71[154] (100 µg/day), while other research concludes that, in healthy adults, sustained intake of more than 1250 μg/day (50,000 IU) can produce overt toxicity after several months and can increase serum 25-hydroxyvitamin D levels to 150 ng/mL and greater”

              So more than double the dose used in this study, daily, for “several months” can produce overt toxicity.

      • Bob76 says:

        Well, notice that they did not give vitamin D3. They wrote:

        Our study does not include a comparison with cholecalciferol, the native vitamin D3 form and nutritional substrate for calcifediol, so that we cannot conclude that calcifediol is superior to vitamin D itself. Nevertheless, calcifediol may have some advantages over native vitamin D. It has a more reliable intestinal absorption (close to 100 %) and can rapidly restore serum concentrations of 25OHD as it does not require hepatic 25-hydroxylation.

        I have no idea how rapid the hepatic 25-hydroxylation proceeds, what impairs it, or even what it is.

        There is a chance that the described treatment is either (1) more expensive than vitamin D or (2) works better for some or all patients.


        • Good catch. They gave calcifediol, which is the “active form” and created from Vit D3 by hydroxylation in the liver. according to wikipedia: “At a typical daily intake of vitamin D3, its full conversion to calcifediol takes approximately 7 days.[4]”

          So, this would rapidly boost circulating active blood levels compared to giving D3.

          On the other hand, even if it were $100 a dose, it’d be extraordinarily cheap compared to an ICU stay.

          If the mechanism actually works, then telling people to take vit D3 at the 4000 IU per day rate in the general population, prior to COVID exposure would be a cheap and easy public health intervention, which could lead to much lower incidence of COVID complications. But it wouldn’t necessarily be a sufficiently quick acute intervention to give D3

          Again, good catch.

          • Bob76 says:

            Your observation that this study indicates that it would be beneficial for the general population to take substantial vitamin D supplements prior to exposure makes sense to me. It seems to me that the relevant cost/benefit calculation is trivial.

            There appears to be a growing literature supporting the view that the severity of a covid-19 infection is inversely correlated with a patients vitamin D level at diagnosis/admission.

            The IOM website has a 2010 webpage discussing vitamin D. They recommend an upper limit of 4,000 IU (100 micrograms) per day for adults.


            Given (1) the correlation, (2) the relatively high upper limit, and (3) the low cost of vitamin D supplements, it appears that encouraging widespread consumption of high doses of vitamin D would be sound public policy.

            PS I take a modest vitamin D supplement daily, but I’ve ordered some 4000 IU pills. I intend to begin taking them unless my physician strongly advises against it.

    • jim says:

      “one must ask if sufficient clinical equipoise remains to ethically justify conducting the larger trial”

      Absolutely it does. It *always* does, since there is *always* the significant possibility that the study was improperly executed. If it were my call, I’d say two more similar studies at different institutions conducted by different groups. That would be the minimum.

      Whether the treatment is harmless or not, it’s critical to know if it’s actually doing anything because at the very least it’s consuming time and resources that could be dedicated to other things. I know right now Daniel will say it’s not very much time or resources, but I disagree. It’s a new curve in the road, and that requires substantial effort to institute. Once it’s been standard protocol for a few months, sure, it might fit in smoothly but until then it will be a distraction.

      • I don’t think it requires “substantial effort to institute”. You could buy vit D pills in bulk for around $0.01/pill. Handing 10 of them out to people who seek COVID tests would be basically cost free. The fact is there are lots of theoretical reasons to believe it could be helpful (including that it’s a well known deficiency, and the deficiency gets worse through the fall and winter). In fact just knowing that it’s a well known deficiency means it would be a cost effective public health intervention even without COVID.

        • jim says:

          I knew you’d say that.

          You’re working in a system. The system has protocols. Tens of thousands of people work in their local systems and have refined the system and protocols in their little world to make everything go as smooth as possible. Adding a new protocol changes the dynamic of the system and creates a distraction in a situation where minutes count. It better be worth it.

          *who* could buy it in bulk? The buyer, who already has 10,000 items to buy and keep track of. Who will stock it? The stock dude, who also has 10,000 items to keep track of. Where will it go so everyone knows where it is?

          On paper everything is simple.

          • So we can pop up tens of thousands of PCR testing labs with hundreds of thousands of employees and purchase swabs and PPE by the bazillions but we can’t order bottles of pills?

            I mean you’re probably right, the US is so dysfunctional it can’t find it’s ass with both hands. But in Germany they’d get it done by this afternoon. In Spain it’d take them a day. In Sweden they’re likely already doing it since they know vit D is a big issue.

  5. Gordon says:

    The other thing about this trial is that the huge effect size is actually in line with much of the other observational data.

    Here’s a list of all data that’s been collected so far:

    The other vitamin D intervention trial which wasn’t randomized also saw a ‘too good to be true’ effect size:

  6. samuel says:

    “The purpose of a pilot study is to demonstrate the feasibility of an experiment, not to estimate the treatment effect.”

    Of course not, estimating a treatment effect from a pilot study would be crazy! No, the purpose of a pilot study is to estimate all the OTHER parameters that go into the sample size calculation for your full-size study.

  7. Anoneuoid says:

    What is the point of randomization if we are going to reject it whenever it results in imbalance? Better to purposefully choose balanced samples then right?

    • Andrew says:


      Randomization is just fine. It’s also fine to adjust for pretreatment predictors, even with a randomized design. And it’s fine to do restricted randomization so as to assure approximate balance of pretreatment predictors of interest.

      • Anoneuoid says:

        But apparently randomization is not fine when it results in obvious large imbalances. In this case we just happen to know about the blood pressure and diabetes imbalances and that casts doubt on whether we can attribute the difference to the intervention. We do not know BMI and many others though.

        • Randomization only ensures balance in the limit of fairly large samples. So when we’re talking 76 people randomized into 50 and 26 person groups, it’s not surprising that there’s a bunch of imbalance. Fitting a good predictive model which accounts for imbalanced factors seems like a good plan in all cases, unless you’re really looking at large randomized studies. If it were 50,000 and 26,000 people randomized with an appropriate RNG you could be pretty darn sure that everything was balanced within epsilon.

          • Anoneuoid says:

            So when we’re talking 76 people randomized into 50 and 26 person groups, it’s not surprising that there’s a bunch of imbalance.

            Then basically we should expect preclinical research to be imbalanced by default since it pretty much ranges from n = 3 to n = 30.

            If it were 50,000 and 26,000 people randomized with an appropriate RNG you could be pretty darn sure that everything was balanced within epsilon.

            To get such large samples usually you have to start averaging over many relevant subgroups and it washes out the effect.

            I just dont see much use for randomization then.

            • Randomization is definitely useful. One way in which it’s useful is to eliminate any choice on the part of the recruiters/researchers. If the reason you’re in the drug arm vs placebo arm is that someone chose you, compared to a RNG chose you, well you can at least eliminate any bias caused by human “desire”. So that’s advantageous even if it doesn’t lead to perfect balance.

              Another reason is that you’ll wind up converging towards balance as sample size gets bigger. The degree to which you rely on “statistical adjustments” will decrease as sample size increases, so even if you don’t get precise balance, you rely less on adjustment.

          • Anoneuoid says:

            I swear there was a Bradley Efron paper on this but cannot find it. Maybe it was someone else.

        • jim says:

          “apparently randomization is not fine when it results in obvious large imbalances.”

          Likely the studies where randomization results in large imbalances are in the studies with small n, like this study.

  8. Nick Adams says:

    Lot’s of problems here.
    This is not a pilot study, it’s just the first 76 patients from a planned 1000+ subject study according to the clinical trials registry. Clearly then it is very low-powered (at least according to the values they used to calculate the power of their larger study). Wild fluctuations in effect size and direction are common early in a data collection series. Also lack of blinding. Primary endpoint is meant to be mortality with 11 secondary endpoints, one of which is ICU admission. Unclear why an interim analysis was done after 76 patients. I could go on.
    The mechanistic and ecological evidence that Gordon above has collated is hardly impressive.
    The history of the attempts to use vitamins (C,D, and E) therapeutically in the absence of a deficiency state is pretty dismal.
    It’s also worth noting that like many things, blood levels of vit D correlate poorly with total body stores. Current consensus is not to measure vit D (although lots of doctors still do it) for this reason.
    The general fate of small, imperfect studies showing huuuge effect sizes is that they fail to replicate.

    • Anoneuoid says:

      The history of the attempts to use vitamins (C,D, and E) therapeutically in the absence of a deficiency state is pretty dismal.

      Since I “discovered” vitamin C after it helped me during a very bad “flu” last November I have looked at the literature. It is filled to the brim with really bad studies.

      Theoretically, it’s primary role is to prevent oxidative damage due to free radicals and the like (labile ferric iron). How many studies can you find that choose a dosing schedule to match the metabolism of vitamin C and give it before most of the oxidative damage has already been done (ie, the house is almost completely burned down)?

      I have not looked so close at the other vitamin studies but I would assume the methodology is just as lacking.

      • jim says:

        What I read is that in large trials vitamin supplements do exactly zero. That would be consistent with my own anecdotal evidence, in which my use of vitamins has always resulted in…nothing. I don’t get the purported boost of energy that many people report; I don’t feel better or feel worse, although my wallet feels lighter, and I guess I feel slightly worse for having let myself be duped into buying into the vitamin meme.

        • Anoneuoid says:

          Yep, they do nothing or even harm until something happens to make you go deficient. Just like spraying water on your house when it is not on fire.

        • I agree with Anoneuoid. Large studies of vitamin supplements are largely studies among people who don’t have mechanisms whereby they need more vitamins than average + a small group that benefits. The net result is “no effect” because the small group is way too small to detect within the dominant zero effect group.

          People with allergies, autoimmune disorders, frequent / chronic infections, or genetic polymorphisms that result in lower activity of various enzymes etc can benefit greatly. But something like 10% of the country suffers from chronic sinusitis for example. That group will benefit. A large group of people suffer from seasonal allergies, they’ll benefit *during their allergy season*. A large group of people live in northern lattitudes, they’ll benefit from vit D in the wintertime…

          But it’s like Anoneuoid says, if you do stupid RCT on the general population type studies, you’ll find that water doesn’t help put out fires, and paint doesn’t prevent rust.

          • confused says:

            Hmm, allergies?

            Maybe I should have been taking vitamin D supplements all along, especially as I sunburn ridiculously easily…

            • Anoneuoid says:

              Maybe I should have been taking vitamin D supplements all along, especially as I sunburn ridiculously easily…

              Can you explain your reasoning behind this statement?

              • confused says:

                Which part?

                I have annoying seasonal allergies, so if vitamin D really helps with that, it would make sense to take.

                If you’re asking about the correlation with sunburn, I was saying perhaps I don’t get enough vitamin D naturally as I tend to try to stay in the shade outside of the winter.

              • I think it’s quite plausible. My own seasonal allergies are helped by a variety of things, allergy shots being an important one, but vit D, vit C, NAC, Niacin, and Mg (epsom salt bath) are all worth trying.

              • Anoneuoid says:

                If you’re asking about the correlation with sunburn, I was saying perhaps I don’t get enough vitamin D naturally as I tend to try to stay in the shade outside of the winter.

                Thanks. I just slowly ramp up my exposure in the spring and then dont get sunburned on the areas that got tanned. I’ve heard taking enough vitamin C or eating an all meat diet also makes it very difficult to get burned.

          • Martha (Smith) says:

            Daniel L said,
            “But something like 10% of the country suffers from chronic sinusitis for example. That group will benefit. ”

            Will benefit from what? I’d love to know, because I think I have chronic sinusitis, if I understand the term correctly. But maybe I don’t? What I have is chronic sinus drainage on one side; the drainage sometimes produces a burning feeling, and sometimes causes dizziness and/or mild nausea.

            • I definitely had chronic sinusitis for several years, also eczyma and a few other issues. Diagnosis is basically “Atopic Dermatitis” but it’s basically just a way of saying over-active immune response. It was extremely bad for a few years between about 2014 and 2017. It’s a disorder which is similar to several other inflammatory disorders. There’s a strong component associated with not having enough healthy sinus bacteria, often because people take antibiotics to treat the initial few infections. But the initial infections are basically caused by swelling and discharge from allergic response usually.

              There are a number of chronic inflammatory or autoimmune diseases that all have similar issues: asthma, eczema, irritable bowel disorder, chronic gastritis, rheumatoid arthritis, Multiple Sclerosis. There’s some thought that alzheimers and type 2 diabetes are related inflammatory disorders etc.

              In all chronic inflammatory conditions, the immune system is hyper-active. One of the main activities of the immune system is to go around dumping oxidative chemicals on things. There’s a number of anti-inflammatory treatments that can help with this:

              1) treatments that reduce the hyperactivity of the immune system: allergy shots, vit D, magnesium salts, possibly melatonin

              2) Treatments that support anti-oxidative-stress processes: N-acetylcysteine which is a precursor to glutathione, Niacin which is used in nicotinamide adenide dinucleotide (NAD) to transport electrons and recycle other antioxidants, vitamin C which is a strong antioxidant, and vit D which is a cell-membrane antioxidant

              3) Reduced calorie intake, or intermittent fasting: a fast of 3 days causes most circulating immune system cells to die off, within a day of starting eating again your body produces massive quantities of immune cells from the stem cells in the bone marrow. This “immune reset” is definitely implicated in reducing inflammation (research out of USC gerontology dept). Other forms of fast are also beneficial, such as half-normal-calorie for a week or even some people only eat between the hours of say 8am and 4pm on a routine basis, so the body has 16 hours each day with no food…

              4) Reduction of exposure to irritants/allergens: HEPA air filters, water filters to avoid chlorine in your tap water, change in diet to avoid foods that you don’t tolerate.

              It’s been relatively long established that people with chronic inflammatory issues, such as hay fever, eczema, etc tend to have dramatically reduced levels of circulating antioxidants, this is basically because they’re getting used up at dramatically higher rates than the rest of the population.

              If you study those people, giving them antioxidant vitamins such as C, D, E, Niacin, etc, you’ll see improvements in skin condition, sinus symptoms, etc that simply wouldn’t be detected if you took a random sample of people and then only had say 5-10% of them had these kind of conditions.

          • Bob76 says:

            MS is a relatively infrequent disorder. There is a North-South gradient of its frequency in the population. (Actually, equator-pole gradient). Vitamin D has long been suspected as a contributing factor—the theory being that low vitamin D in childhood increases the likelihood of a person developing MS.

            To do an RCT of the effects of vitamin D, you would need a large cohort of subjects—recruited shortly after being born. You would have to have to track their vitamin D or provide supplements for about 15 years, and then track their health status (MS?) for another 25 years.

            Here’s a general discussion.

            I think this example illustrates how hard it would be to measure any effect.


            • Certainly it’d be hard for MS. But there are many other inflammatory disorders that are easier to study. For example eczema, hay-fever, irritable bowel disorder, type 2 diabetes, chronic gastritis. People have successfully treated for example eczyma with UV irradiation in the past, is that basically just a way to increase vit D? I don’t think we know. The poor design of studies as mentioned by Anoneuoid results in lots of “there is no effect” articles which should actually be written “we were too ignorant to formulate a proper hypothesis first, so we did a noisy study and didn’t find any statistical significance, we then incorrectly declared no effect when what we should have declared is that we wasted money on a terrible study”

    • jim says:

      Nick Adams says: “The general fate of small, imperfect studies showing huuuge effect sizes is that they fail to replicate.”

      Ding ding ding! Winner!!

    • Mendel says:

      > Unclear why an interim analysis was done after 76 patients.

      My guess at the reason would be “ethics”. If you see a huge effect like that, you’ve got to consider terminating the study and just giving everyone vitamin D, and also spreading the word, because you would be denying people help and treatment if you didn’t.

    • Carlos Ungil says:

      > Primary endpoint is meant to be mortality with 11 secondary endpoints, one of which is ICU admission.

      It’s not as bad as that. Death and ICU admission are the 2 primary endpoints (there are also 11 secondary endpoints).

    • Esteban says:

      Interesting. This thread brings to mind the Vit C/Cancer/Pauling/Cameron/Mayo saga from decades past. With respect to the recent controversies over the potential efficacy of HCQ/Az and now supplemented with calcifediol against COVID-19 associated morbidities there arises the following question: does there exist a hegemonic medical orthodoxy that enjoys and zealously protects a cognitive authority (Gieryn) over medical knowledge? Remdesivir’s rapid and somewhat tenuous ascendency for COVID-19 therapy vs. HCQ is eerily reminiscent of the interferon for cancer vs. IV-administered Vit C Pauling story. The primacy of the well-designed clinical trial in the final adjudication of medical evidence in what passes for widely acceptable knowledge vs. “fake news”. Incidentally, recent studies on IV Vit C appear to indirectly support some of Pauling and Cameron’s positions.

  9. Stevec says:

    I’m glad to see this paper highlighted. Yes, small scale but given the well-known role of vitamin D in the immune system, and the well-known fact that large numbers of people in the US and UK (don’t know about the rest of Europe) are vitamin D deficient in the winter, someone (WHO, CDC) should have organized large scale trials long ago.

    I recommend:
    – Vitamin D supplementation to prevent acute respiratory tract infections: systematic review and meta-analysis of individual participant data, Adrian R Martineau et al, BMJ 2016
    – Vitamin D deficiency in African Americans is associated with a high risk of severe disease and mortality by SARS-CoV-2, Virna Margarita Martín Giménez et al, Journal of Human Hypertension, 2020

    • Stevec says:

      I was trying to remember another paper when I wrote the above comment.
      Also worth a read:

      Association of Vitamin D Status and Other Clinical Characteristics With COVID-19 Test Results
      David O. Meltzer et al, JAMA Network Open, 2020 –

      Hopefully the statistics gurus here can comment.

      The simple version:

      Question Is vitamin D status, reflecting vitamin D levels and treatment, associated with test results for coronavirus disease 2019 (COVID-19)?

      Findings In this cohort study of 489 patients who had a vitamin D level measured in the year before COVID-19 testing, the relative risk of testing positive for COVID-19 was 1.77 times greater for patients with likely deficient vitamin D status compared with patients with likely sufficient vitamin D status, a difference that was statistically significant.

      Meaning These findings appear to support a role of vitamin D status in COVID-19 risk; randomized clinical trials are needed to determine whether broad population interventions and interventions among groups at increased risk of vitamin D deficiency and COVID-19 could reduce COVID-19 incidence.

      • Yes, randomized clinical trials are needed to determine the role of Vitamin D. A very thorough literature needs to be conducted whether there are in fact some good previous studies on Vitamin D.

        There are several physician/nutrition gurus that offer their views on YouTube. So far Dean Ornish and Neal Barnard have been very convincing about the role of nutrition to build one’s immune response. Though neither take large dose of Vitamin D.

        I rely on sunlight as it is claimed to be the safest source for Vitamin D as long as we start out slowsly to avoid sunburns.

        As you age, your absorbtion of Vitamin D is more and more impaired. In any case most people are deficient according to some British and American studies.

        • confused says:

          RCTs may be needed to provide a really solid answer, but when the risk of the intervention is really very low, as it seems to be in this case, are they needed to start *recommending action*?

          The risk/benefit balance for vitamin D seems very different than for something like hydroxychloroquine.

  10. Wesley says:

    There is no good evidence that vitamin D is an effective treatment for covid-19. Unfortunately, I do not think this study showed that vitamin D is an effective treatment for covid-19.

    There is a pretty big confounder in the clinical trial above: all patients received hydroxychloroquine. Some quick Google scholar searches suggest that use of HCQ is often associated with lowered vitamin D levels and that supplementing with high-dose vitamin D can be very beneficial. For example, two case studies here:

    It’s apparently also known that HCQ inhibits vitamin D production in vitro and in certain patient subpopulations:

    HCQ is basically ineffective for covid (, and it’s potentially dangerous due to these side effects (and more!). Seems more like the conclusion should be “if you’re going to use HCQ in covid treatment, you better supplement your patients with vitamin D.”

    Now it’s certainly possible that HCQ+vitamin D is a good treatment, but this study didn’t test that.

  11. confused says:

    I really don’t like the headline that “Vitamin D can likely end the COVID-19 pandemic”, though. I believe pandemic really means ‘unusual/epidemic-level worldwide transmission’, not necessarily ‘unusually deadly’. 2009 H1N1 is considered a pandemic despite being less deadly than the average flu season.

    So if universal Vitamin D supplementation reduced ICU/death from COVID dramatically, down to seasonal flu levels or below, it would mean it wouldn’t warrant a lot of public-health action… but not technically ‘end the pandemic’.

    I do think these sorts of definitions need to be looked at/revisited by WHO or someone like that, at least in terms of ‘terminology for public consumption’, after the dust has settled.

    • Agreed, a better headline would be something like “Preemptive Vitamin D supplementation may dramatically reduce the morbidity and mortality of COVID-19”

    • jim says:

      The risk of farting around with something like vitamin D is that time is wasted and hopes are spent on something that in almost complete certainty is totally worthless.

      Look: if vitamin D is some miracle drug, we’d damn well know it by now. It’s almost 100% certain that this study has fatal flaws. Just the same, it would be reasonable for the study to be replicated by a reputable research group, if for no other reason than to seal one more miracle cure in the coffin.

      • confused says:

        Why is it almost completely certain to be totally worthless? I am certainly skeptical of vitamin = miracle cure claims, but there are rather odd disparities in COVID mortality that could conceivably be explained by something like this (e.g. why are African-Americans badly affected while people in sub-Saharan Africa seem lightly affected?)

        And given low cost/very low risk, the effect doesn’t have to be huge to be worth pursuing.

      • Anoneuoid says:

        Why would correcting a vitamin deficiency be worthless? That is a bizarre idea that ignores like 100 years of science.

  12. Bob76 says:

    I noticed that the UK NICE has published a quick reaction to the Castillo et al. paper.

    Bottom line. There are possible issues with the paper. It would be premature to change clinical practice based on it. But be sure you have enough vitamin D—400 IU per day.

    Of course, this piece does not appear to be peer reviewed. So, given the research incumbency rule we can ignore it


Leave a Reply