Thursday 10 March 2022

Covid-19 and Vitamin D – study from Liverpool


Covid-19 and Vitamin D – a study from Liverpool

Liverpool water-front

We have just seen the publication of another paper concerning the outcome of Covid-19 in relation to the blood level of Vitamin D. This study was undertaken in Liverpool, UK.

The results are not entirely straightforward as the study also investigated several parameters of Vitamin D. I will try to extract what I regard as the most important points and integrate them with other studies.

The study was of 992 patients with Covid-19 admitted to hospital in Liverpool between March 18th and November 2nd 2020, early in the pandemic. Of these, 472 were studied in detail because following initial blood testing, there was sufficient surplus blood serum to allow retrospective blood testing for Vitamin D.

Vitamin D levels

From the recent study from Israel, I find it compelling  that Vitamin D deficiency should be defined as less than 30ng/ml, 75 nmol/L. Less than this blood level indicates a high risk of severe or critical Covid-19, whereas above this level can be regarded as a safe range. In the study from Israel about 50% were deficient by this definition, and in a study from Heidelberg, Germany, in 2020 the proportion deficient of Vitamin D in was much higher at about 80%, in Figure 1.

Figure 1. Vitamin D in patients with Covid-19, 2020

In my personal research in Blackburn, UK, 20 years ago the proportion with Vitamin D deficiency as defined above using the Israel data, was 682 in the ethnic white sample of 818 (83%), and a very disturbing 1514 in the ethnic South Asian group of 1574 (96%). 

I have shown this data previously, but as it was two years ago I will display the bar-charts below. Each vertical bar represents the blood vitamin D level of one person. As the distribution is not "normal" or symmetrical, the median is shown. This is the halfway point, 50% of the subjects below or above this level, the vertical orange line. The horizontal yellow line indicates the blood level of the median. The horizontal red line indicates blood vitamin D level of 30ng/ml, 75nmol/L and we can immediately see how few of the subjects were above this safe line.

Figure 2. Distribution of blood level of Vitamin D in 818 ethnic white people

Figure 3. Distribution of blood level of Vitamin D in 1574 ethnic South Asian people

In the ethnic white group the median blood level of Vitamin D is 18ng/ml, 45nmol/L.
In the ethnic South Asian group he median blood level of Vitamin D is 9ng/ml, 22.5nmol/L.

Liverpool experience

Once again we have the problem of two units of measurement of Vitamin D in current use. The paper from Liverpool uses the unit "nmol/L". I will express both even though this might appear cumbersome. 1ng/ml = 2.5nmol/L

The Liverpool study regarded 20ng/ml, 50nmol/L as being the lower limit of the ideal range, and therefore below this to be VItamin D deficiency. However the study from Israel indicated that 30ng/ml, 75 nmol/L would better be regarded as the lower limit of the safe range, based on biological advantage.

Of the 472 Covid-19 patients studied in Liverpool, we can see in Figure 4 the numbers with various blood vitamin D levels.

Figure 4. Distribution of Vitamin D in 472 Covid-19 patients

83% were deficient of Vitamin D with blood levels less than 30ng/ml, 75nmol/L. 
96% had sub-optimal blood levels of less than 40ng/ml, 100ng/L. 
These results indicate a serious but reversible public health problem.

We can see that in these 472 Covid-19 patients, there is an asymmetrical distribution, and so once again an arithmetic mean is not of particular value. The distribution is skewed to the lower range of blood levels of Vitamin D and the median, the middle of the range, was expressed in the study report. 

The proportion with suboptimal levels (as defined by high illness incidence) is the most useful way of expressing Vitamin D status of a population.

The pattern is similar to but not quite as dramatic as the Heidelberg study ahown in Figure 1. In this, there are more sub-divisions of Vitamin D level than in Figure 4.


Of the 472 Covid-19 patients, 112 died within 28 days, 23.7% overall, 28% in men, 23.5% in women. This proportion of deaths was the general experience during 2020, when the highly vulnerable members of the population died.

Of those who died:

Average (median) blood level of vitamin D was 15.8ng/ml, 39.5nmol/L.

Of those who survived:

Average (median) blood level of vitamin D was 17.3ng/ml, 43nmol/L.

Those who survived had on average a slightly higher blood vitamin D level than those who died, in keeping with other studies, but the difference was not dramatic. The difference did not achieve "statistical significance", meaning that it could have been a chance finding. Whether or not there is clinical significance is a matter of judgement, a balance of potential benefits and risks of using or not using Vitamin D in clinical practice.

Statistical significance is influenced to a major extent in sample size. In investigating the effect of Vitamin D in Covid-19, it is operationally difficult to use a large sample size, unlike a pharmaceutical trial in a sample of normal people in the community.

More detail is shown in Figure 5. Look at it carefully to see which green columns (died) were higher than the blue (survived).

Figure 5. Covid-19 survival or death and Vitamin D

We can see that the chance of death is greatest in those with the lowest blood levels of VItamin D, but also in those with the highest. 

Severe disease

The infuence of Vitamin D on severe Covid-19 was expressed as Odds Ratio, the relative chance of developing severe Covid-19 based on blood level of Vitamin D. The reference was a blood level of 20–30ng/ml, 50–74 nmol/L, as considered to be middle of the range found in the general population.

The odds ratio for severe disease:

for Vitamin D 20–30ng/ml, 50–74 nmol/L OR=1 (reference)

for Vitamin D <10ng/ml, <25nmol/L OR=2.37

There is no surprise here: severe disease is much more likely in those with the the very low blood levels of VItamin D, less than 10ng/ml, 25nmol/L.

The study therefore identified something expected. But there is more to consider.

High blood VItamin D

There were 18 patients with Vitamin D greater than 40ng/ml, 100nmol/L. For these the odds ratio of severe Covid-19 was high, with OR=4.65. This was an unexpected finding.

We know about Vitamin D "intoxication", hypervitaminosis D, in which excess Vitamin D causes an increase in the amount of calcium in the blood and in the urine. This can occur only with very high blood levels of Vitamin D, greater than about 200ng/ml, 500nmol/L.

But in this study there was no hypervitaminosis D, as judged by high calcium level in the blood, but we do see a disadvantage of a blood Vitamin D level just greater than 40ng/ml, 100nmol/L.

The report is of quadruple (OR=4.65) the risk of severe Covid-19 in this group with Vitamin D level >40ng/ml, 100nmol/L compared to the reference range of 20–30ng/ml, 50–74 nmol/L.

An explanation

This finding came as a surprise to me as this result has not appeared in other studies. Is it just a chance finding, a sort of mistake? All findings need to be replicated before they are generally accepted, or there must be a plausible mechanism, some connection that makes sense. But initial findings such as this must not be dismissed.

The paper from the study looked into a plausible explanation. 

We must remember that Vitamin D produced in the skin or taken by mouth must be converted in the liver into 25(OH)D, the form which it circulates in the blood. 

In response to infection and to enable immune response, 25(OH)D is taken up by the immunity cells and is activated into 1,25(OH)D, which then unlocks VDR. A molecule of 1,25(OH)D can be used only once, and then it is converted into the inactive 24,25(OH)D.


25(OH)D is converted into the active 1,25(OH)D by the enzyme 1-alpha hydroxylase. Now, it appears that when the blood level of 25(OH)D increases above 40ng/ml, 100nmol/L, "Fibroblast Growth Factor 23" (FGF23) is induced, and this leads to the suppression of the enzyme 1-alpha hydroxylase. Activation of 25(OH)D to 1,25(OH)D will thereby be suppressed, and as a result there will be a diminished immune response.

This is a perfect example of the interaction of enzyme systems,  feedback control mechanisms, that became essential in evolution. The inactivation of 1,25(OH)D to 24,25(OH)D is an excellent example of an essential control system, and the suppression of the enzyme 1-alpha hydroxylase would be another.

Free and bio-available Vitamin D

Other Vitamin D parameters were investigated in the Liverpool study. When we assess blood levels of Vitamin D we measure it in its form that has been processed in the liver, 25(OH)D (also known as calcidiol or calcifediol). This is carried in the blood as a reserve, ready for use as and when necessary. In the blood it is bound to a specific Vitamin D Binding Protein (DBP), the amount of which can vary from person to person. This is because DBP is genetically determined, and the gene penetration is variable, determined by genetic polymorphisms, differences. The vast majority of 25(OH)D is bound to DBP and only 0.03% of it "free", unbound. Did this make a difference? 

This was investigated  and it was found that there was no relationship between DBP, and free D with mortality. Measuring 25(OH)D seems to be a satisfactory way to assess Vitamin D status.


The pandemic of Covid-19 has given huge research opportunity to investigate the optimisation of natural defensive immunity, with the role of Vitamin D based on extensive research during the previous forty years. 

Taking together the Israel and Liverpool studies, we can conclude that to minimise severe or fatal Covid-19 it is necessary for the blood level of Vitamin D to be between 30ng/ml, 75nmol/L and 40ng/ml, 100nmol/L.

It is clear that it is necessary to monitor regularly the blood level of Vitamin D so as to determine the optimal dose. 

We must monitor the blood parameters when correcting Vitamin D deficiency just as we monitor blood parameters when we treat diabetes and thyroid disorders.

I have noted many Comments. I have attempted to reply to them, but unfortunately, and for reasons that I do not understand, my replies do not attach. Sorry about this.