The ongoing coronavirus pandemic has seen a wide variety of untested off-the-shelf COVID treatments suggested—from Trump’s infamous disinfectants to more traditional herbal remedies, the idea of a simple protection from the deadly disease has proven an appealing one.
Now however, a new study from geneticists at Brunel University London, published in BMJ Nutrition, Prevention and Health, has found there to be no evidence that a person’s vitamin D levels affects how severely they suffer from contracting COVID-19.
Vitamin D is created naturally by the body using sunlight and is particularly important for bone health.
Whilst some previous studies have suggested a link between vitamin D deficiency and COVID-19 mortality rates, the NHS currently stops short of recommending it as a treatment against the disease, although they do recommend that people should consider taking vitamin D supplements to make up for a lack of sunshine during lockdown
“All of our work is based on the idea that correlation is not causation,” said project-lead Dr. Fotios Drenos, a senior lecturer in genetic epidemiology at Brunel.
“For example, there are studies that say that those that die have lower levels of vitamin D, and that’s true.
But those that die also tend to be older – they’re kept at home and they’re inside. They’re not young, active people spending their day outside. That was the problem we were trying to address, what is correlation and what is causation.”
Using data from the COVID-19 Host Genetics Initiative and the UK Biobank, and a process known as Mendelian Randomisation, the team set out to determine whether an individual’s genetic predisposition for high or low levels of vitamin D had an effect on how badly they suffered the effects of COVID-19.
Whereas the ‘gold standard’ test of a randomised controlled trial would split a real-world group in two and give half vitamin D and half a placebo, Mendelian Randomisation creates its split using people’s inbuilt genetics. The scientists identify the genes associated with vitamin D—some people are predisposed to lower levels, whilst others naturally have higher levels—and then investigate whether those two groups are affected by COVID differently.
“So, essentially you have two groups that have been randomly selected by nature, as it were, because of their genes,” said Hasnat Amin, a doctoral researcher at Brunel.
“We can then compare the rates of COVID-19 between those two groups, and if we find an association we can say it is causal, because there are no other differences between the two groups.
“If the only difference is that one group has good vitamin D genetics and the other group has bad vitamin D genetics, and there is a difference in how severely they get COVID, you can say that there is evidence of a causal relationship, and if there isn’t a difference, you can say there is no evidence of a causal relationship. We found no evidence of a causal relationship.”
The study only covers people of European ancestry due to the datasets that were available to the researchers, who say their results aren’t necessarily applicable to other groups because of how a person’s skin-colour naturally affects their levels of vitamin D.
The study of genetics can be very highly skewed towards people from European ancestry as the data used is most frequently collected from Europeans or Americans of European decent.
“For many things, what we see in one population is relatively similar to the other populations,” said Dr. Drenos. “But especially in the case of vitamin D, skin colour, cultural differences and how people are exposed to the sun makes it difficult to say if our findings apply to everyone.”
The team are keen to stress though that they don’t seek to discourage people taking vitamin D supplements, which can have other positive health effects: “But if they do choose to take vitamin D, there is no evidence it’ll protect them from COVID 19 and they should keep following existing guidance to stay safe,” said Dr. Drenos.
In the ongoing COVID-19 pandemic, significant numbers of fatalities appeared to occur in the northern hemisphere with far fewer in the tropics and southern hemisphere. Past coronavirus and influenza epidemics and pandemics have displayed very strong seasonality [1].
Seasons are determined by latitude, and winter provides a more favourable environment for viral transmission: enveloped viruses are fragile and more easily destroyed by heat and UV light than other viruses [2]. Therefore, one would expect transmission rates to vary by latitude, however, overall COVID-19 data suggest that case fatality rates (CFRs) also appeared to be strongly governed by latitude, and this is highly unexpected.
A possible causal role for vitamin D in global COVID-19 outcomes was proposed as early as 12th March 2020 [3] by Dr Garami suggesting it should be tried, and the idea began to be discussed and gain support [4]. Dr Davies and Dr Byers had simultaneously been researching this area and together we jointly published a summary of early findings for medical professionals on 19th March [5], calling for hospitals to begin to treat, test, measure and report.
Vitamin D is a logical candidate hypothesis since it is produced in the skin in sufficiently strong sunlight and deficiency is well-known to depend on latitude and season [6][7]. Vitamin D is a steroid hormone with receptors in most tissue cell types [8], influences a large number of biological pathways, and deficiency is associated with increased risk for many chronic diseases including autoimmune diseases, some cancers, cardiovascular disease, infectious disease, schizophrenia and type 2 diabetes. Furthermore, Italy, the country worst affected by COVID-19 in Europe from February through April, has a very high prevalence of vitamin D deficiency [8].
Risk factors for low vitamin D are old age, winter, living at higher latitudes, darker skin pigmentation, less sunlight exposure, dietary habits, obesity, malabsorption syndromes and absence of vitamin D fortification [9][10].
COVID-19 deaths in BAME communities have been alarmingly high in the UK with data from intensive care units indicating that over a third of patients are from BME backgrounds [11,12] and 63.2% (67/106) of the first 106 health and social care staff known to have died from the virus were Black or Asian [13]. The New York Department of Health and Mental Hygiene also reported that Black/African American and Hispanic/Latino populations are suffering up to twice the case fatality rates of white New Yorkers [14].
There is a large literature base spanning decades showing strong correlations between vitamin D and all-cause morbidity, all-cause mortality, respiratory infections, acute respiratory distress syndrome (ARDS), as well as with comorbidities that have shown associations with COVID-19 such as diabetes mellitus [15].
Some small studies have demonstrated a causal role with respect to specific illnesses such as dementia [16] and stroke [17]. Conflicting trial results and study design flaws have led to controversy and raised the question of whether vitamin D plays a causal role or is simply a bystander, and is merely a biomarker for overall disease-health status.
Given the difficulties of running large randomized controlled trials (RCTs), more recent efforts have turned to Causal Inference methods and modern techniques such as Mendelian Randomisation making it possible to answer the question of causality using observational data sets. A combined sample from three European cohort studies used Mendelian Randomisation to provide further support for a causal relationship between vitamin D deficiency and increased all-cause mortality.
However, the study size, (~10,000 participants), was underpowered and larger studies on genetics and mortality were called for [18]. Meta-analyses showing causal relationships between vitamin D supplementation and morbidity and mortality exist. Intake of ordinary doses of vitamin D supplements seems to be associated with decreases in total mortality rates [19]. A more recent 2019 study of higher dose random control trials showed that while vitamin D supplementation did not reduce total cancer incidence, it significantly reduced total cancer mortality [20].
The vast scale of the COVID-19 pandemic provides a unique opportunity for high-level causal inference models involving a wide set of independent inputs (“exogenous variables”, or root causes). This makes it possible to do a simple but powerful qualitative analysis of model predictions from which we are able to draw firm causal conclusions by comparing those predictions with observed data. “Randomisation” in Causal Inference is generally performed by nature rather than the experimenter, but even non-randomised observational data can be used to infer causality using Bayesian analysis to remove selection bias [21].
The objectives of this study are two-fold. In the first section, we analyse and quantify severity of global COVID-19 outbreaks to define clearly the data that require explanation. In section two, we then use a causal inference framework to answer the binary question: has vitamin D played a causal role in outbreak severity or not?
Readers unfamiliar with Causal Inference may find it useful to first read Appendix 1 – Background on Causal Inference to familiarise themselves with the techniques used in this analysis.
reference link: https://www.medrxiv.org/content/10.1101/2020.05.01.20087965v2.full
More information: Hasnat A Amin et al. No evidence that vitamin D is able to prevent or affect the severity of COVID-19 in individuals with European ancestry: a Mendelian randomisation study of open data, BMJ Nutrition, Prevention & Health (2021). DOI: 10.1136/bmjnph-2020-000151