Study population and setting
This prospective cohort study included 838 adults (18+ years old) who tested positive for SARS-CoV-2 by PCR, had not used Vitamin D in the previous three months, and were admitted to one of eight COVID-19 wards at a hospital in Barcelona, Spain between March 1 and May 31, 2020 based on bed availability. Five wards added calcifediol (Vitamin D3) — 532 mcg on day 0 and 266 mcg on days 3, 7, 15, and 30 — to the standard treatment, whereas three wards did not. Participants admitted to all included wards received empiric treatment with hydroxychloroquine and azithromycin, with ceftriaxone (an antibiotic) if they had a suspected bacterial superinfection and dexamethasone (a corticosteroid) if they had severe or critical COVID-19. The study team collected baseline Vitamin D levels (deficient or replete), age, sex, ethnicity, high blood pressure, obesity (yes or no), dyslipidemia, cardiovascular disease, kidney disease, diabetes, respiratory disease, cancer, viral infection, and autoimmune disease status. Participants were followed until medical discharge or death, and the outcomes of interest included intensive care unit (ICU) admission and mortality. The primary analysis used multivariable logistic regression to calculate the odds ratios for ICU admission and death by initial treatment assignment adjusted for baseline covariates.
Summary of Main Findings
Of the 838 included participants, 447 were treated in one of the five wards in which vitamin D was used as part of standard COVID-19 care and 391 were treated in the three wards where it was not. The authors reported lower baseline Vitamin D levels in participants admitted to the non-Vitamin D ward (56% in the non-Vitamin D ward had <20 ng/mL, vs. 71% in the treated ward), but participants were otherwise similar by ward treatment status. There were fewer ICU admissions (4.5%  versus 21% ) and deaths (4.7%  versus 15.9% ) – and shorter times to ICU admission and death – in those assigned to wards that added Vitamin D to the standard of care. After adjusting for baseline covariates, the risk of ICU admission (OR 0.13, 95% CI: 0.07, 0.23) and death (OR 0.21, 95% CI: 0.10, 0.43) was lower among participants admitted to wards that added Vitamin D to standard of care.
This study attempted to measure the impact of a frequently used and fairly accessible supplement on COVID-19 outcomes.
While the authors report that participants went to each ward based on bed availability, this is not an appropriate replacement for randomization, although the authors treated it as such. Not accounting for ward-level treatment protocols in their analysis resulted in inappropriate measurements of statistical significance and artificially narrow confidence intervals that make it impossible to make a conclusion about the relationship between Vitamin D and ICU admission and death from COVID-19 from these data. With appropriate statistical adjustments, it would not be possible to reach any reasonable level of statistical certainty with only 8 wards. Additionally, the authors did not report or adjust for clinical severity or alternative treatments (e.g., corticosteroids) by ward which likely influenced outcomes regardless of vitamin D treatment status. Unusually, more people originally admitted to Vitamin D treatment wards died of COVID-19 than were transferred to the ICU during the study period. This is unexpected given that unstable patients are generally transferred to the ICU in an attempt to avert death; however, this could reflect health system strain and limited ICU space in the early days of the pandemic.
This article was originally published as a preprint, where the authors stated that participants were randomly assigned to treatment. The preprint, hosted on SSRN, was retracted seemingly because it was unclear if the authors received ethical approval for conducting a randomized trial prior to study initiation. The final publication, which does not reference the preprint, correctly states that this is an observational study and adds another reference to an ethical approval, although this approval is dated after data collection finished. The final version also excluded 92 participants who reported Vitamin D use in the preceding three months, which altered the sample size from 930 participants total (551 in the treatment group, 379 in the control group) to 838 participants (447 treatment, 391 control). The authors did not explain the source of the additional 12 participants in the control group.
This review was posted on: 19 July 2021