Low vitamin D levels do not aggravate COVID-19 risk or death, and vitamin D supplementation does not improve outcomes in hospitalized patients with COVID-19: a meta-analysis and GRADE assessment of cohort studies and RCTs

Background The associations between vitamin D and coronavirus disease 2019 (COVID-19) infection and clinical outcomes are controversial. The efficacy of vitamin D supplementation in COVID-19 is also not clear. Methods We identified relevant cohort studies that assessed the relationship between vitamin D, COVID-19 infection and associated death and randomized controlled trials (RCTs) that reported vitamin D supplementation on the outcomes in patients with COVID-19 by searching the PubMed, EMBASE, and medRxiv databases up to June 5th, 2021. Evidence quality levels and recommendations were assessed using the GRADE system. Results Eleven cohort studies with 536,105 patients and two RCTs were identified. Vitamin D deficiency (< 20 ng/ml) or insufficiency (< 30 ng/ml) was not associated with an significant increased risk of COVID-19 infection (OR for < 20 ng/ml: 1.61, 95% CI: 0.92–2.80, I2 = 92%) or in-hospital death (OR for < 20 ng/ml: 2.18, 95% CI: 0.91–5.26, I2 = 72%; OR for < 30 ng/ml: 3.07, 95% CI: 0.64–14.78, I2 = 66%). Each 10 ng/ml increase in serum vitamin D was not associated with a significant decreased risk of COVID-19 infection (OR: 0.92, 95% CI: 0.79–1.08, I2 = 98%) or death (OR: 0.65, 95% CI: 0.40–1.06, I2 = 79%). The overall quality of evidence (GRADE) for COVID-19 infection and associated death was very low. Vitamin D supplements did not significantly decrease death (OR: 0.57, I2 = 64%) or ICU admission (OR: 0.14, I2 = 90%) in patients with COVID-19. The level of evidence as qualified using GRADE was low. Conclusions Current evidence suggested that vitamin D deficiency or insufficiency was not significantly linked to susceptibility to COVID-19 infection or its associated death. Vitamin D supplements did not significantly improve clinical outcomes in patients with COVID-19. The overall GRADE evidence quality was low, we suggest that vitamin D supplementation was not recommended for patients with COVID-19. Supplementary Information The online version contains supplementary material available at 10.1186/s12937-021-00744-y.


Introduction
Recent studies have highlighted that the mean plasma vitamin D level is significantly lower in patients who tested positive for COVID-19 than in patients who tested negative [1]. Rhodes et al. [2] found that patients with COVID-19 residing in all countries that lie below 35 degrees north latitude have relatively low death. These results suggest that low vitamin D levels are associated with increased COVID-19 infection rates and worse outcomes. Several studies have reported that serum vitamin D deficiency is associated with an increased risk of COVID-19 positivity and worse outcomes (e.g., severe COVID-19 and in-hospital death) [3][4][5]. However, a larger cohort-based study from the UK Biobank showed a nonsignificant association after adjustment for confounders (COVID-19 infection HR: 1.00; p = 0.89; death HR: 0.98; p = 0.69) [6]. Two other studies also found no association between vitamin D and COVID-19 positivity [7,8]. Therefore, the impact of vitamin D on COVID-19 risk and clinical outcomes remains controversial, and a definite conclusion has not been reached. Thus, we performed a meta-analysis to clarify the association between serum vitamin D level, COVID-19 risk and associated death and assessed the effect of vitamin D supplements on clinical outcomes in patients with COVID-19 by pooling current evidence from clinical trials.

Methods
The present study was performed according to Preferred Reporting Items for Systematic reviews and Meta-Analyses Statement (PRISMA) guidelines (Supplemental Table S1). The protocol of this meta-analysis was not registered.

Literature search and study selection
Two authors (X.L. and J.C.) independently searched several databases (PubMed, EMBASE, and medRxiv) using the following groups of keywords with no language restrictions up to June 5th, 2021,: 2019-novel coronavirus, SARS-CoV-2, COVID-19, 2019-nCoV, vitamin D, death, severe and ICU. The details of the search strategy are described in Supplemental Table S2. We also searched the reference lists of relevant publications to identify further studies. The following inclusion criteria were used: 1) human studies that were published as original articles; 2) reports designed as cohorts with estimated effects (multivariate-adjusted) and 95% confidence interval (CI) results that reported the association between vitamin D, COVID-19 risk, and death; and 3) clinical trials (randomized controlled design) that assessed vitamin D supplementation on clinical outcomes in patients with COVID-19. Case-control, cross-sectional were excluded. When multiple papers reporting on the same study were identified, the most informative or complete article was included.

Data extraction and statistical analysis
Two authors independently extracted all data from the included studies. Discrepancies and disagreements were resolved via collegial discussion. The following information was extracted: first author, country, publication year, gender, mean or median age, study design, sample size, vitamin D level, OR or RR with the 95% CI for each category (results adjusted according to most potential confounders), and adjusted variables.
Effect measures were transformed to their natural logarithms (logOR), and the standard errors (SElog[OR]) were calculated from the corresponding 95% CI. We calculated study-specific slopes (vitamin D per 10 ng/ml increase), and 95% CIs from the natural logs of the reported RRs and CIs across categories of vitamin D levels [9,10]. Cochran Q and I2 statistics were used to detect statistical heterogeneity between studies. The overall quality of the included studies was assessed with the Newcastle-Ottawa quality assessment scale (NOS), and an NOS score ≥ 6 was considered high quality [11]. The risk of bias for the trials was assessed with the Cochrane risk of bias tool. The GRADE methodology was used to evaluate the quality of the body of retrieved evidence (GRADEpro, https:// gdt. grade pro. org/ app/# proje cts). All statistical analyses were performed using Review Manager version 5.3 (The Cochrane Collaboration 2014; Nordic Cochrane Center Copenhagen, Denmark). All statistical tests were doublesided, and P < 0.05 was considered statistically significant.

Publication bias
Publication bias was not assessed because of the limited studies (n < 10) according to the guidelines [25].

Grade
The overall evidence for the RCTs and observational studies was qualified using GRADE. Very little confidence showed that low vitamin D supplementation contributed to an increased risk of COVID-19 or COVID-19-related death. Furthermore, only low-quality evidence supports the benefits of vitamin D supplementation on death or ICU admission in patients with COVID-19. The GRADE tables are described in detail in Supplemental Tables S6  and S7.

Discussion
The present study showed that: (i) vitamin D deficiency (< 20 ng/ml) or insufficiency (< 30 ng/ml) was not associated with a significantly increased risk of COVID-19 infection or in-hospital death (P = 0.56). (ii) A 10 ng/ml increase in serum vitamin D was not significantly linked to an increased risk of COVID-19 infection or in-hospital death. (iii) Vitamin D supplements did not improve clinical outcomes in patients with COVID-19. Overall, our study suggested no significant association between vitamin D level, COVID-19 infection, and outcomes and no benefit of vitamin D supplementation in hospitalized patients with COVID-19.

Comparisons with previous studies and further research
In contrast, several meta-analysis studies found a positive correlation between low serum vitamin D levels and worse clinical outcomes [26,27]. However, they included case-control studies or reported unadjusted estimate effects, which might cause greater bias. For example, findings from the UK Biobank found a positive association between low vitamin D and COVID-19 infection, but the association was not significant after adjusting for confounders [6]. The present study included only cohort studies and multivariate-adjusted studies, which should reduce the potential bias. Evidence from a few Mendelian randomization studies also showed that vitamin D status did not causally affect susceptibility to and the severity of COVID-19 infection [28,29]. Overall, these results strongly suggested no association between serum vitamin D and COVID-19.
Our results also support the recommendation of joint guidance of insufficient evidence of vitamin D for the treatment of COVID-19 [30]. Although a small random controlled trial showed that oral vitamin D supplementation helped achieve SARS-CoV-2 RNA negativity in greater proportion and decreased inflammatory markers [21], we did not find a significant benefit of vitamin  Hazard ratio and incidence rate ratio were treated as odd ratio D supplements by combining available evidence from RCTs on clinical outcomes in patients with COVID-19, which was consistent with two quasi-experimental studies [31,32]. The overall quality of evidence for death or ICU admission was low, suggesting no recombination of vitamin D supplementation for hospitalized patients with COVID-19. However, all of the trials included a low or unclear percentage of patients with 25-hydroxyvitamin D deficiency. The benefits of vitamin D supplementation in patients with vitamin D deficiency should be further studied. Further studies with higher quality should also determine whether preventive or early vitamin D3 supplementation would be useful. (NCT04535791; NCT04482673; NCT04407286).

Strengths and limitations
The strength of the present study lies in the study design, which included only cohort studies and RCTs. This study is the first meta-analysis to use the GRADE system to evaluate the quality of the evidence. The present study also has several limitations. Our meta-analysis had high heterogeneity, which might be derived from the study design and variability in baseline characteristics.
For example, several studies have shown that the death rate for black patients with COVID-19 was higher than the rate for white patients with COVID-19 [33,34]. Second, many factors modulate vitamin D status, including genetic polymorphisms, age, health, sun exposure behavior, and season [11]. Although we included only studies that performed multivariable analysis, some potential risk factors were not fully adjusted, which affected our results. Therefore, further research should adjust for additional confounding factors to verify the results.

Conclusion
Based on current evidence, vitamin D deficiency or insufficiency was not significantly linked to susceptibility to COVID-19 infection or its associated death. Vitamin D supplements did not significantly improve clinical outcomes in patients with COVID-19, and the overall GRADE evidence quality was low, which suggested that vitamin D supplementation was not recommended for patients with COVID-19.
Additional file 1: Table S1. Table S2. Detailed description of the search strategy. Supplemental Table S4. Quality assessment of the included  Table S5. Quality assessment of the included RCT trials. Table S6. GRADE table observational studies.  Table S7. GRADE table for RCTs.