Efficacy of Vitamin D Supplements in Prevention of Acute Respiratory Infection: A Meta-Analysis for Randomized Controlled Trials

Background: Previous systematic reviews and meta-analyses of randomized controlled trials (RCTs) have reported inconsistent results regarding the efficacy of vitamin D supplements in the prevention of acute respiratory infections (ARIs). Methods: We investigated these efficacy results by using a meta-analysis of RCTs. We searched PubMed, EMBASE, and the Cochrane Library in June 2021. Results: Out of 390 trials searched from the database, a total of 30 RCTs involving 30,263 participants were included in the final analysis. In the meta-analysis of all the trials, vitamin D supplementation showed no significant effect in the prevention of ARIs (relative risk (RR) 0.96, 95% confidence interval (CI) 0.91–1.01, I2 = 59.0%, n = 30). In the subgroup meta-analysis, vitamin D supplementation was effective in daily supplementation (RR 0.83, 95% CI, 0.73–0.95, I2 = 69.1%, n = 15) and short-term supplementation (RR 0.83, 95% CI, 0.71–0.97, I2 = 66.8%, n = 13). However, such beneficial effects disappeared in the subgroup meta-analysis of high-quality studies (RR 0.89, 95% CI, 0.78–1.02, I2 = 67.0%, n = 10 assessed by the Jadad scale; RR 0.87, 95% CI, 0.66–1.15, I2 = 51.0%, n = 4 assessed by the Cochrane’s risk of bias tool). Additionally, publication bias was observed. Conclusions: The current meta-analysis found that vitamin D supplementation has no clinical effect in the prevention of ARIs.


Introduction
Acute respiratory infection (ARI) is classified into an upper respiratory tract infection (URI) and a lower respiratory tract infection (LRI). URIs include common cold (nasopharyngitis), sinusitis, pharyngitis, laryngitis, and laryngotracheitis [1]. The common cold, as a frequent cause of URIs, is caused by viral infections such as rhinovirus, coronavirus, influenza virus A/B/C, respiratory syncytial virus, parainfluenza virus, and adenovirus [1]. URIs are a common disease, of which adults experience 2-4 episodes a year on average, and children 7-12 episodes [2]. LRIs are mostly caused by viruses such as the influenza virus and respiratory syncytial virus. Moreover, they are caused by bacterial infections such as S. aureus, S. pneumoniae, and H. influenza, tuberculosis infections, fungal infections, and parasite infections [3]. LRIs have been the fifth leading cause of death and responsibility for mortality in adults older than 70 years worldwide since 1990, accounting for up to 94.6 per 1000 global deaths [4].

Data Sources and Search
We searched the Cochrane Library, Embase, and PubMed in order to retrieve articles about the effect of vitamin D supplementation in the prevention of ARIs from inception to June 2021. Common keywords used for searching were as follows: "vitamin D," for an intervention variable, "respiratory tract infections," for a disease variable, and "randomized controlled trial" for a study design.

Data Selection and Quality
We selected RCTs that met all the following criteria: reported the efficacy of vitamin D supplementation in the prevention of ARIs; reported outcome measures with dichotomous variables. We excluded studies targeting participants in pregnancy and prenatal periods. Regarding studies using shared data from the identical population, we selected a more comprehensive study or a study with a longer follow-up period. Two authors (H.-E. Cho and H. Cho) independently evaluated the suitability of an individual study using the above-described selection criteria. Discrepancies between authors with the selection were solved with discussion and consultation with the third author (S.-K. Myung).

Assessment of Risk of Bias
The risk of bias was estimated based on both the Jadad score [47] and the Cochrane risk of bias tool [48] by two authors (H.-E. Cho and H. Cho). Studies were considered as having high quality if they had ≥5 items in the Jadad scale or ≥6 items in the Cochrane risk of bias tool because the mean score for the Jadad scale was 4.5 and the Cochrane risk of bias tool was 5.

Main and Subgroup Meta-Analysis
In the main analysis, we investigated the association between vitamin D supplementation and the incidence of ARIs as a risk. Subgroup analyses were conducted according to various factors as follows: duration of vitamin D supplementation (≤11 weeks and >11 weeks), dosage (daily, weekly, monthly, >2000 IU, and ≤2000 IU), type of disease (URIs and LRIs), number of the study participants (>1000 vs. ≤1000), region of the study (America, Europe, Asia, and Oceania), mean age (≤18 vs. >18), supply source for supplements (pharmaceutical company vs. non-pharmaceutical company), use of placebo, and quality of the study (Jadad score and Cochrane risk of bias).

Statistical Analysis
Values in cells of a 2 × 2 table based on an intention-to-treat analysis were used to calculate a relative risk (RR) with its 95% confidence interval (CI) in an individual study. Then, we calculated a pooled RR with its 95% CI in the random-effects meta-analysis.
To test the heterogeneity across studies, Higgins I 2 , which measures the percentage of total variation across studies [49], was used. I 2 calculated by a formula as follows where Q is the Cochrane's heterogeneity statistic and df means the degree of freedom. The negative predictive values of the I 2 were set at zero. An I 2 value ranges from 0% (no observed heterogeneity) to 100% (maximal heterogeneity), and those greater than 50% indicate substantial heterogeneity [50]. In this study, because individual trials were conducted in the different populations, we used a random-effects model meta-analysis.
The publication bias was evaluated by using the Begg's funnel plot and Egger's test. If Begg's funnel plot shows asymmetry or the p-value of the Egger's test is below 0.05, it indicates the existence of publication bias in the study. We used Stata MP version 17.0 software package (StataCorp., College Station, TX, USA) for all the statistical analyses. Figure 1 shows how we selected relevant articles, out of a total of 390 articles initially searched from the three databases. After excluding 141 duplicated articles, two authors independently reviewed 249 articles based on the title and abstract. Among them, 196 articles that did not meet the pre-determined selection criteria were excluded. For the remaining 53 articles, we reviewed the full text of the trials and excluded 23 articles because of the following reasons: four articles were irrelevant, five were replies or comments, and 14 had insufficient data. A total of 28 randomized double-blind placebo-controlled trials (RDBPCTs) and two open-label, randomized controlled trials (OLRCTs)  were included in the final analysis. Table 1 shows the general characteristics of the clinical trials included in the final analysis. Studies were published between 2009 and 2021, spanning 12 years. The total number of the study participants were 30,263 with 4259 in an intervention group and 4069 in a control group. The number of the study participants ranged from 49 to 8117. For studies reporting the information of age, the mean age of the participants was 36.6 years old (from 3 to 81). The main outcome measures were URIs (n = 23), LRIs (n = 6), and both URIs and LRIs (n = 1). The periods of supplementation or follow-up ranged from 1 week to 60 weeks.

General Characteristics of Trials
The dosage regimens for vitamin D supplements used in the trials were as follows: 300, 400, 500, 600, 1000, 1200, 2000, 4000, 10,000 IU daily, 14,000, 50,000 IU weekly, 60,000, 100,000, 120,000, 200,000 IU monthly, 100,000, or 300,000 IU quarterly. Out of 28 trials reporting their funding sources, eight trials were supplied vitamin D supplements from pharmaceutical companies. The remaining 20 trials were funded by mainly public or governmental organizations or independent scientific foundations.

Association between Vitamin D Supplementation and Prevention of ARIs
As shown in Figure 2, a random-effects meta-analyses of RCTs showed that vitamin D supplementation did not significantly lower the risk of ARIs (RR 0.96, 95% CI 0.91-1.01, I 2 = 59.0%, n = 30).   Table 1 shows the general characteristics of the clinical trials included in the final analysis. Studies were published between 2009 and 2021, spanning 12 years. The total number of the study participants were 30,263 with 4259 in an intervention group and 4069 in a control group. The number of the study participants ranged from 49 to 8117. For studies reporting the information of age, the mean age of the participants was 36.6 years old (from 3 to 81). The main outcome measures were URIs (n = 23), LRIs (n = 6), and both URIs and LRIs (n = 1). The periods of supplementation or follow-up ranged from 1 week to 60 weeks.    mainly public or governmental organizations or independent scientific foundations.

Association between Vitamin D Supplementation and Prevention of ARIs
As shown in Figure 2, a random-effects meta-analyses of RCTs showed that vitamin D supplementation did not significantly lower the risk of ARIs (RR 0.96, 95% CI 0.91-1.01, I 2 = 59.0%, n = 30).
As shown in Figure 5, publication bias was observed: the Begg's funnel plot was asymmetrical, and the Egger's p for bias was 0.048 (p < 0.05).

Discussion
In the current study, we found that the use of vitamin D supplements had no preventive effect on ARIs in the meta-analysis of 30 RCTs. Vitamin D supplementation was efficacious in the prevention of ARIs in the subgroup meta-analyses in daily supplementation and its short-term use. However, the subgroup meta-analyses of the high-quality studies in each category showed that the use of vitamin D supplements has no statistically signif-

Discussion
In the current study, we found that the use of vitamin D supplements had no preventive effect on ARIs in the meta-analysis of 30 RCTs. Vitamin D supplementation was efficacious in the prevention of ARIs in the subgroup meta-analyses in daily supplementation and its short-term use. However, the subgroup meta-analyses of the high-quality studies in each category showed that the use of vitamin D supplements has no statistically significant effect in the prevention of ARIs.
There are several biological mechanisms that could explain the preventive effect of vitamin D supplements on ARIs. It has been reported that vitamin D modulates both the adaptive immune and innate immune systems from in vitro studies and animal studies. First, vitamin D could work as a direct and indirect regulator of T cells [7]. Vitamin D regulates T cells directly by inhibiting T cell proliferation, Interleukin-2 (IL-2) and Interferon-γ (INF-γ) transcription, and IL-17 secretion by helper T 17 cells. Additionally, the vitamin D receptor (VDR) is expressed in both the innate and the adaptive immune cells [7]. The VDR mediates 1,25(OH)2D to suppress helper T 1 cell proliferation that produces inflammatory cytokines, thus decreasing the production of INF-γ and IL-2 [51,52]. Moreover, vitamin D induces the development of IL-10 and regulatory T cells [8]. Second, vitamin D fortifies the antibacterial responses of the innate immune response by the toll-like receptors (TLRs) and the 1,25(OH)2D/VDR signaling [7]. The TLRs, which are expressed on macrophages, polymorphonuclear cells, monocytes, and epithelial cells play a key role in the innate immune system [50]. Some of the antimicrobial peptides that demonstrate antiviral effects are associated with the TLRs, and their expression is affected by 1,25(OH)2D [7,50]. In addition, several TLRs are affected by the VDR stimulation [50]. Finally, the gene expression of the antibacterial agents, cathelicidin, and human β-defensin are induced by 1,25(OH)2D/VDR signaling [7]. Cathelicidin is an antimicrobial peptide induced by the TLR 1/2 activation, and human β-defensin acts as a chemoattractant for neutrophils and monocytes [50]. In the animal study, the lungs of the 25(OH)D3-fed mice had a significantly lower viral titer than the lungs of the control mice. After influenza virus infection, the proinflammatory cytokines, IL-5 and INF-γ, significantly decreased in 25(OH)D3-fed mice compared with the control mice. 25(OH)D3 was found to reduce viral replication and inflammatory cytokines, and then decreased the clinical manifestation of influenza virus infection in a mouse model [11]. In other words, vitamin D deficiency is associated with an increased risk of infections of bacterial and viral origin through decreased innate immunity [53].
In the meantime, previous RCTs and meta-analyses have reported inconsistent findings about the preventive effect of vitamin D supplements on ARIs [6,[43][44][45][46]. Among them, only one study reported consistent findings with ours [42], and the others reported a preventive effect of vitamin D on ARIs [6,[44][45][46]. Xiao et al.'s [43] systematic review in 2015 concluded that there was no efficacy of vitamin D supplementation for the prevention of childhood ARIs. Martineau et al. [44] and Pham et al. [45] reported that high levels of serum 25(OH)D are associated with the prevention of ARIs. Abioye et al. [6] reported that micronutrients including vitamin D, vitamin C, and zinc reduced the occurrence of ARIs and the duration of the symptoms. Jolliffe et al. [46] suggested that although the heterogeneity across the trial was significant, the vitamin D supplementation slightly reduced the risk of ARIs compared to the control group.
Compared to the previous meta-analyses, our study has several strengths. We conducted subgroup meta-analyses by important factors that affect individual results and found out that the preventive effect of vitamin D supplements on ARIs was associated with the quality of the studies. In the subgroup meta-analysis, a significant preventive effect of vitamin D supplementation on ARIS was observed in daily supplementation and in the use of supplements during the short-term period. However, such beneficial effects disappeared in the subgroup meta-analysis of high-quality studies. That is, we think that the inconsistent findings of the previous meta-analyses might be associated with the study quality. Moreover, we used both the Jadad scale and Cochrane risk of bias tool to assess the methodological quality of the trials. Because the Jadad scale, which is a simple tool for assessing study quality, has been criticized by its generic problems of scale, we also used the Cochrane risk of bias tool for accuracy.
There are some limitations in this study. First, it would be ideal to investigate the efficacy of vitamin D supplementation on ARIs considering the baseline concentration of the 25(OH)D. However, this was unavailable in most of the studies included in our analysis. Thus, we could not investigate the differences in the preventive effect on ARIs between people with vitamin D deficiency and normal vitamin D levels. Further clinical trials with the data of baseline 25(OH)D levels are warranted to confirm our findings. Second, publication bias was found in this study, which means that trials showing an increasing risk of or no effect on ARIs by vitamin D supplementation might not be published. This favors our conclusion that there is no preventive effect of vitamin D supplements on ARIs. Finally, several RCTs included in the current study were not designed specifically to investigate the efficacy of vitamin D supplements on ARIs as a primary endpoint. Findings in the secondary endpoint might be due to chance.

Conclusions
The current meta-analysis of RCTs shows that the use of vitamin D supplements has no efficacy in the prevention of ARIs. Data Availability Statement: The authors used published data from the individual studies and declare that the data supporting the findings of this study are available within the article.

Conflicts of Interest:
All authors declare no conflict of interest.