Micronutrient perspective on COVID-19: Umbrella review and reanalysis of meta-analyses

Abstract Introduction Micronutrients are clinically important in managing COVID-19, and numerous studies have been conducted, but inconsistent findings exist. Objective To explore the association between micronutrients and COVID-19. Methods PubMed, Web of Science, Embase, Cochrane Library and Scopus for study search on July 30, 2022 and October 15, 2022. Literature selection, data extraction and quality assessment were performed in a double-blinded, group discussion format. Meta-analysis with overlapping associations were reconsolidated using random effects models, and narrative evidence was performed in tabular presentations. Results 57 reviews and 57 latest original studies were included. 21 reviews and 53 original studies were of moderate to high quality. Vitamin D, vitamin B, zinc, selenium, and ferritin levels differed between patients and healthy people. Vitamin D and zinc deficiencies increased COVID-19 infection by 0.97-fold/0.39-fold and 1.53-fold. Vitamin D deficiency increased severity 0.86-fold, while low vitamin B and selenium levels reduced severity. Vitamin D and calcium deficiencies increased ICU admission by 1.09 and 4.09-fold. Vitamin D deficiency increased mechanical ventilation by 0.4-fold. Vitamin D, zinc, and calcium deficiencies increased COVID-19 mortality by 0.53-fold, 0.46-fold, and 5.99-fold, respectively. Conclusion The associations between vitamin D, zinc, and calcium deficiencies and adverse evolution of COVID-19 were positive, while the association between vitamin C and COVID-19 was insignificant. REGISTRATION: PROSPERO CRD42022353953.


Introduction
The coronavirus disease 2019 (COVID-19) is caused by severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2), which has triggered a global pandemic (Wu, Leung, et al. 2020).SARS-CoV-2 is highly contagious and spreads mainly through respiratory droplets and contact (Alpdagtas et al. 2020).And people are generally susceptible to it, with a median age of about 50 (Wu and McGoogan 2020;Wang et al. 2020).Since the first report in December 2019 in Wuhan, China (Holmes et al. 2021;Wu, Leung, et al. 2020), there have been over 620 million confirmed cases and 6.5 million deaths reported worldwide (as of November 2022) (World Health Organization 2022).The spike (S) protein of SARS-CoV-2 can bind to the angiotensin-converting enzyme 2 (ACE2) receptor and membrane fusion ensues (Jackson et al. 2022;Singh et al. 2021;Walls et al. 2020), affecting not only the respiratory system, leading to pneumonia, but also the digestive system, nervous system, and cardiovascular system (Pascarella et al. 2020), ultimately leading to respiratory failure, multiple organ failure and death (Chen et al. 2020;Wiersinga et al. 2020).The most common clinical symptoms of COVID 19 are fever, fatigue and dry cough, as well as less common symptoms such as headache, diarrhea, and vomiting (Guan et al. 2020;Huang et al. 2020).
Micronutrients are a general term for vitamins, minerals and trace elements that affect the body's immune system and metabolism (Grober, Schmidt, and Kisters 2020).A deficiency of micronutrients affects innate and adaptive immunity, leading to immune suppression and thus increasing the risk of various infections (Gombart, Pierre, and Maggini 2020;Pae and Wu 2017;Toubal et al. 2020).Studies have shown that the micronutrients most needed to maintain immune capacity include vitamins A, C, D, E, B2, B6 and B12, folic acid, β-carotene, iron, selenium and zinc (Albers et al. 2013;Bego et al. 2022;Maggini, Pierre, and Calder 2018).For example, vitamin D promotes the release of antimicrobial peptides and the secretion of defensins, and plays an immunomodulatory role in improving innate immunity (Greiller and Martineau 2015;Wang et al. 2010).On the other hand, vitamin D prevents the adaptive immune system from overreacting with inflammation, preventing further damage to cells or tissues (Schwalfenberg 2011).Vitamin C improves the immune system by promoting the differentiation and proliferation of T and B lymphocytes (Carr and Maggini 2017).Furthermore, zinc has garnered much attention for its dual-immunomodulatory and anti-viral properties (Yao et al. 2021;Read et al. 2019;Bonaventura et al. 2015).
To date, numerous studies have shown that the micronutrients are related to the prevention, infection and prognosis of COVID-19, among which vitamin D is the most discussed micronutrient (Pereira et al. 2022;Akbar et al. 2021;Al Kiyumi et al. 2021), but their findings are not consistent.Damayanthi and Prabani (2021) described that low vitamin D, vitamin B12, Mg, and Se status are associated with malnutrition, oxygen therapy, intensive care support and survival of COVID-19 patients.Ghasemian et al. (2021, Fakhrolmobasheri et al. 2021) described lower vitamin D and Se levels in patients with COVID-19.For primary prevention of COVID-19, some studies have shown that low vitamin D levels is associated with an increased risk of COVID-19 infection (Ghasemian et al. 2021;Szarpak, Pruc, et al. 2021).In contrast, supplement or adequate vitamin D levels may reduce infection (Crafa et al. 2021;Teshome et al. 2021).In improving symptoms and prognosis, lower vitamin D levels are associated with increased COVID-19 severity, mortality, mechanical ventilation (MV), hospitalization and intensive care unit (ICU) admission (Hariyanto et al. 2022;Petrelli et al. 2021), and low Ca, Zn levels are associated with higher hospitalization (Yang et al. 2021;Allard et al. 2020).Vitamin D supplement may ameliorate adverse outcomes in patients with COVID-19 (Nikniaz et al. 2021;Hosseini, El Abd, and Ducharme 2022).However, other studies have shown that deficiency or insufficiency of vitamin D levels is not significantly associated with major outcomes (Ghasemian et al. 2021;Chen et al. 2021;Hu et al. 2022), nor does vitamin D supplement reduce COVID-19 infection or improve these outcomes (Rawat, Roy, Maitra, Gulati, et al. 2021).Some studies also have been inconclusive on the benefits and harms of micronutrient supplements in treating COVID-19 (Stroehlein et al. 2021;Balboni et al. 2022).
As more systematic reviews and meta-analyses have revealed the relationship between micronutrients and COVID-19, there is an urgent need for further evidence integration.This study aimed to summarize the extensive evidence available on the role of micronutrients in COVID-19 in order to draw more convincing conclusions and provide a more convincing basis for developing clinical practice policies.

Study design
Umbrella review aims to synthesize all types of research evidence relevant to a research topic, including various forms of systematic reviews and meta-analyses, to make recommendations for practice and research based on the results of systematic evaluations of data collection (The Joanna Briggs Institute 2014;Nakagawa et al. 2019).In this review, we explored the evidence for the association between micronutrients and COVID-19.The protocol has been registered with PROSPERO (CRD42022353953).

Data source and search strategy
We completed a search through PubMed, Web of Science, Embase, Cochrane Library and Scopus on July 30, 2022 for a combination of medical subject headings and free-text terms for the conditions of interest.The search strategy was developed around the key terms: COVID-19, SARS-CoV-2, micronutrients, meta-analysis, systematic review, without language and time restrictions.We also conducted a manual search of the reference lists of eligible reviews to identify additional documents.Appendix A, supplementary materials, provides a detailed search strategy.

Eligibility criteria
Based on our research question we identified the key elements (PICOS) of the umbrella review in terms of prevention and treatment: Participant: adults > 19 years, excluding minors and animal studies.

Intervention/Exposure:
• Prevention: sufficiency and/or deficiency of any micronutrient singly or in combination with other micronutrients; • Treatment: treatment using any micronutrient singly or in combination with other micronutrients or therapeutical drugs.
Comparison: a single control for the above intervention or exposure.
Study types: systematic reviews and meta-analysis, excluding guidelines, narrative reviews, scoping reviews, comments, letters, summaries, conference papers, etc.

Study selection and data extraction
The search results were imported into the Rayyan for literature screening for further research (Ouzzani et al. 2016).Two independent researchers first performed a blind screening based on literature titles and abstracts.After cross-checking, literatures that met the criteria moved to the full-text selection stage and final determination of inclusion.Data extraction continued independently by both researchers, extracting general characteristics of the studies, search information, quality assessment and data analysis results.For unclear items, we described them as not applicable.The data extraction form used for this study was optimized regarding the Joanna Briggs Institute (Appendix B, supplementary materials) (The Joanna Briggs Institute 2014).Throughout the literature screening and data extraction process, if disagreements were encountered, the group discussed and negotiated.A third researcher stepped in to arbitrate if necessary to reach a consensus conclusion.

Quality assessment
We used A Measurement Tool to Assess systematic Reviews (AMSTAR 2) to assess the methodological quality of the included systematic reviews and to provide an overall rating (Shea et al. 2017).The AMSTAR 2 assessment scale contains 16 items relevant to the systematic review, seven of which are considered critical (Appendix C1.1, supplementary materials).The key items are: a clear protocol; adequate literature search; reasons for excluding individual studies; risk of bias assessment; appropriate statistical methods for meta-analysis; consideration of the risk of bias in interpreting results; assessment of publication bias.Three items are not applicable in a systematic review without meta-analysis (appropriate statistical methods in meta-analysis, influence of bias on results, and assessment of publication bias), which we described as not applicable.Item evaluation criteria were divided into "yes," "partly yes" and "no." Four categories are based on "overall confidence": high, medium, low, and very low.For systematic reviews that met the inclusion criteria, we assigned two researchers to perform an independent quality assessment followed by cross-checking, with a third researcher participating in arbitration to reach consensus in case of disputes.

Managing overlapping systematic reviews
As the number of published systematic reviews increases (Page et al. 2016), the assessment associations in reviews that address an identical or similar research question is overlapping, as evidenced by the inclusion of many of the same underlying original studies.Direct inclusion of review outcomes with overlapping associations may result in multiple inclusion of the original studies, leading to a large bias in the pooled findings and effect estimates (Pollock et al. 2020;Senn 2009).We used corrected covered area (CCA) to quantify the degree of overlap for some meta-analysis with overlapping associations (Pieper et al. 2014).CCA is an effective method to quantify the degree of overlap between two or more reviews and helps provide a theoretical basis for dealing with overlap, which is calculated based on the citation matrix generated by the overlapping systematic reviews (Pieper et al. 2014).The citation matrix is a two-dimensional graphical cross tabulation of systematic reviews (columns) and included original studies (rows).CCA is expressed as a percentage and classifies the degree of overlap into four grades: slight (CCA = 0-5%), moderate (CCA = 6-10%), high (CCA = 11-15%), or very high (CCA > 15%) (Pieper et al. 2014).
Effect estimates from all non-overlapping systematic reviews that met the inclusion criteria were directly included in the analysis, with overlapping reviews managed as follows: • When the degree of overlap was slight (CCA ≤5%), the bias caused by direct pooling results was acceptable, and the effect estimates of systematic reviews were directly included in the analysis.• When the degree of overlap was more than moderate (CCA ≥6%), the bias caused by directly pooling the results was large.To enhance the comprehensiveness of evidence sources for effect estimates, we reintegrated data from all relevant original studies in the existing meta-analysis.The quality assessment of the included original studies was at least moderate in quality.If the quality assessment of the original studies in the existing meta-analysis was inconsistent or had no assessment, our researchers performed a reassessment and were included.

Data synthesis
Standard-compliant systematic reviews and meta-analysis, as well as selected original studies, constituted the unit of analysis.The characteristics and results of the reviews were tabularly presented.We used a random-effects model for pooling results of meta-analysis, which is more conservative for pooling studies with high heterogeneity (Hunter 2000).
Results were presented as forest plots, and effects were estimated using odds ratio (OR) or weighted mean difference (WMD) with a 95% confidence interval (CI).We used the I 2 index to quantify the heterogeneity between studies, with I 2 <25% generally considered low heterogeneity, I 2 >75% considered high heterogeneity, and 25%-75% considered moderate heterogeneity (Higgins et al. 2003).Begg's funnel plot and Egger's test were used to detecting publication bias.Subgroup analyses were also performed according to micronutrient level.Sensitivity analysis were also performed to evaluate the robustness of the results.Statistical software Stata MP 17.0 and R-4.2.2 were used for statistical analysis and data validation.

Studies selection and study characteristics
Through database and citation searching we identified 3,688 records.After removing duplications, title and abstract screening, 76 records were eligible for full-text screening.
Finally, we identified 60 records, which included 57 studies.
Figure 1 summarizes the study selection process, as well as Appendix E, supplementary materials provides a list of records for full-text screening.
Of the 57 included systematic reviews, 12 reviews provided narrative descriptions, and the remaining 45 reviews conducted meta-analysis as the primary form of evidence synthesis.Table 1 demonstrates the brief general characteristics of these studies, with detailed information in Table S1.All studies evaluated the effect of micronutrient deficiency and/ or supplements on the prevention and/or prognosis of COVID-19.Of these, 51 studies examined single micronutrients, including vitamin D (n = 40), vitamin C (n = 4), zinc (n = 3), ferritin (n = 2), calcium (n = 1), and selenium (n = 1),  with the remaining 6 studies involving multiple micronutrients.Authors from 20 countries and regions participated in the above studies, with the most studies from Iran (n = 7), India (n = 6), China and Indonesia (n = 5 each).There were differences in the number of participants included, the type of original study, and the summary obtained.
A supplementary search of the latest studies yielded 8,679 records, and 57 studies were identified for subsequent analysis after a double-blinded two-stage screening.

Methodological quality
Of the 57 reviews included in the analysis, 21 were rated as high or moderate in quality, 22 were rated as low in quality, and 14 were rated as very low in quality (Appendix C2.1, supplementary materials).Of the 20 studies that were inconsistently assessed or not assessed for original study quality in the systematic reviews, we reassessed them and the quality was all above moderate.Among the quality assessment for 57 latest studies, 53 studies were rated as above moderate in quality.Of the four rated as low in quality, one was a randomized controlled trial and three were cross-sectional studies (Appendix C2.2, supplementary materials).

Overlapping and non-overlapping associations
About 43 meta-analysis reported overlapping associations, and Appendix F provides citation matrix to assess the degree of overlap.Overlapping associations included: differences in vitamin D levels between healthy people and patients (n = 6) and in different clinical progression (n = 4); associations between vitamin D and COVID-19 infection (participants were healthy people and patients, n = 10; participants were healthy people, n = 11); associations between vitamin D and severity (n = 16), ICU admission (n = 13), MV (n = 9), and mortality (n = 27); associations between vitamin C and MV (n = 2), mortality (n = 5); and association between zinc and mortality (n = 5).Appendix G, supplementary materials, presents reviews of outcomes with non-overlapping associations, covering five micronutrients associated with 11 outcomes.

Associations between micronutrients and COVID-19
Tables S2 and S3 provide the findings of the narrative syntheses and meta-analysis from the systematic reviews, respectively.We re-performed the quantitative analysis based on the high degree of overlapping associations of some outcomes.Supplementary figures provide pooled forest plots.

COVID-19 infection.
The infection rate of COVID-19 in people with vitamin D deficiency was 1.97-fold higher than that in people with normal vitamin D levels (95% CI: 1.68, 2.31; Figure S3) and 1.39-fold higher than that in people with vitamin D supplement (95% CI: 1.13, 1.72; Figure S4).Severity.Our pooled analysis showed that vitamin D deficiency was associated with an increase in severity (OR: 2.02; 95% CI: 1.38, 2.96; Figure S5).Subgroup analysis showed that the degree of vitamin D deficiency was directly proportional to severity.And it suggested that vitamin D supplement was associated with a reduction in severity of patients with COVID-19 (OR: 1.54; 95% CI: 1.09, 2.17; Figure S6).The severity of COVID-19 in people with vitamin D deficiency was 1.86-fold higher than in people without vitamin D deficiency (Figure 2).

ICU admission.
A higher rate of severe COVID-19 was observed in patients with low serum vitamin D levels (OR: 1.52; 95% CI: 1.06, 2.18; Figure S7).However, when vitamin D levels were not below 20, ICU admissions did not seem to increase (OR: 1.07; 95% CI: 0.71, 1.60; Figure S7).The analysis showed that vitamin D supplement was associated with a reduction in ICU admission rate (OR: 3.25; 95% CI: 1.91, 5.54, Figure S8).Pooled analysis showed that the rate of ICU admission was 2.09-fold higher in the vitamin D-deficient population than in the non-deficient population (Figure 2).

MV.
Pooling data revealed a trend for an increased need for MV in patients with COVID-19 with vitamin D deficiency (OR: 1.34; 95% CI: 1.01, 1.78; Figure S9).However, the subgroup analysis was not statistically significant (Figure S9).Four original studies evaluated the effect of vitamin D supplement on MV in patients with COVID-19 and showed that vitamin D supplement might reduce the need for MV (OR: 1.91; 95% CI: 1.07, 3.39; Figure S10).The pooled effect sizes suggested a higher demand for MV in patients with COVID-19 with low levels of vitamin D (OR: 1.40; 95% CI: 1.09, 1.81; Figure 2).

Vitamins A, B, and C
Vitamin A. The two vitamin A-related outcomes were reported in one original study each.There was no statistical difference in vitamin A level differences between severe and non-severe patients (WMD: 0.05; 95% CI: −0.10, 0.20; Figure S13) and in the association with severity (OR: 0.97; 95% CI: 0.57, 1.64; Figure S14).
Cuprum.Differences in cuprum levels varied in opposite ways in different comparison groups.One original study reported higher levels of cuprum in patients than in the healthy population (WMD: 0.67; 95% CI: 0.32, 1.02; Figure S34), while a pooled analysis of the two studies showed higher levels of cuprum in non-severe patients than in severe patients (WMD: −0.09; 95% CI: −0.19, 0.00; Figure S35).

Publication bias and sensitivity analysis
We tested publication bias for meta-analysis that combined more than 10 original studies, including Begg's funnel plot and Egger's quantitative test.Analysis revealed non-significant publication bias for vitamin D level differences between severe and non-severe patients (p = 0.711, Figure S37), the association between vitamin D and COVID-19 infection (study participants were healthy people) (p = 0.961, Figure S39), the association between vitamin D and ICU admission (p = 0.222, Figure S41) and for the association between vitamin D and MV (p = 0.851, Figure S42).However, there was publication bias for vitamin D level differences between patients and healthy people (p = 0.025, Figure S36), the association between vitamin D and COVID-19 infection (study participants were patients and healthy people) (p = 0.037, Figure S38), the association between vitamin D and severity (p = 0.001, Figure S40) and the association between vitamin D and mortality (p = 0.043, Figure S43).
Sensitivity analysis and subgroup analysis of the pooled estimate effects showed good robustness of the results (Figures S44-S54).

Discussion
We conducted an umbrella review combining all relevant research evidence from the micronutrient perspective for COVID-19.The results obtained from our integration represent the most comprehensive evidence currently available on the association between micronutrient and COVID-19.Evidence came from 57 systematic reviews and 53 latest original studies with good methodological quality.The results of the re meta-analyses showed that vitamin D levels were lower in patients with COVID-19 than in healthy people and that the effect was more pronounced in severe cases and deaths.Vitamin D deficiency has been linked to an increased risk of SARS-CoV-2 infection.More importantly, our meta-analysis suggested that vitamin D deficiency increased mortality, severity, MV, and ICU admission rate in patients with COVID-19.Vitamin D supplement could improve these COVID-19-related outcomes.Vitamin A and C did not appear to have a significant effect on clinical outcomes in patients with COVID-19, while vitamin B levels were lower in patients with COVID-19 and low levels of vitamin B appear to be a protective factor in reducing clinical severity.Zinc was a well-studied mineral, and pooled analysis showed that its levels were lower in patients, especially in severe patients.Concurrent zinc supplement decreased the rate of COVID-19 infection and mortality, but their contribution to the disease's evolution was unclear.Both selenium and ferritin showed lower levels in patients, but the former reduced clinical severity.Cuprum levels were lower in severe patients.Finally, a meta-analysis showed that calcium supplement reduced ICU admission and mortality in patients with COVID-19.Before the COVID-19 outbreak, researchers had studied the relationship between vitamin D and respiratory infections, which found that vitamin D deficiency is common in bronchiectasis and acute respiratory distress syndrome (ARDS) patients, and the need to correct the vitamin D deficiency is emphasized (Chalmers et al. 2013;Dancer et al. 2015).Previous studies have raised questions about whether there is a similar link between vitamin D and COVID-19, but the summaries were still controversial.
For serum vitamin D levels, Alvares et al. (2022) declared that it was impossible to establish a relationship between serum levels of vitamin D and clinical prognosis by COVID-19, while Ghasemian et al. (2021) observed that most patients with COVID-19 were suffering from vitamin D insufficiency/deficiency.Our meta-analysis of 27 original studies showed that patients with COVID-19 were indeed more prone to vitamin D deficiency than healthy individuals.Age, season and other factors can affect vitamin D levels in the body (Liu et al. 2021).In the Northern Hemisphere, COVID-19 is usually more prevalent in winter when Ultraviolet (UV) rays are weak.In addition, vitamin D levels are lower in the elderly, who are more vulnerable to the virus due to weakened immune function.This may explain why people with COVID-19 are more prone to vitamin D deficiency (Liu et al. 2021).
Based on the available evidence, we observed that people with low vitamin D levels were 1.39-fold more likely to develop COVID-19 than people with high vitamin D levels.This may be related to the inhibition of viral replication by active metabolites of vitamin D, leading to a decrease in SARS-CoV-2 viral titers and thus reducing the risk of positive detection (Mok et al. 2020).Pereira et al. (2022) suggested Vitamin D deficiency was not significantly observed to be associated with increased COVID-19 infection.For this, we included 267,526 patients with COVID-19 and 773,747 healthy individuals in our study and the effect size OR was 1.97 (95% CI: 1.68, 2.31), and the sample sizes were sufficiently powered.
In line with our study, several earlier meta-analyses also showed that low vitamin D levels were associated with increased mortality, ICU admission, MV, and disease severity (Hariyanto et al. 2022;Hosseini, El Abd, and Ducharme 2022;Setyarini, Manikam, and Mudjihartini 2021).Several mechanisms that could explain how vitamin D deficiency increases and improve these clinical outcomes.Vitamin D promotes the transcription of cathelicidins (IL-37) and defensins in some human cell lines thereby increasing the physical antiviral effect of the cells (Schwalfenberg 2011;Gombart 2009;Liu et al. 2006;Grant et al. 2020).In addition, Vitamin D reduces inflammation levels by inhibiting Th1 and Th17 cells activity and inducing INF-I production (Almeida Moreira Leal et al. 2020;Cantorna et al. 2015;Kralj and Jakovac 2021).Furthermore, vitamin D inhibits the renin angiotensin aldosterone system (RAAS) and ACE2, thereby achieving anti-inflammatory and protective effects on lung tissue (Machado et al. 2019;Mitchell 2020;Kurniawan et al. 2020).Severe patients with COVID-19 have higher levels of various inflammatory markers and are more prone to cytokine storms, which accelerates the poor prognostic course of the disease (Buonaguro et al. 2020).Vitamin D receptor (VDR) is possessed by most immune cells and its activation regulates macrophage responses and prevents the overproduction of cytokines and chemokines (van Harten et al. 2018).Thus, vitamin D is particularly important in reducing the risk of death in COVID-19 through immunomodulation and inhibition of cytokine storm (Li et al. 2020).
On the other hand, severe COVID-19 is associated with a hypercoagulable state, and vitamin D has the potential role of promoting the expression of antithrombin and thrombomodulin genes, thereby inhibiting this hypercoagulable state and having a significant protective effect during severe COVID-19 (Pereira et al. 2022;Kralj and Jakovac 2021).A cohort study showed that three single nucleotide polymorphisms (SNP) in VDR are associated with upper respiratory infections (URI), suggesting that genetic variation is also important (Jolliffe et al. 2018).A meta-analysis by Laplana, Royo, and Fibla (2018) examining the role of VDR gene polymorphisms in susceptibility to enveloped viruses highlighted the T allele as a risk factor for enveloped virus infection, including respiratory viruses.The same may be true of SARS-CoV-2 as an enveloped virus.This may be related to the flavobacterium okeanokoites (FokI) polymorphism producing shorter VDR proteins, resulting in higher NF-κB driven transcription rates, increased IL-12, and higher lymphocyte proliferation (Jolliffe et al. 2018).These mechanisms may explain the benefit of vitamin D supplementation on COVID-19 outcomes.
We observed that vitamin C levels did not benefit relevant clinical outcomes in patients with COVID-19.Vitamin C has a variety of physiological functions (Carr and Maggini 2017;Holford et al. 2020), including antioxidant, anti-inflammatory, antithrombotic, and immunomodulatory functions that appear to be associated with COVID-19related cytokine storms, sepsis, and ARDS (Wiersinga et al. 2020;Bastard et al. 2020).Research suggested that the anti-inflammatory effects of vitamin C may reduce inflammation in patients with COVID-19 and prevent a transition to a severe state or death (Chen et al. 2014).But in our review, we found few significant effects of high vitamin C levels in patients with COVID-19.A meta-analysis involving five RCTs and six observational trials also suggested that vitamin C supplement may not be associated with reduced mortality in critically ill patients with COVID-19 (Gavrielatou et al. 2022).In addition, multiple studies have shown that low vitamin C levels were not associated with COVID-19 outcomes (Ao et al. 2022;Kwak, Choo, and Chang 2022;Rawat, Roy, Maitra, Gulati, et al. 2021).There was no clear link between vitamin A and patients with COVID-19.It was special that vitamin B levels in severe patients with COVID-19 were significantly lower than those in non-severe patients with COVID-19, but low vitamin B reduced the occurrence of severe cases.In common, vitamins A and B also have anti-inflammatory and immunomodulatory properties (Dawson 2000;Mikkelsen and Apostolopoulos 2018).However, Akasov et al. (2022) showed that vitamin B supplement was associated with the normalization of clinically relevant immune markers, but did not improve clinical adverse progression The same opinion has been shown in other studies (Nimer et al. 2022).Further research should need to find out the reason.
Zinc deficiency is common in patients with COVID-19.Two meta-analyses by Hyewon, Soyeon, and An (2021) and Tabatabaeizadeh (2022) showed that zinc supplement was associated with lower mortality in patients with COVID-19.Still, the analysis by Szarpak, Pruc, et al. (2021) and Beran et al. (2022) were not significant.Our integrated results also suggested that zinc supplement was a protective factor against COVID-19 deaths, although the role of zinc in other COVID-19 clinical progression was not significant.It has been shown that zinc can inhibit coronavirus RNA polymerase activity thereby blocking viral replication and can shorten the course of acute viral respiratory infections (Te Velthuis et al. 2010;Hunter et al. 2021).In addition, zinc can reduce IL-1 and IL-6 and enhance macrophage function to achieve antiviral effects.Although the above mechanism was not confirmed in the SARS-CoV-2 study, it is also consistent with the results of COVID-19 infection.The pooled results showed low selenium levels in patients and differences at different stages of clinical progression, which was consistent with the results obtained from the systematic review by Fakhrolmobasheri et al. (2021).There is a large geographic influence on the selenium levels in humans, and the statistical results suggested that the different levels of soil selenium may be an important factor in the differences in COVID-19 mortality (Ulfberg and Stehlik 2021).Studies on coxsackievirus viruses have shown a higher rate of virion group mutations in patients with selenium deficiency (Guillin et al. 2019).Although this point has not been further investigated on SARS-CoV-2, it is a direction that should be looked at.It is worth noting that selenium has a similar amount of adaptations and metrics and should be supplemented with caution (Bodnar et al. 2016).
Current studies on ferritin were limited to concentration differences, which vary significantly in the different clinical stages of COVID-19.Current studies on ferritin were limited to level differences, which vary considerably in the different clinical stages of COVID-19.Ferritin was involved in the composition of hemoglobin affecting the oxygenation of the organism, and a meta-analysis by Taneri et al. (2020) showed that disease severity and prognosis in COVID-19 depended on hemoglobin levels.Data on serum cuprum status in patients with COVID-19 were limited, but levels were lower in severe symptomatic patients.Raha et al. (2020) hypothesized that cuprum could protect against COVID-19 by enhancing autoimmunity, but further pathological studies are needed to confirm this.Al-Saleh et al. (2022) found elevated cuprum levels in mild to moderate patients compared to asymptomatic patients, suggesting that cuprum may be involved in inflammation and oxidative stress in the development of COVID-19, which brings controversy to whether to cuprum supplement.A meta-analysis by Alemzadeh et al. (2021) showed a significant correlation between calcium and COVID-19 mortality and ICU admission.Cytokine disruption of calcium receptor expression may lead to calcium imbalance (Klein 2018); and the interaction between calcium and the immune system could explain the poor evolution of COVID-19 (Hendy and Canaff 2016).
Significant heterogeneity in some results was identified during the meta-analyses.The sources of heterogeneity may be attributed to the following aspects: 1) Clinical heterogeneity: Firstly, there were differences in participants enrollment across the original studies, such as age, gender, ethnicity, and comorbid underlying diseases.Second, there were differences in baseline information of participants between different original studies at the time of pooling, such as length of disease, severity of disease, etc.Finally, differences in treatment, such as the dose of drugs, the manufacturer, and the duration of follow-up, may also cause heterogeneity to some extent.2) Methodological heterogeneity: Although we assessed the methodological quality and selected original studies of moderate or high quality for inclusion in the quantitative analysis, the heterogeneity caused by factors such as design protocols and measurement methods still existed.3) Clinical heterogeneity and methodological heterogeneity together led to statistical heterogeneity.Moreover, some of the results had publication bias, which may be because negative studies are less likely to be published.

Strengths and limitations
This umbrella review has several advantages.A broad and sensitive search strategy was used to identify evidence for associations between micronutrient and COVID-19 comprehensively.Secondary integration was done on the results of overlapping associations after the assessment of CCA, while incorporating the latest studies to ensure a comprehensive and broad source of evidence for the integrated findings.Throughout the process we did methodological quality control as well as supporting analysis for the results (including subgroup analysis and sensitivity analysis), which provided some help for the reliability of the findings.
Several limitations arose Heterogeneity was found when conducting meta-analyses, which may be caused by participants, definitions of outcomes, differences of co-administration, study duration, status monitoring, and lack of standardization in clinical design and implementation, and most of which we had difficulty avoiding.And since most of our data came from observational studies, a significant proportion of these studies were retrospective observational designs, which may hinder inferences of causality, these results should be interpreted with caution.In addition, most of the studies were conducted in hospitalized patients, which to some extent affects the generalizability of our findings.Finally, funnel plots and Egger's tests suggested that publication bias may be a source of influence on the reliability of our results.

Implications for practice and future research
Implications for practice include the prevention of COVID-19 and reduction in the adverse progression of COVID-19.Current evidence suggests that vitamin D and zinc are effective in preventing COVID-19 infection; and vitamin B, vitamin D, zinc, selenium, and calcium could reduce the progression of COVID-19 to some extent.The role of micronutrients in preventing and treating COVID-19 can be achieved by improving the education of patients and clinicians.
As mentioned previously, the evidence sources were mostly observational studies with low levels of evidence.In contrast, the completed RCTs were limited by small sample sizes, low incident rates of micronutrients supplement, large baseline differences in participants, and many confounding factors.Therefore, more high-quality RCTs and intrinsic mechanism studies are expected to provide high-level evidence on the role of micronutrients in the management of patients with COVID-19.

Conclusion
The current integrated evidence suggested that vitamin B, vitamin D, zinc, selenium, ferritin, and cuprum levels were low in patients with COVID-19.Vitamin D and zinc deficiency could be risk factors for COVID-19 infection.Supplementing vitamin D, zinc, and calcium to maintain normal levels would reduce the adverse evolution of COVID-19, including severity, ICU admission, MV, and mortality.Vitamin C did not appear to be associated with clinical outcomes in COVID-19.However, further RCTs and mechanistic studies are needed to provide insight evidence into the use of micronutrients in managing of COVID-19.
COVID-19 has been a continuous outbreak for three years, and research on various aspects of COVID-19 has made rapid progress (Pasin and Pasin 2021).To overcome the shortcomings of insufficient coverage in published systematic reviews, we performed a supplementary search for the most recent original studies (Pollock et al. 2020).Appendix D, supplementary materials describes the search strategy used to identify newly published studies.Literature selection, data extraction, and quality assessment were still conducted independently by two researchers and a third researcher intervened to resolve disagreements.Studies with quality assessments above moderate quality were considered eligible for inclusion in the statistical analysis, and we used different quality assessment tools for different types of studies (Appendices C1.2-1.4):• Randomized controlled trials: Cochrane Risk of Bias tool 2 (ROB 2) (Sterne et al. 2019) • Case-control study and cohort study: Newcastle-Ottawa Scale (NOS) (Wells et al. 2000) • Cross-sectional study: Agency for Healthcare Researchand Quality scale (AHRQ)(Viswanathan et al. 2017)

Figure 1 .
Figure 1.Flow diagram of study selection.

Table 1 .
Brief general characteristics of systematic reviews and meta-analysis included in umbrella review.