Vitamin D is a family of fat-soluble molecules that are important micronutrients for humans, with two forms vitamin D2 and vitamin D3 playing a central role in bone growth by increasing the uptake of calcium from the gut. Vitamin D is thus especially important in growing children and a deficiency can lead to rickets, which is characterised by weak and deformed bones.
Humans can obtain both vitamin D2 and D3 from their diet, with fish liver oils, eggs and milk being particularly rich. But we obtain most of our vitamin D, in the form of vitamin D3, by synthesising it directly when our skin is exposed to sunlight. This powers a photochemical reaction in which a derivative of cholesterol is converted into pre-vitamin D3, which is then transformed into vitamin D3 by the heat of the skin (Wagner 2008).
Both vitamin D2 and D3 do not dissolve in water and so are bound to carrier proteins, mainly vitamin D binding protein, before they can be transported in the blood to the liver. Here, they are converted to 25D (25-hydroxyvitamin D), which is the major circulating form of vitamin D. In the kidneys, 25D is converted to 1,25D (1,25-dihydroxyvitamin D) via the action of an enzyme known as 1α hydroxylase (Holick 2008; White 2008). This is the active form of vitamin D in the body and is thus the true hormonal form of vitamin D. It binds to the vitamin D receptor (VDR), which is found on the nuclear membrane of many cells (Walker 2009).
Vitamin D levels in the body are best measured using the concentration of 25D in blood serum. Generally, normal (sufficient) concentrations are greater than 30 to 32 ng/mL (75 to 80 nM). People with levels below 20 ng/mL (50 nM) are deficient, while those with between 20 ng/mL (50 nM) and 30 to 32 ng/mL (75 to 80 nM) are termed vitamin D insufficient (Walker 2009). It is estimated that around 1 billion people in the world may be vitamin D insufficient or deficient. Lack of sunlight (especially during winter months), vegetarian diets, a dark pigmented skin (as melanin acts as a natural sunscreen), increased pollution, and wearing long-sleeved garments or clothes completely covering the body are the major risk factors (Williams 2008). Modern recommendations to avoid the sun to prevent skin cancers may also be contributing to a deficiency in vitamin D (Misra 2008).
Over the past 20 years, much attention has been paid to recognizing vitamin D insufficiency and deficiency in children worldwide. Childhood vitamin D deficiency is very prevalent in many developing countries, even those with abundant sunlight such as Turkey (Ozgur 1996), Iran (Salimpour 1975), Saudi Arabia (Elidrissy 1984), India (Ghai 1991; Wayse 2004), China (Zhao 1991; Zhao 1992; Du 2001), Algeria (Garabedian 1991), and Nigeria (Akpede 1999; Akpede 2001). A study on the health status of children in low- and middle-income countries reported that 73.1% of underprivileged children were 25D deficient (< 8 ng/mL) and 23.1% were 25D insufficient (8 to 15 ng/mL) (Manaseki-Holland 2008). A high occurrence of insufficiency or deficiency of vitamin D in infants and children has also been reported in many other countries, including industrialized ones (Prentice 2008) like the US (Mansbach 2009), the UK (Lawson 1999), Greece (Nicolaidou 2006), Finland Lehtonen-Veromaa 1999), Canada (Ward 2007), and New Zealand (Grant 2009).
Fortified foods such as infant formulas, breakfast cereal, cheese, and milk are the major dietary source of vitamin D in children, but they may not be consumed in sufficient quantities. Furthermore, dairy products may not be fortified in all countries. Moreover, diet contributes less than 10-20% of an adult’s vitamin D stores, so this proportion may be even smaller in children (Sichert-Hellert 2006). Since vitamin D is a fat-soluble vitamin, a diet that is extremely low in fat can also impair its absorption. Breast milk can be a poor source of vitamin D, particularly if the mother has clinical or subclinical vitamin D deficiency.
As well as causing rickets, vitamin D deficiency has been associated with tuberculosis (TB) and influenza. It has also been linked with various other inflammatory and long-term diseases, including cardiovascular diseases (myocardial infarction), multiple sclerosis, asthma, rheumatoid arthritis, type 1 and type 2 diabetes (Wagner 2008a), and cancers such as breast, ovarian, colorectal, and prostate (Cavalier 2009).
In the past, clinicians have primarily focused on the role of vitamin D in preventing and treating rickets. A few early researchers, however, realized that children with rickets were more likely to have respiratory infections. Initially, clinicians presumed these infections were caused by poor lung function as a result of bone deformities and the overall compromised nutritional status associated with rickets. Several studies, including case-control and case series, link rickets with pneumonia and respiratory tract infections, while the connection between vitamin D insufficiency and TB has also been widely studied. Many epidemiological studies have recently been conducted in children to observe the link between inadequate vitamin D concentrations and respiratory infections, including TB (Karatekin 2009; McNally 2009; Muhe 1997; Najada 2004; Nnoaham 2008; Roth 2009; Salimpour 1975; Wayse 2004; Williams 2008). A recent small randomized controlled trial of vitamin D supplementation among children with pneumonia was associated with a reduction in repeat episodes of pneumonia (Manaseki-Holland 2010).
The precise molecular mechanisms by which vitamin D helps defend against infectious disease are now being elucidated. It has become clear that 1,25D plays a role not only in calcium homeostasis and bone metabolism, but also in the integrity of the innate immune system (Bhutta 2008; Wagner 2008a). Acting via the VDR, 1,25D alters the activity of many immune system cells, including macrophages, regulatory T cells and natural killer cells.
Because of the wide-ranging health benefits of vitamin D, the American Academy of Pediatrics now recommends 400 International Units (IU) daily of vitamin D supplements to begin in the first few days after birth and to be continued throughout childhood and adolescence (Wagner 2008). All breastfed infants, regardless of whether they are being supplemented with formula, should be given 400 IU of vitamin D, because it is unlikely they will consume 1 L of formula per day, which would provide 400 IU of vitamin D. Health Canada also recommends 400 IU/day for all exclusively breastfed, healthy infants; this should be continued until the infants' diet provides at least 400 IU from other sources (Health Canada). Vitamin D is generally well tolerated when given at appropriate doses. However, when given orally in excessive amounts, it can cause high blood calcium levels that in the long run could lead to kidney stones.
We are conducting this review to assist physicians in making a practical decision as to whether children under five years of age should receive vitamin D supplements and whether doing so can have any benefit in terms of preventing infectious diseases, especially diarrhoea and pneumonia.
How the intervention might work
Biological and behavioural basis of the intervention
Vitamin D influences the action of more than 200 human genes in a wide range of tissues and displays as many molecular mechanisms (Cannell 2008). In particular, it interacts with the human immune system in a wide variety of ways, helping to protect against infectious diseases.
For example, it has been known for 20 years that exposure to 1,25D stimulates anti-mycobacterial activity in human monocytes and macrophages. Recent research suggests that this is down to vitamin D helping to generate antimicrobial peptides (AMPs) like cathelicidin and some β defensins. These AMPs then lead to enhanced killing of intracellular Mycobacterium tuberculosis by direct membrane damage and also by acting as chemoattractants for monocytes (White 2008).
Recent research also indicates that a sufficient intake of vitamin D is essential for killer T cells to fend off serious infections, by controlling T cell antigen receptor (TCR) signalling and the activation of human T cells. Besides this, 1,25D also suppresses an overzealous adaptive immune response to pathogens that may be difficult for macrophages to handle efficiently (Walker 2009).
The relationship between vitamin D and infectious diseases is also supported by genetic studies of polymorphisms in the gene for the VDR. Researchers have found a significant link between single nucleotide polymorphisms of genes related to the innate immune function, including the VDR, and genetic susceptibility to respiratory syncytial virus (RSV) bronchiolitis (Janssen 2007).
While there is much research on the beneficial effects of vitamin D for TB infections, data are emerging from various sources about its role in fighting other bacterial and viral pathogens. Apart from the synthesis of AMPs (cathelicidin and defensins), the binding of activated vitamin D to the VDR can modulate viral lower respiratory tract disease. One of the defensins, retrocyclin-2, inhibits infection with the influenza virus by blocking its fusion with cell membranes (Leikina 2005). Respiratory tract infections peak during the winter season when there is less sunlight and so the lack of vitamin D during this season may enhance the infectivity of influenza viruses. A randomized controlled trial on 208 African-American post-menopausal women showed that vitamin D supplements had a beneficial effect on cold and influenza symptoms during the three-year study period (Aloia 2005).
Cystic fibrosis (CF) is an inherited disorder seen in children, characterized by pancreatic insufficiency and recurrent infections. CF patients have inadequate fat-soluble vitamins, including vitamin D. Yim et al. showed there was an enhanced antibacterial activity against airway pathogens such as Pseudomonas aeruginosa and Bordetella bronchiseptica in both normal and CF bronchial epithelial cells in participants supplemented with 1,25D. They also witnessed 1,25D-induced production of cathelicidin in this cell type (Yim 2007).
Why it is important to do this review
Around 11% of all deaths of children under five are associated with four micronutrient deficiencies (vitamin A, zinc, iron, and iodine) (Black 2008). Although rickets is the most well-known medical condition associated with vitamin D deficiency, it has also been linked to various infectious diseases, especially respiratory infections like pneumonia, TB and bronchiolitis. A study in the UK on a small cohort reported that the cost of preventing vitamin D deficiency in a high-risk population of Asian children was much less than the cost of treating the general health issues linked with chronic vitamin D deficiency (Zipitis 2006).
Vitamin D supplementation is a relatively simple intervention that might decrease the incidence of many infections. Because of its cost effectiveness and ease of administration, it can easily be applied on a large scale to children in communities or in health facilities. Vitamin D could therefore help to prevent the enormous burden of morbidity and mortality in children.
Criteria for considering studies for this review
Types of studies
Randomized controlled trials.
Types of participants
Children under five years of age.
Types of interventions
We shall compare synthetic oral vitamin D supplementation with a control (placebo or no intervention), including trials of various doses and frequencies. The co-interventions (for example, multiple vitamins or adjunct mineral and nutrient supplementation), must be identical in both groups ie vitamin D should be the only difference between the intervention and control groups. We shall exclude studies evaluating the effects of food fortification or the consumption of vitamin D-rich foods. If a trial includes more than one eligible intervention group (for example, differing in dose), we shall combine the groups for the main analysis, although they may be treated separately for subgroup analyses.
The comparisons to be included in this review are:
Comparisons that will be excluded are:
vitamin D plus 'other micronutrient' versus placebo or no treatment;
vitamin D plus 'other micronutrient' versus 'different other micronutrient';
we shall include only trials that have used supplementation for at least 2 weeks.
Types of outcome measures
1. Incidence (number of cases or episodes per total child-years) of:
1. Incidence (number of cases or episodes per total child-years) of:
2. Incidence (cases or episodes per total child-years) of febrile illness.
3. Duration (mean number of days of all episodes) of:
4. Severity of infection:
5. All-cause mortality rate.
6. Cause-specific mortality (as defined by authors) due to:
7. Hospital admission rate due to infections.
8. Change in mean serum vitamin D levels.
9. New cases per total children of:
We shall include only trials reporting on at least one of the review-defined outcomes related to infections. Trials reporting only non-infectious outcomes will be excluded. If a trial has information on infection-related outcomes in the full text, it will be included, but the authors will not be contacted for trials where this information is not present in the full text.
Time of outcome assessment
We shall group outcomes by time: 0-12 months, 13-60 months, and over 60 months. When trials report multiple time points, we shall extract the longest outcome interval in a given period.
Search methods for identification of studies
We shall attempt to identify all relevant studies regardless of language or publication status (published, unpublished, in press, ongoing).
We shall search the following databases: Cochrane Infectious Disease Group Specialized Register, Cochrane Central Register of Controlled Trials (CENTRAL), published in The Cochrane Library, MEDLINE, EMBASE, and LILACS, using the search terms detailed in Appendix 01. We shall also search the metaRegister of Controlled Trials (mRCT) using 'vitamin D’ and ‘child* OR infant*’ as search terms.
Researchers, organizations, and pharmaceutical companies
We shall contact researchers in the field to identify additional studies that may be eligible for inclusion.
We shall also check the reference lists of all studies identified by the above methods.
Data collection and analysis
Selection of studies
Two review authors will independently assess studies for inclusion in the review. The review authors will select potentially relevant studies by screening the titles and abstracts, if available, of studies. We shall retrieve and review full texts when their relevance cannot be ascertained from their titles or abstracts. Two review authors will independently assess the eligibility of all potentially relevant studies by reviewing their full texts and filling out eligibility forms designed in accordance with the specified inclusion criteria. We shall resolve differences of opinion about studies' suitability for inclusion by discussion. If a decision is not reached, we shall consult a third review author. We shall present excluded studies that appear to meet the inclusion criteria but on further investigation do not, in the 'Characteristics of excluded studies' table, along with the reasons for their exclusion. In the case of conference abstracts, if additional data are not forthcoming, we shall use the information provided in the abstract for review purposes. We shall also attempt to contact the trial authors regarding eligibility for cases where eligibility is not clear.
Data extraction and management
We shall use a data extraction sheet to extract the following information from each study:
Intervention and comparison:
For each intervention and comparison group of interest:
For each outcome of interest:
Two review authors will independently extract data from the studies; we shall resolve discrepancies through discussion. Incidence and duration are usually count data and mortality is usually dichotomous data. Therefore, for dichotomous outcomes, the number of participants experiencing the condition and the total number of participants in each treatment group will be extracted. For count outcomes we shall extract the number of events in the treatment and control group, and the total person time at risk in each group or the rate ratio, and a measure of variance eg standard error (SE), directly from the trial report.
For count data, where applicable, arithmetic means and standard deviations for each treatment group together with the numbers of participants in each group will be extracted. If the data have been reported using geometric means, then this information will be recorded and standard deviation extracted on the log scale. If medians have been used, then medians and ranges, if available, will be extracted.
Assessment of risk of bias in included studies
Two authors will independently assess methodological quality using the Cochrane Collaboration’s tool for assessing risk of bias (Higgins 2008). We shall assess eligible studies based on the following six components: method of sequence generation; allocation concealment; blinding of participants, providers, and outcomes assessors; incomplete outcome data, selective outcome reporting, and other biases (including detection bias, for example, differential effort to locate death records for the intervention and control groups). We shall present findings in a 'Risk of bias' table where, for each question-based entry, the judgement of the authors (‘Yes’ for low risk of bias; ‘No’ for high risk of bias; or ‘Unclear’ for unclear or unknown risk of bias) will be followed by a text box providing details on the available information that lead to each judgement. The results will also be reported in the form of 'Risk of bias' graph. For information not clear from the full text, we shall attempt to contact the authors for clarification. Any disagreements between the two assessors will be resolved by consulting a third author. Further details about the risk of bias tool is included in the Cochrane Handbook of Systematic Reviews of Interventions (Higgins 2008).
We shall be attentive to sources of bias originating from differences between individual (RCTs) and cluster randomized trials. For example, the relationship between allocation concealment and recruitment bias may be greater in cluster randomized trials.
Measures of treatment effect
For dichotomous data (for example, mortality) and count data (incidence and duration), all participants randomized will be included in the denominators whether total children, or total child-years, respectively. For dichotomous outcomes, risk ratios will be used to summarize results, while rate ratios will be used for count data. These ratios will be reported with 95% CI. If count data are summarized by arithmetic means and standard deviations data, then we shall combine them using the mean differences. Where count data are summarized using geometric means, we shall combine them on the log scale using the mean difference. Medians and ranges will be reported in a table.
In the case of cluster RCTs, we shall make adjustments by inflating the SE, or by using adjusted estimates provided by the author (see below).
Unit of analysis issues
In studies randomizing units other than individuals (ie clusters), the results should be presented with controls for clustering (for example, robust SEs or hierarchical linear models). Where the results do not control for clustering, we shall contact the authors to ask for an estimate of the intra-cluster correlation coefficient (ICC). If the authors are unable to provide an ICC, we shall estimate it from the other published values. We shall analyse clustered data using procedures outlined in the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2008). That is, we shall inflate the SEs of effect sizes by multiplying them by the square root of the design effect. If it is impossible to adjust for clustering, we shall analyse the outcomes from cluster-randomized trials using individuals as the unit of analysis, and we shall describe the risk of the overestimating treatment effect in the text. We shall analyse the effect of clustering through sensitivity analyses that explore the effect of larger and smaller ICCs.
Dealing with missing data
We shall describe missing data, including dropouts. Differential dropout rates can lead to biased estimates of the effect size, and bias may arise if the reasons for droppping out differ across groups. We shall report the reasons for dropout. If data are missing for some cases, or if the reasons for dropping out are not reported, we shall contact the authors. If there are missing data for dichotomous outcomes, a complete case analysis will be performed. The reason for this is that if participants are missing and the number randomized is used as the denominator, then an assumption is made about the missing participants i.e. that they did not have the condition. But if there are no missing data, then we shall do an 'intention to treat' analysis. When analyses are reported for completers as well as controlling for dropout (for example, imputed using regression methods), we shall extract the latter.
Assessment of heterogeneity
We shall assess included studies for clinical, methodological, and statistical heterogeneity. We shall assess clinical heterogeneity by comparing the distribution of important factors, such as the study participants, study setting, dose and duration of the intervention and co-interventions. We shall evaluate methodological heterogeneity on the basis of factors such as the method of sequence generation, allocation concealment, blinding of outcome assessment, and losses to follow up. We shall assess statistical heterogeneity among the trials by visual inspection of forest plots, by performing a Chi2 test (assessing the P value) and by calculating the I2 statistic (calculated as I2 = 100% x (Q-df ) / Q; where Q is Cochrane’s heterogeneity statistic and df is the degree of freedom). If the P value is less than 0.10 and I2 exceeds 50% and visual inspection of forest plots is indicative, we shall consider heterogeneity to be substantial and seek reasons for it.
Assessment of reporting biases
If we find sufficient studies, we shall draw funnel plots to help assess the possibility of bias.
We plan to perform the meta-analyses by using both fixed-effect and random-effects models. We shall perform all the meta-analyses using Review Manager Software Version 5. If we include a cluster randomized trial in meta-analysis, we shall adjust the SE of the effect size of that study, as described above.
We shall use risk ratios of the original data to combine dichotomous outcomes and Hedges g for continuous outcomes; we shall report both with 95% CI. We shall calculate overall effects using inverse variance methods.
We shall use fixed-effect models for the main analysis; although there may be some differences across trials (for example, dose and population), the biological mechanism should be similar across trials and we shall explore differences through analyses described elsewhere.
Subgroup analysis and investigation of heterogeneity
We plan to carry out the following three pre-specified subgroup analyses for incidence of infections.
Subgroup analysis according to the age of participants: < 1 year and 1-5 years.
Subgroup analysis to examine the possibility that there will be a variable response according to the dosages of vitamin D supplementation as follows: standard dosages versus high (greater than standard).
Subgroup analysis according to the duration of vitamin D supplementation as follows: low (six months or less) versus high (greater than six months).
We shall also explore the contribution of these variables to heterogeneity by meta-regression.
We shall perform sensitivity analyses according to the following factors: