Scolaris Content Display Scolaris Content Display

Cochrane Database of Systematic Reviews Protocol - Intervention

Vitamin D supplementation for women during pregnancy

Authors' declarations of interest

Version published: 08 December 2010 Version history

https://doi.org/10.1002/14651858.CD008873

This is not the most recent version

view the current version 26 July 2019
Collapse all Expand all

Abstract

This is a protocol for a Cochrane Review (Intervention). The objectives are as follows:

This review aims to assess the effects and safety of vitamin D supplementation in pregnancy and to examine whether supplementation with vitamin D alone or in combination with calcium and other vitamins and minerals given to women during pregnancy can safely improve pregnancy outcomes.

Background

Description of the condition

Vitamin D metabolism

Vitamin D is a fat‐soluble vitamin which comes primarily from exposure to sunlight, and is found naturally only in a few foods, such as fish‐liver oils, fatty fish, mushrooms, egg yolks, and liver (Holick 2007a; Holick 2008). There are two physiologically active forms of vitamin D collectively called calciferol: D2 and D3. Vitamin D2 (also called ergocalciferol) is synthesized by plants while vitamin D3 (also called cholecalciferol) is subcutaneously produced from 7‐dehydrocholecalciferol upon exposure to ultraviolet light B (UVB) radiation (DeLuca 2004). Vitamin D in supplements is found as either vitamin D2 or D3. The latter could be three times more effective than vitamin D2 in raising serum concentrations of vitamin D and maintaining those levels for a longer time; also, its metabolites have superior affinity for vitamin D‐binding proteins in plasma (Armas 2004; McCullough 2007). As vitamin D has a short half‐life, adequate vitamin D intake is necessary in order to ensure sustained circulating levels.

Both D2 and D3 forms share a similar metabolism. They are first hydroxylated in the liver to form 25 hydroxyl vitamin D (25(OH)D or calcidiol), and then in the kidney to 1,25 di hydroxyl vitamin D (1,25‐di(OH) D or calcitriol) in response to parathyroid hormone (PTH) levels. Calcitriol is considered an important pre‐hormone with active metabolites that are involved in metabolic processes including bone integrity and calcium homeostasis (Wagner 2008).

The major sites of vitamin D action include the skin, intestine, bone, parathyroid gland, immune system, and pancreas as well as the small intestine and colon in the human fetus (Theodoropoulos 2003). Additionally, vitamin D helps maintain normal levels of glucose in the blood, by binding to its receptors in the pancreatic beta cells, to regulate the release of insulin in response to the level of circulating glucose (Clifton‐Bligh 2008; Maghbooli 2008; Palomer 2008).

There is a unique relationship between vitamin D and calcium. The PTH is responsible for raising the calcium concentration in the blood through bone resorption, while calcitriol inhibits PTH and allows an increase of serum calcium concentration from sources other than the bone. In the presence of calcitriol, renal and intestinal calcium and phosphorus absorption is augmented leading to an improved calcium status.

Vitamin D status

Serum calcidiol or 25(OH)D can be used to assess vitamin D status, as it reflects the sum of the vitamin D produced cutaneously and that obtained from foods and supplements (Jones 2008). This metabolite is difficult to measure, with large variations between methods and among laboratories even when the same methods are used (Hollis 2004).

Currently, there is no consensus on the optimal levels of serum calcidiol for promoting health. In general, levels lower than 50 nmol/L (or lower than 20 ng/mL) are considered inadequate (Institute of Medicine 1997). Levels around 80 nmol/l (32 ng/ml) could be considered optimal, since they suppress PTH levels and lead to the greatest calcium absorption and the highest bone mass, reducing the rates of bone loss, falls, and fractures (Dawson‐Hughes 2005; Dawson‐Hughes 2008). Individuals with plenty of sun exposure have circulating calcidiol levels ranging from 54 to 90 ng/ml (135 to 225 nmol/L) (Haddock 1985). Whether the optimal levels proposed for non‐pregnant adults are adequate for pregnant women remains uncertain.

Vitamin D status is affected by factors that regulate its production in the skin (i.e. skin pigmentation, latitude, dressing codes, season, aging, sunscreen use, and air pollution) and by factors affecting its absorption or metabolism (Holick 2007b; Maghbooli 2007). Melanin acts as a filter for ultraviolet (UV) rays hence reducing the skin production of vitamin D. Hispanic and black populations in the United States may have higher melanin content, and thus have reduced vitamin D photosynthesis (Clemens 1982), explaining the variation in vitamin D concentration among ethnic groups living in the same geographical areas (Brooke 1980; Egan 2008; Matsuoka 1991; Nesby‐O'Dell 2002; Rockell 2005). Differences in latitude have also been shown to influence the concentration of vitamin D, and individuals from countries in high and low latitudes have lower vitamin D levels. The importance of UV rays is further shown by the seasonal variation in the concentration of vitamin D between summer and winter, with changes during the summer compared to the winter months (Holick 2007b; Levis 2005). Vitamin D metabolism is also affected in obese individuals, as vitamin D is deposited in body fat stores, making it less bioavailable (Arunabh 2003). It has been shown that low calcidiol levels are more prevalent among overweight and obese individuals compared to normal weight individuals (Vilarrasa 2007; Wortsman 2000). In the same context, sedentary activity is also associated with low vitamin D levels as it may be linked with diminished sunlight exposure (Ohta 2009).

Magnitude of vitamin D deficiency (VDD)

it has been estimated that one billion people worldwide have vitamin D deficiency or insufficiency and about 40% to 100% of elderly men and women still living in the community in the United States and Europe are deficient in vitamin D (Holick 2007a). VDD is a common health problem both in children and adults, affecting an estimated 30% to 50% of the global population (Bandeira 2006; Holick 2007a). Low concentrations of vitamin D have been found in all age groups in various countries including some in the Middle East (Fuleihan 2001; Sedrani 1984), the United States (Gordon 2004; Lips 2001; Sullivan 2005; Tangpricha 2002), India (Farrant 2009; Marwaha 2005), Japan (Sato 2005) and Australia (McGrath 2001b).

In pregnancy, vitamin D deficiency or insufficiency is also thought to be common. A study in black and white pregnant women residing in the northern United States found that approximately 29% of black pregnant women and 5% of white pregnant women had VDD (defined as serum 25(OH)D less than 37.5 nmol/L); whereas 54% of black women and 47% of white women had vitamin D insufficiency (defined as serum 25(OH)D levels 37.5 to 80 nmol/lL) (Bodnar 2007). Similar results have been found in pregnant African‐American adolescents (Davis 2010), in pregnant Asian women (Alfaham 1995), in Iranian pregnant women (Kazemi 2009), in veiled or dark‐skinned pregnant women (Grover 2001), in Indian pregnant women (Sachan 2005), in non‐Western pregnant women in the Netherlands (Van der Meer 2006), and in pregnant women from Pakistan, Turkey and Somalia (Madar 2009). Recent studies in white pregnant women also show high prevalence of VDD in the UK (Holmes 2009) and Ireland (O'Riordan 2008).

Seasonal variation increases the risk of VDD in pregnancy, with greater prevalence of VDD during the winter months compared to the summer months (Nicolaidou 2006; O'Riordan 2008). Differences in latitude have also been shown to influence the concentration of vitamin D in a majority of pregnant women (Sloka 2009).

Vitamin D status and health outcomes

Vitamin D status and hypertensive disorders during pregnancy

Maternal VDD in pregnancy has been associated with an increased risk of pre‐eclampsia (new‐onset gestational hypertension and proteinuria for the first time after 20 weeks of gestation), a condition associated with an increase in maternal and perinatal morbidity and mortality (Bodnar 2007; Holick 2008; Li 2000; MacKay 2001; Xiong 1999). Women with pre‐eclampsia have lower concentrations of calcidiol compared with women with normal blood pressure (Diaz 2002; Frenkel 1991; Halhali 1995; Halhali 2000; Tolaymat 1994). The low levels of urinary calcium (hypo calciuria) in women with pre‐eclampsia may be due to a reduction in the intestinal absorption of calcium impaired by low levels of vitamin D (August 1992; Halhali 1995). Additionally, pre‐eclampsia and VDD are directly and indirectly associated through biologic mechanisms including immune dysfunction, placental implantation, abnormal angiogenesis, excessive inflammation, and hypertension (Bodnar 2007; Cardus 2006; Evans 2004; Hewison 1992; Li 2002).

Vitamin D status and other maternal conditions

Maternal VDD in early pregnancy has been associated with elevated risk for gestational diabetes mellitus, although findings are still not consistent (Farrant 2008; Zhang 2008). Poor control of maternal diabetes in early pregnancy is inversely correlated with low bone mineral content in infants, as is low maternal vitamin D status (Namgunga 2003). VDD may lead to a high bone turnover, bone loss, osteomalacia and myopathy in the mother in addition to neonatal and infant VDD (Glerup 2000; Lips 2001).

An adequate vitamin D status may also protect against other adverse pregnancy outcomes. For example, maternal VDD has been linked to cesarean section in a single recent study (Merewood 2009) but the mechanisms involved are unclear.

Low prenatal and perinatal maternal vitamin D concentrations can affect the function of other tissues, leading to a greater risk of multiple sclerosis, cancer, insulin‐dependent diabetes mellitus, and schizophrenia later in life (McGrath 2001a).

Vitamin D status and preterm birth and low birthweight

The potential association between maternal vitamin D status and preterm birth (less than 37 weeks' gestation) has been reported (Dawodu 2010; Morley 2006). In comparison, not all the studies show significant associations between maternal calcidiol levels and any measure of the child's size at birth or during the first months of life (Bodnar 2010; Farrant 2009; Gale 2008; Morley 2006). There is not much information on maternal vitamin D status and low birthweight or preterm birth in children born from HIV‐infected pregnant women (Mehta 2009).   

Vitamin D status and postnatal growth

Some observational studies suggest that vitamin D levels during pregnancy influence fetal bone development and children's growth (Bodnar 2010; Brooke 1980; Mahon 2010; Morley 2006). While head circumference in children nine years of age has been significantly associated with maternal calcidiol levels (Gale 2008), there is still inconsistent information about the association of maternal vitamin D status and infants' bone mass (Akcakus 2006; Javaid 2006; Viljakainen 2010).

It is not clear if maternal VDD leads to neonatal rickets, since rickets is usually identified later in childhood. Early studies indicate a possible risk for neonatal rickets in the offspring of women with osteomalacia, abnormal softening of the bone by deficiency of phosphorus, calcium or vitamin D (Ford 1973). More recent studies have found that VDD (serum levels less than 25 nmol/L) was identified in 92% of rachitic Arab children and 97% of their mothers compared with 22% of nonrachitic children and 52% of their mothers (Dawodu 2005). A positive correlation was found between maternal and child vitamin D levels.

Vitamin D status and immune response

Vitamin D has direct effects on both adaptive and innate immune systems (Miller 2010; Walker 2009). In children vitamin D insufficiency is linked to autoimmune diseases such as type 1 diabetes mellitus, multiple sclerosis, allergies and atopic diseases (Bener 2009; Miller 2010; Pierrot‐Deseilligny 2010). Various studies have also shown that vitamin D deficiency is strongly associated with tuberculosis, pneumonia, and cystic fibrosis (Chocano‐Bedoya 2009; Hall 2010; Williams 2008) and both prenatal and perinatal vitamin D deprivation might influence early‐life respiratory morbidity as this vitamin is important for lung growth and development. (Devereux 2007; Litonjua 2009).

Vitamin D may have positive effects on the immune system by up‐regulating the production of the antimicrobial peptides by macrophages and endothelial cells (Wang 2004), which may inactivate viruses and suppress inflammation (Cantorna 2008), which may reduce the severity of infections.

Vitamin D toxicity

There has been little toxicity reported in adults taking doses as high as 10,000 IU/d (250 µg/d) of vitamin D (Hathcock 2007; Heaney 2008; Vieth 1999), although toxicity becomes generally present at 20,000 IU/d (500 µg/d). Vitamin D toxicity leads to hypercalcaemia (serum calcium higher than 10.6 mg/L), hypercalciuria (fasting urinary calcium/creatinine ratio of higher than 0.16 ng/mg) and an upper limit of serum 25(OH)D levels of 200 nmol/L (Aloia 2008). Hypercalciuria has been associated with renal and kidney stones (Heaney 2008).

Description of the intervention

Many health organizations recommend vitamin D supplementation during pregnancy and lactation. However, there are some variations in the recommended dose for supplementation ranging from 200 to 400 IU/d (Canadian Paediatric Society 2007; Hollis 2004; Institute of Medicine 1997; UK Department of Health 2009; WHO/FAO 2004) and some authors have suggested that requirements could be even greater (Hollis 2004). The American Academy of Pediatrics (Wagner 2008) suggests that healthcare professionals who provide obstetric care should consider monitoring maternal vitamin D status by measuring its concentrations in pregnant women.

The intake of vitamin D necessary to achieve a blood concentration considered optimal (80 nmol/l or 32 ng/ml) is greater than current recommendations. The debate on adequate values stems from the variability in baseline serum calcidiol levels and the laboratory techniques used. Studies have found that to increase serum calcidiol levels in 1.2 nmol/l, a supplemental dose of vitamin D of 400 IU/d is needed in those with low serum calcidiol levels, while those with better baseline levels have smaller increments with the same dose. It has been suggested that a supplemental dose of vitamin D of 1000 to 1600 IU might be necessary (Dawson‐Hughes 2005). However, the dose of vitamin D needed to have an effect during pregnancy or to prevent or treat VDD is not clear. Some have suggested that doses around 1000 IU/d may be needed in order for pregnant women to maintain a blood concentration of vitamin D of more than 50 nmol/l (Heaney 2003; Hollis 2004; Hollis 2007; Vieth 2001). Others have suggested providing vitamin D as a weekly dose of 5000 IU (125 ug/wk) (Utiger 1998) or a single dose of 200,000 IU or greater (Mallet 1986; Sahu 2009; Yu 2009).

Since vitamin D can also be synthesized by the skin upon exposure to sunlight, increasing casual sun exposure for reaching the optimal serum levels (Holick 2002) has been recommended. However, since excessive UV radiation is a carcinogen, it might be better to obtain additional vitamin D from foods or supplements.

How the intervention might work

Vitamin D supplementation improves maternal vitamin D status during pregnancy (Delvin 1986; Yu 2009), which in turn has direct influence on the fetal and neonatal supply of vitamin D (Brooke 1980). The potential effect of gestational vitamin D supplementation in preventing preterm birth (less than 37 weeks' gestation) and low birthweight (less than 2500 g) has been suggested (Maxwell 1981); although there is not much information on the additional benefit of vitamin D supplementation over other nutritional interventions during pregnancy such as iron and folic acid supplementation on the risk of low birthweight (Christian 2003). There is also a potential effect of maternal vitamin D supplementation on neonatal growth (Marya 1988). In addition to adequately exposing the infant to sunlight, vitamin D supplementation during pregnancy may be necessary to assure adequate concentration of vitamin D in breast milk during lactation (Butte 2002).

Some agencies have indicated that ensuring adequate vitamin D status with conventional prenatal vitamin D supplements should be encouraged as gestational vitamin D deficiency is common (Arden 2002; Namgunga 2003; Specker 1994; WHO/FAO 2004).

Why it is important to do this review

This review will update a previously published review (Mahomed 1999) and will incorporate new evidence on the effects and safety of vitamin D supplementation in pregnancy for the well being of the mother and newborn.

Objectives

This review aims to assess the effects and safety of vitamin D supplementation in pregnancy and to examine whether supplementation with vitamin D alone or in combination with calcium and other vitamins and minerals given to women during pregnancy can safely improve pregnancy outcomes.

Methods

Criteria for considering studies for this review

Types of studies

We will include both randomized and quasi‐randomized trials. If we identify crossover trials we will include only the first period. We will include cluster‐randomized trials if they are otherwise eligible. We will not include other levels of evidence (e.g. cohort or case‐control studies) in this meta‐analysis but we will consider such evidence in the discussion where relevant.

Types of participants

Pregnant women of any gestational or chronological age, parity and number of fetuses.

Types of interventions

Vitamin D supplementation during pregnancy irrespective of dose, duration or time of commencement of supplementation. We will include trials testing vitamin D in combination with other micronutrients in this review as long as the intervention and the control group were treated similarly except for additional vitamin D to the intervention group. Specifically, we will be assessing the following comparisons.

  1. Vitamin D alone versus no treatment/placebo (no vitamins or minerals)

  2. Vitamin D + calcium versus no treatment/placebo (no vitamin or minerals)

  3. Vitamin D + calcium versus calcium (but no vitamin D)

  4. Vitamin D + calcium + other vitamins and minerals versus calcium + other vitamins and minerals (but no vitamin D)

Types of outcome measures

Maternal antenatal clinical and laboratory outcomes and infant clinical and laboratory outcomes as described below.

Primary outcomes
Maternal

  1. Pre‐eclampsia (as defined by trialists)

  2. Gestational diabetes (as defined by trialists)

Infant

  1. Preterm birth (less than 37 weeks' gestation)

  2. Low birth weight (less than 2500 g)

Secondary outcomes
Maternal

  1. Vitamin D status (25(OH)D nmol/L)

  2. Impaired glucose tolerance (as defined by trialists)

  3. Caesarean section

  4. Gestational hypertension (as defined by trialists)

  5. Side effects (e.g. hypercalcaemia, kidney stones)

  6. Maternal death

Infant

  1. Length at birth (cm)

  2. Head circumference at birth (cm)

  3. Weight at birth (g)

  4. Admission to intensive care unit during the neonatal period

  5. Stillbirth (as defined by trialists)

  6. Neonatal death (as defined by trialists)

  7. APGAR score less than seven at five minutes

  8. Neonatal infection (e.g. respiratory infections)

  9. Very preterm birth (less than 34 weeks' gestation)

Search methods for identification of studies

Electronic searches

We will contact the Trials Search Co‐ordinator to search the Cochrane Pregnancy and Childbirth Group’s Trials Register. 

The Cochrane Pregnancy and Childbirth Group’s Trials Register is maintained by the Trials Search Co‐ordinator and contains trials identified from:

  1. quarterly searches of the Cochrane Central Register of Controlled Trials (CENTRAL);

  2. weekly searches of MEDLINE;

  3. handsearches of 30 journals and the proceedings of major conferences;

  4. weekly current awareness alerts for a further 44 journals plus monthly BioMed Central email alerts.

Details of the search strategies for CENTRAL and MEDLINE, the list of handsearched journals and conference proceedings, and the list of journals reviewed via the current awareness service can be found in the ‘Specialized Register’ section within the editorial information about the Cochrane Pregnancy and Childbirth Group.

Trials identified through the searching activities described above are each assigned to a review topic (or topics). The Trials Search Co‐ordinator searches the register for each review using the topic list rather than keywords. 

We will not apply any language restrictions.

Searching other resources

Additionally, we will contact different institutions including World Health Organization (WHO), UNICEF, the Micronutrient Initiative (MI) and the International Micronutrient Malnutrition Prevention and Control Program (IMMPaCt) from the US Centers for Disease Control and Prevention (CDC) for the identification of ongoing and unpublished studies.

We will search the international clinical trials registry platform (ICTRP) for any ongoing or planned trials.

Data collection and analysis

We will assess trials for methodological quality using the criteria in the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2009) for adequate, unclear and inadequate allocation concealment.

Two review authors will independently assess the eligibility of identified studies. We will include cluster‐randomized studies and adjust their samples sizes (Higgins 2009) if sufficient information is available to allow for this. If possible, we will estimate the intracluster correlation coefficients for each outcome from original data provided by the authors.

We will describe the proportion of variability in the data due to between‐study heterogeneity using the I² statistic test available in The Cochrane Collaboration's Review Manager software (RevMan 2008). If there is a high heterogeneity among trials, we will pool trial results using a random‐effects model and be cautious in our interpretation of the pooled results.

Selection of studies

Two review authors will independently assess for inclusion the abstracts of all the studies identified as a result of the search strategy. We will resolve any disagreement through discussion or, if required, we will consult a third author.

If studies are published only as abstracts, or study reports contained little information on methods, we will attempt to contact the authors to obtain further details of study design and results; if there is insufficient information for us to be able to assess risk of bias, we will designate studies as awaiting assessment until further information is published, or made available to us.

Data extraction and management

We will design a form to extract data. For eligible studies, two review authors will extract the data using the agreed form. We will resolve discrepancies through discussion or, if required, we will consult a third author. We will enter data into Review Manager software (RevMan 2008) and check for accuracy.

When information regarding any of the above is unclear, we will attempt to contact authors of the original reports to provide further details.

We will analyze dichotomous data in terms of risk ratio and we will analyse continuous data in terms of mean difference, unless the trials used in a particular analysis reported outcomes on different scales that could not be converted to a common scale, in which case, we will use the standard mean difference.

Assessment of risk of bias in included studies

Two authors will independently assess risk of bias for each study using the criteria outlined in the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2009). We will resolve any disagreement by discussion or by involving a third assessor.

(1) Sequence generation (checking for possible selection bias)

We will describe for each included study the method used to generate the allocation sequence. We will assess the method as:

  • adequate (any truly random process, e.g. random number table; computer random number generator);

  • inadequate (any non random process, e.g. odd or even date of birth; hospital or clinic record number); or

  • unclear.   

 (2) Allocation concealment (checking for possible selection bias)

We will describe for each included study the method used to conceal the allocation sequence and will assess whether intervention allocation could have been foreseen in advance of, or during recruitment, or changed after assignment.

We will assess the methods as:

  • adequate (e.g. telephone or central randomization; consecutively numbered sealed opaque envelopes);

  • inadequate (open random allocation; unsealed or non‐opaque envelopes, alternation; date of birth);

  • unclear.   

(3) Blinding (checking for possible performance bias)

We will describe for each included study the methods used, if any, to blind study participants and personnel from knowledge of which intervention a participant received. We will assess blinding separately for different outcomes or classes of outcomes and we will note where there is partial blinding.

We will assess the methods as:

  • adequate, inadequate or unclear for women;

  • adequate, inadequate or unclear for clinical staff;

  • adequate, inadequate or unclear for outcome assessors.

We will classify blinding "inadequate" if the blinding status of a trial is unclear or the trial is open.

(4) Incomplete outcome data (checking for possible attrition bias through withdrawals, dropouts, protocol deviations)

We will assess losses to follow‐up and post‐randomization exclusions systematically for each trial.

We will describe for each included study, and for each outcome or class of outcomes, the completeness of data including attrition and exclusions from the analysis. We will note whether attrition and exclusions are reported, the numbers included in the analysis at each stage (compared with the total randomized participants), reasons for attrition or exclusion where reported, and whether missing data are balanced across groups or are related to outcomes. We will assess methods as:

  • adequate;

  • inadequate:

  • unclear.

We will consider follow‐up to be adequate if more than 80% of participants initially randomized in a trial were included in the analysis, unclear if the percentage of initially randomized participants included in the analysis is unclear, and inadequate if less than 80% of those initially randomized are included in the analysis.

(5) Selective reporting bias

We will describe for each included study how we investigate the possibility of selective outcome reporting bias and what we find.

We will assess the methods as:

  • adequate (where it is clear that all of the study’s pre‐specified outcomes and all expected outcomes of interest to the review had been reported);

  • inadequate (where not all the study’s pre‐specified outcomes had been reported; one or more reported primary outcomes were not pre‐specified; outcomes of interest were reported incompletely and so could not be used; study failed to include results of a key outcome that would have been expected to have been reported);

  • unclear.

(6) Other sources of bias

We will note for each included study any important concerns we have about other possible sources of bias.

We will assess whether each study was free of other problems that could put it at risk of bias:

  • yes;

  • no;

  • unclear.

(7) Overall risk of bias

We will make explicit judgements about whether studies are at high risk of bias, according to the criteria given in the Handbook (Higgins 2009) and will explore the impact of the level of bias through undertaking sensitivity analyses ‐ seeSensitivity analysis. 

Measures of treatment effect

Dichotomous data

For dichotomous data, we will present results as summary risk ratio with 95% confidence intervals. 

Continuous data

For continuous data, we will use the mean difference if outcomes are measured in the same way between trials. We will use the standardized mean difference to combine trials that measure the same outcome, but use different methods.

Unit of analysis issues

Cluster‐randomized trials

We will include cluster‐randomized trials in the analyses along with individually randomized trials. We will adjust the standard errors of the results from cluster‐randomized studies using the methods described in the Handbook (Higgins 2009) if sufficient information is available to allow for this. We will use an estimate of the intracluster correlation co‐efficient (ICC) derived from the trial (if possible), or from another source. If ICCs from other sources are used, we will report this and will conduct sensitivity analyses to investigate the effect of variation in the ICC.

Where we identify both cluster‐randomized trials and individually‐randomized trials we will consider that it is reasonable to combine the results from both if there is little heterogeneity between the study designs and the interaction between the effect of intervention and the choice of randomization unit is considered to be unlikely.

Crossover trials

If we identify any crossover trials on this topic, and such trials are deemed eligible for inclusion, we will include them in the analyses with parallel group trials, using methods described by Elbourne 2002.

Dealing with missing data

For included studies, we will note levels of attrition. We will explore the impact of including studies with high levels of missing data in the overall assessment of treatment effect by using sensitivity analysis.

For all outcomes, we will carry out analyses, as far as possible, on an intention‐to‐treat basis, i.e. we will attempt to include all participants randomized to each group in the analyses, and analyze all participants in the group to which they were allocated, regardless of whether or not they received the allocated intervention. The denominator for each outcome in each trial will be the number randomized minus any participants whose outcomes are known to be missing.

Assessment of heterogeneity

We will assess statistical heterogeneity in each meta‐analysis using the T², I² and Chi² statistics. We will regard heterogeneity as substantial if I² is greater than 30% and either T² is greater than zero, or there is a low P value (less than 0.10) in the Chi² test for heterogeneity. 

Assessment of reporting biases

If there are 10 or more studies in the meta‐analysis we will investigate reporting biases (such as publication bias) using funnel plots. We will assess funnel plot asymmetry visually, and use formal tests for funnel plot asymmetry. For continuous outcomes we will use the test proposed by Egger 1997, and for dichotomous outcomes we will use the test proposed by Harbord 2006. If we detect asymmetry in any of these tests or by a visual assessment, we will perform exploratory analyses to investigate it.

Data synthesis

We will carry out statistical analysis using the Review Manager software (RevMan 2008). We will use fixed‐effect meta‐analysis for combining data where it is reasonable to assume that studies are estimating the same underlying treatment effect: i.e. where trials are examining the same intervention, and the trials’ populations and methods are judged sufficiently similar. If there is clinical heterogeneity sufficient to expect that the underlying treatment effects differ between trials, or if substantial statistical heterogeneity is detected, we will use random‐effects meta‐analysis to produce an overall summary if an average treatment effect across trials is considered clinically meaningful. We will treat the random‐effects summary as the average range of possible treatment effects and we will discuss the clinical implications of treatment effects differing between trials. If the average treatment effect is not clinically meaningful we will not combine trials.

If we use random‐effects analyses, we will present the results as the average treatment effect with its 95% confidence interval, and the estimates of  T² and I².

Subgroup analysis and investigation of heterogeneity

If we identify substantial heterogeneity, we will investigate it using subgroup analyses and sensitivity analyses. We will consider whether an overall summary is meaningful, and if it is, use random‐effects analysis to produce it. We plan to carry out the following subgroup analyses, when information is available:

  1. by total dose of supplementary vitamin D during pregnancy: 56,000 IU vitamin D or less versus more than 56,000 and less than 200,0000 IU versus more than 200,000 IU of vitamin D (the lowest cutoff is based on the highest daily supplemental dose during pregnancy, 400 IU/d times 140 days in 20 weeks of gestation; the highest cutoff is based on the usual single dose during gestation);

  2. by start of supplementation: less than 20 weeks versus more than 20 weeks of pregnancy;

  3. by pre‐gestational body mass index: low, normal, overweight, unknown/mixed;

  4. by scheme/regimen: single versus daily versus weekly;

  5. by skin pigmentation based on Fitzpatrick skin tone chart (Fitzpatrick 1988): three or less versus four or more;

  6. by latitude: between Tropics of Cancer and Capricorn versus north of the Tropic of Cancer or South of the Tropic of Capricorn;

  7. by season at the start of pregnancy: summer versus winter versus unknown.

We will use only the primary outcomes in subgroup analysis.

For fixed‐effect inverse variance meta‐analyses we will assess differences between subgroups by interaction tests. For random‐effects and fixed‐effect meta‐analyses using methods other than inverse variance, we will assess differences between subgroups by inspection of the subgroups’ confidence intervals; non‐overlapping confidence intervals indicate a statistically significant difference in treatment effect between the subgroups.

Sensitivity analysis

We will conduct a sensitivity analysis based on the quality of the studies. We will consider a study to be of high quality if it was graded as adequate in both the randomization and allocation concealment and in either blinding or loss to follow‐up. We will conduct an available case analysis and reinstate previously excluded cases when possible.