Cochrane Review: Vitamin D supplementation for improving bone mineral density in children

Authors


Abstract

Background

Results of randomised controlled trials (RCTs) of vitamin D supplementation to improve bone density in children are inconsistent.

Objectives

To determine the effectiveness of vitamin D supplementation for improving bone mineral density in children, whether any effect varies by sex, age or pubertal stage, the type or dose of vitamin D given or baseline vitamin D status, and if effects persist after cessation of supplementation.

Search methods

We searched the Cochrane Central Register of Controlled Trials (CENTRAL Issue 3, 2009), MEDLINE (1966 to present), EMBASE (1980 to present), CINAHL (1982 to present), AMED (1985 to present) and ISI Web of Science (1945 to present) on 9 August 2009, and we handsearched key journal conference abstracts.

Selection criteria

Placebo-controlled RCTs of vitamin D supplementation for at least three months in healthy children and adolescents (aged from one month to < 20 years) with bone density outcomes.

Data collection and analysis

Two authors screened references for inclusion, assessed risk of bias, and extracted data. We conducted meta-analyses and calculated standardised mean differences (SMD) of the percent change from baseline in outcomes in treatment and control groups. We performed subgroup analyses by sex, pubertal stage, dose of vitamin D and baseline serum vitamin D and considered these as well as compliance and allocation concealment as possible sources of heterogeneity.

Main results

We included six RCTs (343 participants receiving placebo and 541 receiving vitamin D) for meta-analyses. Vitamin D supplementation had no statistically significant effects on total body bone mineral content (BMC), hip bone mineral density (BMD) or forearm BMD. There was a trend to a small effect on lumbar spine BMD (SMD 0.15, 95% CI -0.01 to 0.31, P = 0.07). There were no differences in effects between high and low serum vitamin D studies at any site though there was a trend towards a larger effect with low vitamin D for total body BMC (P = 0.09 for difference). In low serum vitamin D studies, significant effects on total body BMC and lumbar spine BMD were approximately equivalent to a 2.6% and 1.7 % percentage point greater change from baseline in the supplemented group.

Authors' conclusions

These results do not support vitamin D supplementation to improve bone density in healthy children with normal vitamin D levels, but suggest that supplementation of deficient children may be clinically useful. Further RCTs in deficient children are needed to confirm this.

Plain Language Summary

Vitamin D for improving bone density in children

This summary of a Cochrane Review, presents what we know from research about the effect of vitamin D supplements on bone density in children.

The review shows that in healthy children generally, vitamin D supplementation does not improve bone density at the hip, lumbar spine, forearm or in the body as a whole.

Some of the evidence suggests that vitamin D supplements may improve bone density in children who have low levels of vitamin D but this is uncertain.

We do not have precise information about side effects and complications but the available information suggests that vitamin D supplements are well tolerated.

What is osteoporosis and what is vitamin D?

Osteoporosis is a condition where bones are weak, brittle and break easily. The risk of osteoporosis and fractures (breaks) in later life depends on how much bone is built when a child and how much bone is lost when an adult. One way to prevent osteoporosis and fractures in later life is to build stronger bones when young. Vitamin D plays an important role in improving the body's absorption of calcium from food, reducing losses of calcium from the body and getting calcium deposited into to bone to improve the quantity of bone developed. Therefore it is thought that if vitamin D levels in the body are low in childhood, less bone will be developed and that improving vitamin D levels by supplements would result in more bone being developed. Bone density is a major measure of bone strength and the amount of bone mineral present at different sites and so is used to measure the effects of interventions, like vitamin D supplementation, to improve bone health.

Summary of findings for the main comparison [Explanation]

original image

Background

Osteoporosis is a systemic skeletal disorder characterised by low bone mass and microarchitectural deterioration of bone tissue, with a consequent increase in bone fragility and susceptibility to fracture (NIH 1993). In other words, there is both less bone and poor bone quality. Osteoporosis is a major and growing public health problem, particularly in women (Jones 1994), with an estimated 9.0 million osteoporotic fractures worldwide occurring in the year 2000 (Johnell 2006). It is a costly global problem. For example in the United States, more than two million incident fractures are predicted at a cost of $17 billion, and fracture costs are expected to rise by 50% by 2025 (Burge 2007). High costs are also reported in other countries (Access 2001; Finnern 2003).

Low bone mineral density (BMD) is a major risk factor for osteoporotic fracture (Marshall 1996). It is well accepted that childhood factors are likely to have an impact on future risk of osteoporosis (NIH 2000). At least 90% of peak bone mass is obtained by the age of 18 years (Bailey 1999). BMD in later life is a function of peak bone mass and the rate of subsequent bone loss (Hansen 1991). Peak bone mass and rate of bone loss are equally important as risk factors for fracture in later life (Riis 1996). Peak bone mass is influenced by genetic factors, but also modifiable lifestyle factors such as adequate nutrition, body weight and physical activity (Javaid 2002). Maximizing peak bone mass is therefore potentially a way to reduce the impact of age-related bone loss. It is estimated that a 10% increase in peak bone mass reduces the risk of an osteoporotic fracture in adult life by 50% (Cummings 1993). In addition, there is evidence that low BMD is a risk factor for fracture in childhood (Goulding 1998; Goulding 2001; Ma 2003), suggesting that optimising age-appropriate bone mass may have a more immediate effect. Fractures in later life would be the ideal outcome measure in intervention studies for osteoporosis prevention, however for intervention studies in children this would require following large numbers of subjects for decades and these studies have not been performed. Therefore, BMD is commonly used as a surrogate outcome for intervention studies in children (Gilsanz 1998) and is the outcome used in this review.

Strategies to maximise peak bone mass in children have been identified as a priority area for research (NIH 2000). While there is significant evidence from randomised controlled trials (RCTs) that physical activity interventions can improve bone health in children (Hind 2007), achieving increased physical activity in children outside of RCTs of imposed exercise interventions is problematic, both if specifically aimed at osteoporosis prevention (Ritenbaugh 2003; French 2005) and more generally. Even complex interventions in children and adolescents do not necessarily improve levels of physical activity (Timperio 2004), making it difficult to implement the evidence supporting a positive effect of physical activity on BMD into practice. While some individual RCTs have reported that calcium supplementation is effective at improving peak bone mass, in our meta-analysis of calcium supplement trials in children (Winzenberg 2006a; Winzenberg 2006b), we found that overall, calcium supplementation did not affect BMD at the two most important sites for fracture in later life, namely the femoral neck and lumbar spine. Although there is a small improvement in BMD at the upper limb (approximately 1.7%) from calcium supplementation, this is unlikely to result in a clinically significant decrease in fracture risk, either in children or in later life. We concluded that alternative nutritional or supplement interventions need to be explored, an opinion which has been supported by others (Lanou 2005).

Vitamin D and bone health in children

Vitamin D has a key role in bone metabolism and its impact on bone health in adults is well accepted (ANZBMS 2005). Overt vitamin D deficiency in children leads to rickets and there is increasing evidence that sub-clinical vitamin D deficiency may also affect bone mineralisation (Jones 1998; Zamora 1999; Outila 2001; Lehtonen 2002; Cheng 2003; Jones 2005). Vitamin D deficiency is best diagnosed by measurement of serum 25-hydroxyvitamin D (Munns 2006). Serum levels above 50 nmol/L are considered normal and largely prevent secondary hyperparathyroidism and elevated bone specific alkaline phosphatase levels (Munns 2006), although there remains debate about the level of this threshold (Holick 2007). Vitamin D deficiency is considered mild at 25 to 50 nmol/L, moderate at 12.5 to 25 nmol/L and severe at < 12.5 nmol/L (Munns 2006). Evidence is growing that low vitamin D levels in children are common enough to be considered a significant public health issue in many parts of the world and across a range of latitudes, including numerous European countries (Zamora 1999; Guillemant 2001; Outila 2001; Lehtonen 2002; Cheng 2003; Tylavsky 2006), the USA (Looker 2002), Lebanon (El-Hajj 2001), Australia (Jones 1999; Jones 2005) and New Zealand (Rockell 2005; Munns 2006; Rockell 2006).

While treatment of moderate to severe vitamin D deficiency with the aim of reaching serum levels of 50 nmol/L is accepted practice (Munns 2006), the effectiveness of vitamin D supplementation for improving bone density outcomes is unclear. Trials of vitamin D supplementation in healthy children measuring bone outcomes have inconsistent results (Ala-Houhala 1988; El-Hajj 2004; Du 2004; Cheng 2005; El-Hajj 2006; Viljakainen 2006). There are methodological differences in the studies, in particular in the choice of dose, compliance and whether subjects were identified as vitamin D deficient or replete. The effect sizes in those studies which do show an effect also vary, ranging from 1.3% and 5% at different sites. While individual RCTs suggest that vitamin D supplementation could increase bone density in children, there has been no systematic review performed previously and it is unclear whether there is an effect overall and, if so, what the effect size is and if methodological differences affect the effect size. There are no studies examining vitamin D supplementation in children with fracture outcomes.

Objectives

  • To determine the effectiveness of vitamin D supplementation for improving bone mineral density in children.

  • To determine if any effect of vitamin D supplementation varies by sex, age or pubertal stage, the type or dose of vitamin D given or baseline vitamin D status.

  • To determine if any effect of vitamin D supplementation persists after cessation of the intervention.

Methods

Criteria for considering studies for this review

Types of studies

Randomised controlled trials of vitamin D supplementation compared with placebo, with a treatment period of at least three months (Munns 2006) were included.

Types of participants

Trials in children and adolescents (aged < 20 years) without co-existent medical conditions or treatments causing osteoporosis were to be included; we excluded studies performed exclusively in neonates (aged < 1 month).

Types of interventions

Trials of vitamin D supplementation regardless of regardless of type or dose of vitamin D supplement or method of administration, compared with placebo. We excluded trials with a treatment period of less than three months.

Types of outcome measures

Main outcome measures were areal or volumetric bone mineral density (BMD and VBMD respectively), or bone mineral content. Quantitative ultrasound measures would also have been included but no studies used these measures. We included studies only if they measured a main outcome.

As discussed above, while fractures in later life would be the ideal outcome measure, for intervention studies in children this would require following large numbers of subjects for decades and these studies have not been performed and there are no studies with childhood fracture outcomes. Therefore, in this review BMD was used as a surrogate outcome, as is commonly seen in intervention studies in children (Gilsanz 1998).

We extracted data on change in areal bone mineral density (BMD) and bone mineral content (BMC) (where possible taken as percent change from baseline) measured a minimum of six months after the study commenced. Measurement sites included femoral neck, total hip, total body, lumbar spine, proximal and distal forearm. Methods of measurement included dual x-ray absorptiometry (DXA), single photon absorptiometry, dual photon absorptiometry and peripheral quantitative computerised tomography. Studies must also have had a measure of variance for outcome measures to be included in the meta-analysis.

We also, where possible, determined sex, age, pubertal stage, baseline serum vitamin D and vitamin D assay used, physical activity, baseline height, baseline weight, dietary calcium intake including calcium supplement use, vitamin D intake, levels of sun exposure, ethnicity, type and dose of vitamin D given, levels of compliance and follow up after cessation of treatment as possible effect modifiers.

We collected data on adverse effects where available.

Search methods for identification of studies

The electronic literature search was last updated on 9 August 2009. The search strategies included a search of the Cochrane Central Register of Controlled Trials (CENTRAL Issue 3, 2009), MEDLINE (1966 to present), EMBASE (1980 to present), CINAHL (1982 to present), AMED (1985 to present) and ISI Web of Science (1945 to present). We handsearched conference abstract issues of key journals (Osteoporosis International, Journal of Bone and Mineral Research, Asia Pacific Journal of Clinical Nutrition, Journal of the American Dietetic Association, Proceedings of the Nutrition Society, Journal of Nutrition) for 2007-08 to identify recent trials that have not yet been published in full. There were no language restrictions.

The search strategy used for MEDLINE is outlined in Appendix 1. The strategy was adapted as appropriate for other databases.

We examined the reference lists and ISI citations of all included studies.

Data collection and analysis

Two review authors (TW, KS) assessed all potentially relevant articles against the study inclusion/exclusion criteria. Two review authors extracted data independently (TW, SP for all studies except Andersen 2008 with extraction by TW/KS). We extracted details regarding the study population, treatment periods, baseline demographic data and baseline and end of study outcomes independently. Two review authors assessed each trial for risk of bias independently (TW, SP for all studies except Andersen 2008 with extraction by TW/KS), addressing randomisation, allocation concealment, blinding of those providing treatment and of treatment subjects, completeness of outcome assessment, selective reporting and other potential sources of bias. When necessary, we contacted the authors of the primary studies to obtain additional information.

We converted outcome measures to standardised mean differences (SMD). For bone density, we calculated a standardised mean difference of the percent change from baseline in treatment and control groups for the various outcomes.

We calculated statistical heterogeneity of the data using a Chi2 test on N-1 degrees of freedom, with significance conservatively set at 0.10. We also assessed inconsistency I2 = [(Q-df)/Q] x 100%, where Q is the Chi2 statistic and df is its degrees of freedom, to describe the percentage of the variability in effect estimates that is due to heterogeneity. A value greater than 50% was considered substantial heterogeneity.

We conducted meta-analysis according to a fixed-effect model. Where heterogeneity was considered substantial, we performed subgroup analyses which had been specified a priori to explore its causes where data were available, i.e. subgroups by sex, pubertal stage, dose of vitamin D and baseline vitamin D levels, compliance and adequacy of allocation concealment. There were insufficient data to consider ethnicity and sun exposure as possible sources of heterogeneity, as planned in the original protocol. Where there was substantial heterogeneity between studies, if there were not clear clinical reasons or study methodology reasons for this, we proceeded to meta-analysis using a random-effects model.

The a priori subgroup analyses by sex, pubertal stage, dose of vitamin D and baseline vitamin D levels were also used to determine whether the effects of supplementation varied by these factors. Where possible, we based the analyses on intention-to-treat data from the individual clinical trials, but if these were not available, in order of preference, data from available data or per protocol analysis. We performed a funnel plot for assessment of publication bias.

Where there were more than one vitamin D dosage group in the studies, we analysed the data after combining all dosage groups together, compared to placebo. One study (Cheng 2005) had low variance compared to other studies in its outcome measures. We confirmed with the authors that the reported variance measure was a standard deviation. We also performed sensitivity analyses omitting this study.

One study was a cluster-randomised controlled with average cluster size of approximately 87 (Du 2004). The data used in the meta-analysis were based on an analysis which did not take into account the effect of clustering. Subsequently, the authors published the intra cluster correlation for total body BMC in the study (Du 2005) at 0.011. We performed a sensitivity analysis making an approximate correction for the effect of clustering by calculating the design effect as 1+(M-1)ICC where M = cluster size and ICC = intra cluster correlation (87 and 0.011 respectively) and correcting the sample size by the calculated design effect of 1.946 (Analysis 2.1).

For subgroup analyses by vitamin D dose, for those studies with an intervention group falling into each subgroup, we used the individual intervention group data and for the controls used the control group mean and standard deviation for each subgroup comparison but reduced the number in the control groups by half for each comparison.

Grading of evidence

We graded evidence using the GRADE system as described in the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2009).

Clinical relevance tables

The results of this review were not amenable to presentation in clinical relevance tables. Where appropriate, we used the SMD effect size to estimate an absolute benefit in mg/cm2 by estimating the pooled standard deviation (SD) from the means of the SD of the outcomes in treatment and control groups for each study, and multiplying the SMD by this (Alderson 2002).

Results

Description of studies

See: Characteristics of included studies; Characteristics of excluded studies.

Our initial search identified 1652 references to potential studies. A further reference to a potential additional study was obtained through author contact (El Hajj Fuleihan 2006a) making a total of 1653 potential references (Figure 1). Of these, we excluded 1599 after initial independent title and abstract screening (kappa = 0.66) with 25 disagreements. The disagreements were resolved by consensus and all 25 were excluded while still at abstract screening stage. Six references were in German and lacked an English abstract. These six abstracts were assessed by a fluent German speaking rheumatologist in conjunction with one author (TW) and all were excluded.

Figure Figure 1.

PRISMA flow diagram

The remaining 23 references required full-text review, from which 12 references to seven studies were included. Reasons for excluding studies at full-text review are given in the Characteristics of excluded studies table. We excluded four as they were not randomised controlled trials (Compston 1998; Calvo 2000; Flynn 2003; Clutterbuck 2005), three as participants were not aged between one month and 18 years of age (Fehily 1992; Venkataraman 1992; Matsumoto 2008), two as primary bone density outcomes were not measured (Ho 1985; Duhamel 2000, one each for not studying a vitamin D intervention (Feliciano 1994) and one as no placebo was used (Guillemant 2001).

One study did not provide sufficient published data to be included in the meta-analysis (Ala-houlala 1988). This study was small (n = 60) and reported no effect of supplementation for BMC at the distal one-third radius.

Characteristics of included studies are given in Characteristics of included studies.

The meta-analysis used data from six studies (Du 2004; Cheng 2005; El-Hajj Fuleihan 2006b; El Hajj Fuleihan 2006a; Viljakainen 2006; Andersen 2008) and included a total of 343 participants receiving placebo and 541 participants receiving vitamin D supplementation. Children in the studies ranged in age from eight to 17 years. One study was performed in Hong Kong Chinese (Du 2004), two in white populations (Cheng 2005; Viljakainen 2006), one in children of Pakistani origin (Andersen 2008) and two studies were performed in Lebanon but ethnicity of participants was not reported (El Hajj Fuleihan 2006a; El-Hajj Fuleihan 2006b). Bone density was measured by dual energy x-ray absorptiometry in all studies with the exception of Cheng 2005, in which volumetric BMD was measured using peripheral quantitative CT at the distal radius.

All studies administered vitamin D3, with the dose administered ranging from 133 IU daily (Du 2004) to 14000 IU per week (El Hajj Fuleihan 2006a; El-Hajj Fuleihan 2006b). Two studies had study groups not included in this review in which calcium supplementation was administered (Cheng 2005; Du 2004). Supplementation was for one year (El Hajj Fuleihan 2006a; El-Hajj Fuleihan 2006b; Viljakainen 2006; Andersen 2008) or two years (Du 2004; Cheng 2005). One study provided follow-up data after supplementation ceased (Du 2004) with five years follow up and this study included a co-intervention of calcium in both vitamin D and comparator groups. Mean baseline serum vitamin D levels ranged from 17.7 to 49.5 nmol/L. Results from available data analysis were extracted for all studies except Du 2004 which reported per protocol analysis, excluding 33 participants with poor compliance. No study reported a true intention-to-treat analysis with imputation of missing data.

Risk of bias in included studies

One study had clearly documented and adequate sequence generation, allocation concealment, blinding and reasons for withdrawals (Cheng 2005).

Two other studies described adequate sequence generation and allocation concealment (El Hajj Fuleihan 2006a; El-Hajj Fuleihan 2006b) and in the remaining two studies this was unclear (Du 2004; Viljakainen 2006; Andersen 2008). All studies stated that they were double-blind, though the blinding process was not fully described. However, in any case it is unlikely that bone density outcomes could be influenced by lack of blinding and so we assessed the risk of bias arising from any deviations from a strict blinding process as very low. No study imputed for missing data in the analysis. However, loss to follow up was for the most part either evenly distributed across groups (El Hajj Fuleihan 2006a; El-Hajj Fuleihan 2006b; Andersen 2008) or reasons for withdrawal were stated and unlikely to related to the true outcome (Cheng 2005; Viljakainen 2006). Incomplete outcome data posed a risk of bias in one study (Du 2004).

Though the number of studies were small, we performed funnel plots to examine for potential publication bias. These suggested that this was not present (Figure 2; Figure 3; Figure 4; Figure 5).

Figure Figure 2.

Funnel plot of comparison: 1 Vitamin D supplementation vs placebo combining dosage groups, all data, outcome: 1.1 % Change total body bone mineral content from baseline.

Figure Figure 3.

Funnel plot of comparison: 1 Vitamin D supplementation vs placebo combining dosage groups, all data, outcome: 1.2 % Change hip bone mineral density from baseline.

Figure Figure 4.

Funnel plot of comparison: 1 Vitamin D supplementation vs placebo combining dosage groups, all data, outcome: 1.3 % Change lumbar spine bone mineral density from baseline.

Figure Figure 5.

Funnel plot of comparison: 1 Vitamin D supplementation vs placebo combining dosage groups, all data, outcome: 1.4 % Change forearm bone mineral density from baseline.

Effects of interventions

See: Summary of findings for the main comparison Vitamin D supplementation for improving bone mineral density in children

Overall, vitamin D supplementation did not have statistically significant effects on percentage change in total body BMC (SMD 0.10, 95% CI −0.06 to 0.26) (Figure 6), hip BMD (SMD 0.06, 95% CI −0.18 to 0.29) (Figure 7) or forearm BMD (SMD 0.04, 95% CI −0.36 to 0.45) (Figure 8). There was a trend to a small effect on percentage change in lumbar spine BMD (SMD 0.15, 95% CI −0.01 to 0.31, P = 0.07) (Figure 9). There was statistical heterogeneity in the results for percentage change in hip BMD (Chi2 = 6.09, df = 3, P = 0.11, I2 = 51%) and forearm BMD (Chi 2 = 7.82, df = 2, P = 0.02, I2 = 74%).

Figure Figure 6.

Forest plot of comparison: 1 Vitamin D supplementation vs placebo combining dosage groups, all data, outcome: 1.1 % Change total body bone mineral content from baseline.

Figure Figure 7.

Forest plot of comparison: 1 Vitamin D supplementation vs placebo combining dosage groups, all data, outcome: 1.2 % Change hip bone mineral density from baseline.

Figure Figure 8.

Forest plot of comparison: 1 Vitamin D supplementation vs placebo combining dosage groups, all data, outcome: 1.4 % Change forearm bone mineral density from baseline.

Figure Figure 9.

Forest plot of comparison: 1 Vitamin D supplementation vs placebo combining dosage groups, all data, outcome: 1.3 % Change lumbar spine bone mineral density from baseline.

The use of data corrected for cluster effects in the study of Du (Du 2004) did not materially effect the result for total body BMC (Analysis 2.1).

Exploring heterogeneity

Sex, baseline serum vitamin D and adequacy of allocation concealment did not explain the heterogeneity between studies seen for analyses of percentage change in hip and forearm BMD. There was a sole study in males. Studies in the female subgroups still had significant heterogeneity (Hip: Chi2 = 4.94, df = 2, P = 0.08; I2 = 60%; forearm: Chi2 = 4.32, df = 1, P = 0.04; I2 = 60%) (Analysis 5.2; Analysis 5.4). There was a single study with participants with a mean baseline serum 25(OH)D less than 35 nmol/L with hip and forearm BMD outcomes. Significant heterogeneity remained in those studies whose participants had a baseline serum 25(OH)D above this level (Hip: Chi2 = 4.54, df = 2, P = 0.10; I2 = 56%; forearm: Chi2 = 7.66, df = 1, P = 0.006; I2 = 87%) (Analysis 7.2; Analysis 7.4). Subgroup analysis by adequacy of allocation concealment (adequate compared to inadequate or unclear allocation concealment) did not account for heterogeneity between studies for hip BMD (Analysis 12.2). There were no studies with inadequate or unclear allocation concealment with forearm BMD outcomes, so a subgroup analysis could not be performed.

Subgroup analyses by compliance and by pubertal status of participants in the study as a whole resulted in identical subgroups. The only study purely in prepubertal children was that of Cheng 2005 and this was also the only study with low compliance. The others all had mixed pubertal status - no study was performed purely in post pubertal children. In the subgroup of studies with mixed pubertal stage/high compliance, there was no statistically significant heterogeneity at either the hip or forearm (Hip: Chi2 = 2.55, df = 2, P = 0.28; I2 = 22%; forearm: Chi2 = 0.64, df = 1, P = 0.42; I2 = 0%).(Analysis 9.2; Analysis 9.4; Analysis 11.2; Analysis 11.4).

We were unable to perform a meaningful exploration of heterogeneity by vitamin D dose (> 200 IU daily versus ≤ 200 IU daily) or by pubertal status within individual studies, as extracting data on both these variables increased the degrees of freedom for the Chi2 test for heterogeneity and decreased inconsistency. Therefore, comparisons with the heterogeneity observed in Analysis 1.2 and Analysis 1.4 are not interpretable.

Subgroup analyses for differences in effects

There were no statistically significant differences in the effects of supplementation at any site by dose of vitamin D administered (> 200 IU daily versus ≤ 200 IU daily) (Analysis 3.1; Analysis 3.2; Analysis 3.3; Analysis 3.4) or by sex (Analysis 5.1; Analysis 5.2; Analysis 5.3; Analysis 5.4). However, in females there was a statistically significant effect on lumbar spine BMD (SMD 0.20, 95% CI 0.01 to 0.39, P = 0.04) but not in males (SMD 0.01, 95% CI -0.31 to 0.33, P = 0.93).

There were no statistically significant differences in effects between high and low baseline serum 25(OH)D studies at any site (Analysis 7.1; Analysis 7.2; Analysis 7.3; Analysis 7.4) though there was a trend for larger effect with low serum vitamin for total body BMC (serum vitamin D ≤ 35 nmol/L: SMD 0.21 (95% CI 0.01 to 0.41); vitamin D group > 35 nmol/L: SMD -0.07 (95% CI -0.33 to 0.18), P = 0.09 for difference). In the low serum vitamin D studies, there was a significant effect for total body BMC (P = 0.04) and the lumbar spine BMD (SMD 0.31, 95% CI 0.00 to 0.61, P = 0.05).

As described above, subgroups of pre- versus mixed pubertal stage studies and low and high compliance studies were identical. Vitamin D supplementation had a greater effect in studies in prepubertal/low compliance children for forearm BMD (prepubertal: SMD 0.51 (95% CI 0.08 to 0.93), mixed pubertal stage: SMD -0.16 (95% CI -0.38 to 0.07), P = 0.02 for effect in prepubertal/low compliance children) (Analysis 9.4; Analysis 11.4). The effects were in the opposite direction for hip and lumbar spine BMD (hip: prepubertal/low compliance SMD -0.30 (95% CI -0.72 to 0.13); mixed pubertal stage/high compliance: SMD 0.14 (95% CI -0.06 to 0.34); lumbar spine : prepubertal/low compliance SMD -0.19 (95% CI -0.61 to 0.23; mixed pubertal stage/high compliance: SMD 0.21 (95% CI 0.04 to 0.38), P = 0.09 for difference and P = 0.02 for lumbar spine effect in mixed pubertal stage/high compliance children) (Analysis 9.2; Analysis 9.3; Analysis 11.2; Analysis 11.3). There was no difference in effects for total body BMC between prepubertal/low compliance and mixed pubertal stage/high compliance groups, though the direction of effect was the same as that seen at the hip and lumbar spine (prepubertal/low compliance SMD -0.22 (95% CI -0.64 to 0.20); mixed pubertal stage/high compliance: SMD 0.15 (95% CI -0.02 to 0.32), P = 0.11 for difference) (Analysis 9.1; Analysis 11.1).

We also compared the effects in Tanner stage 1 or 2 children with Tanner stage 3 and 4 children derived from within individual studies. These did not demonstrate any significant differences in effects between groups at any site (Analysis 10.1; Analysis 10.2;Analysis 10.3; Analysis 10.4).

Sensitivity analyses

In light of the above findingswe repeated (as post hoc analyses) subgroup analyses by sex (Analysis 6.1; Analysis 6.2; Analysis 6.3; Analysis 6.4), baseline serum vitamin D (Analysis 8.1;Analysis 8.2; Analysis 8.3; Analysis 8.4), vitamin D dose (Analysis 4.1; Analysis 4.2; Analysis 4.3; Analysis 4.4) and adequacy of allocation concealment (Analysis 13.2) as sensitivity analyses omitting Cheng 2005 if there was heterogeneity in any subgroup and there were sufficient studies to allow for the analysis.

In the analysis for hip BMD by sex, heterogeneity was reduced to non-significant levels by omitting Cheng 2005. An effect on hip BMD in females not observed with the inclusion of Cheng 2005 was found (SMD 0.29, 95% CI 0.08 to 0.50, P = 0.006) but there were still no sex differences in effect size (Analysis 6.2). Results of analyses for hip BMD by serum vitamin D level and allocation concealment were similar to those of the analyses including Cheng 2005.

Benefits after supplementation ceases

The only study (Du 2004) which provided follow-up data after supplementation ceased reported that there were no significant differences in total body BMC between the calcium supplemented milk group and the calcium-vitamin D supplemented milk group by three years after cessation of supplementation (percentage change since baseline 71.9 (SD 1.5) and 70.8 (SD 1.5) respectively).

Adverse events

There was limited reporting of adverse events (Table 1) and data were insufficient for meta-analysis, but published reasons for drop-outs and withdrawals suggest that supplementation was well-tolerated (see Characteristics of included studies). One study reported that no children developed hypercalcaemia (Andersen 2008). In one study, four children dropped out through disease onset unrelated to the trial, but otherwise not described (Cheng 2005). In another, one child withdrew due to skin allergy, but the intervention arm to which the child belonged and potential relationship to vitamin D was not reported (Du 2004). One study reported that treatment was very well-tolerated (El Hajj Fuleihan 2006a; El-Hajj Fuleihan 2006b) with two children in the placebo group having elevated serum calcium levels at the end of one year, and one girl receiving 1400 IU vitamin D3 per week developing post-streptococcal glomerulonephritis. In boys, three children in the placebo group and 1 in each vitamin dose group had elevated serum calcium levels at 1 year (El Hajj Fuleihan 2006a; El-Hajj Fuleihan 2006b). The final study reported that 16 children dropped out for reasons not related to the study protocol, but gave no further details (Viljakainen 2006).

Table 1. Adverse events
StudyAdverse events
Ala-houlala 1988 Not reported
Andersen 2008 No hypercalcaemia, otherwise not reported
Cheng 2005 No drop-outs due to adverse events, otherwise not reported
Du 2004 1 withdrawal due to skin allergy - intervention arm not reported, otherwise not reported
El Hajj Fuleihan 2006a 3 boys in placebo and 1 in each vitamin D dosage arm had hypercalcaemia at 1 year, otherwise not reported
El-Hajj Fuleihan 2006b 2 subjects in placebo arm with hypercalcaemia; 11 withdrawals evenly distributed across 3 treatment arms with reasons for withdrawal unrelated to adverse events in 10 cases; 1 girl developed poststreptococcal glomerulonephritis
Viljakainen 2006 16 drop-outs for reasons unrelated to study protocol, otherwise not reported

Discussion

Summary of main results

This review, the first systematic review examining the effectiveness of vitamin D supplementation for improving bone density in children known to the authors, failed to demonstrate statistically significant effects of vitamin D supplementation when applied to healthy children generally. However, subgroup analyses suggest that this may be explained in part by the lack of targeting of at-risk children, i.e. those with known low serum vitamin D levels. Compliance and pubertal status may also affect the effectiveness of supplementation. However, the making of a definitive recommendation regarding these aspects of vitamin supplementation is hampered by the paucity of evidence available. This was a significant limitation, particularly for subgroup analyses of possible important sources of clinical heterogeneity.

While it is difficult to extrapolate bone density outcomes to fracture risk reduction, particularly bone density in childhood to fractures in adulthood, the review results suggest that it is possible that vitamin D supplementation can deliver clinically significant improvements in bone health in children with low serum vitamin D. For example, the magnitude of the standardised mean difference (SMD) of effects in children with low baseline serum vitamin D levels were total body bone mineral content (BMC) 0.21, lumbar spine bone mineral density (BMD) 0.31 and hip BMD 0.25, all which were at least 0.2 SMD higher than in the high baseline vitamin D groups. This is approximately equivalent to a 2.6% percentage point greater change in total body BMC from baseline in the vitamin D than the control group. Equivalent estimates were 1% for hip BMD and 1.7% for lumbar spine BMD. Effects at the forearm were neither statistically nor clinically significant. The available studies do not allow us to ascertain whether the effects of supplementation accumulate while supplementation continues or if the effect plateaus over time. However, if 10% increase in peak bone mass can, as postulated, reduce the risk of an osteoporotic fracture in adult life by 50% (Cummings 1993), it is certainly possible that an effect on fracture risk in later life of both clinical and public health significance is achievable. This is particularly so if effects do accumulate with ongoing supplementation. In addition, in a case-control study of fractures in children, children with wrist and forearm fractures had BMD at the total body, femoral neck and lumbar spine (Ma 2003) ranging from 1% to 5% lower than controls, depending on site. Similar magnitude differences have been observed in other studies (Goulding 1998;Goulding 2001). From Ma 2003, we estimate that a 5% increase in total body BMD and in femoral neck or lumbar spine BMD could decrease the relative risk of upper limb fracture in childhood by approximately 17% and 9% respectively. Thus, there is potential for vitamin D supplementation to have a substantial effect on adult fracture rates, as well as an effect in childhood. The cut-off of 35 nmol/L to define low serum vitamin D in this subgroup analysis was arbitrarily chosen based on the availability of adequate data - a lower cut-off would not have enabled the subgroup analysis to be performed at all. However, this means that the only study contributing data to the low baseline vitamin D subgroups for forearm and hip, and the only study of any size for lumbar spine, includes around 20% children with baseline vitamin D above 50 nmol/L. Thus, it is possible that the effect sizes observed in this review underestimate the possible benefits of supplementation in deficient children at these three sites.

Other potential modifiers of the effect of vitamin D supplementation include pubertal status and compliance. Unfortunately, subgroups studies by pre-pubertal versus mixed pubertal stage and low and high compliance were identical and the prepubertal/low compliance subgroup consisted only of one study (Cheng 2005) making it difficult to separate these effects and making it necessary to interpret the results of the analysis with caution. The study Cheng 2005 was identified prior to any analysis as having a low standard deviation compared to the other included studies. Pre-specified analyses to explore heterogeneity identified that this study may be a major contributor to the heterogeneity observed in the hip and forearm analyses. Potential causes for this include the narrow age range in the study, the fact that it is the only study with only prepubertal participants and the only study with low compliance in the subjects analysed. Other factors could include the use of vBMD measures at the forearm compared to DXA in other studies, and the fact that percentage change figures extracted from Cheng were based on models adjusted for baseline value and Tanner stage (1 and 2). However, when study participants were divided within studies into pre and post pubertal groups, there were no differences in effects between subgroups, nor were there any effects of supplementation at any site within the subgroups. Moreover, there was persistent heterogeneity at the hip and forearm in the prepubertal subgroups, which included the study Cheng 2005. This suggests that compliance, rather than pubertal stage, may be the factor more likely to modify the effects of supplementation. The direction of effect was inconsistent at the forearm compared to other sites in the analysis by pubertal status/compliance subgroups. For forearm BMD there was an overall effect of 0.67 favouring the control group in the prepubertal/low compliance study (Cheng 2005) but the direction of effects favoured vitamin D supplementation at the hip, lumbar spine and total body BMC. This could reflect clinical and methodological differences between studies at the different sites. It could also reflect genuine differences in the response of cortical bone (predominates at distal 1/3 radius) compared to trabecular bone (predominates at lumbar spine, femoral neck and total body BMC). It is, however, unclear why the control group would demonstrate an effect at site. The available data do not allow this to be adequately addressed, and this adds further to the need to be cautious in interpreting this analysis.

Overall completeness and applicability of the evidence

The numbers of studies identified in this review were small (seven) and only five of these contributed a significant amount of data to the meta-analysis (Du 2004; Cheng 2005; El Hajj Fuleihan 2006a; El-Hajj Fuleihan 2006b; Viljakainen 2006). Overall, the data available for this review were not adequate to address important clinical factors that might influence the review results. For example, compliance is a factor that potentially affect the results of studies, but the available data did not allow us to separate clearly the effects of compliance from those of pubertal stage, nor examine the potential interplay between compliance, the dose of supplement given and baseline vitamin D levels. Other major gaps in the evidence are the very limited amounts of data in males, in different ethnic groups and describing whether or not the benefits of supplementation persist after supplementation ceases.

Quality of the evidence

The quality of the evidence from this review is assessed as gold.

Authors' conclusions

Implications for practice

The available evidence does not support the use of vitamin D supplementation to improve bone health in healthy children with normal vitamin D levels, but does suggest that supplementation given to vitamin D deficient children may have clinically useful benefits for peak bone mass. However, this remains to be clearly demonstrated.

Implications for research

Further randomised controlled trials in children with vitamin D deficiency are needed to determine definitively what benefits can be obtained, whether benefits continue to be accrued with increasing duration of supplementation, below what level of serum vitamin D can benefits on bone from supplementation be seen and whether any such benefits persist after supplementation ceases. On the current evidence, it may be that females with vitamin D deficiency are the subgroup likely to have the greatest response to supplementation, but the paucity of data in males means that this is by no means certain.

Acknowledgements

Thank you to Louise Falzon for undertaking the electronic searches for this review.

Characteristics of studies

Characteristics of included studies [ordered by study ID]

Ala-houlala 1988
Methods STUDY DESIGN: randomised controlled trial LOCATION AND SETTING: school children, Tampere, Finland; latitude 61 degrees north DURATION OF SUPPLEMENTATION:  13 months DURATION OF FOLLOW-UP: 13 months TYPE OF ANALYSIS: per protocol COMPLIANCE: 4/30 (13%) of vitamin D supplement group excluded as had taken < 5 capsules per week, else not stated CONFOUNDERS MEASURED: none noted
Participants N SCREENED: unknown N RANDOMISED: 60 N COMPLETED: 57 but further 6 excluded (compliance poor = 4, pubertal stage advanced = 1, multivitamin use = 1) M = 45% at completion, after exclusions F = 55% at completion, after exclusions ETHNICITY: not stated MEAN BASELINE AGE (yrs): not stated BASELINE AGE RANGE (yrs): 8 to 10 INCLUSION CRITERIA: 8 to 10 years old, prepubertal EXCLUSION CRITERIA: history of malabsorption, hepatic, renal or endocrine diseases, use of vitamin pills BASELINE SERUM 25 hydroxy vitamin D (nmol/L): 49 nmol/L in supplement group, 46 nmol/L in placebo group CALCIUM INTAKE: not stated PUBERTAL STATUS: prepubertal
Interventions 1. 400 IU vitamin D2 daily 2. Placebo CO-INTERVENTIONS: nil
Outcomes BONE MEASURES: SITES AREAL BONE MINERAL DENSITY: nil BONE MINERAL CONTENT: 1/3 from distal end of diaphysis of non-dominant radius BONE AREA: nil METHOD OF MEASUREMENT: single photon absorptiometry FOLLOW-UP ASSESSMENT POINTS (yrs): 0, 8/12, 13/12 OTHER OUTCOMES MEASURED: weight, height, serum vitamin D metabolites, calcium, albumin, inorganic phosphorus, alkaline phosphatase, parathyroid hormone BONE MEASURES INCLUDED IN STUDY ANALYSES: BONE MINERAL CONTENT: 1/3 from distal end of diaphysis of non-dominant radius SUB-GROUPS IDENTIFIED: Nil
Notes
Risk of bias
ItemAuthors' judgementDescription
Adequate sequence generation? Unclear Not described but described as "randomised controlled trial"
Allocation concealment? Unclear Not described
Blinding? All outcomes Yes Described as "double blind", placebo used and bone density outcomes not likely to be influenced by lack of blinding
Incomplete outcome data addressed? Bone density outcomes No "At the end of the study, after control counting of the remaining capsules, we also had to exclude 4 children in the vitamin D supplemented group who had consumed less than 5 capsules weekly."
Free of selective reporting? Unclear No information available
Free of other bias? Yes The study appears to be free of other sources of bias
Andersen 2008
Methods STUDY DESIGN: randomised controlled trial LOCATION AND SETTING: community setting, Copenhagen area, Denmark DURATION OF SUPPLEMENTATION: 1 year DURATION OF FOLLOW UP: 1 year TYPE OF ANALYSIS: available data analysis (intention to treat stated but no indication of imputation of missing data) COMPLIANCE: 85% CONFOUNDERS MEASURED: age, BMI, dietary calcium intake, dietary vitamin D intake, height, weight
Participants N SCREENED: not specified N RANDOMISED: 26 in total; 8 to placebo, 9 each to 5 and 10 microgram groups N COMPLETED: 7 in each group M = 0% F = 100% ETHNICITY: Pakistani MEDIAN BASELINE AGE (yrs): 12.2 BASELINE AGE RANGE (yrs): 10.1 to 14.7 INCLUSION CRITERIA: Pakistani origin (immigrants or descendants with Pakistani parents) EXCLUSION CRITERIA: serious illness, medication known to affect bone metabolism or 25(OH)D concentrations, pregnancy, breastfeeding, planning pregnancy within 1 year, serum ionised calcium concentrations > 1.5 mmol/L BASELINE SERUM Vitamin D (nmol/L): 7.3, 16.9 and 8.8 nmol/L for placebo, 5 and 10 microgram groups respectively CALCIUM INTAKE: 510 to 656 mg/day (median) PUBERTAL STATUS: not stated
Interventions 1. 400 IU vitamin D3 2. 200 IU vitamin D3 3. Placebo CO-INTERVENTIONS: nil
Outcomes BONE MEASURES: SITES AREAL BONE MINERAL DENSITY: whole body and L2-4 vertebrae BONE MINERAL CONTENT: whole body and L2-4 vertebrae BONE AREA: whole body and L2-4 vertebrae FOLLOW-UP ASSESSMENT POINTS (yrs): 0, 0.5 and 1 year (bone measures 0 and 1 year) OTHER OUTCOMES MEASURED: weight, height, serum 25 hydroxy D, intact parathyroid hormone, osteocalcin, urinary pyridinoline and deoxy pyridinoline, serum ionised calcium BONE MEASURES INCLUDED IN STUDY ANALYSES: AREAL BONE MINERAL DENSITY: whole body and L2-4 vertebrae BONE MINERAL CONTENT: whole body and L2-4 vertebrae BONE AREA: whole body and L2-4 vertebrae SUB-GROUPS IDENTIFIED: nil
Notes
Risk of bias
ItemAuthors' judgementDescription
Adequate sequence generation? Unclear Stated randomisation but no description
Allocation concealment? Unclear Not described
Blinding? All outcomes Unclear Stated study was double-blind and used a placebo but inadequate description. Bone density outcomes unlikely to be influenced by lack of blinding.
Incomplete outcome data addressed? Bone density outcomes Unclear 19% (5/26) drop-outs, no description given and no method of imputing missing data described
Free of selective reporting? Unclear No information available
Free of other bias? Unclear The study appears to be free of other sources of bias.
Cheng 2005
Methods STUDY DESIGN: randomised controlled trial LOCATION AND SETTING: schools, Jyvaskyla, Finland DURATION OF SUPPLEMENTATION: 2 years DURATION OF FOLLOW UP: 2 years TYPE OF ANALYSIS: available data and per protocol analyses COMPLIANCE: 70% for calcium plus vitamin D, 68% for calcium alone CONFOUNDERS MEASURED: age, BMI, physical activity, Tanner stage, calcium intake, dietary vitamin D intake
Participants N SCREENED: 1367 N RANDOMISED: 195 to all 4 study arms; 49 to calcium + vit D, 49 to Calcium N COMPLETED: Ca+D= 46; Ca =41 M = 0% F = 100% ETHNICITY: white MEAN BASELINE AGE (yrs): 11.1 BASELINE AGE RANGE (yrs): 10 to 12 INCLUSION CRITERIA: no history of serious medical conditions, history of medication known to affect bone metabolism, age 10 to 12, Tanner stage I-II, dietary calcium intake < 900 mg/day EXCLUSION CRITERIA: Tanner stage > II, dietary calcium intake > 900 mg/day, race other than white BASELINE SERUM Vitamin D (nmol/L): 49.5 nmol/L CALCIUM INTAKE: 666 mg/day baseline; 1623 mg/day in trial PUBERTAL STATUS: prepubertal (Tanner stage I-II)
Interventions 1. 1000 mg calcium carbonate plus 200 IU vitamin D3 2. 1000 mg calcium carbonate plus vitamin D3 placebo CO-INTERVENTIONS: the 2 other study arms not included in this review were: 3. 1000mg calcium daily from supplemented dairy products and 4. Calcium placebo with vitamin D3 placebo
Outcomes BONE MEASURES: SITES AREAL BONE MINERAL DENSITY: whole body, femoral neck, total femur and L2-4 vertebrae BONE MINERAL CONTENT: whole body, femoral neck, total femur and L2-4 vertebrae BONE AREA: whole body, femoral neck, total femur and L2-4 vertebrae VOLUMETRIC BMD: distal radius, left tibial shaft METHOD OF MEASUREMENT: DXA and pQCT FOLLOW-UP ASSESSMENT POINTS (yrs): 0, 0.5, 1, 1.5 and 2 years (bone measures 0, 1 and 2 years) OTHER OUTCOMES MEASURED: weight, height, lean tissue mass, fat mass, cross-sectional area and moment of inertia of distal radius, left tibial shaft, cortical thickness of left tibial shaft, serum 25 hydroxy D, intact parathyroid hormone, leptin, bone resorption and formation markers, urinary calcium excretion BONE MEASURES INCLUDED IN STUDY ANALYSES: AREAL BONE MINERAL DENSITY: whole body, femoral neck, and L2-4 vertebrae BONE MINERAL CONTENT: whole body, femoral neck, and L2-4 vertebrae BONE AREA: whole body, femoral neck, and L2-4 vertebrae VOLUMETRIC BMD: distal radius, left tibial shaft SUB-GROUPS IDENTIFIED: nil
Notes
Risk of bias
ItemAuthors' judgementDescription
Adequate sequence generation? Yes "assignments were generated by a computer program in blocks of randomly varying size"
Allocation concealment? Yes "Study group assignments were placed in double sealed envelopes"
Blinding? All outcomes Yes "The investigators were unblinded at the conclusion of the trial"
Incomplete outcome data addressed? Bone density outcomes Yes Reasons for withdrawals stated and unlikely to be related to true outcome
Free of selective reporting? Yes Outcomes reported on all bone density measures given in the methods; a priori power calculations reported for BMC outcomes
Free of other bias? Yes The study appears to be free of other sources of bias
Du 2004
Methods STUDY DESIGN: randomised controlled trial LOCATION AND SETTING: schools, China DURATION OF SUPPLEMENTATION: 2 years DURATION OF FOLLOW UP: 5 years TYPE OF ANALYSIS: per protocol with available data COMPLIANCE: 33/757 excluded from analysis for failure to drink milk for ? 4 days. Remainder close to 100% compliance and average dose vitamin D received 3.33 microgram/day CONFOUNDERS MEASURED: calcium intake, physical activity, dietary vitamin D intake, pubertal status, UV exposure
Participants N SCREENED: not stated N RANDOMISED: 757 in total, 498 to treatment arms included in review N COMPLETED: 698 in total, 451 in treatment arms included in review M = 0% F = 100% ETHNICITY: Chinese MEAN BASELINE AGE (yrs): 10.1 BASELINE AGE RANGE (yrs): not stated INCLUSION CRITERIA: 10 years old EXCLUSION CRITERIA: any disease that might affect bone development BASELINE SERUM Vitamin D (nmol/L): 17.7 nmol/L (calcium group); 20.6 nmol/L (calcium plus vitamin D) CALCIUM INTAKE: 418 mg/day (baseline); 649 to 660 mg/day during trial PUBERTAL STATUS: Tanner I 40.6% to 42%; Tanner II 49.3% to 49.8%, Tanner III-IV 8.3% to 9.6%
Interventions 1. 330 ml calcium fortified milk (calcium dose 560 mg) daily on school days (245 mg/day average) 2. 330 ml calcium fortified milk (calcium dose 560 mg) with 5 or 8 microgram vitamin D3 on school days (average 245 mg calcium and 3.33 microgram vitamin D3 daily) 3. Control group with normal diet (not included in review) CO-INTERVENTIONS: 330 ml calcium fortified milk (calcium dose 560 mg) daily on school days (245 mg/day average)
Outcomes BONE MEASURES: SITES AREAL BONE MINERAL DENSITY: distal and proximal forearm of non-dominant arm, whole body BONE MINERAL CONTENT: distal and proximal forearm of non-dominant arm, whole body BONE AREA: distal and proximal forearm of non-dominant arm, whole body METHOD OF MEASUREMENT: DXA FOLLOW-UP ASSESSMENT POINTS (yrs): 0, 1, 2 OTHER OUTCOMES MEASURED: weight, height, serum 25 hydroxy vitamin D, serum intact parathyroid hormone, serum total calcium (uncorrected for albumin, urinary calcium:creatinine ratio (mmol/mmol), bone age BONE MEASURES INCLUDED IN STUDY ANALYSES: AREAL BONE MINERAL DENSITY: distal radial metaphysis, diaphysis of radius, femoral neck, femoral trochanter, femoral diaphysis and L2-4 vertebrae, mean of the 6 sites BONE MINERAL CONTENT: L2-4 vertebrae, mean of all 6 sites BONE AREA: femoral diaphysis and L2-4 vertebrae, mean of the 6 sites SUB-GROUPS IDENTIFIED: pre and post menarche
Notes Cluster-randomised by school. In first reference (Du, British Journal of Nutrition 2004) data was not analysed taking into account clustering, corrected in letter to editor (Du, British Journal of Nutrition 2005). Whole body measures performed on a random selection of half the participants.
Risk of bias
ItemAuthors' judgementDescription
Adequate sequence generation? Unclear Stated randomisation but no description
Allocation concealment? Unclear Not described
Blinding? All outcomes Yes "the identity of the supplement being unknown to both subjects and investigators during the course of the study"
Incomplete outcome data addressed? Bone density outcomes No Participants (n = 33) who were noncompliant were excluded from analysis, and then only full data sets on 681 of the 698 completing the study were included in analysis
Free of selective reporting? No BMC, bone area and BMD were measured at the distal and proximal forearm of the non-dominant arm, but results were not reported
Free of other bias? Yes The study appears to be free of other sources of bias
El Hajj Fuleihan 2006a
Methods STUDY DESIGN: randomised controlled trial LOCATION AND SETTING: Beirut, schools DURATION OF SUPPLEMENTATION: 1 year DURATION OF FOLLOW UP: 1 year TYPE OF ANALYSIS: available data COMPLIANCE: 97% to 98% of dose taken CONFOUNDERS MEASURED: calcium intake, sun exposure, exercise (hours per week), age, height, weight, Tanner stage, baseline vitamin D levels
Participants N SCREENED: not known N RANDOMISED: 184 N COMPLETED: 172 M = 100% F = 0% ETHNICITY: Not stated MEAN BASELINE AGE (yrs): 13.0 years BASELINE AGE RANGE (yrs): 10 to 17 years INCLUSION CRITERIA: healthy EXCLUSION CRITERIA: any disorders or medications known to affect bone metabolism BASELINE SERUM Vitamin D (nmol/L): 40 CALCIUM INTAKE: mg/day (baseline); 775 PUBERTAL STATUS: Tanner stage 1-2 = 92, Tanner 3-4 = 80
Interventions 1. 1400 IU vitamin D3 per week 2. 14,000 IU vitamin D3 per week 3. Placebo CO-INTERVENTIONS: nil
Outcomes BONE MEASURES: SITES AREAL BONE MINERAL DENSITY: lumbar spine, total hip, femoral neck, trochanter and one-third radius and subtotal (whole body excluding head) BONE MINERAL CONTENT: lumbar spine, total hip, femoral neck, trochanter and one-third radius and subtotal (whole body excluding head) BONE AREA: lumbar spine, total hip, femoral neck, trochanter and one-third radius and subtotal (whole body excluding head) METHOD OF MEASUREMENT: DXA FOLLOW-UP ASSESSMENT POINTS (yrs): 0, 1 OTHER OUTCOMES MEASURED: weight, height, grip strength, sick days, fractures, serum calcium, phosphorus and alkaline phosphatase and 25 hydroxy D, subtotal fat mass and lean mass (non-fat soft tissue mass) BONE MEASURES INCLUDED IN STUDY ANALYSES: AREAL BONE MINERAL DENSITY: lumbar spine, total hip, femoral neck, trochanter and one-third radius and subtotal (whole body excluding head) BONE MINERAL CONTENT: lumbar spine, total hip, femoral neck, trochanter and one-third radius and subtotal (whole body excluding head) BONE AREA: lumbar spine, total hip, femoral neck, trochanter and one-third radius and subtotal (whole body excluding head) SUB-GROUPS IDENTIFIED: Tanner I-II versus III-IV boys
Notes Risk of bias assessment based on methods given in references to El Hajj Fuleihan (F) 2006 Data provided are unpublished
Risk of bias
ItemAuthors' judgementDescription
Adequate sequence generation? Yes "The randomisation sequence, stratified by socioeconomic status, was generated by a computer at Merck headquarters, mailed to the study centre and administered by a senior pharmacist"
Allocation concealment? Yes "The randomisation sequence, stratified by socioeconomic status, was generated by a computer at Merck headquarters, mailed to the study centre and administered by a senior pharmacist"
Blinding? All outcomes Yes "The subjects were randomly assigned in a double blind manner" "All students received identical bottles of an oily solution". Bone density outcomes not likely to be influenced by lack of blinding.
Incomplete outcome data addressed? Bone density outcomes Unclear 93% of subjects included in analysis; no data on distribution of loss to follow up across groups and reasons for loss to follow up not published
Free of selective reporting? Unclear Study protocol not available. A priori power calculations for lumbar spine bone density outcomes described, but not for total body outcomes.
Free of other bias? Yes The study appears to be free of other sources of bias
El-Hajj Fuleihan 2006b
Methods STUDY DESIGN: randomised controlled trial LOCATION AND SETTING: Beirut schools, Lebanon DURATION OF SUPPLEMENTATION: 1 year DURATION OF FOLLOW UP: 1 year TYPE OF ANALYSIS: available data COMPLIANCE: 97% to 98% in the 168 girls completing the study and included in the analysis CONFOUNDERS MEASURED: Calcium intake, sun exposure, exercise (hours per week), age, height, weight, Tanner stage, baseline vitamin D levels
Participants N SCREENED: 219 N RANDOMISED: 179 N COMPLETED: 168 M = 0% F = 100% ETHNICITY: not stated MEAN BASELINE AGE (yrs): 13.2 (girls) BASELINE AGE RANGE (yrs): 10 to 17 INCLUSION CRITERIA: healthy EXCLUSION CRITERIA: any disorders or medications known to affect bone metabolism BASELINE SERUM Vitamin D (nmol/L): 34.9 nmol/L CALCIUM INTAKE: 677 mg/day (girls, baseline); 798 mg/day (boys, baseline) PUBERTAL STATUS: 34 premenarcheal, 134 post menarcheal (of 168 girls completing study)
Interventions 1. 1400 IU vitamin D3 per week 2. 14,000 IU vitamin D3 per week 3. Placebo CO-INTERVENTIONS: nil
Outcomes BONE MEASURES: SITES AREAL BONE MINERAL DENSITY: lumbar spine, total hip, femoral neck, trochanter and one-third radius and subtotal (whole body excluding head) BONE MINERAL CONTENT: lumbar spine, total hip, femoral neck, trochanter and one-third radius and subtotal (whole body excluding head) BONE AREA: lumbar spine, total hip, femoral neck, trochanter and one-third radius and subtotal (whole body excluding head) METHOD OF MEASUREMENT: DXA FOLLOW-UP ASSESSMENT POINTS (yrs): 0, 1 OTHER OUTCOMES MEASURED: weight, height, grip strength, sick days, fractures, serum calcium, phosphorus and alkaline phosphatase and 25 hydroxy D, subtotal fat mass and lean mass (non-fat soft tissue mass). BONE MEASURES INCLUDED IN STUDY ANALYSES: AREAL BONE MINERAL DENSITY: lumbar spine, total hip, femoral neck, trochanter and one-third radius and subtotal (whole body excluding head) BONE MINERAL CONTENT: lumbar spine, total hip, femoral neck, trochanter and one-third radius and subtotal (whole body excluding head) BONE AREA: lumbar spine, total hip, femoral neck, trochanter and one-third radius and subtotal (whole body excluding head) SUB-GROUPS IDENTIFIED: males and females; pre and post menarcheal girls, Tanner I-II versus III-IV boys
Notes Unpublished and published data used
Risk of bias
ItemAuthors' judgementDescription
Adequate sequence generation? Yes "The randomisation sequence, stratified by socioeconomic status, was generated by a computer at Merck headquarters, mailed to the study centre and administered by a senior pharmacist"
Allocation concealment? Yes "The randomisation sequence, stratified by socioeconomic status, was generated by a computer at Merck headquarters, mailed to the study centre and administered by a senior pharmacist"
Blinding? All outcomes Yes "The subjects were randomly assigned in a double blind manner". "All students received identical bottles of an oily solution". Bone density outcomes not likely to be influenced by lack of blinding.
Incomplete outcome data addressed? Bone density outcomes Yes 94% of subjects included in analysis; loss to follow up even across groups and reasons for loss to follow up not likely to be related to true outcome
Free of selective reporting? Unclear Study protocol not available. A priori power calculations for lumbar spine bone density outcomes described, but not for total body outcomes.
Free of other bias? Yes The study appears to be free of other sources of bias
Viljakainen 2006
  1. BMC = bone mineral content

  2. BMD = bone mineral density

  3. BMI = body mass index

  4. Ca = calcium

  5. DXA = dual x-ray absorptiometry

  6. pQCT = peripheral quantitative computerised tomography

Methods STUDY DESIGN: randomised controlled trial LOCATION AND SETTING: schools, Finland, 60 degrees north latitude DURATION OF SUPPLEMENTATION: 1 year DURATION OF FOLLOW UP: 1 year TYPE OF ANALYSIS: available data and per protocol COMPLIANCE: 90% of participants completing had compliance > 80% CONFOUNDERS MEASURED: pubertal stage, dietary vitamin D and calcium intake, physical activity, time spent outdoors
Participants N SCREENED: not stated N RANDOMISED: 228 N COMPLETED: 212 M = 0% F = 100% ETHNICITY: white MEAN BASELINE AGE (yrs): 11.4 BASELINE AGE RANGE (yrs): 11 to 12 INCLUSION CRITERIA: healthy EXCLUSION CRITERIA: use of medicines known to affect calcium metabolism BASELINE SERUM Vitamin D (nmol/L): 47 nmol/L CALCIUM INTAKE: 1198 mg/day baseline PUBERTAL STATUS: mixed (n = 10/112/71/7/12 for Tanner I-V respectively)
Interventions 1. 400 IU vitamin D3 2. 200 IU vitamin D3 3. Placebo CO-INTERVENTIONS: nil
Outcomes BONE MEASURES: SITES AREAL BONE MINERAL DENSITY: total hip and L2-4 vertebrae BONE MINERAL CONTENT: total hip and L2-4 vertebrae BONE AREA: total hip and L2-4 vertebrae SIZE ADJUSTED BMC: total hip and L2-4 vertebrae METHOD OF MEASUREMENT: DXA FOLLOW-UP ASSESSMENT POINTS (yrs): 0, 1 OTHER OUTCOMES MEASURED: weight, height, serum osteocalcin, urinary pyridinoline, deoxypyridinoline, serum 25 hydroxy vitamin D, serum intact parathyroid hormone, serum and urinary calcium, phosphate and creatinine BONE MEASURES INCLUDED IN STUDY ANALYSES: AREAL BONE MINERAL DENSITY: total hip and L2-4 vertebrae. BONE MINERAL CONTENT: total hip and L2-4 vertebrae BONE AREA: total hip and L2-4 vertebrae SIZE ADJUSTED BMC: total hip and L2-4 vertebrae  SUB-GROUPS IDENTIFIED: Tanner I-II versus Tanner III-V
Notes Published and unpublished data used
Risk of bias
ItemAuthors' judgementDescription
Adequate sequence generation? Unclear "stratified randomisation process". "Stratification factor was pubertal development"
Allocation concealment? Unclear "The randomisation was done by a person not involved in the project" but allocation was not described
Blinding? All outcomes Yes Stated study was double-blind and used a placebo but inadequate description. Bone density outcomes unlikely to be influenced by lack of blinding.
Incomplete outcome data addressed? Bone density outcomes Yes 7% dropped out but for reasons not related to true outcomes
Free of selective reporting? Unclear No information available
Free of other bias? Yes The study appears to be free of other sources of bias

Characteristics of excluded studies [ordered by study ID]

Study Reason for exclusion
Calvo 2000 Not a randomised controlled trial
Clutterbuck 2005 Not a randomised controlled trial
Compston 1998 Not a randomised controlled trial
Duhamel 2000 No primary outcomes were measured
Fehily 1992 All participants aged < 1 month
Feliciano 1994 Not a vitamin D intervention
Flynn 2003 Not a randomised controlled trial
Guillemant 2001 No placebo in control group
Ho 1985 No bone density outcomes
Matsumoto 2008 Participants aged > 18 years
Venkataraman 1992 All participants aged < 1 month

Data and analyses

Download statistical data

Table Comparison 1.. Vitamin D supplementation vs placebo combining dosage groups, all data
Outcome or subgroup title No. of studies No. of participants Statistical method Effect size
1 % Change total body bone mineral content from baseline 5 672 Std. Mean Difference (IV, Fixed, 95% CI) 0.10 [-0.06, 0.26]
2 % Change hip bone mineral density from baseline 4 639 Std. Mean Difference (IV, Random, 95% CI) 0.06 [-0.18, 0.29]
3 % Change lumbar spine bone mineral density from baseline 5 660 Std. Mean Difference (IV, Fixed, 95% CI) 0.15 [-0.01, 0.31]
4 % Change forearm bone mineral density from baseline 3 427 Std. Mean Difference (IV, Random, 95% CI) 0.04 [-0.36, 0.45]
Figure Analysis 1.1.

Comparison 1 Vitamin D supplementation vs placebo combining dosage groups, all data, Outcome 1 % Change total body bone mineral content from baseline.

Figure Analysis 1.2.

Comparison 1 Vitamin D supplementation vs placebo combining dosage groups, all data, Outcome 2 % Change hip bone mineral density from baseline.

Figure Analysis 1.3.

Comparison 1 Vitamin D supplementation vs placebo combining dosage groups, all data, Outcome 3 % Change lumbar spine bone mineral density from baseline.

Figure Analysis 1.4.

Comparison 1 Vitamin D supplementation vs placebo combining dosage groups, all data, Outcome 4 % Change forearm bone mineral density from baseline.

Table Comparison 2.. Sensitivity analysis - vitamin D supplementation vs placebo combining dosage groups, all data, correcting Du 2004 for clustering
Outcome or subgroup title No. of studies No. of participants Statistical method Effect size
1 % Change total body bone mineral content from baseline 5 563 Std. Mean Difference (IV, Fixed, 95% CI) 0.09 [-0.08, 0.26]
Figure Analysis 2.1.

Comparison 2 Sensitivity analysis - vitamin D supplementation vs placebo combining dosage groups, all data, correcting Du 2004 for clustering, Outcome 1 % Change total body bone mineral content from baseline.

Table Comparison 3.. Vitamin D supplementation - high dose vs low dose
Outcome or subgroup title No. of studies No. of participants Statistical method Effect size
1 % change total body bone mineral content from baseline 5 672 Std. Mean Difference (IV, Fixed, 95% CI) 0.10 [-0.05, 0.26]
1.1 High dose (> 200 IU daily) 3 192 Std. Mean Difference (IV, Fixed, 95% CI) 0.06 [-0.24, 0.37]
1.2 Low dose (≤ 200 IU daily) 4 480 Std. Mean Difference (IV, Fixed, 95% CI) 0.12 [-0.07, 0.30]
2 % change hip bone mineral density from baseline 4 639 Std. Mean Difference (IV, Random, 95% CI) 0.09 [-0.10, 0.28]
2.1 High dose (> 200 iu daily) 3 281 Std. Mean Difference (IV, Random, 95% CI) 0.11 [-0.22, 0.44]
2.2 Low dose (≤ 200 iu daily) 4 358 Std. Mean Difference (IV, Random, 95% CI) 0.07 [-0.18, 0.33]
3 % change lumbar spine bone mineral density from baseline 5 659 Std. Mean Difference (IV, Fixed, 95% CI) 0.11 [-0.05, 0.28]
3.1 High dose (> 200 IU daily) 4 301 Std. Mean Difference (IV, Fixed, 95% CI) 0.18 [-0.06, 0.42]
3.2 Low dose (≤ 200 IU daily) 4 358 Std. Mean Difference (IV, Fixed, 95% CI) 0.06 [-0.16, 0.28]
4 % change forearm bone mineral density from baseline 3 427 Std. Mean Difference (IV, Random, 95% CI) 0.10 [-0.15, 0.34]
4.1 High dose (> 200 IU daily) 2 171 Std. Mean Difference (IV, Random, 95% CI) -0.06 [-0.43, 0.32]
4.2 Low dose (≤ 200 IU daily) 3 256 Std. Mean Difference (IV, Random, 95% CI) 0.20 [-0.12, 0.52]
Figure Analysis 3.1.

Comparison 3 Vitamin D supplementation - high dose vs low dose, Outcome 1 % change total body bone mineral content from baseline.

Figure Analysis 3.2.

Comparison 3 Vitamin D supplementation - high dose vs low dose, Outcome 2 % change hip bone mineral density from baseline.

Figure Analysis 3.3.

Comparison 3 Vitamin D supplementation - high dose vs low dose, Outcome 3 % change lumbar spine bone mineral density from baseline.

Figure Analysis 3.4.

Comparison 3 Vitamin D supplementation - high dose vs low dose, Outcome 4 % change forearm bone mineral density from baseline.

Table Comparison 4.. Vitamin D supplementation - high dose vs low dose; sensitivity - omit Cheng 2005
Outcome or subgroup title No. of studies No. of participants Statistical method Effect size
1 % change total body bone mineral content from baseline 4 585 Std. Mean Difference (IV, Fixed, 95% CI) 0.15 [-0.02, 0.32]
1.1 High dose (> 200 IU daily) 3 192 Std. Mean Difference (IV, Fixed, 95% CI) 0.06 [-0.24, 0.37]
1.2 Low dose (≤ 200 IU daily) 3 393 Std. Mean Difference (IV, Fixed, 95% CI) 0.19 [-0.01, 0.40]
2 % change hip bone mineral density from baseline 3 552 Std. Mean Difference (IV, Random, 95% CI) 0.16 [-0.02, 0.34]
2.1 High dose (> 200 IU daily) 3 281 Std. Mean Difference (IV, Random, 95% CI) 0.11 [-0.22, 0.44]
2.2 Low dose (≤ 200 IU daily) 3 271 Std. Mean Difference (IV, Random, 95% CI) 0.20 [-0.05, 0.45]
3 % change lumbar spine bone mineral density from baseline 4 572 Std. Mean Difference (IV, Fixed, 95% CI) 0.17 [-0.01, 0.34]
3.1 High dose (> 200 IU daily) 4 301 Std. Mean Difference (IV, Fixed, 95% CI) 0.18 [-0.06, 0.42]
3.2 Low dose (≤ 200 IU daily) 3 271 Std. Mean Difference (IV, Fixed, 95% CI) 0.15 [-0.10, 0.40]
4 % change forearm bone mineral density from baseline 2 340 Std. Mean Difference (IV, Random, 95% CI) -0.01 [-0.24, 0.21]
4.1 High dose (> 200 IU daily) 2 171 Std. Mean Difference (IV, Random, 95% CI) -0.06 [-0.43, 0.32]
4.2 Low dose (≤ 200 IU daily) 2 169 Std. Mean Difference (IV, Random, 95% CI) 0.03 [-0.29, 0.35]
Figure Analysis 4.1.

Comparison 4 Vitamin D supplementation - high dose vs low dose; sensitivity - omit Cheng 2005, Outcome 1 % change total body bone mineral content from baseline.

Figure Analysis 4.2.

Comparison 4 Vitamin D supplementation - high dose vs low dose; sensitivity - omit Cheng 2005, Outcome 2 % change hip bone mineral density from baseline.

Figure Analysis 4.3.

Comparison 4 Vitamin D supplementation - high dose vs low dose; sensitivity - omit Cheng 2005, Outcome 3 % change lumbar spine bone mineral density from baseline.

Figure Analysis 4.4.

Comparison 4 Vitamin D supplementation - high dose vs low dose; sensitivity - omit Cheng 2005, Outcome 4 % change forearm bone mineral density from baseline.

Table Comparison 5.. Vitamin D supplementation - females vs males; all data
Outcome or subgroup title No. of studies No. of participants Statistical method Effect size
1 % Change total body bone mineral content from baseline 5 672 Std. Mean Difference (IV, Fixed, 95% CI) 0.10 [-0.06, 0.26]
1.1 Females 4 500 Std. Mean Difference (IV, Fixed, 95% CI) 0.13 [-0.05, 0.31]
1.2 Males 1 172 Std. Mean Difference (IV, Fixed, 95% CI) 0.01 [-0.31, 0.33]
2 % Change hip bone mineral density from baseline 4 639 Std. Mean Difference (IV, Random, 95% CI) 0.06 [-0.18, 0.29]
2.1 Females 3 467 Std. Mean Difference (IV, Random, 95% CI) 0.09 [-0.21, 0.40]
2.2 Males 1 172 Std. Mean Difference (IV, Random, 95% CI) -0.07 [-0.39, 0.25]
3 % Change lumbar spine bone mineral density from baseline 5 660 Std. Mean Difference (IV, Fixed, 95% CI) 0.15 [-0.01, 0.31]
3.1 Females 4 488 Std. Mean Difference (IV, Fixed, 95% CI) 0.20 [0.01, 0.39]
3.2 Males 1 172 Std. Mean Difference (IV, Fixed, 95% CI) 0.01 [-0.31, 0.33]
4 % Change forearm bone mineral density from baseline 3 427 Std. Mean Difference (IV, Random, 95% CI) 0.04 [-0.36, 0.45]
4.1 Females 2 255 Std. Mean Difference (IV, Random, 95% CI) 0.20 [-0.35, 0.76]
4.2 Males 1 172 Std. Mean Difference (IV, Random, 95% CI) -0.25 [-0.57, 0.07]
Figure Analysis 5.1.

Comparison 5 Vitamin D supplementation - females vs males; all data, Outcome 1 % Change total body bone mineral content from baseline.

Figure Analysis 5.2.

Comparison 5 Vitamin D supplementation - females vs males; all data, Outcome 2 % Change hip bone mineral density from baseline.

Figure Analysis 5.3.

Comparison 5 Vitamin D supplementation - females vs males; all data, Outcome 3 % Change lumbar spine bone mineral density from baseline.

Figure Analysis 5.4.

Comparison 5 Vitamin D supplementation - females vs males; all data, Outcome 4 % Change forearm bone mineral density from baseline.

Table Comparison 6.. Vitamin D supplementation - females vs males; all data; sensitivity omit Cheng 2005
Outcome or subgroup title No. of studies No. of participants Statistical method Effect size
1 % Change total body bone mineral content from baseline 4 585 Std. Mean Difference (IV, Fixed, 95% CI) 0.15 [-0.02, 0.32]
1.1 Females 3 413 Std. Mean Difference (IV, Fixed, 95% CI) 0.21 [0.01, 0.41]
1.2 Males 1 172 Std. Mean Difference (IV, Fixed, 95% CI) 0.01 [-0.31, 0.33]
2 % Change hip bone mineral density from baseline 3 552 Std. Mean Difference (IV, Random, 95% CI) 0.14 [-0.06, 0.34]
2.1 Females 2 380 Std. Mean Difference (IV, Random, 95% CI) 0.24 [0.03, 0.45]
2.2 Males 1 172 Std. Mean Difference (IV, Random, 95% CI) -0.07 [-0.39, 0.25]
3 % Change lumbar spine bone mineral density from baseline 4 573 Std. Mean Difference (IV, Fixed, 95% CI) 0.21 [0.04, 0.38]
3.1 Females 3 401 Std. Mean Difference (IV, Fixed, 95% CI) 0.29 [0.08, 0.50]
3.2 Males 1 172 Std. Mean Difference (IV, Fixed, 95% CI) 0.01 [-0.31, 0.33]
4 % Change forearm bone mineral density from baseline 2 340 Std. Mean Difference (IV, Random, 95% CI) -0.16 [-0.38, 0.07]
4.1 Females 1 168 Std. Mean Difference (IV, Random, 95% CI) -0.06 [-0.38, 0.26]
4.2 Males 1 172 Std. Mean Difference (IV, Random, 95% CI) -0.25 [-0.57, 0.07]
Figure Analysis 6.1.

Comparison 6 Vitamin D supplementation - females vs males; all data; sensitivity omit Cheng 2005, Outcome 1 % Change total body bone mineral content from baseline.

Figure Analysis 6.2.

Comparison 6 Vitamin D supplementation - females vs males; all data; sensitivity omit Cheng 2005, Outcome 2 % Change hip bone mineral density from baseline.

Figure Analysis 6.3.

Comparison 6 Vitamin D supplementation - females vs males; all data; sensitivity omit Cheng 2005, Outcome 3 % Change lumbar spine bone mineral density from baseline.

Figure Analysis 6.4.

Comparison 6 Vitamin D supplementation - females vs males; all data; sensitivity omit Cheng 2005, Outcome 4 % Change forearm bone mineral density from baseline.

Table Comparison 7.. Vitamin D supplementation - by baseline serum vitamin D levels; all data
Outcome or subgroup title No. of studies No. of participants Statistical method Effect size
1 % Change total body bone mineral content from baseline 5 672 Std. Mean Difference (IV, Fixed, 95% CI) 0.10 [-0.06, 0.26]
1.1 High (> 35 nmol/L) 2 259 Std. Mean Difference (IV, Fixed, 95% CI) -0.07 [-0.33, 0.18]
1.2 Low (< 35 nmol/L) 3 413 Std. Mean Difference (IV, Fixed, 95% CI) 0.21 [0.01, 0.41]
2 % Change hip bone mineral density from baseline 4 639 Std. Mean Difference (IV, Random, 95% CI) 0.06 [-0.18, 0.29]
2.1 High (> 35 nmol/L) 3 471 Std. Mean Difference (IV, Random, 95% CI) -0.02 [-0.31, 0.28]
2.2 Low (< 35 nmol/L) 1 168 Std. Mean Difference (IV, Random, 95% CI) 0.25 [-0.07, 0.58]
3 % Change lumbar spine bone mineral density from baseline 5 660 Std. Mean Difference (IV, Fixed, 95% CI) 0.15 [-0.01, 0.31]
3.1 High (> 35 nmol/L) 3 471 Std. Mean Difference (IV, Fixed, 95% CI) 0.09 [-0.10, 0.28]
3.2 Low (< 35 nmol/L) 2 189 Std. Mean Difference (IV, Fixed, 95% CI) 0.31 [0.00, 0.61]
4 % Change forearm bone mineral density from baseline 3 427 Std. Mean Difference (IV, Random, 95% CI) 0.04 [-0.36, 0.45]
4.1 High (> 35 nmol/L) 2 259 Std. Mean Difference (IV, Random, 95% CI) 0.12 [-0.62, 0.85]
4.2 Low (< 35 nmol/L) 1 168 Std. Mean Difference (IV, Random, 95% CI) -0.06 [-0.38, 0.26]
Figure Analysis 7.1.

Comparison 7 Vitamin D supplementation - by baseline serum vitamin D levels; all data, Outcome 1 % Change total body bone mineral content from baseline.

Figure Analysis 7.2.

Comparison 7 Vitamin D supplementation - by baseline serum vitamin D levels; all data, Outcome 2 % Change hip bone mineral density from baseline.

Figure Analysis 7.3.

Comparison 7 Vitamin D supplementation - by baseline serum vitamin D levels; all data, Outcome 3 % Change lumbar spine bone mineral density from baseline.

Figure Analysis 7.4.

Comparison 7 Vitamin D supplementation - by baseline serum vitamin D levels; all data, Outcome 4 % Change forearm bone mineral density from baseline.

Table Comparison 8.. Vitamin D supplementation - by baseline serum vitamin D levels; sensitivity omit Cheng 2005
Outcome or subgroup title No. of studies No. of participants Statistical method Effect size
1 % Change total body bone mineral content from baseline 4 585 Std. Mean Difference (IV, Fixed, 95% CI) 0.15 [-0.02, 0.32]
1.1 High (> 35 nmol/L) 1 172 Std. Mean Difference (IV, Fixed, 95% CI) 0.01 [-0.31, 0.33]
1.2 Low (< 35 nmol/L) 3 413 Std. Mean Difference (IV, Fixed, 95% CI) 0.21 [0.01, 0.41]
2 % Change hip bone mineral density from baseline 3 552 Std. Mean Difference (IV, Random, 95% CI) 0.14 [-0.06, 0.34]
2.1 High (> 35 nmol/L) 2 384 Std. Mean Difference (IV, Random, 95% CI) 0.09 [-0.21, 0.38]
2.2 Low (< 35 nmol/L) 1 168 Std. Mean Difference (IV, Random, 95% CI) 0.25 [-0.07, 0.58]
3 % Change lumbar spine bone mineral density from baseline 4 573 Std. Mean Difference (IV, Fixed, 95% CI) 0.21 [0.04, 0.38]
3.1 High (> 35 nmol/L) 2 384 Std. Mean Difference (IV, Fixed, 95% CI) 0.16 [-0.05, 0.37]
3.2 Low (< 35 nmol/L) 2 189 Std. Mean Difference (IV, Fixed, 95% CI) 0.31 [0.00, 0.61]
4 % Change forearm bone mineral density from baseline 2 340 Std. Mean Difference (IV, Random, 95% CI) -0.16 [-0.38, 0.07]
4.1 High (> 35 nmol/L) 1 172 Std. Mean Difference (IV, Random, 95% CI) -0.25 [-0.57, 0.07]
4.2 Low (< 35 nmol/L) 1 168 Std. Mean Difference (IV, Random, 95% CI) -0.06 [-0.38, 0.26]
Figure Analysis 8.1.

Comparison 8 Vitamin D supplementation - by baseline serum vitamin D levels; sensitivity omit Cheng 2005, Outcome 1 % Change total body bone mineral content from baseline.

Figure Analysis 8.2.

Comparison 8 Vitamin D supplementation - by baseline serum vitamin D levels; sensitivity omit Cheng 2005, Outcome 2 % Change hip bone mineral density from baseline.

Figure Analysis 8.3.

Comparison 8 Vitamin D supplementation - by baseline serum vitamin D levels; sensitivity omit Cheng 2005, Outcome 3 % Change lumbar spine bone mineral density from baseline.

Figure Analysis 8.4.

Comparison 8 Vitamin D supplementation - by baseline serum vitamin D levels; sensitivity omit Cheng 2005, Outcome 4 % Change forearm bone mineral density from baseline.

Table Comparison 9.. Vitamin D supplementation vs placebo by pubertal stage of entire study
Outcome or subgroup title No. of studies No. of participants Statistical method Effect size
1 % change total body bone mineral content from baseline 5 672 Std. Mean Difference (IV, Fixed, 95% CI) 0.10 [-0.06, 0.26]
1.1 Prepubertal 1 87 Std. Mean Difference (IV, Fixed, 95% CI) -0.22 [-0.64, 0.20]
1.2 Mixed 4 585 Std. Mean Difference (IV, Fixed, 95% CI) 0.15 [-0.02, 0.32]
2 % change hip bone mineral density from baseline 4 639 Std. Mean Difference (IV, Random, 95% CI) 0.06 [-0.18, 0.29]
2.1 Prepubertal 1 87 Std. Mean Difference (IV, Random, 95% CI) -0.30 [-0.72, 0.13]
2.2 Mixed 3 552 Std. Mean Difference (IV, Random, 95% CI) 0.14 [-0.06, 0.34]
3 % change lumbar spine bone mineral density from baseline 5 660 Std. Mean Difference (IV, Fixed, 95% CI) 0.15 [-0.01, 0.31]
3.1 Prepubertal 1 87 Std. Mean Difference (IV, Fixed, 95% CI) -0.19 [-0.61, 0.23]
3.2 Mixed 4 573 Std. Mean Difference (IV, Fixed, 95% CI) 0.21 [0.04, 0.38]
4 % change forearm bone mineral density from baseline 3 427 Std. Mean Difference (IV, Random, 95% CI) 0.04 [-0.36, 0.45]
4.1 Prepubertal 1 87 Std. Mean Difference (IV, Random, 95% CI) 0.51 [0.08, 0.93]
4.2 Mixed 2 340 Std. Mean Difference (IV, Random, 95% CI) -0.16 [-0.38, 0.07]
Figure Analysis 9.1.

Comparison 9 Vitamin D supplementation vs placebo by pubertal stage of entire study, Outcome 1 % change total body bone mineral content from baseline.

Figure Analysis 9.2.

Comparison 9 Vitamin D supplementation vs placebo by pubertal stage of entire study, Outcome 2 % change hip bone mineral density from baseline.

Figure Analysis 9.3.

Comparison 9 Vitamin D supplementation vs placebo by pubertal stage of entire study, Outcome 3 % change lumbar spine bone mineral density from baseline.

Figure Analysis 9.4.

Comparison 9 Vitamin D supplementation vs placebo by pubertal stage of entire study, Outcome 4 % change forearm bone mineral density from baseline.

Table Comparison 10.. Vitamin D supplementation - by pubertal status (pre- vs post-pubertal); all data
Outcome or subgroup title No. of studies No. of participants Statistical method Effect size
1 % Change total body bone mineral content from baseline 4 643 Std. Mean Difference (IV, Fixed, 95% CI) 0.08 [-0.08, 0.24]
1.1 Prepubertal 4 318 Std. Mean Difference (IV, Fixed, 95% CI) -0.01 [-0.24, 0.22]
1.2 Post pubertal 3 325 Std. Mean Difference (IV, Fixed, 95% CI) 0.18 [-0.05, 0.40]
2 % Change hip bone mineral density from baseline 4 639 Std. Mean Difference (IV, Random, 95% CI) 0.07 [-0.14, 0.27]
2.1 Prepubertal 4 335 Std. Mean Difference (IV, Random, 95% CI) 0.00 [-0.36, 0.37]
2.2 Post pubertal 3 304 Std. Mean Difference (IV, Random, 95% CI) 0.17 [-0.07, 0.40]
3 % Change lumbar spine bone mineral density from baseline 4 639 Std. Mean Difference (IV, Fixed, 95% CI) 0.08 [-0.08, 0.25]
3.1 Prepubertal 4 335 Std. Mean Difference (IV, Fixed, 95% CI) 0.01 [-0.22, 0.24]
3.2 Post pubertal 3 304 Std. Mean Difference (IV, Fixed, 95% CI) 0.16 [-0.07, 0.40]
4 % Change forearm bone mineral density from baseline 3 427 Std. Mean Difference (IV, Random, 95% CI) 0.10 [-0.19, 0.39]
4.1 Prepubertal 3 213 Std. Mean Difference (IV, Random, 95% CI) 0.13 [-0.45, 0.72]
4.2 Post pubertal 2 214 Std. Mean Difference (IV, Random, 95% CI) 0.06 [-0.23, 0.34]
Figure Analysis 10.1.

Comparison 10 Vitamin D supplementation - by pubertal status (pre- vs post-pubertal); all data, Outcome 1 % Change total body bone mineral content from baseline.

Figure Analysis 10.2.

Comparison 10 Vitamin D supplementation - by pubertal status (pre- vs post-pubertal); all data, Outcome 2 % Change hip bone mineral density from baseline.

Figure Analysis 10.3.

Comparison 10 Vitamin D supplementation - by pubertal status (pre- vs post-pubertal); all data, Outcome 3 % Change lumbar spine bone mineral density from baseline.

Figure Analysis 10.4.

Comparison 10 Vitamin D supplementation - by pubertal status (pre- vs post-pubertal); all data, Outcome 4 % Change forearm bone mineral density from baseline.

Table Comparison 11.. Vitamin D supplementation - by compliance (< 80% versus more than 80%); all data
Outcome or subgroup title No. of studies No. of participants Statistical method Effect size
1 % Change total body bone mineral content from baseline 5 672 Std. Mean Difference (IV, Fixed, 95% CI) 0.10 [-0.06, 0.26]
1.1 Low compliance 1 87 Std. Mean Difference (IV, Fixed, 95% CI) -0.22 [-0.64, 0.20]
1.2 High compliance 4 585 Std. Mean Difference (IV, Fixed, 95% CI) 0.15 [-0.02, 0.32]
2 % Change hip bone mineral density from baseline 4 639 Std. Mean Difference (IV, Random, 95% CI) 0.06 [-0.18, 0.29]
2.1 Low compliance 1 87 Std. Mean Difference (IV, Random, 95% CI) -0.30 [-0.72, 0.13]
2.2 High compliance 3 552 Std. Mean Difference (IV, Random, 95% CI) 0.14 [-0.06, 0.34]
3 % Change lumbar spine bone mineral density from baseline 5 660 Std. Mean Difference (IV, Fixed, 95% CI) 0.15 [-0.01, 0.31]
3.1 Low compliance 1 87 Std. Mean Difference (IV, Fixed, 95% CI) -0.19 [-0.61, 0.23]
3.2 High compliance 4 573 Std. Mean Difference (IV, Fixed, 95% CI) 0.21 [0.04, 0.38]
4 % Change forearm bone mineral density from baseline 3 427 Std. Mean Difference (IV, Random, 95% CI) 0.04 [-0.36, 0.45]
4.1 Low compliance 1 87 Std. Mean Difference (IV, Random, 95% CI) 0.51 [0.08, 0.93]
4.2 High compliance 2 340 Std. Mean Difference (IV, Random, 95% CI) -0.16 [-0.38, 0.07]
Figure Analysis 11.1.

Comparison 11 Vitamin D supplementation - by compliance (< 80% versus more than 80%); all data, Outcome 1 % Change total body bone mineral content from baseline.

Figure Analysis 11.2.

Comparison 11 Vitamin D supplementation - by compliance (< 80% versus more than 80%); all data, Outcome 2 % Change hip bone mineral density from baseline.

Figure Analysis 11.3.

Comparison 11 Vitamin D supplementation - by compliance (< 80% versus more than 80%); all data, Outcome 3 % Change lumbar spine bone mineral density from baseline.

Figure Analysis 11.4.

Comparison 11 Vitamin D supplementation - by compliance (< 80% versus more than 80%); all data, Outcome 4 % Change forearm bone mineral density from baseline.

Table Comparison 12.. Vitamin D supplementation - by study quality (adequate allocation concealment); all data
Outcome or subgroup title No. of studies No. of participants Statistical method Effect size
1 % Change total body bone mineral content from baseline 5 672 Std. Mean Difference (IV, Fixed, 95% CI) 0.10 [-0.06, 0.26]
1.1 Adequate allocation concealment 3 427 Std. Mean Difference (IV, Fixed, 95% CI) 0.06 [-0.14, 0.26]
1.2 Inadequate or unclear allocation concealment 2 245 Std. Mean Difference (IV, Fixed, 95% CI) 0.16 [-0.09, 0.41]
2 % Change hip bone mineral density from baseline 4 639 Std. Mean Difference (IV, Random, 95% CI) 0.06 [-0.18, 0.29]
2.1 Adequate allocation concealment 3 427 Std. Mean Difference (IV, Random, 95% CI) -0.02 [-0.32, 0.29]
2.2 Inadequate or unclear allocation concealment 1 212 Std. Mean Difference (IV, Random, 95% CI) 0.23 [-0.06, 0.51]
3 % Change lumbar spine bone mineral density from baseline 5 660 Std. Mean Difference (IV, Fixed, 95% CI) 0.15 [-0.01, 0.31]
3.1 Adequate allocation concealment 3 427 Std. Mean Difference (IV, Fixed, 95% CI) 0.08 [-0.12, 0.28]
3.2 Inadequate or unclear allocation concealment 2 233 Std. Mean Difference (IV, Fixed, 95% CI) 0.29 [0.01, 0.56]
Figure Analysis 12.1.

Comparison 12 Vitamin D supplementation - by study quality (adequate allocation concealment); all data, Outcome 1 % Change total body bone mineral content from baseline.

Figure Analysis 12.2.

Comparison 12 Vitamin D supplementation - by study quality (adequate allocation concealment); all data, Outcome 2 % Change hip bone mineral density from baseline.

Figure Analysis 12.3.

Comparison 12 Vitamin D supplementation - by study quality (adequate allocation concealment); all data, Outcome 3 % Change lumbar spine bone mineral density from baseline.

Table Comparison 13.. Vitamin D supplementation - by study quality (adequate allocation concealment); sensitivity analysis omit Cheng 2005
Outcome or subgroup title No. of studies No. of participants Statistical method Effect size
1 % Change total body bone mineral content from baseline 4 585 Std. Mean Difference (IV, Fixed, 95% CI) 0.15 [-0.02, 0.32]
1.1 Adequate allocation concealment 2 340 Std. Mean Difference (IV, Fixed, 95% CI) 0.15 [-0.08, 0.37]
1.2 Inadequate or unclear allocation concealment 2 245 Std. Mean Difference (IV, Fixed, 95% CI) 0.16 [-0.09, 0.41]
2 % Change hip bone mineral density from baseline 3 552 Std. Mean Difference (IV, Random, 95% CI) 0.14 [-0.06, 0.34]
2.1 Adequate allocation concealment 2 340 Std. Mean Difference (IV, Random, 95% CI) 0.09 [-0.23, 0.41]
2.2 Inadequate or unclear allocation concealment 1 212 Std. Mean Difference (IV, Random, 95% CI) 0.23 [-0.06, 0.51]
3 % Change lumbar spine bone mineral density from baseline 4 573 Std. Mean Difference (IV, Fixed, 95% CI) 0.21 [0.04, 0.38]
3.1 Adequate allocation concealment 2 340 Std. Mean Difference (IV, Fixed, 95% CI) 0.16 [-0.07, 0.38]
3.2 Inadequate or unclear allocation concealment 2 233 Std. Mean Difference (IV, Fixed, 95% CI) 0.29 [0.01, 0.56]
Figure Analysis 13.1.

Comparison 13 Vitamin D supplementation - by study quality (adequate allocation concealment); sensitivity analysis omit Cheng 2005, Outcome 1 % Change total body bone mineral content from baseline.

Figure Analysis 13.2.

Comparison 13 Vitamin D supplementation - by study quality (adequate allocation concealment); sensitivity analysis omit Cheng 2005, Outcome 2 % Change hip bone mineral density from baseline.

Figure Analysis 13.3.

Comparison 13 Vitamin D supplementation - by study quality (adequate allocation concealment); sensitivity analysis omit Cheng 2005, Outcome 3 % Change lumbar spine bone mineral density from baseline.

Appendices

Appendix 1. MEDLINE search strategy

1. osteoporosis/

2. osteoporo$.tw.

3. exp bone density/

4. bone loss$.tw.

5. (bone adj2 densit$).tw.

6. or/1-5

7. exp vitamin d/

8. vitamin d.tw.

9. vitamin d2.tw.

10. vitamin d3.tw.

11. exp Ergocalciferols/

12. ergocalciferol$.tw.

13. exp Cholecalciferol/

14. cholecalciferol.tw.

15. hydroxycholecalciferol.tw.

16. calcitriol.tw.

17. dihyroxyvitamin D3.tw.

18. alphacalcidol.tw.

19. or/8-18

20. 6 and 19

21. clinical trial.pt.

22. randomized.ab.

23. placebo.ab.

24. dt.fs.

25. clinical trials/

26. randomly.ab.

27. trial.ti.

28. groups.ab.

29. or/21-28

30. animals/

31. humans/

32. 30 and 31

33. 30 not 32

34. 29 not 33

35. 20 and 34

36. limit 35 to "all child (0 to 18 years)"

What's new

DateEventDescription
15 May 2008 Amended CMSD ID: A036-P

History

Protocol first published: Issue 1, 2008

Review first published: Issue 10, 2010

DateEventDescription
15 May 2008 Amended Converted to new review format.

Contributions of authors

Tania Winzenberg - wrote review protocol, performed data extraction and quality assessment of articles, performed analyses and wrote the discussion of review results with the input of other authors. She will also be responsible for regularly updating and improving the review, as per Cochrane requirements.

Kelly Shaw - reviewed articles to decide on inclusion, had input into writing of the protocol and was involved in the writing of discussion of review results.

Sandi Powell - had input into writing of the protocol, reviewed initial search for articles to decide on inclusion, performed data extraction and quality assessment of articles, and was involved in the writing of discussion of review results. She also has content knowledge in the field of vitamin D.

Graeme Jones - is the content expert in paediatric bone health for the review. He provided input into design of the protocol, and assisted with the analysis and with writing the discussion of review results.

Declarations of interest

None known.

Sources of support

Internal sources

  • Menzies Research Institute, Australia.

External sources

  • National Health and Medical Research Council, Australia.

Differences between protocol and review

See Methods. We did not perform meta-regression due to the low number of studies included.

Ancillary