A systematic review of pediatric clinical trials of high dose vitamin D

Introduction

Vitamin D is a steroid-based hormone familiar to health care providers, patients and the media. It is well established that appropriate body stores of vitamin D are essential to musculoskeletal health (Parfitt et al., 1982; Beck-Nielsen et al., 2009; Melamed & Kumar, 2010). As reliance solely on sun (UV) exposure results in high rates of vitamin D deficiency (Robinson et al., 2006; Ahmed et al., 2011; Merewood et al., 2012; Thacher, Fischer & Pettifor, 2014), multiple scientific agencies have recommended daily supplementation with small quantities of vitamin D (Ross et al., 2011). Despite the success of this approach in reducing the incidence of vitamin D deficiency related electrolyte disturbances and rickets, there continues to be significant interest in alternative high dose vitamin D supplementation strategies. Potential explanations include concern that specific pediatric populations remain at risk for vitamin D deficiency despite recommended dosing (Aguirre Castaneda et al., 2012) and that higher doses of vitamin D may protect against or improve outcomes for a wide range of non-musculoskeletal diseases involving the immune, respiratory, and cardiovascular systems (Brehm et al., 2010; Levin et al., 2011; Gray et al., 2012; Abrams, Coss-Bu & Tiosano, 2013; McNally et al., 2015; Tomaino et al., 2015; Cadario et al., 2015). Presumably due to uncertainty surrounding the benefit of and best approach to vitamin D supplementation, there has been a growing body of pediatric clinical trial literature. This work spans time, geography, populations (disease states), dosing regimens, and outcome measures. These factors, combined with the large number of adult trials, animal studies and observational literature make the available evidence difficult to find, synthesize and translate to clinical practice or cutting edge research. To assist clinicians and researchers we have sought to identify, describe, and quantify the existing clinical trial literature of high-dose vitamin D supplementation in children through the completion of a systematic review.

The objective of this systematic review was to describe the populations, dosing regimens, methodologies and outcome measures and evaluate how they have varied across geography and time. In addition, we sought to determine the areas where there may be sufficient quantity of high quality evidence to evaluate the benefits of high-dose vitamin D on clinically relevant outcomes. Finally, to assist end user groups in the identification of trials relevant to their specific patient populations, policy development or research areas, we sought to develop an online trial database searchable by keyword and study characteristics.

Material and Methods

Study protocol and objectives were established a priori (PROSPERO protocol registration number: CRD42015016242) and reported here according to the PRISMA guidelines of systematic reviews (Table S1) (Moher et al., 2009).

Results

Search results

Figure 1 demonstrates the flow of studies identified by the search strategy. In total, 2,304 unique records were retrieved from the original electronic search with an additional 146 citations found in the reference lists of systematic reviews and eligible articles. Updating the search in January 2015 added an additional 129 records. Of the 2,579 articles, 2,188 were excluded at level one, with an additional 135 excluded at level two screening. In total, we identified 256 publications that reported on the results of a clinical trial administering any dose of ergocalciferol or cholecalciferol to children. From these 256 articles, 169 articles met eligibility criteria (Appendix S3). Further inspection identified 6 instances of double publications of the same trial with reporting of nearly identical findings (Table S3). These duplicate publications were not included in the study, resulting in a total of 163 publications.

Study design, participant number and quality

The 163 publications evaluated 181 distinct study populations, included 365 separate arms, and enrolled a total of 18,539 children (Fig. S1). As shown in Table 1, RCTs contributed to the majority of the trials (n = 108∕163, 66%) and patients (n = 15,728, 84.8%). Assessment of trial quality determined that 23% (n = 38∕163) and 42% (n = 69∕163) were at low or medium/uncertain risk for bias, respectively. Of the 365 study arms, 263 (72.1%) administered one or more doses of vitamin D meeting our eligibility criteria, on a total of 11,947 children. The median number of participants in the high dose arms was 25 (IQR: 14–42). The 163 trials were published over a 46-year period between 1969 and 2014 (inclusive). The rate of trial publication changed significantly over time (p < 0.001), increasing from 1 trial per decade (1960–1969) to 15 trials in 2014 alone (Fig. 2A). Of the 163 trials, almost half (n = 72, 44%) have been published in the last 5 years (2010–2014). Compared to a linear model, the change over time better fits an exponential function with the number of trials doubling every 12.7 years (R² = 0.85 vs. 0.96 respectively).

Table 1:

Assessment of study design and methodological quality.

Trials enrolled a total of 18,539 patients and a median (IQR) of 49 (25–94). High dose arms enrolled 11,947 patients with a median of 25 (14–42).

Study characteristic	Trials/populations	%a
Study designb
RCT/qRCT	108	66
Single arm	42	26
Controlled, other	13	8
Randomized trial qualityb,c
Low risk	38	23
Medium risk/unclear	69	42
High risk	56	34
Cochrane risk of biasb,c,d
Generation adequate	56/57	34/35
Concealment adequate	50/62	31/38
Blinding adequate	47/25	29/15
Outcome report complete	124/36	76/22
Outcome not selective	89/6	55/4
Age groupse
Neonates	58	32
Infants	35	19
Toddlers	61	34
Schoole rs	104	57
Adolescents	82	45

DOI: 10.7717/peerj.1701/table-1

Notes:

Abbreviations

(q)RCT

(Quasi randomized controlled trial)

aBecause of rounding, percentages may not total 100.

bValues represent the number of trials, and the percentage out of the 163 identified trials.

cStudies were assessed using Cochrane risk of bias tool (Higgins & Green, 2011).

dFor the Cochrane assessment, the number of trials where the risk of bias was unclear, we indicated their numbers after the ‘/’.

eNumbers of populations in each age group out of 181 populations. Numbers will add up to more than 181 populations as some included children from two or more age groups.

Geographical regions

Trials were categorized by geographical region (Fig. 3A) (p < 0.001). The area with the most published trials was Europe (n = 46∕163, 28%) with North America and the Middle East each contributing 35 trials (21%). For comparison, the number of total participants by geographical region is shown in Fig. 3B (p < 0.001). Difference in dosing regimen preference by geographical region was also evident. Doses higher than 40,000 IU were sparse in Europe (n = 14∕263, 5.3%) and North America (n = 9∕263, 3.4%), but represented the most commonly used regimen in every other region (Fig. S3). In addition to dose, statistically significant differences were observed between North America and Europe when compared to the rest of the world for route, form and dosing frequency (Fig. 3). A comparison of the population types studied in these two continents, with the remainder of the world, did not identify statistically significant differences (p = 0.81) (Fig. 3G).

Discussion

This systematic review sought to identify all published pediatric clinical trials of high dose vitamin D supplementation, and determined this to be a large and rapidly expanding area of clinical research. Descriptive analysis identified heterogeneity in important trial design characteristics, including a recent significant increase in studies evaluating a wide range of populations and outcomes not classically related to vitamin D. A comprehensive searchable online database was developed to aid clinicians and researchers in the identification and evaluation of trials relevant to their patient population or area of interest.

Our literature search identified 256 pediatric clinical trials of vitamin D supplementation, of which 169 included one or more study arms meeting our definition of high dose. Evaluation of publications over time further demonstrated high-dose vitamin D supplementation to be a rapidly expanding area of clinical research. With the exception of the 1990s, where a brief decline in publications was observed, the trial number has almost doubled each decade. The decline in publications on high-dose vitamin D may relate to a late 1980’s publication reporting high hypercalcemia rates (34%) in young infants receiving 600,000 IU (Markestad et al., 1987). A detailed comparison of the change in publication rate with other areas in pediatric research was limited by the lack of well-done systematic reviews of similar design and scope. However, one excellent comparator is the recent review published by Duffett and colleagues (2013), wherein they demonstrated a constant linear rise in clinical trials in the pediatric critical care setting. Taken together, the results suggest that pediatric clinical trial literature is steadily increasing. The exponential growth of the randomized controlled trial literature has been previously demonstrated (Tsay & Yang, 2005; Bastian, Glasziou & Chalmers, 2010). Part of the value of systematic reviews is in summarizing the literature, as the rate of publication of RCTs makes it impractical for clinicians to keep up with the primary publications.

Further comparison of publication rates between these two pediatric studies suggested a faster rate of rise in trials on high dose vitamin D. For perspective, if the current rate is maintained there will have been more trials published between 2010 and 2019 then in the preceding 5 decades combined. Evaluation of study characteristics, including the change over time, provided some insight into why the publication rate may be rising faster than other areas. Importantly, this evaluation demonstrated that an increasing number of trials are focusing on populations or outcomes not classically related to vitamin D (Canadian Agency for Drugs and Technologies in Health, 2015). This shift is consistent with the substantial growth in observational literature over the past two decades linking vitamin D to a widespread number of disorders involving the immune, neurological, respiratory, and cardiovascular systems (Brehm et al., 2010; Levin et al., 2011; Gray et al., 2012; McNally et al., 2012; Abrams, Coss-Bu & Tiosano, 2013; Cadario et al., 2015). The decision to pursue clinical trials of high dose supplementation in these populations may relate to postulations made that higher 25OHD levels, relative to those achieved with RDA, may be required to achieve maximal benefit for non-musculoskeletal outcomes (Hathcock et al., 2007). Further, available literature also suggests that when compared to healthy patients, those with acute and chronic disease may have a blunted response to usual doses of vitamin D (McNally et al., 2015).

As part of the descriptive analysis we also sought out heterogeneity in other trial design features, including dosing regimen characteristics. The main dosing regimen characteristics where uniformity was evident include use of the oral route for drug administration and choice of the cholecalciferol form. Consistency in these areas is expected, given that cholecalciferol has been suggested to have favorable metabolism and greater biological activity, and 25OHD is well accepted as the best biological marker of vitamin D status (Melamed & Kumar, 2010). Outside of route and form, significant heterogeneity in regimen dose and frequency was evident. Further exploration determined that some of the heterogeneity could be explained by the geographical origin of the trial. In North America and Europe, daily supplementation was by far the most common regimen, while the remainder of the world predominantly used single or divided loading doses well in excess of the IOM Daily Upper Tolerable Intake Level. These differences might explain why, in contrast to countries like Australia and New Zealand, guidelines originating out of North America do not offer any strategies based on weekly or less frequent loading doses (Munns et al., 2006; Godel, 2007; Manaseki-Holland et al., 2010). Similar to the shift from ergocalciferol to cholecalciferol over time, a study characteristic anticipated to be changing was the proportion of dosing regimens incorporating patient age or weight into dose selection (Aguirre Castaneda et al., 2012; McNally et al., 2015). Despite being well recognized in other areas of pediatric research, the need for age- or weight-based vitamin D dosing has only recently been acknowledged by agencies such as the IOM. Our analysis demonstrated that 90% of trials used a constant dose, with no evidence of a recent shift towards age or weight based practise.

Our study findings strongly suggest that identification and synthesis of the clinical trial literature on high dose vitamin D supplementation has been and will continue to be a challenge for clinicians, researchers and policy makers. First, we demonstrated that there is a significant and rapidly expanding body of clinical trial literature. Second, our descriptive analysis identified significant heterogeneity in multiple relevant study design characteristics, including population (age, disease), dosing regimen, and outcome selection. As many of these characteristics are often poorly described in titles and abstracts, end-users may struggle to not only locate relevant citations but also to determine whether the study will address the question(s). A solution to these two problems was provided as part of this study: a comprehensive online open access searchable database of pediatric clinical trials of high dose vitamin D, similar to what has been generated for other areas, including Eczema (Nankervis, Maplethorpe & Williams, 2011; Nankervis et al., 2015). Further, we completed a validation study proving the database to have excellent sensitivity, containing 36 of 38 of the pediatric high dose trials reported within 13 published independent systematic reviews. The validation work also identified that the database contained an additional 18 trials that met one or more of the systematic review eligibility criteria. This last finding suggests that researchers performing systematic reviews of vitamin D supplementation may already be struggling with the volume of literature and poor description of trials in titles, abstracts and keywords. In addition to being comprehensive, the database was further desi gned to assist the end user with heterogeneity. Not only was the database designed to be searchable by key study characteristics, but also the outcome page was set-up to present the user with the study design, population, dosing regimen, and outcome data required to evaluate trial relevance. Utility of the database was also evaluated, by comparing with 4 pediatric systematic reviews, and when combined with the search functions, it was determined that the number of full text articles requiring review could have been reduced by 85% (Das et al., 2013; Fares et al., 2015; Riverin, Maguire & Li, 2015; Ali & McDevitt, 2015). These observations, combined with the fact that the database would reduce the time associated with developing and performing a literature search, indicate the database to be a beneficial resource for the field.

Finally, the availability of a comprehensive validated database allowed us to identify those areas where there may be sufficient evidence to answer questions regarding the clinical benefits of high-dose vitamin D. Considering all low-risk of bias studies, regardless of size, there were only two areas (respiratory infection/asthma, n = 2,166 and prematurity/low birth weight, n = 2,127) with more than 100 total children enrolled in the high-dose arms. Compared to reviews of other areas of pediatric research, the average number of children recruited per study was smaller (Hamm et al., 2010). The reason for this is unclear, but given the recent interest in the non-classical roles of vitamin D, much of the identified work may represent pilot work intended to precede large phase III studies (Randolph & Lacroix, 2002; Nicholson et al., 2003). Of the two areas with a relative breadth of evidence, pediatric respiratory illness and asthma have been recognized by research groups, culminating in multiple systematic reviews (Charan et al., 2012; Das et al., 2013; Pojsupap et al., 2015). Due to heterogeneity in population, dosing regimen and outcome characteristics these reviews were suggestive, but not definitive, for benefit and further research is underway. In contrast to the area of pediatric respiratory illness, there have not been any recent attempts to systematically synthesize the significant clinical trial literature in the area of prematurity and/or low birth weight. A systematic review of the effectiveness and safety in this population may be worthwhile as metabolic bone disease remains a problem and recent observational data suggests that vitamin D deficiency may augment non-musculoskeletal pathophysiology (Onwuneme et al., 2012). This work would benefit nutrition guidelines in NICU and inform dosing regimens for future phase III studies. Finally, it is important to note that with somewhere between one and two dozen new publications per year the areas with sufficient high-quality evidence to address clinical efficacy could change quickly.

Although this review has many strengths, a number of important limitations should be acknowledged. First, for the majority of the trials information was not available on potentially relevant study characteristics including race, UV exposure, diet, drug compliance and blood collection techniques. Second, the large volume of studies and significant heterogeneity in relevant study characteristics presented considerable challenges for synthesis and presentation of the studies in a manner useful to all clinicians and researchers. As has been successfully performed in other clinical areas, including eczema and pediatric intensive care, we sought to address this problem through the creation of a comprehensive accessible online searchable database of identified trials (Nankervis, Maplethorpe & Williams, 2011; Duffett et al., 2013). Using this database readers and end-users should be able to quickly locate and evaluate the specific population, dosing regimen, outcome and/or study design features of interest. The searchable database also helps address another important study limitation created by the lack of an accepted definition or algorithm for high dose vitamin D. Recognizing the challenges involved with developing such a definition, we chose to use a very inclusive or sensitive threshold. We recognize that the true definition of high dose also incorporates dosing and patient factors including age, disease status, and dose frequency and duration. We have carefully designed the database to ensure that users have the ability to rapidly evaluate these characteristics and apply their own thresholds or algorithm for defining high dose. Finally, a significant limitation of the database is that it is at risk for going out of date. Our goal is to update this list yearly, and provide a data entry page for research groups to provide relevant information on their study.

Conclusion

This systematic review identified 169 publications reporting on pediatric trials of high-dose vitamin D, and made relevant trial information available as part of an online searchable database. Importantly, this field has seen an increase in published trials over the past decade, spanning a wide range of populations, dosing regimens and outcomes. To assist the field we developed, validated and demonstrated utility of an online database of pediatric clinical trials of high dose supplementation vitamin D development. The availability of this data- base, combined with search functions and extracted data, should aid clinicians, researchers and policy makers. Using this resource, clinicians will be able to quickly and comprehen- sively identify and evaluate the level of clinical trial evidence for a particular patient population. Finally, the availability of an up-to-date list of published trials, combined with extracted information on population, eligibility criteria, dosing regimen and outcomes should expedite the systematic review process for researchers and policy makers.

Supplemental Information