Vitamin D status and blood pressure in children and adolescents: a systematic review of observational studies

Background Childhood hypertension is a growing public health problem. Simultaneously, hypovitaminosis D is widespread in this population and could be associated with hypertension. This study systematically reviewed the literature on the relationship between vitamin D status and blood pressure (BP) in children and adolescents. Methods Following the PRISMA guidelines, PUBMED, MEDLINE, CINAHL, EMBASE, Cochrane Library, and ClinicalTrials.gov and the gray literature without language or time restrictions were searched. We included observational studies, assessed their risk of bias, and extracted data on population characteristics, vitamin D status and BP measurements, and the association between the two variables. A narrative analysis of the studies was performed. Results In total, 85 studies were included. Prospective cohort studies showed no association between vitamin D and BP, and generally, they were flawed. Also, the majority of non-prospective cohort studies (cross-sectional, retrospective, case-control) did not report an association between vitamin D and BP. They were mostly flawed regarding BP measurement and adjusting to potential confounders. Conclusion The results on the relationship between vitamin D status and BP in children and adolescents varied between the studies, and mainly pointed towards lack of association. Supplementary Information The online version contains supplementary material available at 10.1186/s13643-021-01584-x.


Background
Childhood hypertension is a serious public health burden, of considerable consequences [1]. In 2018, the global prevalence of childhood hypertension was estimated to be 4%, representing a substantial rise during the past two decades [2]. Between 2000 and 2015, the prevalence of hypertension showed a substantial increase among children aged 6 to 19 years [2]. This trend of increasing prevalence is expected to persist in the future, putting children at danger and further burdening health care systems [2]. Primary hypertension in childhood is commonly associated with cardiovascular risk factors, obesity, left ventricular hypertrophy, retinal vascular, and cognitive changes [3] and is associated with essential hypertension in adulthood [4,5]. The increasing prevalence of hypertension is multifaceted, and its main driver is a higher body mass index (BMI) [2].
Simultaneously, hypovitaminosis D or low serum levels of hydroxyvitamin D (25(OH)D), is widespread in children and adolescents worldwide, even in countries which have plentiful sunlight all-year round [6][7][8]. The most common determinants of deficiency include limited exposure to sunlight, low dietary vitamin D intake, and sequestration in fat tissue especially among obese children and adolescents [9][10][11][12].
There is a general consensus that sufficient vitamin D levels during childhood promote skeletal growth and development. Yet, attention is being increasingly given to the extraskeletal benefits attained by having adequate vitamin D status [13]. Emerging evidence suggests that low serum vitamin D level is associated with poor health outcomes in the pediatric population, specifically obesity-related chronic health conditions, most notably hypertension [14,15]. Obese youth with lower vitamin D levels have showed increased odds for hypertension, and this association remained significant even after adjusting for either BMI or total fat mass [16]. Moreover, an inverse association between 25(OH)D and systolic blood pressure (SBP) has been noted [17]. Several biologically plausible hypotheses support the beneficial effect of adequate vitamin D status on blood pressure (BP), including the role of vitamin D in improving endothelial function, decreasing proinflammatory cytokine levels, and regulating the renin-angiotension-aldosterone system [18][19][20][21][22][23].
To date, human interventional studies have failed to produce conclusive evidence pertaining to vitamin D as a potential antihypertensive supplement. A recent systematic review conducted by Abboud [24] revealed no benefit of vitamin D supplementation on decreasing SBP or diastolic blood pressure (DBP). This could be explained by the fact that other factors, rather than simply dietary intake, determine vitamin D status and its health implications. Vitamin D status can vary quite markedly in groups of people with apparently similar input level and is affected by calcium intake, some therapeutic agents, adiposity levels, and exercise [25]. Attaining adequate vitamin D status for preventing hypertension could thus be of public health relevance. Therefore, this study aims to systematically review the literature to decipher the relationship between vitamin D status and BP in children and adolescents.

Review design
This is a systematic review conducted according to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) statement [26] (see Additional file 1) and following a predefined protocol that was registered with the International Prospective Register of Systematic Reviews (PROSPERO) (CRD42020167550). Ethical approval was not required for this purpose.
The databases PUBMED, MEDLINE (Ovid), CINAHL (EBSCO), EMBASE (Ovid), the Cochrane Library, and http://www.ClinicalTrials.gov, were searched as well as the references of included articles, and previous reviews of vitamin D and BP in children and adolescents identified by the search. The search included observational studies reporting on the association between vitamin D status and BP (systolic, diastolic, or mean arterial pressure (MAP)) in children and adolescents. There were no language or time restrictions to eligible reports.

Search strategy
Our search strategy included three key concepts: (1) vitamin D, (2) blood pressure, and (3) children and adolescents, and for each concept, we mapped Medical Subject Headings (MeSH) and keywords. The search included terms such as vitamin D, cholecalciferol, ergocalciferol, or calcidol, combined with BP or hypertension, and pediatric, child, adolescent, youth, or teenage. We did not apply time restrictions to the search, i.e., the databases were searched from their inception date through January 17, 2020. In addition, we conducted a search update on June 09, 2020. A medical information specialist validated the electronic search strategy (see Additional file 2). The references retrieved from scientific databases were managed using EndNote software, version X6.

Study selection
The selection included prospective cohort, crosssectional, case-control, or retrospective studies including children and adolescents as defined by the studies (e.g., aged less than 18 years). Observational studies evaluating the relationship between vitamin D status (e.g., 25(OH)D blood level) and BP were also considered. The outcomes of interest included the difference in measured BP readings, or the odds of increased BP, or the differences in the prevalence of hypertension between participants with suboptimal vitamin D status, e.g., deficiency, and those with adequate levels of vitamin D. The outcomes of interest also included the correlation between vitamin D levels and BP levels, and the difference in vitamin D levels across groups of BP (e.g., normotensive vs. hypertensive). Mixed studies (including children, adolescents, and adults) were included if there was a subgroup analysis for children and/or adolescents only.
This review excluded studies evaluating the effect of vitamin D supplementation on BP reduction as well as studies investigating the association between hypervitaminosis D and BP. Also excluded were studies conducted on adult participants as defined by the studies (e.g., aged 18 years and above), pregnant women, neonates (aged 0 to 30 days), and infants (aged 1 month to 2 years), as classified by the World Health Organization, participants with diseases affecting vitamin D metabolism (e.g., chronic kidney disease, dialysis, liver disease, parathyroid abnormality, and vitamin D-dependent rickets types 1 and 2), participants taking medications known to interfere with vitamin D metabolism (e.g., phenytoin, phenobarbital, carbamazepine, and rifampin).
Screening of titles and/or abstracts retrieved by the search was done in duplicate on EndNote software, version X6, and eligible studies were identified. Two pairs of authors (M.A. and R.R.; N.M. and S.H.) then retrieved the full texts of these studies and assessed them for eligibility in duplicate. Disagreements were solved through discussion. A calibration exercise was conducted before study selection to ensure the validity of the process.

Data extraction
Data from eligible studies were extracted in duplicate by two pairs of authors (M.A. and R.R.; N.M. and S.H.) on Microsoft Excel 2016, following a data extraction standard form. A calibration exercise was first conducted. Disagreements were resolved through discussion The following details were retained: characteristics of the study, details of the population included, the studied exposures and outcomes, and the main findings and adjustments to the analyses, as applicable. Serum 25(OH)D measures were converted to nanomoles per liter whenever it was reported as nanograms per milliliter, by multiplying by a factor of 2.496.

Risk of bias assessment
After an initial calibration exercise, two pairs of authors (M.A. and R.R.; N.M. and S.H.) collaboratively assessed in duplicate the risk of bias of included studies, solving disagreements between them through discussion. We used a modified version of the Cochrane Risk of Bias tool [27] that is designed to assess to risk of bias of observational studies. Each potential source of bias was graded as low, high, or unclear risk. The criteria for judging a high risk of bias included failure to develop and apply appropriate eligibility criteria; flawed measurement of both exposure and outcome; failure to adequately control confounding; and incomplete follow-up (only for prospective cohort studies).

Data synthesis
Separate narrative analyses of the findings of prospective cohort and non-prospective cohort studies were performed. Furthermore, we provided separate analyses for non-prospective studies based on the level of adjustment for confounding factors. We opted for this method of analysis because failure to take into account confounding factors decreases the quality of the evidence generated by the study [27].

Search results
The details of the search process are detailed in Fig. 1. A total of 85 studies [14][15][16] were included in the systematic review.
Prospective cohort studies Characteristics of the studies We identified three prospective cohort studies [60,99,102], which characteristics are detailed in Table 1. These studies were published between 2014 [102] and 2019 [99], and were conducted in Australia [60], United States of America (USA) [99], and England [102].

Results of the studies
The findings from the prospective cohort studies are also presented in Table 1. Two [60,102] out of three studies showed no association between vitamin D and BP (SBP and DBP), whereas one [99] showed an inverse association between low vitamin D status during early childhood and SBP a decade later, which was rendered not significant after adjustment for weight status. In details, Ke et al. [60] examined 25(OH)D concentrations in an Australian cohort of 8-year-olds (n = 249) followed up at age 15 (n = 162) and explored associations between 25(OH)D with cardiovascular disease (CVD) risk factors, including BP, in these populations. On the cross-sectional analyses of 8-and 15-year-olds, SBP and DBP were not found to be significantly associated with 25(OH)D concentrations. Of interest, prospectively there was no association between 25(OH)D at age 8 and SBP nor DBP at age 15, both in boys and in girls. Similarly, Williams et al. [102] compared prospective associations of two analogs of childhood 25(OH)D (25(OH)D2 and 25(OH)D3) at ages 7-12 years with SBP and DBP measured in adolescence. The analyses were conducted on 2470 participants of the Avon Longitudinal Study of Parents and Children (ALSPAC)-a prospective birth cohort that recruited pregnant women in the former county of Avon, South West England. The results of this study showed that there were no associations between vitamin D measured during childhood with SBP and DBP at mean age of 15.5 years. Finally, among 775 children from the Children's Health Study (CHS)-a prospective birth cohort study recruiting children from the original Boston Birth Cohort at Boston Medical Center, Wang et al. [99] investigated whether vitamin D status in early life can affect SBP a decade later and reported not association between the two variables at ages 3 to 5 years. Further, Wang et al. [99] reported higher odds of elevated SBP at age 6 to 18 years among children with lower vitamin D during early childhood; however, this result was rendered not significant when adjusted for current weight status. Among the three studies, only Wang et al. [99] studied vitamin D levels across SBP status groups (< 75 th percentile vs. ≥ 75 th percentile) during childhood and reported no association between the two variables (p = 0.05).

Assessment of risk of bias
The assessment of risk of bias of prospective cohort studies is presented in Fig. 2. Developing and/or applying appropriate eligibility criteria were appropriate in the three studies. The assessment of vitamin D was adequate in Ke et al. [60] and Wang et al. [99], yet not reported in Williams et al. [102]. The assessment of BP was adequate in Ke et al. [60], yet, it was not reported in Williams et al. [102]. In Wang et al. [99], the number of BP measurements was not reported; thus, the risk of bias was unclear. The three studies extensively adjusted to potential confounders regarding the association between vitamin D and BP. Finally, there was no suspicion of differential loss to follow-up in Ke et al. [60] and Wang et al. [99]; however, in Williams et al. [102], there were numerous significant differences in the characteristics of included and excluded participants because of missing data on one or more variable, which might imply a high risk of bias.
Regarding the four studies [46,54,68,107] that adjusted for age and/or gender, all of them [46,54,68,107] found a significant inverse association between vitamin D and SBP. Also, three [46,54,68] reported a negative relationship with DBP; yet, only one study [107] did not find any association between vitamin D and DBP. One study [46] reported on higher prevalence of hypertension with lower vitamin D levels. The only study [58] that adjusted for BMI and physical activity found a higher SBP with vitamin D deficiency, but did not find any association between vitamin D and DBP, as well as the prevalence of hypertension.
Out of the six studies [14,35,37,61,80,84] which adjusted for age, gender, and/or anthropometric measurements including BMI, three [35,61,80] found a significant inverse association between vitamin D and SBP, whereas another three [14,37,84] did not find such a relationship. Two studies [61,80] reported a significant negative association between vitamin D and DBP, whereas four [14,35,37,84] did not. The only study [14] that assessed high BP did not report any significant association with vitamin D.
Regarding the three studies [53,62,85] that adjusted for age, gender, anthropometric measurements, and/or sexual maturation level, one [85] found a significant inverse association with SBP, while two [53,62] did not find such an association. Further, one [85] found a significant inverse association with DBP, whereas two [53,62] did not find any association. The only study [85] that assessed high BP status, reported an inverse relationship with vitamin D.

Assessment of risk of bias
The assessment of risk of bias of non-prospective cohort studies is presented in Fig. 3. Developing and/or applying appropriate eligibility criteria were flawed in seven studies [16,39,49,66,71,95,103] and were unclear in another eight [43,44,51,59,73,81,97,104]

Discussion
Vitamin D deficiency is a global and pressing public health problem even in countries that have an abundance of sunshine throughout the year [110], and in countries where vitamin D supplementation has been implemented for years [8,111]. Despite the striking lack of data pertaining to children and adolescents worldwide [112], available evidence pinpoints a widespread vitamin D deficiency specifically in this age group [113][114][115][116][117][118]. This deficiency can be attributed to atmospheric and environmental determinants such as high latitude and air pollution; endogenous characteristics such as genetics, skin pigmentation, obesity, and limited physical activity; and behavioral factors, such as sun avoidance, reduced outdoor activities, and use of sunscreen [119]. Similarly, the prevalence of elevated BP in this age group has been on the rise [2]. Interestingly, factors associated with Choi, 2014 [44] DeMoraes, 2014 [45] Depiero Belmont, 2015 [46] Dura-Trave, 2020 [47] Ganji, 2011 [15] Ghobadi, 2019 [48] Ha, 2013 [50] Hannesdottir, 2017 [51] Hassan, 2015 [52] Fig. 3 Risk of bias of the non-prospective cohort studies decreased skin production of vitamin D, including high latitude, industrialization, and dark skin, were also shown to be associated with increased BP levels [120]. Thus, the association between vitamin D and BP gained increasing attention over the last couple of years. Accumulating evidence ranging from molecular mechanisms to biochemical (physiological) and clinical data suggests that vitamin D deficiency contributes to hypertension and proposes an antihypertensive effect of this vitamin, through vasculo-and renoprotective effects, suppression of the renin-angiotensin-aldosterone system, and effects on calcium homeostasis, among others [120]. Yet, to date, epidemiological and experimental data has not provided conclusive evidence about the association between BP and 25(OH)D concentrations, and available results are inconsistent with neither inverse nor an independent relationship reported [120,121]. In order to thoroughly investigate the association between vitamin D and BP in children and adolescents, we conducted a systematic review of observational studies. We provided separate analyses for prospective cohort and nonprospective cohort studies (cross-sectional, retrospective, and case-control). We opted this approach since in nonprospective cohort studies, the temporal relationship between exposure and outcome can often not be determined; thus, it is not possible to determine the direction of causality of the reported relationship [122]. In contrast, although observational cohort studies hold a lower position in the evidence rankings compared with randomized controlled trials (RCTs), their longitudinal findings may shed light on the direction of causality [122] and would, therefore, facilitate causal inference [123].
Our search identified three prospective cohort studies that explored the association between childhood 25(OH)D status and BP during later childhood and adolescence. The three studies showed no association between vitamin D and BP. However, due to the limited number of studies and limited sample size, the conclusions that can be drawn from this analysis are considerably limited, and the clinically relevant information provided is not novel.
The vast majority of the studies identified in our review were non-prospective cohort, which mainly did not show a consistent inverse association between 25(OH)D and SBP or DBP levels, or hypertensive status. Possible reasons for the inconsistency in the findings of these studies might relate to the different methodologies used to evaluate vitamin D status and BP, as well as the flawed evaluation of the latter in some studies, the wide difference in the sample size, seasonality, and difference in maturity (Tanner stage) of the participants, in addition to other physiological and environmental factors [6,11,12,25,124,125].
In specific, for the majority of these studies, BP evaluation was mainly based on the average of two readings. This method yields a crude estimation of the average BP level and may be subject to artifacts such as regression to the mean. Recent data among children suggest that the first three measurements differ significantly from the average BP, while the fourth till tenth readings do not, and suggest that in children, the fourth BP reading might be used as a reliable approximation, and BP measurement might be improved by including ten measurements [124]. Only two studies [41,93] adopted ambulatory blood pressure monitoring (ABPM), which is a superior technique [125] and is considered the gold standard of BP measurements in children and adolescents, as it precisely characterizes changes in BP throughout daily activities, correlates more with target organ damage [126,127]. Benzato et al. [41] evaluated the relationship between 24-h BP patterns and vitamin D levels in 32 obese children and showed that low levels of vitamin D were associated with a higher BP burden, especially at night. In contrast, Skrzypczyk et al. [93] evaluated 49 children with arterial hypertension and found that vitamin D status did not correlate with office BP nor ABPM parameters except for heart rate, suggesting negative influence of vitamin D on arterial wall, which requires further investigations. Up-till-now, ABPM and other reliable estimates of average BP, such as repeated measurements, were rarely used in large epidemiological and clinical studies in children and even in adults [120].
The lack of a significant association between vitamin D status and BP might be attributable to the recruitment of normotensive participants in the vast majority of the studies [121]. Yet, even in studies investigating vitamin D levels across groups of BP status, the results were inconsistent. And, while the majority these studies did not find a difference in vitamin D level among normotensive participants and those with high BP, four found a lower vitamin D level in participants with elevated BP compared with their counterparts. Further investigating the reason behind these mixed results through prospective cohort studies is warranted, especially that Reis et al. [91] who used data from a large sample from NHANES 2001-2004 demonstrated a lower mean vitamin D among subject with high BP. These same inconsistent findings were reported in the adult population [120].
Through a systematic review of the literature investigating the pediatric population, Abboud [24] showed no effect of vitamin D supplementation on SBP or DBP in RCTs, and only a significant small decrease in DBP in non-randomized trials. The quality of evidence of this analysis ranged between low and moderate. Yet, the author acknowledged the small number of included studies, and the fact that all participants were normotensive at baseline, which might mitigate any effect of vitamin D on reducing BP. Similarly, Hauger et al. [128], through a systematic review of RCTs, found no effect of vitamin D supplementation on cardiometabolic health in childhood and adolescence, except for a beneficial impact of increasing 25(OH)D above 70 nmol/L on insulin resistance only in obese subjects, which still needs to be confirmed in adequately powered, high-quality RCTs. The authors acknowledged the low risk of bias in the included studies, but advised caution when interpreting their results as they are based on a relatively limited number of available RCTs. These results are in line with the findings reported among adults showing that vitamin D supplementation did not affect cardiometabolic outcomes, including BP [129][130][131]. Only, Witham et al. [132], through a systematic review of the literature and a meta-analysis, showed a small but significant fall in DBP with vitamin D supplementation only among adults with elevated BP, whereas no reduction in BP was observed in people who were normotensive at baseline. Accordingly, while observational studies might show associations between low 25(OH)D concentrations and elevated BP, interventional studies have not confirmed that optimizing vitamin D status through supplementation has antihypertensive effects. It could be deduced that associations between 25(OH)D and BP are not causal. This was also suggested by a large systematic review and meta-analysis on the relationship between vitamin D status and a wide range of acute and chronic health disorders [133].
Finally, it is always worthy to note that when discussing the effects of vitamin D on BP, one must consider the extremely rare yet relevant issue of toxicity with mega doses of vitamin D and associated-hypercalcemia [134], which might lead to reversible hypertension [135]. Data from Tomaino et al. [97] indicate a U-shaped relationship between SBP and 25(OH)D, and inverse Jshaped relationships between DBP and MAP with serum 25(OH)D status.

Strengths and limitations
To our knowledge, this is the first review to systematically assess the association between vitamin D status and BP in children and adolescents, according to a predefined protocol, and following standard methods for reporting systematic reviews [26]. We searched multiple databases and did not set limits to publication language or time, to increase the comprehensiveness of our search. In addition, we critically appraised included studies. Yet, one limitation of our work relates to the quality of the included records. We included observational studies, only three of them were prospective cohort. The remaining included studies were non-prospective cohort (cross-sectional, retrospective, case-control), among which more than half did not adequately control for significant potential confounders. Although the majority of the studies assessed serum concentration of 25(OH)D which is considered the best determinant of vitamin D status [136], few studies adopted rigorous methodology for measuring BP, such as the use of random-zero sphygmomanometers, multiple readings, or automated measurement [132], which may lead to a suboptimal assessment of BP. The quality of evidence of our review is limited by the variable study quality, significant heterogeneity of outcome assessment methods, and study populations. All of which limit our ability to draw a solid conclusion regarding the relationship between vitamin D status and BP. Accordingly, our results suggest of a lack of association, which are certainly not robust enough and should be interpreted with caution.