Diet

There are two main forms of vitamin D, namely vitamin D₂ (ergocalciferol) and vitamin D₃ (cholecalciferol), both of which are found in the diet in predominantly plant-based and animal-based foods, respectively. While there are few foods which are a natural-rich source of vitamin D, good sources include oily fish and egg yolks (vitamin D₃) and wild mushrooms (vitamin D₂)⁽Reference Holick¹⁾. Foods can also be fortified, on a voluntary or mandatory basis, with vitamin D₂ or D₃, although fortification policies vary quite considerably between countries, with minimal voluntary fortification found in the UK and across much of Europe (with the exception of Finland; for review see Kiely and Black⁽Reference Kiely and Black²⁾). Consequently, dietary vitamin D intakes among adolescents are typically low (<3 µg/d). In the UK National Diet and Nutrition Survey (NDNS) rolling programme (2008/09–2011/12), mean vitamin D intakes from food sources alone were 1·9 and 2·4 µg/d in females and males aged 11–18 years, respectively⁽Reference Bates, Lennox and Prentice³⁾. Main food group contributors were meat and meat products (35 %), fat spreads (20 %) and cereal/cereal products (17 %), with eggs and oily fish, both good sources of vitamin D, contributing only 9 % each, emphasising the low consumption of oily fish among adolescents⁽Reference Bates, Lennox and Prentice³⁾. While supplement use can be an effective means of increasing dietary vitamin D intakes, uptake tends to be greater among infants, young children and elderly adults and low among adolescents. Indeed, in the UK NDNS, only 8 % of adolescents aged 11–18 years reported taking any type of supplement (compared with 16, 22 and 41 % of 4–10, 19–64 and 65+ year olds, respectively) and mean vitamin D intakes increased only marginally to 2·1 and 2·6 µg/d in females and males, respectively, when supplements were included⁽Reference Bates, Lennox and Prentice³⁾.

Endogenous synthesis

Vitamin D is unique in that the majority of this nutrient (80–90 %) is synthesised within the human body following skin exposure to sunlight, hence vitamin D's recognition as the ‘sunshine nutrient’. Ultraviolet B (UVB) radiation (wavelength 290–315 nm) mediates the conversion of 7-dehydrocholesterol, a cholesterol precursor contained within the skin, to vitamin D₃. This synthesised vitamin D₃, along with dietary vitamin D₂/D₃, then undergoes a two-step hydroxylation process: first in the liver to form the biologically inactive 25-hydroxyvitamin D (25(OH)D; the main circulating and storage form of vitamin D and that measured to assess vitamin D status), and secondly in the kidneys to the active 1,25-dihydroxyvitamin D (1,25(OH)₂D) when required by the body⁽Reference Holick¹⁾. The amount of vitamin D₃ endogenously produced is a function of the amount of UVB radiation reaching the skin, therefore being affected by a number of environmental and individual factors, including season and latitude, skin pigmentation, concealing clothing, use of sunscreen, time spent outdoors and other individual factors such as obesity and ageing⁽Reference Webb⁴⁾. Season and latitude dramatically affect vitamin D₃ synthesis, and at high latitudes >40°N during the winter time (October–March), there are marked decreases in endogenous vitamin D₃ production, giving rise to striking seasonal variations in vitamin D status throughout the year⁽Reference Bates, Lennox and Prentice^3, Reference Darling, Hart and Macdonald^5, Reference Klingberg, Oleröd and Konar⁶⁾. This is demonstrated in the NDNS data of sex-combined adolescents aged 11–18 years, whose mean plasma 25(OH)D concentrations ranged from 52·3 nmol/l when blood samples were collected during the months of July–September, down to 31·5 nmol/l when sampled during the winter months of January–March⁽Reference Bates, Lennox and Prentice³⁾.

Physiological function

The primary and well-recognised function of the biologically active metabolite 1,25(OH)₂D is in the maintenance of calcium and phosphorus homeostasis via endocrine mechanisms targeting the intestine, kidneys and bone⁽Reference Holick^1, Reference DeLuca⁷⁾. This is essential for skeletal health throughout the life cycle, from bone accretion and growth in infancy, childhood and adolescence, through to maintenance of healthy bones and prevention of bone loss in later adulthood. Vitamin D, along with calcium, is important during the adolescent years when the most rapid bone accrual occurs⁽Reference Weaver, Gordon and Janz⁸⁾. Approximately 80–90 % of peak bone mass is achieved by late adolescence and maximising this may help reduce age-related bone loss in later life⁽Reference Bailey, McKay and Mirwald^9, Reference Henry, Fatayerji and Eastell¹⁰⁾.

In recent years however, more attention has been paid to the paracrine/autocrine functions of vitamin D in the facilitation of gene expression. It was previously believed that the kidneys were the only site of 1,25(OH)₂D synthesis, although it is now recognised that many extra renal tissues (e.g. colon, prostate, breast and immune cells) have the ability to locally convert 25(OH)D to 1,25(OH)₂D due to the presence of the vitamin D receptor and 1 α-hydroxylase enzymes⁽Reference DeLuca^7, Reference Norman¹¹⁾. Epidemiological studies have therefore indicated a broader role of vitamin D in common cancers, CVD and respiratory diseases⁽Reference Bouillon, Schoor and Gielen^12–Reference Martineau, Jolliffe and Hooper¹⁴⁾. Although vitamin D is emerging as a promising nutrient in many extra skeletal health outcomes, it must be borne in mind that the evidence is largely observational and thus further data from robust randomised controlled trials (RCT) is required to help clarify the causal role of vitamin D and the underlying mechanisms.

Defining vitamin D deficiency and adequacy

Currently there is no international consensus on the optimal circulating 25(OH)D concentrations for health, with debate surrounding the cut-off thresholds to be applied to define vitamin D deficiency and adequacy. Table 1 summarises the current cut-off thresholds proposed by various international authoritative bodies and agencies.

Table 1. Circulating 25-hydroxyvitamin D deficiency and adequacy cut-off thresholds currently proposed by various international agencies

ESPGHAN, European Society for Paediatric Gastroenterology, Hepatology and Nutrition.

There is generally good agreement that populations should not have circulating 25(OH)D concentrations below 25–30 nmol/l based on an increased risk of rickets and impaired bone growth in adolescents. At present, the Institute of Medicine (IOM), European Food Safety Authority (EFSA), European Society for Paediatric Gastroenterology, Hepatology and Nutrition and the American Academy of Pediatrics, all suggest that a serum 25(OH)D concentration >50 nmol/l is adequate based on ensuring optimal bone health^(15–Reference Wagner and Greer¹⁸⁾, while others have proposed much greater 25(OH)D sufficiency thresholds. The Society for Adolescent Health and Medicine and the Endocrine Society, for example, consider sufficiency at 25(OH)D concentrations >75 nmol/l and that concentrations <50 nmol/l are indicative of deficiency^(19, Reference Holick, Binkley and Bischoff-Ferrari²⁰⁾. It is worth noting however that the Endocrine Society guidelines target patient care of those at risk of vitamin D deficiency (e.g. obese patients, patients with malabsorption syndromes or those on medication that affects vitamin D metabolism), in contrast to the IOM, EFSA and the UK Scientific Advisory Committee on Nutrition (SACN), who propose recommendations for the general healthy population^{(15, 16, 21)}. Regardless it may be premature to recommend circulating 25(OH)D concentrations >75 nmol/l due to the lack of evidence from RCT to date supporting higher serum 25(OH)D concentrations and health outcomes beyond musculoskeletal health.

Prevalence of vitamin D deficiency among adolescent populations

Adolescents are a population group with a recognised risk of low vitamin D status, and several nationally representative and individual studies have highlighted the global nature of vitamin D deficiency and inadequacy among adolescent populations⁽Reference Smith, Lanham-New and Hart²²⁾. Estimates of vitamin D deficiency (defined across studies as 25(OH)D <22·5–27·5 nmol/l) have ranged from 4–17 % in the USA and Canada⁽Reference Ganji, Zhang and Tangpricha^23–Reference Whiting, Langlois and Vatanparast²⁶⁾, 15–40 % in Europe⁽Reference Bates, Lennox and Prentice^3, Reference González-Gross, Valtueña and Breidenassel^27–Reference Öberg, Jorde and Almås³¹⁾, 13–33 % in Asia (Korea, India and China)⁽Reference Kim, Oh and Namgung^32–Reference Foo, Zhang and Zhu³⁴⁾ and up to 81 % in Saudi Arabia⁽Reference Siddiqui and Kamfar³⁵⁾. If a cut-off of 50 nmol/l is applied, these estimates increase to 19–91 % of the population as having inadequate vitamin D concentrations⁽Reference Smith, Lanham-New and Hart²²⁾. Data from the UK NDNS rolling programme (2008/09–2011/12) demonstrated that 22 % of adolescents aged 11–18 years had year-round plasma 25(OH)D concentrations <25 nmol/l and this increased to 40 % during the winter months (January–March)⁽Reference Bates, Lennox and Prentice³⁾.

Caution must be taken when comparing data from different studies due to variations in participant characteristics (e.g. age, ethnicity, cultural dress), latitude and season of sampling and in the analytical techniques and assays used to measure 25(OH)D (for review of the impact of different analytical techniques on 25(OH)D results, see Fraser and Milan⁽Reference Fraser and Milan³⁶⁾). A recent study, based on standardised serum 25(OH)D data from European cohorts and national surveys (n 55 844), found adolescents aged 15–18 years to have the highest risk of vitamin D deficiency (25(OH)D <30 nmol/l) in the range of 12–40 %, compared with 4–7 % in younger children (1–14 years), 9–24 % in adults (19–60 years) and 1–8 % in older adults (>61 years)⁽Reference Cashman, Dowling and Škrabáková³⁷⁾. The underlying causes of this apparent increased risk and higher prevalence of vitamin D deficiency in adolescents are unclear, although negative associations between pubertal status and circulating 25(OH)D concentrations have been reported⁽Reference El-Hajj Fuleihan, Nabulsi and Choucair^38–Reference Abrams, Hicks and Hawthorne⁴⁰⁾, which may be due to behavioural (e.g. reduced outdoor time, sedentary lifestyles)⁽Reference Absoud, Cummins and Lim⁴¹⁾ and/or biological mechanisms⁽Reference Aksnes and Aarskog⁴²⁾. In order to meet the increased calcium demands for skeletal growth during adolescence, the metabolism of 25(OH)D to the active 1,25(OH)₂D is increased⁽Reference Aksnes and Aarskog⁴²⁾ and a concomitant decrease in circulating 25(OH)D has been reported⁽Reference Aksnes and Aarskog^42, Reference Lehtonen-Veromaa, Möttönen and Irjala⁴³⁾, although not consistently⁽Reference Krabbe, Christiansen and Hummer⁴⁴⁾.

Consequences of vitamin D deficiency and inadequacy during adolescence

This susceptibility to low vitamin D status in adolescent populations is concerning given the important role of vitamin D in promoting normal bone mineralisation and attainment of peak bone mass during periods of rapid growth and development. It is well recognised that prolonged and severe vitamin D deficiency leads to an increased risk of rickets and altered bone development in children and adolescents⁽Reference Pettifor⁴⁵⁾, although the effects of mild vitamin D inadequacy (25(OH)D between 30 and 50 nmol/l), which has been widely reported in adolescent populations worldwide⁽Reference Wahl, Cooper and Ebeling⁴⁶⁾, remains less clear. Observational studies in adolescent females have suggested adverse effects of circulating 25(OH)D concentrations <40–50 nmol/l on bone mass at various skeletal sites⁽Reference Outila, Kärkkäinen and Lamberg-Allardt^29, Reference Cheng, Tylavsky and Kröger^30, Reference Cashman, Hill and Cotter^47, Reference Foo, Zhang and Zhu⁴⁸⁾, although the evidence is inconsistent with others reporting no association between vitamin D status and bone mass in adolescents⁽Reference Kristinsson, Valdimarsson and Sigurdsson^49, Reference Marwaha, Tandon and Reddy⁵⁰⁾. A 3-year prospective study in Finnish females aged 9–15 years found a 4 % difference in lumbar spine bone mineral density accrual between those with serum 25(OH)D concentrations >37·5 nmol/l and those with concentrations <20 nmol/l at baseline⁽Reference Lehtonen-Veromaa, Möttönen and Nuotio⁵¹⁾. Importantly, there is limited data available in male adolescents, a research gap recognised in the National Osteoporosis Foundation's 2016 position statement on peak bone mass development⁽Reference Weaver, Gordon and Janz⁸⁾. Intervention studies, predominantly in white females, have failed to consistently find a beneficial effect of vitamin D supplementation on bone health indices in adolescents and a 2010 Cochrane systematic review concluded that supplementation may only benefit those with inadequate circulating 25(OH)D concentrations below 35 nmol/l⁽Reference Winzenberg, Powell and Shaw⁵²⁾. There is currently no consensus on the 25(OH)D concentration that is optimal for the acquisition of bone mass in adolescents or the outcome measures that should be targeted.

As previously alluded to, there is a growing body of evidence that vitamin D may be associated with other non-musculoskeletal, chronic diseases, such as CVD, cancer, diabetes and autoimmune diseases. However, much of the evidence base for this has been derived from epidemiologic studies of adults and much less is known of the effect of low vitamin D status during adolescence on these health outcomes in the short and long terms⁽Reference Holick⁵³⁾. Cross-sectional studies in children and adolescents aged 2–19 years have found circulating 25(OH)D concentrations to be inversely associated with total and LDL-cholesterol concentrations⁽Reference Petersen, Dalskov and Sørensen^54, Reference Cabral, Araújo and Teixeira⁵⁵⁾, glucose⁽Reference Reis, von Mühlen and Miller^56, Reference Johnson, Nader and Weaver⁵⁷⁾ and systolic blood pressure⁽Reference Petersen, Dalskov and Sørensen^54, Reference Reis, von Mühlen and Miller^56, Reference Ganji, Zhang and Shaikh^58, Reference Williams, Fraser and Lawlor⁵⁹⁾ and positively correlated with HDL-cholesterol⁽Reference Ganji, Zhang and Tangpricha^23, Reference Johnson, Nader and Weaver^57, Reference Williams, Fraser and Lawlor⁵⁹⁾, although not consistently⁽Reference Nam, Kim and Cho^60, Reference Black, Burrows and Lucas⁶¹⁾. Furthermore, vitamin D status below 50 nmol/l has been associated with an increased prevalence of metabolic syndrome in adolescents aged 12–19 years in the USA⁽Reference Reis, von Mühlen and Miller^56, Reference Ganji, Zhang and Shaikh⁵⁸⁾; however, no association was found in Portuguese adolescents aged 13 years⁽Reference Cabral, Araújo and Teixeira⁵⁵⁾. A 2013 systematic review concluded that there was no consistent association between 25(OH)D and lipid and glucose concentrations in adolescents and that systolic blood pressure was inversely associated with 25(OH)D in cross-sectional studies, but no association was found in prospective cohort studies⁽Reference Dolinsky, Armstrong and Mangarelli⁶²⁾.

Development and re-evaluation of the UK recommendations

In 1991, the UK Department of Health published dietary reference values for most nutrients and energy⁽⁶⁸⁾. At this time, it was assumed that free-living individuals aged between 4 and 64 years of age would synthesise sufficient vitamin D from summer sun exposure to ensure adequate vitamin D throughout the winter months to avoid deficiency (25(OH)D <25 nmol/l). Thereby no reference nutrient intake was set for these age groups, although intake recommendations were determined for population sub-groups perceived to be at risk of vitamin D deficiency, namely infants, those aged ≥65 years and pregnant and lactating women (Table 2). Subsequent reviews in 1998 and 2007 by the Department of Health and SACN, respectively, concluded that these recommendations should remain unchanged while further evidence was generated from RCT^{(63, 69)}. In 2010, SACN agreed to reassess the evidence relating to vitamin D and health on the basis of new data that had become available since the previous review.

The revised dietary intake recommendations for vitamin D were released by SACN in 2016 and a daily intake of 10 µg was set for those aged ≥4 years to ensure a serum 25(OH)D concentration >25 nmol/l, which was deemed protective against adverse musculoskeletal health outcomes⁽²¹⁾. This year-round recommendation also includes those of ethnic minority groups and with limited sun exposure (e.g. institutionalised individuals). A precautionary safe intake of 8·5–10 µg/d, rather than a reference nutrient intake, was set for infants 0 to <4 years of age due to insufficient evidence and uncertainties with specifying an intake recommendation for this age group.

Other notable recommendations

The 2011 IOM dietary reference intakes was the first landmark publication of vitamin D (and calcium) requirements, based on a risk assessment framework and provided a comprehensive review of vitamin D requirements for health⁽¹⁵⁾. The risk assessment framework, which has previously been reviewed in detail⁽Reference Cashman and Kiely^70, Reference Cashman and Kiely⁷¹⁾, has since been adopted in more recent re-evaluations of vitamin D requirements by EFSA, SACN and the Nordic Nutrition Recommendations^{(16, 21, 64)}. Briefly, the risk assessment framework involves undertaking independent systematic evidence-based reviews, which are then used by the committee to identify, describe and rate potential indicators (e.g. rickets, osteomalacia, calcium absorption and fracture risk) in order to derive population targets for 25(OH)D status. The associated dietary vitamin D intake requirements were then established by the IOM via a simulated dose–response meta-regression exercise using group mean/median 25(OH)D response data from selected RCT (winter-based at latitudes >49°N)⁽¹⁵⁾. Based upon this method, the IOM committee proposed a RDA of 15 µg/d for those aged 1–70 years (20 µg/d for those >70 years), which corresponds to a serum 25(OH)D concentration of 50 nmol/l that meets the needs of 97·5 % of the population. Additionally, an estimated average requirement (EAR) of 10 µg/d for those aged ≥1 year was set to maintain serum 25(OH)D of 40 nmol/l in 50 % of the population⁽¹⁵⁾.

Interestingly, using the same simulated dose–response meta-regression method, the Nordic Nutrition Recommendations set a recommended intake (RDA equivalent) of 10 µg/d to maintain serum 25(OH)D >50 nmol/l, using data from winter-based RCT conducted at latitudes covering the Nordic region (49·5–60°N)⁽⁶⁴⁾.

Of note are also the recent EFSA vitamin D dietary reference values (15 µg/d to maintain serum 25(OH)D >50 nmol/l), which defined an adequate intake instead of an average requirement, based on the committee's assertion that there was insufficient evidence to allow for an average requirement to be established for all population groups⁽¹⁶⁾.

In contrast to the meta-regression approach adopted by the IOM, EFSA and the Nordic Nutrition Recommendations as described earlier, SACN generated its vitamin D intake estimates by modelling individual participant-level data and it has been suggested that this method may lead to improved dietary requirement guidelines⁽Reference Vale, Rydzewska and Rovers^72, Reference Cashman, Ritz and Kiely⁷³⁾. This may also be an underlying reason for the variability in the current recommendations and is discussed in more detail later.

Meta-regression v. individual participant-level modelling in determining vitamin D intake requirements

The use of meta-regression and individual participant-level modelling and the impacts on dietary vitamin D intake estimates has been described previously in detail⁽Reference Cashman, Fitzgerald and Kiely^74, Reference Cashman⁷⁵⁾ and so will be reviewed briefly here. The meta-regression approach used by the IOM, and more recently EFSA, in establishing vitamin D requirements used group mean/median data from intervention arms of selected RCT, together with estimates of vitamin D intakes (from diet and supplements) in a simulated dose–response relationship^{(15, 16)}. While this avoids over-reliance on data from any particular RCT, the disadvantage is that data are combined from different RCT that use a variety of analytical methods to measure 25(OH)D concentrations⁽Reference Fraser and Milan³⁶⁾. The use of group mean/median data and the resulting regression line and 95 % CI in the meta-regression model generates average responses, and as such the intake estimates of 10 and 15 µg/d might only be expected to offer protection for 50 % of the population (EAR-type estimate) instead of the intended 97·5 % (RDA-type estimate). Furthermore, the meta-regression approach does not take into consideration the inter-individual variability in 25(OH)D response to increasing vitamin D intakes. This can be overcome with the use of individual participant-level data and 95 % prediction intervals (Fig. 1), as applied by SACN⁽²¹⁾. Individual data from three winter-based RCT in female adolescents aged 11–12 years⁽Reference Cashman, FitzGerald and Viljakainen⁷⁶⁾ and adults aged 20–40 and ≥64 years⁽Reference Cashman, Hill and Lucey^77, Reference Cashman, Wallace and Horigan⁷⁸⁾ was modelled by SACN in order to derive vitamin D intake estimates for the UK population. With this method, it is possible to estimate with more confidence the distribution of intakes required to achieve specific serum 25(OH)D concentrations.

Fig. 1. Relationship between serum 25-hydroxyvitamin D (25(OH)D) concentration and total vitamin D intake using data from randomised controlled trials. The central thick solid line is the regression line and the two curved lines represent the 95 % CI around the mean. The outer two dashed lines represent the 95 % prediction intervals. IU, international units (to convert IU to μg/d divide by 40). Reprinted from Journal of Steroid Biochemistry and Molecular Biology, 148, Cashman KD, Vitamin D: dietary requirements and food fortification as a means of helping achieve adequate vitamin D status, 19–26, Copyright (2015), with permission from Elsevier.

Evidence-based dietary vitamin D requirements in adolescence: use of individual participant-level data to fill knowledge gaps

While vitamin D intake recommendations have been set for the adolescent age group (Table 2), the regulatory authorities tasked with establishing these recommendations have highlighted the significant knowledge gaps in relation to winter-based dose–response RCT specifically designed to estimate dietary vitamin D intakes required to maintain 25(OH)D above certain cut-off thresholds in younger populations. Consequently, intake recommendations for adolescents are often extrapolated from adult data in the absence of adolescent-specific data. Of the three RCT in children and adolescents included in the IOM dose–response model, a 1 month long study was conducted in twenty participants (6–14 years)⁽Reference Schou, Heuck and Wolthers⁷⁹⁾ and another was carried out in 1988 in sixty participants (8–10 years) using doses of 0 and 10 µg/d⁽Reference Ala-Houhala, Koskinen and Koskinen⁸⁰⁾. Other under-researched population sub-groups identified with a lack of dose–response experimental data also included younger children, pregnant women and ethnic minority populations. Winter-based dose–response RCT, designed and implemented within the Food-based solutions for optimal vitamin D nutrition through the life cycle (ODIN) project aimed to fill these knowledge gaps by modelling individual participant-level data in order to estimate dietary requirements for vitamin D in these population sub-groups⁽Reference Kiely and Cashman⁸¹⁾.

In the ODIN adolescent study, 110 healthy white males and females aged 14–18 years were randomly allocated to receive 0, 10 or 20 µg vitamin D₃ daily for 20 weeks throughout the winter (October–March)⁽Reference Smith, Tripkovic and Damsgaard⁸²⁾. Via mathematical modelling of individual participant-level 25(OH)D and total vitamin D intake data, the vitamin D intakes required to maintain serum 25(OH)D concentrations above 25, 30, 40 and 50 nmol/l in 50, 90, 95 and 97·5 % of the adolescents were estimated. Of particular note, the vitamin D intakes needed to avoid deficiency and maintain serum 25(OH)D concentrations >25 and >30 nmol/l in 97·5 % of the population were 10·1 and 13·1 µg/d, respectively⁽Reference Smith, Tripkovic and Damsgaard⁸²⁾. The 10·1 µg/d estimate supports the SACN recommendation for the UK population aged ≥4 years. In order to maintain serum 25(OH)D >40 nmol/l in 50 % of the population, corresponding to the EAR-type intake, 6·6 µg/d was estimated for adolescents. While lower than the 10 µg/d recommendation of the IOM, this is close agreement with the 6·3 µg/d estimate in Danish and Finnish females aged 11–12 years in a similarly designed RCT⁽Reference Cashman, FitzGerald and Viljakainen⁷⁶⁾. However, greater discrepancies arise when considering the RDA-type intake (to maintain serum 25(OH)D >50 nmol/l in 97·5 % of the population). The intake estimate of about 30 µg/d in adolescents derived from the ODIN study is considerably higher than the IOM recommendation of 15 µg/d. The large inter-individual variability in 25(OH)D response to vitamin D intakes and the inability of the meta-regression using aggregate data to consider this, explains, in part, the wide variation in these estimates at the 97·5th percentile⁽Reference Cashman, Ritz and Kiely⁷³⁾.

It is important to note that the adolescent trial aforementioned and many of those included by the IOM, SACN and EFSA were conducted in healthy white populations. A winter-based RCT in 8–14 year old black African American and white children and adolescents estimated that vitamin D intakes of 28 and 52 µg/d would maintain serum 25(OH)D >30 and >50 nmol/l, respectively, in 97·5 % of the population⁽Reference Rajakumar, Moore and Yabes⁸³⁾. Unfortunately intake estimates were not determined for the ethnic groups separately, limiting the opportunity to compare intake estimates. In a recent food-based dose–response RCT in Swedish children aged 5–7 years, it was estimated, using 95 % prediction intervals, that vitamin D intakes of 6 and 20 µg/d would maintain 25(OH)D >30 and >50 nmol/l, respectively, in fair-skinned children, while the corresponding intakes in dark-skinned children were 14 and 28 µg/d, respectively⁽Reference Öhlund, Lind and Hernell⁸⁴⁾. These data provide early indications that greater intakes may be required by ethnic minority populations; however, this should be confirmed in further winter-based vitamin D dose–response RCT⁽Reference Smith and Hart⁸⁵⁾.