Vitamin D intake of Dutch infants from the combination of (fortified) foods, infant formula, and dietary supplements

Due to the changed Dutch fortification policy for vitamin D and the changed supplementation advice for infants, a scenario was designed to estimate the potential total vitamin D intake. In this scenario (a), the vitamin D intake from food was estimated for the 2011 situation applying the 2011 fortification practice to the 2002 food consumption data (see below). In an additional scenario (b), next to vitamin D intake from foods (scenario a), all infants were assigned to a 100 % adherence to the use of a daily supplement containing 10 µg vitamin D conforming the supplementation advice for infants in the Netherlands.

To estimate the vitamin D intake for infants 0–6 months of age, the volume of infant formula estimated based on energy requirement and median or 97.5th percentile of body weight distribution was multiplied with the vitamin D content of infant formula. It was assumed that these infants did not consume complementary foods. The estimated vitamin D intake was compared to the UL of 25 µg/day as set by EFSA [3] to assess the risk of excessive intake. For infants 7–19 months of age, the vitamin D intake was calculated by multiplying the consumed amounts of foods with the vitamin D content of these foods, resulting in a total vitamin D intake per subject for each study day. The consumption data were collected over a short-term period (day), and the individual consumption can vary considerably from day to day (i.e. within-person variation). Consequently, without correction for within-person variation, intake measured on a limited number of days per individual will be a poor indicator of the population habitual intake distribution (also referred to as usual intake). The use of such data in the evaluation of the risk of excessive intake may give invalid results. Therefore, the data in our study were statistically corrected for within-person variation to get the habitual vitamin D intake distribution using the Statistical Program to Assess Dietary Exposure (SPADE version 2.25) [24, 25].

In the VIO study, most children consumed infant or follow-on formula on both study days or did not consume it at all. As vitamin D is added to infant and follow-on formula, it contributed largely to total vitamin D intake of consumers of these products. Consequently, the vitamin D intake distribution was multimodal. In the estimation of the habitual intake, it is assumed that the intake can be transformed to a more or less normal distribution, and this was not the case for the multimodal distribution. It was therefore not appropriate to estimate the habitual vitamin D intake distribution from the total vitamin D intake per day [19]. We therefore applied a so-called first-shrink-then-add approach [26‐28] and estimated the habitual intake distribution separately for vitamin D from infant/follow-on formula and for vitamin D from other food sources. For vitamin D from other food sources, a 1-part model was applied, as this was consumed daily by all subjects. With a 1-part model, the habitual intake amount is modelled, not the intake frequency, as all subjects have a daily intake. For vitamin D intake from infant/follow-on formula, a 2-part model was applied as part of the infants did not consume infant/follow-on formula on the studied days. These two habitual intake distributions were combined using Monte Carlo to obtain the habitual total vitamin D intake distribution.

For scenario b, all infants were assigned to take daily a dietary supplement containing 10 µg vitamin D. This amount was added to the vitamin D from other food sources, as it did not contain any within- or between-person variation. The estimation of the habitual total vitamin D intake distribution went via same procedure as described for scenario a.

To get insight into potential differences in habitual total vitamin D intake distributions for infants of different months of ages, the habitual intake was modelled as a function of age. This was performed separately for infants 7–13 and 16–19 months of age, because data on infants 14 and 15 months of age were lacking. Uncertainty in modelling the habitual intake was quantified with bootstrap analyses (1000 iterations) and is reported as 95 % confidence intervals around the point estimates.

The proportion of infants (7–19 months) with habitual total vitamin D intakes above the UL as set by EFSA (i.e. cut-off value of 25 µg/day for infants below 1 year and of 50 µg/day for infants 1 year and older [3]) was calculated.

In 2011, vitamin D was voluntary added to some breakfast cereals, cookies, dairy products, and drinks (Table 2). The labelled vitamin D levels (total of natural vitamin D and added level) varied from 0.74 µg/100 g in soy drink to 16.5 µg/100 g in porridge cereals. In accordance with EU legislation [29], baby foods, other than cereal-based foods, were not fortified with vitamin D. Infant formula (based on 33 products) contained on average 1.2 µg (SD 0.2) vitamin D per 100 ml. Follow-on formula (based on 71 products) contained on average 1.4 µg (SD 1.0) vitamin D per 100 ml. For both types of formula, the levels were below the legal maximum amounts of 2.5 and 3.0 µg/100 kcal, respectively.

Based on energy requirement [15] and body weight reported in the fourth Dutch national growth study [16], an indication of the energy intake was provided for children 0–6 months of age. Taking the median body weight, these intakes varied from 345 kcal/day for girls aged 0–1 month to 635 kcal/day for boys aged 5–6 months (Table 1). This range was 428 kcal/day for girls aged 0–1 month to 794 kcal/day for boys aged 5–6 month when taking the 97.5th percentile of the body weight curve (Table 1). In 2011, the vitamin D level in infant formula intended for children aged 0–6 months was about 1.8 µg/100 kcal, based on label information. Taking this vitamin D level, the intake ranged from 6.4 µg/day for girls aged 0–1 month using median body weight to 14.7 µg/day for boys aged 5–6 months using 97.5th percentile of body weight. For the whole age group 0–6 months, estimated vitamin D intake from infant formula remained below the UL of 25 µg/day. Also with the addition of a daily supplement containing 10 µg vitamin D, intakes were below the UL. However, for children 4–6 months, considering the 97.5th percentile of body weight, vitamin D intake from infant formula and a dietary supplement was close to the UL, 22.7–24.7 µg/day (Table 1).

The percentage of infants 7–19 months of age consuming infant formula decreased with age (Table 3). About 95 % of the infants aged 7–9 months consumed infant formula, compared to 5–16 % of the infants 16–19 months of age. Most infants consuming infant formula did this on both study days. Besides the age-related decrease in the proportion of infants consuming infant formula, also the consumed volume decreased with age (Table 4). Among users of infant formula, the median habitual consumption of infant formula decreased from 592 ml/day for infants aged 7 months to 253 ml/day for infants aged 19 months.

In 2011, the median habitual vitamin D intake from foods only (including infant formula) decreased with age by about 75 %, from 10.9 µg/day for infants 7 months old to about 2.5 µg/day for infants 16–19 months of age (Table 5; Fig. 1). Logically, the habitual vitamin D intake increased with about 10 µg by the addition of the daily-advised dosage vitamin D from dietary supplements (Table 5). The habitual intake was modelled separately including and excluding a daily supplement of 10 µg; therefore, the observed increase was not exactly 10 µg due to modelling uncertainty.

For children 7–11 months old, the vitamin D intake at the 99th percentile was 27–29 µg/day and decreased with age. This suggests that some infants are exceeding the UL at a limited time period of at maximum 5 months and with a limited amount (8–15 % above the UL) which decreased with age. From 1 year onwards, the UL is two times higher, namely 50 µg/day, and no exceedances were observed. The few studies that were used for setting the UL for infants (0–1 year) showed no adverse effects after a duration of about 4 months to 24 weeks [30]. This time period has a similar length as the period in our study at which some infants are at risk of vitamin D intakes above the UL.

An intake above the UL will, in most cases, not directly result in harmful health effects, as in setting the UL uncertainty factors are applied and the UL is generally set for health effects presenting after chronic high exposure. In addition, the risk of harmful health effects will be determined by the duration of high intakes and the actual amounts and varies from person to person [3, 31]. Moreover, the UL for infants is surrounded by uncertainties due to lack of data. This is illustrated by the higher ULs as set by IoM [2] compared to EFSA for children aged 7 months to 3 years. In our study, the 99th percentile of total vitamin D intake remained below the IoM ULs for all infants. Given the above arguments, the results of this study are not directly alarming, but do call for studies on the relationship between high vitamin D intakes during infancy and adverse health effects. In addition, we advise to be alert on potential adverse health effects that may be related to excessive vitamin D intakes in infants and monitor the habitual vitamin D intake regularly.

Other studies have shown that it is not uncommon that fortified foods contain higher amounts of micronutrients than declared on the package [32, 33]. This so-called overage is applied to ensure adequate levels of the nutrient at the end of the shelf life. We are aware of few published data regarding vitamin D overages in fortified foods or dietary supplements. In 1992, vitamin D overages were observed in infant formula in USA [34]. Seven of ten samples contained more than 200 % of the labelled amount. A more recent study in USA showed lower overage levels of 100–180 % of labelled values in infant formula [35]. In a New Zealand study, 12 of the 18 foods (i.e. baby food, food drink, margarine, milk product) had a measured vitamin D concentration that significantly deviated from the label claim [32]. For seven products, this was an overage. In the whole sample, the deviation ranged from −68 to +70 %. In a recent update of the vitamin D content of fortified foods and supplements in the UK Nutrition Survey Nutrient Databank, for each vitamin D level on the label a standard overage of 12.5 % was applied [36]. The overage at time of production was estimated to be on average 25 % (manufacturer and trade association: overages ranged 20–30 % for fortified foods and 20–40 % for supplements). The average remaining overage at time of consumption was assumed at 50 % of this 25 %. Taking into account, this overage of 12.5 % for all vitamin D fortified foods and supplements showed a 6 % increase in mean vitamin D intake in UK population aged 1.5 years and over. This proportion tended to be higher for the youngest age group (1.5–3 year), namely 9 %.

The importance of the fortified foods and supplements in the total vitamin D intake will influence the effect of potential overages on the intake. For infants in the Netherlands using vitamin D supplements as advised and in addition consuming fortified foods, the effect of overages on total intake might be substantial. If in this study the habitual total vitamin D intake from food and supplements will be increased due to overages by 9 % or even 12.5 % in children aged 7 months, the proportion with intakes above the UL would crudely be estimated at 30–35 % as compared to 11 % based on label information. However, even at high percentiles, the crude estimate of the habitual intake would be still below the UL of 37.5 µg/day as set by the IoM. It is recommended to study the overage (at time of consumption) in fortified foods and supplements, especially for products and/or nutrients contributing largely to the total nutrient intake.

In our scenario, all infants were assigned to a 100 % adherence to the vitamin D supplementation advice. Several studies showed that not all infants take supplements according to this advice [37‐40]. In the Netherlands in 2012, the vitamin D supplementation advice for young children was adapted to make it unambiguous. In the new advice, all young children, despite the amounts of infant formula consumed, are advised to take a daily vitamin D supplement. There is no insight into the current vitamin D intake from supplements among children 0–4 year old. In a recent study among children visiting day care (at least 2 and maximum 5 days per week), about 90 % of the parents reported to provide vitamin D supplementation to their infant (10 months to 3 year) [41]. It was unclear whether dosages were provided and whether the supplement was given daily. The risk of exceeding the UL may be an overestimation if not all children use a vitamin D supplement daily, as it was shown that especially the combined intake from foods and supplements resulted in intakes above the UL. On the other hand, our estimate may be an underestimation if children take more vitamin D from supplements than advised [40]. Our study shows the potential risk of exceeding the UL if all children would follow the supplementation advice as is warranted in combination with available fortified foods in a realistic consumption pattern.

The food consumption data used in this scenario study are from 2002. The response rate was high (82.5 %), and the data can be considered representative for the Netherlands at that time. Total energy intakes of the infants indicated that there was no major problem of under- or overestimation at the group level [19]. It is possible that the consumption pattern of infants changed during the last decade; however, no recent data are available for the Netherlands. Also for other European countries, these data are scarce [42]. To get better insight into the food consumption of infants, it is recommended to conduct a food consumption survey regularly among this age group, preferably also including the younger ages 0–6 months and those consuming breast milk.

In our scenario study, for infants aged 0–6 months a crude estimate of vitamin D intake was made based on energy requirement and body weight. It was assumed that these infants did not consume any complementary foods; however, part of the infants younger than 6 months of age may already start consuming these foods [43]. The vitamin D intake estimated in our scenario study may be an underestimation for infants aged 3–6 months. On the other hand, the foods often started with, for example, vegetables and fruit generally do not contain large vitamin D levels. It is therefore expected that this will not affect the estimates of the vitamin D intake largely.

Originally, the VIO study was combined with Dutch food composition data from 2001. For our study, the food consumption data were combined with more recent Dutch food composition data from 2011, to reflect better the actual vitamin D intake. Based on a quick scan of the current market (2015) using similar resources as in 2011, it can be concluded that the current market of vitamin D fortified foods and supplements is comparable with the situation in 2011 (data not shown). As information was collected from several sources, we expect this inventory to be rather complete. For some foods, the highest level in a product group was applied what may lead to an overestimation. In general, the contribution of vitamin D fortified foods, other than infant formula, was low; therefore, the effect of this potential overestimation on the total vitamin D intake is considered minor. To improve the estimation of vitamin D intake, it is recommended to have an up-to-date overview of the vitamin D fortified foods available on the Dutch market including the vitamin D levels, for example, by notification.

In this paper, we presented a scenario study. Scenario analyses are theoretical exercises only. But they provide insight into changes in the exposure distribution, where otherwise no rapid quantitative insight could be given. A number of assumptions are always needed to build scenarios, and it is important to interpret the results always in the light of these assumptions. The results are most predictive with realistic assumptions. From a policy making point of view, scenario analyses are very relevant as it may give at least some insight into what may the population impact of a (potential) change in policy. For instance, will the target population receive the adequate intake level with a specific fortification practice and is there no or limited risk of excessive intakes in the whole population. In addition, signals from scenario studies may be used to get the scope for additional research required to get more precise answers whether there is a health problem [44, 45].

A strength of this study was that the habitual total vitamin D intake from all potential vitamin D sources was estimated using a model that could cope with the multimodality of the data, namely a first-shrink-then-add approach. By splitting up the consumption data in vitamin D from infant/follow-on formula and vitamin D from other sources, two uni-modal distributions were created for which the habitual intake distribution could be estimated without violation of the model assumptions. It is important that the model assumptions are met, otherwise the habitual intake distribution may be estimated invalidly (especially at the tails), and as a result, the proportion at risk of excessive intakes may be invalid [24, 26]. This approach may be applied in research facing multimodal distributions which could be identified by splitting up the data in different sources of intake, such as voluntary fortification practices and intake from dietary supplements.