Scientific opinion on the tolerable upper intake level for vitamin D, including the derivation of a conversion factor for calcidiol monohydrate

Abstract

Following two requests from the European Commission (EC), the EFSA Panel on Nutrition, Novel Foods and Food Allergens (NDA) was asked to deliver a scientific opinion on the revision of the tolerable upper intake level (UL) for vitamin D and to propose a conversion factor (CF) for calcidiol monohydrate into vitamin D₃ for labelling purposes. Vitamin D refers to ergocalciferol (vitamin D₂), cholecalciferol (vitamin D₃), and calcidiol monohydrate. Systematic reviews of the literature were conducted to assess the relative bioavailability of calcidiol monohydrate versus vitamin D₃ on serum 25(OH)D concentrations, and for priority adverse health effects of excess vitamin D intake, namely persistent hypercalcaemia/hypercalciuria and endpoints related to musculoskeletal health (i.e. falls, bone fractures, bone mass/density and indices thereof). Based on the available evidence, the Panel proposes a CF for calcidiol monohydrates of 2.5 for labelling purposes. Persistent hypercalciuria, which may be an earlier sign of excess vitamin D than persistent hypercalcaemia, is selected as the critical endpoint on which to base the UL for vitamin D. A lowest-observed-adverse-effect-level (LOAEL) of 250 μg/day is identified from two randomised controlled trials in humans, to which an uncertainty factor of 2.5 is applied to account for the absence of a no-observed-adverse-effect-level (NOAEL). A UL of 100 μg vitamin D equivalents (VDE)/day is established for adults (including pregnant and lactating women) and for adolescents aged 11-17 years, as there is no reason to believe that adolescents in the phase of rapid bone formation and growth have a lower tolerance for vitamin D compared to adults. For children aged 1-10 years, a UL of 50 μg VDE/day is established by considering their smaller body size. Based on available intake data, European populations are unlikely to exceed the UL, except for regular users of food supplements containing high doses of vitamin D.

Keywords: Ergocalciferol; calcidiol monohydrate; cholecalciferol; conversion factor; tolerable upper intake level.

Figure 1

Chemistry of vitamin D Ergocalciferol (a), cholecalciferol (b), calcidiol monohydrate (c), and 25‐hydroxyvitamin D₃ (d). Source: PubChem (CID 5280793, 5280795, 6441383 and 5283731)

Figure 2

The effect of calcidiol monohydrate on serum 25(OH)D concentration compared to vitamin D₃: (a) difference of means of achieved serum 25(OH)D concentration; (b) difference of means of achieved serum 25(OH)D concentration per μg/day of vitamin D; (c) ratio of means (ROM) of achieved serum 25(OH)D concentration per μg/day of vitamin D Duration of the intervention as reported by the authors. For age, recruitment target range is presented unless otherwise indicated. Achieved concentration refers to serum 25(OH)D concentration at the end of the treatment. Mean difference = mean serum 25(OH)D concentrations achieved with (a) calcidiol minus those achieved with vitamin D₃, (b) per μg/day of vitamin D, and (c) ratio of means. Abbreviations: CI, confidence interval; CH, Switzerland; CLIA, chemiluminescence immune assay; D₃, vitamin D₃; ELISA, enzyme‐linked immunosorbent assay; ES, Spain; F, females; HPLC–MS/MS, high‐performance liquid chromatography tandem mass spectrometry; IE, Ireland; IT, Italy; LC–MS/MS, liquid chromatography tandem mass spectrometry; M, males; MD, mean difference; NL, Netherlands; NR, not reported; RoB, risk of bias; ROM = ratio of means; S‐25(OH)D, serum 25‐hydroxyvitamin D; UK, United Kingdom; UV, ultraviolet; US, United States; 25(OH)D₃, 25‐hydroxyvitamin D₃, i.e., calcidiol. *From hospital admission to 3 months after discharge.

Figure 3

Relative bioavailability of calcidiol monohydrate [25(OH)D₃] compared to vitamin D₃: ratio of means achieved serum 25(OH)D concentration per μg/day of vitamin D administered Meta‐analysis was performed among: (a) RCTs using the same doses of calcidiol and vitamin D₃; (b) all RCTs and intervention arms available; (c) excluding 3 RCTs at high RoB (tier 3); (d) excluding studies using doses ~60 μg/day vitamin D₃ as control. ‘Mean’ and ‘SD’ refer to study means and standard deviations for the achieved S‐25(OH)D concentrations per μg/day of vitamin D, respectively. Abbreviations: CI, confidence interval; μg: micrograms; RCT, randomised controlled trial; ROM, ratio of means; SD, standard deviation; S‐25(OH)D, serum 25‐hydroxyvitamin D; Vit‐D₃, vitamin D₃; 25(OH)D₃, 25‐hydroxyvitamin D₃, i.e., calcidiol.

Figure 4

The effect of calcidiol monohydrate on serum PTH concentration compared to vitamin D₃: (a) achieved serum PTH concentration; b) achieved PTH concentration per μg/day of vitamin D; (c) ratio of means (ROM) of achieved PTH concentration per μg/day of vitamin D Duration as reported by the authors. For age, recruitment target range is presented unless otherwise indicated. Outcome measure: end concentration refers to S‐PTH concentration at the end of the treatment; change from baseline concentration refers to difference between end concentration and baseline concentration [end concentration minus baseline concentration]. Mean difference = mean S‐PTH concentrations achieved with calcidiol minus those achieved with vitamin D₃. Abbreviations: CI, confidence interval; CH, Switzerland; CLIA, chemiluminescence immune assay; D₃, vitamin D₃; EIA, enzyme immunoassay; ES, Spain; F, females; IE, Ireland; IT, Italy; M, males; MD, mean difference; NL, the Netherlands; RoB, risk of bias; S‐PTH, serum parathyroid hormone; UK, United Kingdom; 25(OH)D₃, 25‐hydroxyvitamin D₃, i.e., calcidiol. *From hospital admission to 7 months after discharge.

Figure 5

Distribution of vitamin D content in food supplements as displayed on labels in EU Member States and Norway (μg/serving) Source: Mintel GNPD. Search for vitamin D‐containing supplements available in the EU market in the last 5 years (from November 2017 to November 2022). A total of 2,150 products available in 24 EU Member States and Norway were identified, of which 2,098 contained complete data on μg/serving.

Figure 6

Mean, median, 5th and 95th percentiles of background vitamin D intakes in toddlers (≥ 1 year to < 3 years old), young children (≥ 3 years to < 7 years old), older children (≥ 7 years to < 10 years old), intakes in young adolescents (≥ 10 to < 14 years) and older adolescents (≥ 14 to < 18 years), by sex and country Estimates for females in orange and for males in blue. Squares correspond to medians and stars to means. Lines represent the range between the 5th and 95th percentiles. Estimated intakes from 5th and 95th percentiles are not presented when sample size is below 60 participants. Abbreviations: AT, Austria; BE, Belgium; BG, Bulgaria; CY, Cyprus; CZ, Czech Republic; DE, Germany; DK, Denmark; EE, Estonia; EL, Greece; ES, Spain; FI, Finland; FR, France; HU, Hungary; IT, Italy; LV, Latvia; NL, the Netherlands; PT, Portugal; RO, Romania; SE, Sweden; SI, Slovenia. *Country for which more than one survey was available; estimates presented in the plot are those of the most recent survey; when surveys covered the same period those, with the highest number of participants are displayed.

Figure 7

Mean, median, 5th and 95th percentiles of background vitamin D intakes in adults (≥ 18 years to < 65 years old) and older adults (≥ 65 years), by sex and country. Estimates for females in orange and for males in blue. Squares correspond to medians and stars to means. Lines represent the range between the 5th and 95th percentiles. Estimated intakes from 5th and 95th percentiles are not presented when sample size is below 60 participants. Abbreviations: AT, Austria; BE, Belgium; CY, Cyprus; CZ, Czech Republic; DE, Germany; DK, Denmark; EE, Estonia; EL, Greece; ES, Spain; FI, Finland; FR, France; HR, Croatia; HU, Hungary; IE, Ireland; IT, Italy; LV, Latvia; NL, the Netherlands; PT, Portugal; RO, Romania; SE, Sweden; SI, Slovenia. *Country for which more than one survey was available; estimates presented in the plot are those of the most recent survey; when surveys covered the same period those with the highest number of participants are displayed.

Figure 8

Mean, median, 5th and 95th percentiles of background vitamin D intakes in pregnant and lactating women, by country Squares correspond to medians and stars to means. Lines represent the range between the 5th and 95th percentiles. Estimated intakes from 5th and 95th percentiles are not presented when sample size is below 60 participants. Abbreviations: EE, Estonia; EL, Greece.

Figure 9

The effect of high doses of vitamin D on the odds of developing persistent hypercalcaemia in children and adolescents For age, recruitment target range is presented. Mean baseline S‐25(OH)D concentrations were classified as follows: < 25 nmol/L, 25–49 nmol/L, 50–74 nmol/L, ≥ 75 nmol/L. ‘Cases’ indicates the number of participants who developed hypercalcaemia. All effect sizes were estimated because pre‐calculated effect sizes were not available. Abbreviations: BMI, body mass index; F, females; IQR, inter quartile range; IR, Iran; LB, Lebanon; LK, Sri Lanka; M, males; S‐25(OH)D, serum 25‐hydroxyvitamin D; S‐Ca, serum calcium; OR, odds ratio; RoB, risk of bias; US, United States.

Figure 10

The effect of high doses of vitamin D on the odds of developing persistent (a) hypercalcaemia or (b) hypercalciuria in pregnant and lactating women For age, recruitment target range is presented. Mean baseline S‐25(OH)D concentrations were classified as follows: < 25 nmol/L, 25–49 nmol/L, 50–74 nmol/L, ≥ 75 nmol/L. ‘Cases’ indicates the number of participants who developed hypercalcaemia or hypercalciuria. ‘Ca’ indicates whether calcium was provided as a co‐intervention. Abbreviations: BD, Bangladesh; BMI, body mass index; MDIG, Maternal Vitamin D for Infant Growth; MN, Mongolia; NR, not reported; OR, odds ratio; RoB, risk of bias; S‐25(OH)D, serum 25‐hydroxyvitamin D; S‐Ca, serum calcium; U‐Ca, urinary calcium; US, United States. Note: In the RCT of Hollis and Wagner (2004), 100 μg/day of vitamin D (90 μg/day of vitamin D₂ + 10 μg/day of vitamin D₃) was compared with 50 μg/day of vitamin D (40 μg/day of vitamin D₂ + 10 μg/day of vitamin D₃). Effect sizes for Roth et al., (2018) were estimated because pre‐calculated effect sizes were not available.

Figure 11

The effect of high doses of vitamin D on the odds of developing persistent hypercalcaemia among general adult populations treated with vitamin D for (a) 8–26 weeks and (b) 1 year or more For age, recruitment target range is presented. Mean baseline S‐25(OH)D concentrations were classified as follows: < 25 nmol/L, 25–49 nmol/L, 50–74 nmol/L, ≥ 75 nmol/L. ‘Cases’ indicates the number of participants who developed hypercalcaemia. ‘Ca’ indicates whether calcium was provided as a co‐intervention. ‘Note’ indicate the effect size included in the forest plot. Pre‐calculated effect sizes were used if available without performing conversions; please note that effect sizes can be therefore different from the one expressed in the plot title. ‘Note2’ provides further information on the outcome. ‘Recurrent’ refers to the number of participants who demonstrated elevated S‐Ca at least at two measurement points after baseline. The zero cases refer to no occurrence of hypercalcaemia or to transient hypercalcaemia. Abbreviations: AR, Argentina; AU, Australia; BEST‐D: Biochemical Efficacy and Safety Trial of vitamin D; CA, Canada; CI, confidence interval; D2d: the Vitamin D and Type 2 Diabetes; DK, Denmark; F, females; M, males; NO, Norway; NR, not reported; S‐25(OH)D, serum 25‐hydroxyvitamin D; S‐Ca, serum calcium; SE, Sweden; OR, odds ratio; RoB, risk of bias; UK, United Kingdom; US, United States; ViDOS, Vitamin D supplementation in Older Women. Note: Aloia et al. (2013) used a 2 × 2 factorial design, but the results are shown for the groups with vitamin D and without vitamin D. *The effect size was computed based on case numbers because pre‐calculated effect sizes were not available; **Two study sites applied different cut‐off points.

Figure 12

The effect of high doses of vitamin D on the odds of developing persistent hypercalciuria among general adult populations For age, recruitment target range is presented. Mean baseline S‐25(OH)D concentrations were classified as follows: < 25 nmol/L, 25–49 nmol/L, 50–74 nmol/L, ≥ 75 nmol/L. ‘Cases’ indicates the number of participants who developed hypercalciuria. ‘Ca’ indicates whether calcium was provided as a co‐intervention. ‘Note’ indicates the effect size included in the forest plot. Pre‐calculated effect sizes were used if available, without performing conversions; please note that effect sizes can be therefore different from the one expressed in the plot title. ‘Note2’ provides further information on the outcome. ‘Recurrent’ refers to the number of participants who demonstrated elevated U‐Ca at least at two measurement points after baseline. The zero cases refer to no occurrence of hypercalciuria or to transient hypercalciuria. Abbreviations: AR, Argentina; CA, Canada; Ca, calcium; CI, confidence interval; F, females; IRR, incidence rate ratio; M, males; NR, not reported; OR, odds ratio; Rob, risk of bias; S‐25(OH)D, serum 25‐hydroxyvitamin D; U‐Ca, urinary calcium; U‐Ca/Cr, urinary calcium/creatinine ratio; US, United States. Note: Recurrent hypercalciuria cases for Gallagher et al. (2012, 2014b) were received from prof. Gallagher (personal communication).
*The effect size was computed based on case numbers because pre‐calculated effect sizes were not available.

Figure 13

The effect of vitamin D₂ or vitamin D₃ supplementation on the odds of sustaining at least one fracture For age, recruitment target range is presented, unless otherwise stated. Mean baseline S‐25(OH)D concentrations were classified as follows, unless otherwise stated: < 25 nmol/L, 25–49 nmol/L, 50–74 nmol/L, ≥ 75 nmol/L. ‘Cases’ indicates the number of participants who sustained at least one fracture during the intervention period. ‘Note’ indicates the effect size included in the forest plot. Pre‐calculated effect sizes were used if available without performing conversions; please note that effect sizes can be therefore different from the one expressed in the plot title. Abbreviations: AU, Australia; BEST‐D: Biochemical Efficacy and Safety Trial of vitamin D; CA, Canada; Ca, calcium; CI, confidence interval; F, females; HR: hazard ratio; M, males; OR, odds ratio; RECORD: Randomised Evaluation of Calcium Or vitamin D; RoB: risk of bias; S‐25(OH)D; serum 25‐hydroxyvitamin D; UK, United Kingdom. Note: Burt et al. (2019) provided calcium supplementation to participants with dietary intake of less than 1,200 mg per day. Flicker et al. (2005) started with a vitamin D dose of 36 μg/day [= 250 μg/week], but due to the discontinuation of the preparation of commercial 250 μg‐tablets during the intervention, they switched to 25 μg‐tablets. Hin et al. (2016) combined the results of the groups treated with 50 μg and 100 μg of vitamin D₃ . *Effect size was computed based on the reported case numbers because pre‐calculated effect sizes were not available.

Figure 14

The effect of vitamin D supplements on the odds of falling at least once or twice. For age, recruitment target range is presented, unless otherwise stated. Mean baseline S‐25(OH)D concentrations were classified as follows, unless otherwise stated: < 25 nmol/L, 25–49 nmol/L, 50–74 nmol/L, ≥ 75 nmol/L. ‘Fallers’ indicate the number of participants who encountered one fracture or more. ‘Note’ indicates the effect size included in the forest plot. Pre‐calculated effect sizes were used if available without performing conversions; please note that effect sizes can be therefore different from the one expressed in the plot title. ‘Note2’ indicates the adjusted effect size, if reported in the publication, or further information on the outcome. Abbreviations: AT, Austria; AU, Australia; aOR, adjusted odds ratio; BEST‐D: Biochemical Efficacy and Safety Trial of vitamin D; CA, Canada; CH, Switzerland; CI, confidence interval; DE, Germany; DEX; Vitamin D and Exercise in Falls Prevention; DO‐HEALTH: Vitamin D₃ ‐ Omega3 ‐ Home Exercise ‐ Healthy Aging and Longevity Trial; EX, exercise program; F, females; FA, fatty acids; FI, Finland; FR, France; HR, hazard ratio; M, males; NR, not reported; OR, odds ratio; PT, Portugal; RECORD, Randomised Evaluation of Calcium Or vitamin D; RoB, risk of bias; S‐25(OH)D, serum 25‐hydroxyvitamin D; STURDY, Study to Understand Fall Reduction and Vitamin D in You; UK, United Kingdom; US, United States; ViDOS, Vitamin D supplementation in Older Women. Note: Flicker et al. (2005) started with 250 μg/week [= 36 μg/day], but due to the discontinuation of the preparation of commercial 250 μg‐tablets during the intervention, they switched to 25 μg‐tablets. Hin et al. (2016) and Smith et al. (2017) combined the results of the groups treated with higher vitamin D₃ doses. Uusi‐Rasi et al. (2015) did not report raw data, i.e., the number of fallers. Bischoff‐Ferrari et al. (2022) adjusted for study site, sex, age, previous fall, baseline BMI, and baseline use of walking aids. Prince et al. (2008) adjusted for baseline height as difference was observed between groups. *Effect size was computed based on the reported case numbers because pre‐calculated effect sizes were not available.

Figure 15

The effect of high doses of vitamin D₃ supplements on the areal BMD at the (a) lumbar spine, (b) femoral neck and (c) total hip among general adult populations For age, recruitment target range is presented. Mean baseline S‐25(OH)D concentrations were classified as follows, unless otherwise stated: < 25 nmol/L, 25–49 nmol/L, 50–74 nmol/L, ≥ 75 nmol/L. ‘BMD Result/Change’ indicates the results/changes reported in the publications. ‘Ca’ indicates whether calcium was provided as a co‐intervention. Effect size was standardised mean difference because the outcome metrics were different, except for plot (b) where the effect size was standardised mean difference because all studies reported %‐changes from baseline. Abbreviations: BMD, bone mineral density; CA, Canada; CI, confidence interval; F, females; M, males; MD, mean difference; NO, Norway; LB, Lebanon; RoB: risk of bias; SMD, standardised mean difference; S‐25(OH)D; serum 25‐hydroxyvitamin D; US, United States; ViDOS; Vitamin D supplementation in Older Women. Note: In Rahme et al. (2017), the analysed test dose was close to 100 μg/day, which is why it was decided to include the study in the evidence synthesis.

Figure A.1

Flow chart for the selection of studies on bioavailability of calcidiol and vitamin D₃

Figure A.2

Flow chart for the selection of studies on hypercalcaemia and hypercalciuria

Figure A.3

Flow chart for the selection of studies on bone fractures, falls, BMD/BMC, and bone strength

Figure F.1

Scatter plot of the Ratio of Means (RoM) versus the achieved 25(OH)D per unit dose (nmol/L) of the comparator (vitamin D₃) Legend to Figure F.1: the colours represent studies; the shape indicates the dose of the control group (dot = 20 μg/d, triangle = 25 μg/d, square = 60 μg/d, cross = 62.5 μg/d); the size of the points is proportional to the sample size of the control group (vitamin D₃).

Figure F.2

Funnel plot of the 16 arms from the 10 eligible studies