DIRECTIONS: Please answer the following questions on a separate piece of paper. Include your name and CIN number on your homework assignment. Students are welcome to work in groups to complete homework assignments. However, each student must submit his/her own answers to receive credit.
All of the following questions are in relation to the following journal article which is available on Moodle: Parr CL, Magnus MC, Karlstad O, Holvik K, Lund-Blix NA, Jaugen M, et al. Vitamin A and D intake in pregnancy, infant supplementation and asthma development: the Norwegian Mother and Child Cohort. Am J Clin Nutr 2018;107:789-798.
A
and D intake in pregnancy, infant supplementation, and asthma development: the
rwegian Mother and Child Cohort
Christine L Parr,1,4 Maria C Magnus,1,5,6 Øystein Karlstad,1 Kristin Holvik,1 Nicolai A Lund-Blix,1,7 Margareta Haugen,2 Christian M Page,1 Per Nafstad,1,8 Per M Ueland,9,10 Stephanie J London,11 Siri E Håberg,1,3 and Wenche Nystad1
1Division of Mental and Physical Health; 2Department of Exposure and Risk Assessment; and 3Center for Fertility and Health, Norwegian Institute of Public Health, Oslo, Norway; 4Department of Nursing and Health Promotion, OsloMet–Oslo Metropolitan University, Oslo, Norway; 5Medical Research Council Integrative Epidemiology Unit, University of Bristol, Bristol, United Kingdom; 6Department of Population Health Sciences, Bristol Medical School, Bris tol, United Kingdom; 7Division of Pediatric and Adolescent Medicine, Department of Pediatrics, Oslo University Hospital, Oslo, Norway; 8Department of Community Medicine, University of Oslo, Oslo, Norway; 9Department of Clinical Science, University of Bergen, Bergen, Norway; 10Laboratory of Clini cal Biochemistry, Haukeland University Hospital, Bergen, Norway; and 11Epidemiology Branch, National Institute of Environmental Health Sciences, NIH,
Background: Western diets may provide excess vitamin A, which is potentially toxic and could adversely affect respiratory health and counteract benefits from vitamin D.
Objective: The aim of this study was to examine child asthma at age 7 y in relation to maternal intake of vitamins A and D during preg nancy, infant supplementationwith these vitamins, and their potential interaction.
Design: We studied 61,676 school-age children (born during 2002– 2007) from the Norwegian Mother and Child Cohort with data on maternal total (food and supplement) nutrient intake in pregnancy (food-frequency questionnaire validated against biomarkers) and in fant supplement use at age 6 mo (n= 54,142 children). Linkage with the Norwegian Prescription Database enabled near-complete follow up (end of second quarter in 2015) for dispensed medications to clas sify asthma. We used log-binomial regression to calculate adjusted
s (aRRs) for asthma with 95
CIs.
Results: Asthma increased according to maternal intake of to tal vitamin A [
equivalents (RAEs)]
highest (≥2031 RAEs/d) compared with the lowest
quin tile (aRR: 1.21; 95% CI: 1.05,
and decreased for total vitamin D in the highest (≥
µg/d) compared with the lowest (≤3.5 µg/d) quintile (aRR: 0.81; 95% CI: 0.67,
during pregnancy. No as sociation was observed for maternal intake in the highest quintiles of both nutrients (aRR: 0.99; 95% CI: 0.83, 1.18) and infant supple mentation with vitamin D or cod liver oil.
: Excess vitamin A (≥2.5 times the recommended in take) during pregnancy was associated with increased risk, whereas vitamin D intake close to recommendations was associated with a re duced risk of asthma in school-age children. No association for high intakes of both nutrients suggests antagonistic effects of vitamins A and D. This trial was registered at
http://www.clinicaltrials.gov
as NCT03197233. Am J Clin Nutr 2018;107:
789
–798.
Keywords: food-frequency questionnaire, dietary supplements, pregnant women, infants, vitamin A, vitamin D, pediatric asthma,
prescriptions, Norwegian Prescription Database, Norwegian Mother and Child Cohort
Asthma is currently among the top 5 chronic conditions con tributing to the global burden of disease in children aged 5–14 y (1). Unfavorable changes in diet have been hypothesized to in crease the susceptibility to asthma (2) and dietary exposures in utero and infancy could play a role, in particular for childhood onset of the disease (3).
Fat-soluble vitamins have a broad range of effects related to antioxidant properties (4), immune function (5), and lung devel opment (6). In particular, vitamin D has attracted much interest because of widespread deficiency in Western populations (7).
TheNorwegianMother and Child Cohort Study is supported by theNorwe gian Ministry of Health and Care Services and the Ministry of Education and Research, NIH/National Institute of Environmental Health Sciences (contract no. N01-ES-75558), and NIH/National Institute of Neurological Disorders and Stroke (grant nos. 1 UO1 NS 047537-01 and 2 UO1 NS 047537-06A1). This work was also supported by the Norwegian Research Council (grant no. 221097; to WN) and by the Intramural Research Program of the NIH, Na tional Institute of Environmental Health Sciences (ZO1 ES49019; to SJL).
The funders of the study had no role in study design, data collection, data analysis and interpretation, writing of the report, or the decision to submit the article for publication.
Supplemental Figure 1 and Supplemental Tables 1–8 are available from the “Supplementary data” link in the online posting of the article and from the same link in the online table of contents at
https://academic.oup.com/ajcn/.
Address correspondence to CLP (e-mail:
christine-louise.parr@fhi.no).
Abbreviations used: FFQ, food-frequency questionnaire; MoBa, Nor wegian Mother and Child Cohort Study; NorPD, Norwegian Prescription Database; RAE, retinol activity
Received June 13, 2017. Accepted for publication January 17, 2018. First published online April 20, 2018; doi:
https://doi.org/10.1093/ajcn/
nqy016.
in the public domain in the US.
789
Studies that used Mendelian randomization do not support that genetically lowered 25-hydroxyvitamin D is a risk factor for asthma (8). However, randomized trials (9, 10) and a meta analysis of birth cohort studies (11) suggest that prenatal vita min D supplementation above the regular dose (9, 10), and higher maternal circulating 25-hydroxyvitamin D (11), may reduce the susceptibility to asthma in the offspring, although follow-up of children to school age is not yet available in the trials.
Vitamin A deficiency poses a public health problem in parts of the world, but westernized diets may provide excess vitamin A (12–14) from increasing intakes of animal products and fortified foods and the use of dietary supplements. High dietary vitamin A has been associated with increased asthma severity in a murine model (15), but human studies are limited by potential toxic effects and a lack of feasible biomarkers for assessing adequate or subtoxic status (16). Observational studies, rather than trials, are therefore important to examine unintended health effects of vitamin A excess at the population level. Previous observational studies of vitamin A and asthma have mainly focused on the antioxidant properties of carotenoids (3) and have not included retinol, the most potent form of vitamin A. Vitamin A supplemen tation trials have been conducted in areaswith endemic deficiency (17, 18) where the effects on respiratory outcomes could differ from those in well-nourished populations due to differences in baseline vitamin A status (19). Few studies, to our knowledge, have examined the risk of child asthma in relation to prenatal con centrations of vitamin A, including retinol, outside of deficient populations (20, 21) or the importance of prenatal compared with early postnatal exposure. Furthermore, high vitamin A intake could potentially counteract the beneficial effects of vitamin D, due to competition for the nuclear retinoid X receptor (22).
Our objective was to investigate the association of maternal intakes of vitamins A and D during pregnancy, infant exposure to dietary supplements containing these nutrients, and potential nutrient interaction, with current asthma at school age when the diagnosis is more reliable than at earlier ages. Norway offers advantages for the study of high intakes of vitamin A during pregnancy because of a generally high intake from food sources in addition to the widespread use of cod liver oil as a dietary supplement.
The study included participants in the Norwegian Mother and Child Cohort Study (MoBa), a population-based pregnancy co hort (births during 1999–2009) administered by the Norwegian Institute of Public Health (23, 24). Women were recruited na tionwide (41% participation) at ∼18 wk of gestation when a pre natal screening is offered to all pregnant women. For the cur rent study we linked MoBa file version 9 (115,398 children and 95,248 mothers) with the Medical Birth Registry of Norway (hereafter referred to as the birth registry) and the Norwegian Prescription Database (NorPD), with follow-up to the end of the second quarter of 2015. The current study was registered at http://www.clinicaltrials.gov as NCT03197233. Eligible children (Figure 1) had available data on maternal dietary intake in preg nancy from a validated food-frequency questionnaire (FFQ) ad ministered at ∼20 gestational weeks and prescription follow-up for≥12mo from age 6 y (n= 61,676; born 2002–2007), of whom 89% (n = 55,142) had data on infant supplement use at 6 mo. We used a random subsample of 2244 births from 2002–2003 to compare maternal dietary intake with plasma concentrations of fat-soluble vitamins at 18 gestational weeks.
The MoBa study has been approved by the Norwegian Data Inspectorate (reference 01/4325) and the Regional Committee for Medical Research Ethics (refererence S-97045, S-95). All of the participants gave written informed consent at the time of enrollment. The current study was approved by the Regional Committee for Medical Research Ethics of South/East Norway.
Total (food and supplement) nutrient intakes during pregnancy were estimated from the FFQ, which queried about intake since becoming pregnant. The FFQ has been validated against a 4-d weighed food diary and with selected biomarkers (25, 26). To tal vitamin A (sum of total retinol and total β-carotene) was ex pressed as daily retinol activity equivalents (RAEs) per day by using the conversion factors 1 μg retinol (from diet or supple ments) = 12 μg β-carotene from diet = 2 μg β-carotene from supplements to account for differences in bioavailability (27). To tal vitamin D (micrograms per day) included vitamin D3 from foods and vitamins D2 and D3 from supplements. Nutrient intake was calculated by using the Norwegian Food Composition Ta ble (28) and a compiled database of dietary supplements, mainly based on themanufacturers’ information.Maternal plasma retinol and 25-hydroxyvitamin D2 and D3 were measured at Bevital AS laboratories in Bergen, Norway
(www.bevital.no),
in a single, nonfasting venous blood sample drawn at ∼18 wk of gestation. The frequency of infant supplement use (never, sometimes, or daily) was assessed from a follow-up questionnaire mailed at 6 mo of age. We analyzed the use of the following supplement categories containing vitamins A or D or both: vitamin D only (liquid oil-based formula), cod liver oil, multivitamins, and any vitamin D supplement, excluding multivitamins. The latter cat egory included vitamin D only, cod liver oil, and less common supplements (fish oil with added vitamin D, liquid vitamin A and vitamin D formula, vitamin D with fluoride, and other vitamin D combinations).
Weexamined current asthma in children at∼7 y of age, defined as having≥2 pharmacy dispensations of asthmamedication in the NorPD within a 12-mo interval, the first prescription being dis pensed between ages 6 and 7 y. Noncases were all children who did not meet these criteria. Asthma medications were inhaled β2 agonists, inhaled glucocorticoids, combination inhalers with β2 agonists and glucocorticoids, or leukotriene receptor antagonists.
Potential confounders and covariates were based on data from the birth registry (maternal age at delivery, parity, region of de livery, mode of delivery, child’s sex, birth weight, and gestational
FIGURE 1
age) or MoBa questionnaires completed at approximately gesta tional weeks 18 (inclusion), 20 (FFQ), and 30 and when the child was aged 6 mo.
Because cod liver oil and other omega-3 supplements con tribute to the intake of vitamins A and D in many MoBa women (13), we also evaluated maternal intakes of other nutrients pro vided by these supplements, including vitamin E (preservative, antioxidant) and long-chain n–3 fatty acids (EPA, docosapen taenoic acid, and DHA). In addition, we included vitamin C as a measure of fruit and vegetable intake (29), folate intake (30), and total energy intake. In sensitivity analyses, we also eval uated maternal zinc intake (3) and birth year to control for a potential cohort effect. To assess potential confounding by UV exposure in the analysis of vitamin D intake, we included leisure time physical activity (0, ≤1, 2–4, or ≥5 times/wk) and solar ium use (0, 1–5, or ≥6 total times) in pregnancy, geographical region of delivery within Norway (South and East, West, Mid,
North) as a proxy for latitude of residence, and season of deliv ery (January–March, April–June, July–September, or October– December). Maternal histories of asthma and allergic disorders (separate variables) were defined as ever reports at week 18 of asthma or hay fever, atopic dermatitis, animal hair allergies, or “other” allergies.
Many clinical practice guidelines recommend the use of di etary supplements, including multivitamins, to ensure adequate nutrient supply to low-birth-weight or premature infants (31). To adjust for child frailty, which could be related to both supple ment use (therapeutic or nontherapeutic) and later asthma suscep tibility, we included low birth weight (<2500 g), premature birth (gestational age <37 wk), and postnatal exposures in the
to full breastfeeding (number of months), respiratory tract infections (no or yes), and maternal smoking (no, sometimes, or daily) in the main analysis. In sensitivity analyses, we addition ally included child’s sex, birth season, cesarean delivery (no or yes), and use of paracetamol or acetaminophen (no or yes) and antibiotics (no or yes) in the first 6 mo.
We examined associations of maternal vitamin A and D intake during pregnancy (exposures) and infant supplement use (expo sures) with children’s asthma (outcome) by using log binomial regression. We calculated RRs with 95% CIs on the basis of ro bust cluster variance estimation and controlled for potential con founding by multivariable adjustment. The NorPD linkage en abled near-complete follow-up for asthma.
Our regression models were based on a directed acyclic graph for the hypothesized causal relations (Supplemental Figure 1). According to the graph, the effects of maternal intake and infant supplementation on children’s asthma can be estimated indepen dently when potential confounding factors and mediators are ad justed for. In the analysis ofmaternal intake (model 1), vitaminsA and D were mutually adjusted for (Spearman correlation of 0.53, continuous data), and we additionally adjusted for total intakes of other nutrients (vitamin E; sum of the n–3 fatty acids EPA, do cosapentaenoic acid, and DHA; vitamin C; and folate) and energy during pregnancy, maternal prenatal factors (age at delivery, par ity, prepregnancy BMI, education, history of asthma and atopy, and smoking in pregnancy), and birth weight and prematurity as potential mediators. In the analysis of infant supplementation (model 2), we mutually adjusted for the different supplements given and included all model 1 factors and postnatal child factors (months of full breastfeeding, child respiratory tract infections in the first 6 mo, and maternal smoking since birth).
val ues in individual covariates were <5% (Supplemental Table 1) and handled by multiple imputation by using chained equations (10 imputations). For 10.6% of the main study sample with miss ing questionnaire follow-up at age 6 mo (6534 of 61,676), we assessed the effect of imputing the infant supplement exposure data before performing multivariable adjustments.
All of the maternal nutrient intake variables were included as quintiles to account for a potential nonlinear association with children’s asthma. We tested for linearity by including the quin tile values (ordinal scale) as a continuous variable. To examine the potential interaction between vitamins A and D in the mother, we created a binary variable for high (highest quintile) compared with low (all lower quintiles) intakes of each vitamin and 4 mu tually exclusive exposure categories for the following combina tions: low vitamin A and low vitamin D, high vitamin A and low vitamin D, high vitamin D and low vitamin A, and high vitamin A and high vitamin D. To account for multiple supplement use in children, we created 6 mutually exclusive categories for daily or sometimes compared with never use of the following: 1) vitamin D only; 2) cod liver oil only; 3) multivitamin only; 4) any vitamin D supplement, including cod liver oil, combined with a multivi tamin; 5) multiple vitamin D supplements (e.g., vitamin D only combined with a fish-oil supplement containing vitamin D); and 6) none of the categories (reference).
In sensitivity analyses, we added more covariates to our main multivariable regression models, as described in Results, and we performed propensity score matching as an alternative method of controlling for potential confounding (32). We tested for multi plicative interaction between maternal intakes of vitamin A and vitamin D, taking potential nonlinearity into account by including all spline term combinations from restricted cubic spline models with 4 knots. We also assessed the potential influence of unmea sured confounding by using a recently published framework de veloped by Ding and VanderWeele (33). The significance level was 5% for all tests. The analyses were conducted in Stata 1
(StataCorp LP).
Participant selection is shown in Figure 1, and selected partic ipant characteristics are shown in Table 1 (mothers) and Table 2 (children). Characteristics were similar for the main study sam ple, the subsample with questionnaire follow-up at 6 mo, and the biomarker subsample (Supplemental Table 1).
Associations between maternal characteristics and dietary in take in pregnancy (n = 61,676) were generally in the same di rection for vitamins A and D. High intakes were associated with older age, higher education, primiparity, lower BMI, less smok ing, and supplement use (Table 1).
Supplementation with cod liver oil at age 6 mo was related to high maternal intakes of both vitamins A and D (Table 1) and was higher in children with positive health indicators (birth weight ≥2500 g, term birth, breastfeeding ≥6 mo, and no respiratory tract infections or postnatal maternal smoking) (Table 2). The use of multivitamins (percentage) was much higher among low–birth weight (45%) and premature (31%) children, indicating therapeu tic use according to clinical practice guidelines (31), and was as sociated with shorter breastfeeding and more postnatal maternal smoking (Table 2).
The prevalence of current asthma at age 7 y, based on prescrip tion registry data, was
% (2546 of 61,676). Children born to women in the highest compared with the lowest quintile of total vitamin A intake during pregnancy had a slightly higher preva lence of asthma (
% compared with 4.1%), and the adjusted RR was 20% higher (Table 3). We observed the lowest preva lence of asthma (3.6%) in the second quintile of total vitamin A (780–1102 RAEs/d) in which intake was close to, or slightly above, the public recommendation for pregnant women of 800 RAEs/d in Nordic countries (34), which is similar to other na tional recommendations (35). Relative to the second quintile, the adjusted RR of asthma was 32% higher (95% CI: 1.15, 1.51) in the highest quintile. The effect of total vitamin A (retinol and β carotene) was onlymarginally stronger than for total retinol. Total β-carotene showed a weak, but positive association with asthma after adjustment for total retinol. The adjusted RR for the high est (≥4007 µg/d) compared with the lowest (≤1360 µg/d) quin tile of β-carotene was 1.11 (95% CI: 0.98, 1.27) (Supplemental Table 2). The Spearman correlation between total retinol and to tal β-carotene (continuous data) was 0.12. A high intake of vita min A from food was not associated with asthma when the study sample was restricted to nonusers of retinol-containing supple ments (712 cases; n = 16,924). The adjusted RR was 1.05 (95% CI: 0.81, 1.36) for the highest (≥1462 RAEs/d) compared with
Distribution of maternal characteristics according to the lowest (
) and highest (Q5) quintiles of total vitamin A and D intake in pregnancy1
Vitamin A
Vitamin D3
| Q5 (≥2031 RAEs/d) | Q1 (≤3.5 µg/d) | Q5 (≥13.6 µg/d) | |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| 12,331 | 12,346 | 12,089 | 12,378 | ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| Maternal age at delivery, <25 y | 12.9 | 11.5 | 9.5 | ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| 25–30 y | 43.0 | 42.0 | 4 | 2.3 | 40.1 | ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| >30 y Previous children, % | 44.0 | 46.5 | 44.1 | 50.4 | |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| 0 1 | 43.5 36.8 | 46.3 34.6 | 38.7 39.1 | 4 | 9.2 | 3.2 | |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| ≥2 Maternal education, | 19.7 | 19.2 | 2 | 2.2 | 17.6 | ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| Less than high school High school | 9.5 33.4 | 8.3 30.3 | 10.7 35.4 | 6.5 27.0 | |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| ≤4 y of college | 38.6 | 40.8 | 37.7 | 41.8 | |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| >4 y of college Missing | 18.0 0.5 | 20.1 0.4 | 15.8 0.4 | 2 | 4.3 | ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| Maternal prepregnancy <18.5 | BMI (kg/m2), | 3.1 | |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| 18.5–24.9 25.0–29.9 | 57.8 25.0 | 64.9 20.3 | 56.0 | 67.9 19.5 | |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| ≥30 Missing | 11.8 3.0 | 9.1 2.6 | 12.9 3.1 | 7.1 2.4 | |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| Maternal smoking in No | pregnancy, % | 74.8 | 76.2 | 73.0 | 78.7 | ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| Stopped in pregnancy Yes | 15.6 9.6 | 15.8 8.0 | 16.0 11.0 | 14.7 6.6 | |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| <0.01 | 0.00 | 0.02 | |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| Maternal history of Maternal history of | asthma, % yes atopy, % yes | 7.2 29.9 | 8.0 32.5 | 8.0 30.0 | 7.2 32.1 | ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| Supplement use in | Cod liver oil | pregnancy, % yes | 12.4 | 31.8 | 70.2 | ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| Other n–3 supplement Multivitamin | 28.4 19.2 | 42.8 67.3 | 21.6 10.0 | 22.5 69.4 | |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| Folic acid | 39.7 | 75.9 | 32.3 | 75.7 | |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| Child supplement use Cod liver oil | at 6 mo (n = | 55,142), % yes | 40.9 | 51.4 | 38.4 | 59.3 | |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| Vitamin D drops Multivitamins | 24.5 9.1 | 24.8 | 8.9 | 23.2 9.6 | 2 | 3.9 | |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| 1n = 61,676. Q, | quintile; RAE, |
Q1 (≤779 RAEs/d) 12,331 12.9 43.0 44.0 43.5 36.8 19.7 9.5 33.4 38.6 18.0 0.5 2.3 57.8 25.0 11.8 3.0 74.8 15.6 9.6 <0.01 7.2 29.9 12.4 28.4 19.2 39.7
Q5 (≥2031 RAEs/d) 12,346 11.5 42.0 46.5 46.3 34.6 19.2 8.3 30.3 40.8 20.1 0.4 3.1 64.9 20.3 9.1 2.6 76.2 15.8 8.0 0.00 8.0 32.5 31.8 42.8 67.3 75.9 51.4 24.8 8.9
Q1 (≤3.5 µg/d) 12,089 13.6 42.3 44.1 38.7 39.1 22.2 10.7 35.4 37.7 15.8 0.4 2.2 56.0 25.9 12.9 3.1 73.0 16.0 11.0 <0.01 8.0 30.0 2.2 21.6 10.0 32.3 38.4 23.2 9.6
Q5 (≥13.6 µg/d) 12,378 9.5 40.1 50.4 49.2 33.2 17.6 6.5 27.0 41.8 24.3 0.4 3.2 67.9 19.5 7.1 2.4 78.7 14.7 6.6 0.02 7.2 32.1 70.2 22.5 69.4 75.7 59.3 23.9 7.1
the lowest (≤97 RAEs/d) quintile of food vitamin A intake (re sults not shown).
A high intake of vitamin D during pregnancy was associated with less-frequent asthma (3.9% compared with
% for the highest compared with the lowest quintile), and the adjusted RR was∼20% lower in the highest compared with the lowest quintile (Table 3). We observed no adverse effect of high vitamin A, or a protective effect of vitamin D, for intakes in the highest quintiles of both nutrients (Table 4).
The use of supplements containing retinol, including cod liver oil, was common (73% overall compared with 86% in the high est quintile). The median intake of supplemental retinol among users was ≥300 µg/d in the third through fifth quintiles of to tal vitamin A intake, indicating that many pregnant women take more than the standard daily dose of 250 µg, or combine multi ple supplements. However, food retinol contributed most to total vitamin A (Supplemental Table 3). The main food sources were sandwich meats, including liver spread, fortified margarine, and dairy products. In Norway, dairy products are not fortified with retinol. Low-fat milk is fortified with low amounts of vitamin D, but food intake of vitamin D varied little, and the use of supple mental vitamin D (76% overall compared with 99% in the highest quintile) was an important contributor to total vitamin D intake
In the biomarker subsample (n = 2244), maternal plasma vitamin D3 concentration increased across each quintile of total vitamin D intake (medians: 68, 72, 74, 75, and 82 nmol/L for the first through the fifth quintile, respectively; see Supplemental Table 4). The overall plasma-diet Spearman correlation
TABLE 2
Distribution of child characteristics according to any (sometimes or daily) postnatal supplement use in the first 6 mo1
| % n | Cod liver oil, % | Vitamin D drops, % | Multivitamin, % | ||
| Child birth weight, g <2500 | 1473 2.7 | 43.1 | 12.8 | 44.5 | |
| 2500–4500 ≥4501 | 51,223 92.9 2424 4.4 | 54.3 54.0 | 29.3 | 25.5 | 8.4 8.1 |
| Missing Preterm birth | 22 0.04 | 27.3 | 22.7 | ||
| No (≥37 wk gestation) Yes (≤36 wk gestation) | 52,346 94.9 2577 4.7 | 54.4 46.7 | 29.1 19.9 | 8.3 30.7 | |
| 219 0.4 | 49.8 | 22.8 | |||
| Months of full breastfeeding 0 | 658 1.2 | 47.6 | 18.2 | 13.2 | |
| 1 to <4 | 21,295 38.6 | 27.7 | 11.1 | ||
| 4 to <6 ≥6 | 25,354 46.0 7835 14.2 | 55.2 57.7 | 29.9 27.9 | 8.9 5.9 | |
| Respiratory tract infections | |||||
| first 6 mo No | 50,838 92.2 | 54.1 | 28.9 | ||
| Yes, not hospitalized Yes, hospitalized | 1645 3.0 1033 1.9 | 51.2 50.3 | 26.7 24.2 | 12.2 12.7 | |
| Missing Postnatal maternal smoking | 1626 2.9 | ||||
| No Some | 45,680 82.8 3095 5.6 | 54.9 51.1 | 29.7 28.4 | 8.8 9.6 | |
| Daily Missing | 7.5 4149 2218 4.0 | 46.3 54.4 | 21.0 21.6 | 14.8 11.3 | |
| 1n = 55,142. |
(continuous) for vitamin D varied with the season of blood draw, from 0.15 in summer to 0.32 in winter. Associations with indicators of UV exposure were in the expected direction (Supplemental Table 5): plasma vitamin D3 increased with leisure-time physical activity and tanning bed use in pregnancy and from North to South for geographical region of delivery. The maternal plasma retinol concentration (median: 1.64 µmol/L; IQR: 1.46–1.83 µmol/L) varied little with vitamin A intake
TABLE 3
(see Supplemental Table 3), also as expected, due to its strict homeostatic control.
Daily infant supplementation with vitamin D only or cod liver oil was not associated with the risk of asthma at school age. Daily use of multivitamins was associated with a 19% higher RR after
Total vitamin A and vitamin D intake in pregnancy and RR estimates (95% CIs) for current asthma at age 7 y1 Quintiles of intake
RR
| Total vitamin A (RAEs/d) Q1 (≤779) | 506/12,331 | 1 (ref) | |||||||||
| Q2 (780–1102) | 445/12,323 | 0.88 (0.78, | 1.00) | 0.92 (0.80, | 1.05) | ||||||
| Q3 (1103–1479) Q4 (1480–2030) | 475/12,331 520/12,345 | 3.9 4.2 | 0.94 (0.83, 1.03 (0.91, | 1.06) 1.16) | 0.99 (0.86, 1.08 (0.93, | 1.13) | |||||
| Q5 (≥2031) P-trend | 600/12,346 | 1.18 (1.05, <0.001 | 1.33) | 1.21 (1.05, 0.001 | |||||||
| Total vitamin D (µg/d) Q1 (≤3.5) | 531/12,089 | ||||||||||
| Q2 (3.6–5.7) | 485/12,487 | 0.90 (0.79, | 1.02) | ||||||||
| Q3 (5.8–8.6) Q4 (8.7–13.5) | 496/12,393 556/12,329 | 4.0 4.5 | 0.91 (0.81, 1.03 (0.91, | 1.03) 1.15) | 0.89 (0.77, 0.96 (0.82, | 1.03) 1.12) | |||||
| Q5 (≥13.6) P-trend | 478/12,378 | 0.88 (0.78, 0.46 | 0.99) | 0.81 (0.67, 0.03 |
1n = 61,676. RRs are from a log binomial regression model. Q, quintile; RAE, retinol activity equivalent; ref, reference. 2Adjusted for maternal total intakes of vitamins A or D (mutual adjustment), vitamin E, vitamin C, folate, and sum of n–3 fatty acids (all in quintiles) and
total energy intake (continuous); the following maternal prenatal factors: age at delivery (continuous), parity (0, 1, or ≥2), education (less than high school, high school, ≤4 y of college/university, or >4 y of college/university), prepregnancy BMI (kg/m2; <18.5, 18.5–24.9, 25.0–29.9, or ≥30), history of asthma (no
or yes), history of atopy (no or yes), and smoking in pregnancy (no, quit, or yes); and the following mediators: birth weight (<2500, 2500–4500, or ≥4500 g) and prematurity (no or yes). Missing values in covariates were handled by multiple imputation (m = 10) by using chained equations.
Combined effect of total vitamin A and vitamin D intake in pregnancy and RR estimates (95% CIs) for current asthma at age 7 y1 Total vitamin A (RAEs/d)
Cases/total n
Low (≤2030) High (≥2031) Low (≤2030) High (≥2031)
Total vitamin D (µg/d) Low (≤13.5) Low (≤13.5) High (≥13.6) High (≥13.6)
1687/41,903 381/7395 259/7427 219/4951
Prevalence, % 4.0 5.2 3.5 4.4
Crude RR
1 (ref) 1.28 (1.15, 1.43) 0.87 (0.76, 0.98) 1.10 (0.96,
Adjusted RR2
1 (ref) 1.21 (1.08, 1.36) 0.86 (0.73, 1.00) 0.99 (0.83, 1.18)
1n = 61,676. RRs are from a log binomial regression model. A high intake corresponds to the highest quintile (Q5) and low intake to all lower quintiles (Q1–Q4) in Table 3. Q, quintile; RAE, retinol activity equivalent; ref, reference. 2Adjusted for maternal total intake of vitamins A or D (mutual adjustment), vitamin E, vitamin C, folate, and sum of n–3 fatty acids (all in quintiles) and total energy intake (continuous); the following maternal prenatal factors: age at delivery (continuous), parity (0, 1, or ≥2), education (less than high school, high school, ≤4 y of college/university, or >4 y of college/university), prepregnancy BMI (kg/m2; <18.5, 18.5–24.9, 25.0–29.9, or ≥30), history of asthma (no or yes), history of atopy (no or yes), and smoking in pregnancy (no, quit, or yes); and the following mediators: birth weight (<2500, 2500–4500, or ≥4500 g) and prematurity (no or yes). Missing values in covariates were handled by multiple imputation (m = 10) by using chained equations.
multivariable adjustment (Table 5). However, there was no in creased risk for any (daily or sometimes) use of multivitamins in infants who were given an additional vitamin D–containing sup plement.
Results on maternal intake (Table 3) were robust to a range of sensitivity analyses including additional adjustment for total zinc intake, proxy variables for UV exposure during pregnancy (leisure-time physical activity, tanning bed use, and geographical region of delivery) in the vitamin D analysis, or birth year to
TABLE 5
control for a potential cohort effect (Supplemental Table 6). The results from the nonlinear analysis of multiplicative interaction were not significant (P-interaction from 0.59 to 0.94 in the multivariable model). Confounder adjustment by multivariable regression and propensity score matching gave similar results (Supplemental Table 7). From our main model (Table 3), we estimated the direct effect of maternal intake not mediated through low birth weight and prematurity; however, the total effect, not adjusting for these mediators, was similar (results not shown). Results on infant supplement use (Table 5) were little affected by additional adjustment for indicators of child frailty or asthma susceptibility (child’s sex, birth season, delivery by
Infant supplement use in the first 6 mo and crude and adjusted RR estimates (95% CIs) for current asthma at age 7 y1
| RR3 | RR3,4 | ||||||||
| No Sometimes | 1095/25,365 428/11,579 | 4.3 3.7 | 1 (ref) 0.86 (0.77, | 0.96) | 1 (ref) 0.91 (0.81, | ||||
| Daily Vitamin D only No | 721/18,198 1617/39,343 | 4.0 4.1 | 0.92 (0.84, 1 (ref) | 1.01) | 0.97 (0.87, 1 (ref) | 1.09) | |||
| Sometimes Daily | 152/3746 475/12,053 | 4.1 3.9 | 0.99 (0.84, 0.96 (0.87, | 1.16) 1.06) | 1.02 (0.87, 0.99 (0.90, | 1.19) 1.10) | 1.05 (0.89, 0.97 (0.86, | 1.23) 1.09) | |
| 2008/50,363 | |||||||||
| 81/2129 155/2650 | 3.8 5.9 | 0.95 (0.77, 1.47 (1.25, | 1.19) 1.72) | 0.97 (0.78, 1.45 (1.24, | 1.21) | 0.88 (0.71, 1.19 (1.01, | 1.10) 1.41) | ||
| Combined use (sometimes/daily) Neither category | 410/9397 | ||||||||
| Cod liver oil only Vitamin D only | 936/24,545 524/12,978 | 3.8 4.0 | 0.89 (0.80, 0.95 (0.83, | 1.00) 1.08) | 0.90 (0.80, 0.97 (0.85, | 1.01) 1.10) | 0.97 (0.86, 1.00 (0.88, | 1.09) 1.15) | |
| Multivitamin only Any vitamin D supplement and multivitamin | 149/2493 108/2541 | 6.0 4.3 | 1.40 (1.16, 1.00 (0.81, | 1.69) 1.23) | 1.39 (1.15, 1.03 (0.84, | 1.67) 1.27) | 1.19 (0.98, 0.94 (0.76, | 1.43) 1.15) | |
| Multiple vitamin D supplements | 126/3188 | 0.93 (0.76, | 0.99 (0.81, | 1.02 (0.83, |
1n = 61,676. ref, reference. 2RRs were from a log binomial regression model. Sample included participants with a follow-up questionnaire at 6 mo (n = 55,142). 3RRs were from a log binomial regression model. Analysis included all eligible children (n = 61,676) with child supplement use imputed for 10.6% of
the sample with missing follow-up at age 6 mo. Missing values were handled by multiple imputation (m = 10) by using chained equations.
4Infant supplements (vitamin D only, cod liver oil, multivitamins) were mutually adjusted for with additional adjustments for maternal total intake of vitamins A, D, E, and C; folate; sum of n–3 fatty acids (all in quintiles); and total energy (continuous); the following maternal prenatal factors: age at delivery (continuous), parity (0, 1, or ≥2), education (less than high school, high school, ≤4 y of college/university, or >4 y of college/university), prepregnancy BMI (kg/m2; <18.5, 18.5–24.9, 25.0–29.9, or ≥30), history of asthma (no or yes), history of atopy (no or yes), and smoking in pregnancy (no, quit, or yes); and the following postnatal child factors: birth weight (<2500, 2500–4500, or ≥4500 g), prematurity (no or yes), months of full breastfeeding (0, 1 to <4, 4 to <6, or ≥6 mo), child respiratory tract infections in first 6 mo (no or yes), and maternal smoking since birth (none, sometimes, or daily). Missing values in covariates were handled by multiple imputation (m = 10) by using chained equations.
cesarean section, and antibiotics and paracetamol use) or the exclusion of 5.5% (3399 of 61,676) of premature or low–birth weight children (Supplemental Table 8). Maternal and child risk estimates were also unaffected by the exclusion of 20% of controls (11,828 of 59,130) who had been dispensed any asthma medication by the age of 8 y (see Supplemental Tables 6 and 8).
Ding and VanderWeele’s (33) approach for assessing the po tential influence of unmeasured confounding showed that to completely explain an RR of 1.2 (as we observed for a high maternal intake of total vitamin A and for infant supplemen tation with multivitamins) it would take an unmeasured con founder with a strength ≥1.7, which is stronger than what we observed for all of ourmeasured confounders, except formaternal asthma.
In this large population–based pregnancy cohort study, a high maternal intake of vitamin A during pregnancy was associated with more asthma and a high intake of vitamin D was associated with less asthma in children at age 7 y, independent of infant sup plement use in the first 6 mo. The RR for intake in the highest compared with the lowest quintile was ∼20% higher for vitamin A and 20% lower for vitamin D. In agreement with the hypothesis that vitamin Amay antagonize actions of vitamin D, we observed no protective effect of vitamin D when the intake of vitamin A was high and likewise no adverse effect of high vitamin A in the face of high vitamin D. We found no protective effect of infant supplementation with vitamin D only, or cod liver oil, on asthma at school age.
Total vitamin A intake in the highest quintile (≥2031 RAEs/d), in which we observed more frequent asthma, corresponds to ∼2.5 times the recommended intake for pregnant women of 800 RAEs/d (34, 35). In comparison, the cutoff for the upper quintile of vitamin D (≥13.6 µg/d), in which we observed less asthma, was close to the Nordic (10 µg/d) and US (15 µg/d) rec ommendations for pregnant women.
Few other studies have assessed asthma development in school-age children in relation to pregnancy intake of vitamin A, including retinol, outside of populations at risk of deficiency. In a study from the Danish National Birth Cohort with half the sample size of the current study, the association of total vitamin A intake with the risk of asthma at age 7 y was only borderline significant (21), but the magnitude (8% higher risk per 1000-µg/d increase) is compatible with our finding of a 20% increased risk in the high est quintile. A study from Finland of maternal antioxidant intake during pregnancy showed positive, but nonsignificant, relations of total intake of carotenoids (α and β) and retinol from food (0.2% retinol supplement use was ignored) with child asthma at age 5 y (20). Our results support that intakes of β-carotene or food vitamin A alone (results shown in Supplemental Table 2) are not sufficient, or high enough, in most women, to increase asthma risk. Furthermore, vitamin A supplementation trials conducted in areas of Nepal with endemic deficiency reported better lung func tion in children of supplemented mothers (17, 18). A prospective study in Norwegian adults reported that daily intake of cod liver oil was associated with increased incidence of asthma (36). The authors attributed the association to the high retinol content of Norwegian cod liver oil at the time (1000 µg/5 mL before 1999), combined with a traditional diet rich in vitamin A. Thus, the risk of adverse effects of vitamin A appears to be greater in Western populations who consume supplemental retinol in the face of high food retinol intake.
A high intake of vitamin D from our FFQ was reflected in higher maternal circulating 25-hydroxyvitamin D, which has been associated with a lower risk of asthma in a recent meta analysis of birth cohort studies (11), including a case-cohort study in younger MoBa children (37). The findings of this review and our current study are in keeping with recent reports from 2 tri als of prenatal vitamin D supplementation, which suggest an in verse association between prenatal exposure to vitamin D and child asthma (38). Our results suggested that the protective ef fect of high vitamin D intake was attenuated among those with vitamin A intake in the highest quintile. Likewise, there was no adverse effect of high vitamin A intake when vitamin D intake was high. Other studies support that retinol and vitamin D may have antagonistic effects that affect health outcomes. A large, nested, case-control study of colorectal cancer found that the pro tective effect of high circulating vitamin D disappeared in sub jects with a high retinol intake (≥1000 µg/d) (39); however, vi tamin D may also reduce toxicity from retinol. In a review of case-reports of vitamin A toxicity, the median dose of retinol as sociated with toxicity was higher in cases who had also taken vitamin D (40).
Our study has several strengths. We used a validated FFQ and few previous studies have estimated the total intake of vitamin A from foods and supplements during pregnancy outside of de ficient populations (20, 21). In addition, we had high statistical power (2546 cases) to study asthma, and the prescription reg istry linkage enabled near-complete follow-up to school age. As in other large, nationwide, population-based studies, we were not able to classify asthma on the basis of clinical examination, and we cannot rule out somemisclassification in our asthma outcome. We expect that any bias in our RR estimates would be in the di rection of slight attenuation, because the risk of outcomemisclas sification should be low and independent of maternal exposure (nondifferential error). Norway has universal health care and pre scription coverage, so undiagnosed or untreated asthma should be rare. In addition, in a validation study of the MoBa 7-y question naire items with regard to asthma, we found that even a single dispensing of asthma medication was very rare in the absence of the maternal report of a doctor’s diagnosis of asthma (41). A pre scription for asthmamedication requires a physician’s evaluation, and we required ≥2 prescriptions to increase the positive predic tive value of our asthma definition (42). Furthermore, we would not expect highmaternal intakes of vitamin A and vitamin D to be associated with asthma in opposite directions, if high intakes just reflected differences in health consciousness and health-seeking behavior.
A limitation of this study is that we did not have data on nutrient intake from supplements in infants, but we were able to compare different supplements. Our results suggested more asthma among childrenwhowere givenmultivitamins but not cod liver oil. Both supplements provide similar doses of vitamin A (typically 200–250 µg) and vitamin D (typically the recom mended dose of 10 µg), and cod liver oil also contains vitamin E and n–3 fatty acids. A potential explanation for this difference is that liquid multivitamins for children contain water-miscible or emulsified retinol, which could be more toxic than retinol in oil based solutions such as cod liver oil (40). Interestingly, a Swedish study found an increased risk of asthma and allergy in infants supplemented with vitamins A and D in water-based but not oil based formula (43). In our study, the lack of association between any multivitamin use and the risk of asthma in infants who were given an additional supplement containing vitamin D could be explained by an antagonistic effect of vitamin D on retinol. How ever, it is also possible that these infants had a lower intake of multivitamins than infants in the multivitamin-only category due to alternating use of a vitamin D supplement. Other vitamins or minerals in a multivitamin formula, potentially folic acid, could also affect asthma development (30). Unmeasured confounding is always of concern in observational studies, but Ding and Van derWeele’s (33) framework provides some reassurance that even a modest RR of 1.2 is relatively robust to unmeasured confound ing. Last, we did not assess the potential influence of vitamin A and D exposures at other time points, such as during lactation or after age 6 mo, on our results.
Asthma is characterized by chronic airway inflammation and has been associated with atopy and a T-helper 2 (Th2)–dominated cytokine profile. Vitamin A exerts many of its effects through retinoic acid–mediated gene transcription, and retinoic acid may have a Th2 cell–promoting effect (44). Although vitamin A is mainly stored in the liver, excess vitamin A also accumulates in the lung (15), where retinoid metabolites may cause asthma like symptoms (45). In the rat lung, vitamin A supplementation with higher and intermediate doses increasesmarkers of oxidative stress (46), which also may impair lung function.We found no in dication that antioxidant properties of β-carotene protect against asthma. The effect of β-carotene was weaker but in the same di rection as retinol. However, many aspects related to thematernal– fetal transfer of retinoids and carotenoids, their metabolism in the developing tissues, and homeostatic control in the face of exces sive maternal dietary vitamin A intake are still poorly understood (47). Our results suggest that little, if any, of the effects of vitamin A and D intake during pregnancy on child asthma were mediated through low birth weight or prematurity. We found some indi cation that the adverse effects associated with excess vitamin A were mitigated by having a sufficient intake of vitamin D. This observation is in line with mechanistic studies in myeloid cells, which showed that vitamin D represses retinoic acid transcrip tional activity, but the action is 2 way, which also explains how vitamin A can attenuate vitamin D activity (22).
Conclusions
In this study, we found that a diet naturally high in vitamin A combined with the use of supplements containing retinol during pregnancy place women at risk of vitamin A excess, which was associated with increased susceptibility to asthma in school-age children. We observed this effect for intakes that were≥2.5 times the recommended dose, which is below the tolerable upper intake level for retinol of 3000 µg/d during pregnancy (27). Vitamin D intake close to recommendations was associated with a reduced risk of asthma at school age but not when maternal intake of vi tamin A was high. Thus, the balance of vitamin A and vitamin D intake during pregnancy could be of importance to asthma sus ceptibility in the offspring. A high intake of dietary retinol com bined with a low intake of vitamin D is seen in many Western populations (12) in which child asthma is common.
The authors’ responsibilities were as follows—WNandCLP:were respon sible for the study conception, design, and data acquisition; CLP, ØK,NAL-B, and MH: contributed to the data analysis; CLP: wrote the manuscript and had primary responsibility for the final content; and all authors: contributed to the interpretation of data, revised the manuscript for intellectual content, and read and approved the final manuscript. The authors had no conflicts of interest to disclose.
transrepresses retinoic acid transcriptional activity via vitamin
Association between pre-diagnostic circulating vitaminD concentration
D receptor in myeloid cells. Mol Endocrinol 2004;18(11): 2685–99.
and risk of colorectal cancer in European populations: a nested case- control study. BMJ 2010;340:b5500.
implications for analyses and interpretation. Am J Epidemiol 2005;161(10):987–94.
fetal transfer and metabolism of vitamin A and its precursor beta- carotene in the developing tissues. Biochim Biophys Acta 2012;
intake during pregnancy and childhood asthma in a population-based cohort. Am J Respir Crit Care Med 2017;195(2):221–8. 31. Metodebok i nyfødtmedisin 2012. [Handbook in neonatal medicine 2012.] Tromsø (Norway): University Hospital of North Norway; 2012 (in Norwegian).
32. Sturmer T, Joshi M, Glynn RJ, Avorn J, Rothman KJ, Schneeweiss
S. A review of the application of propensity score methods yielded increasing use, advantages in specific settings, but not substantially different estimates compared with conventional multivariable methods. J Clin Epidemiol 2006;59(5):437–47.
33. Ding P, VanderWeele TJ. Sensitivity analysis without assumptions.
Epidemiology 2016;27(3):368–77.
34. Nordic Nutrition Recommendations 2012: integrating nutrition and physical activity. Copenhagen (Denmark): Nordic Council ofMinisters;
