Plasma vitamin B₁₂ concentration is positively associated with cognitive development in healthy Danish 3-year-old children: the SKOT cohort studies

Abstract

Adequate vitamin B₁₂ (B₁₂) and folate concentrations are essential for neural development in early childhood, but studies in well-nourished children are lacking. We investigated the relation between plasma B₁₂ and folate at 9 and 36 months and psychomotor development at 36 months in well-nourished Danish children. Subjects from the SKOT cohorts with B₁₂ measurement and completed Ages and Stages Questionnaire, 3rd edition (ASQ-3) at 36 months were included (n 280). Dietary intake, B₁₂ and folate concentrations were collected at 9 and 36 months, and ASQ-3 was assessed at 36 months. Associations between B₁₂ and folate at 9 and 36 months and ASQ-3 were analysed using regression models. Associations between diet and B₁₂ were also investigated. No children had insufficient B₁₂ (<148 pmol/l) at 36 months. B₁₂ at 36 month was positively associated with total ASQ-3 corresponding to an increase of 100 pmol/l B₁₂ per 1·5 increase in total ASQ-3 score (P = 0·019) which remained significant after adjustment for potential confounders including 9 months values. B₁₂ at 9 months or folate at any time point was not associated with total ASQ-3. Intake of milk products was associated with B₁₂ at 36 months (P = 0·003) and showed a trend at 9 months (P = 0·069). Intake of meat products was not associated with B₁₂. In conclusion, B₁₂ was positively related to psychomotor development at 3 years in well-nourished children, indicating that the impact of having marginally low B₁₂ status on psychomotor development in well-nourished children should be examined further.

The water-soluble micronutrients vitamin B₁₂ (cobalamin) and folate are necessary for the rapid growth and development during the early years of life⁽¹⁾. The relation between these micronutrients and cognitive and motor development is not fully understood, but the shared metabolism suggests that the status mutually affect the metabolism of the other⁽²⁾. Both are essential cofactors for RNA and DNA synthesis and required for development of and maintaining the nervous system. Vitamin B₁₂ participate in the folate-dependent conversion of homocysteine to methionine and the conversion of methylmalonyl CoA to succinyl CoA⁽³⁾. Deficiency leads to increased homocysteine levels and reduced supply of methyl groups and succinyl CoA that affects the production of myelin. Myelin is an essential component of brain development and maturation and is related to psychomotor development. Thus, vitamin B₁₂ deficiency may lead to disturbed myelin structure and demyelination of the central nervous system which may affect the process of learning and achieving motor abilities^(2,4,5) .

Vitamin B₁₂ is synthesised by micro-organisms and the primary sources of vitamin B₁₂ are animal source foods. Fruit and vegetables are key sources for folate⁽⁶⁾. In low- and middle-income regions with low intake of meat, milk and fish, poor vitamin B₁₂ status is widespread, especially among pregnant women and young children^(7,8) . Deficiency is less common among children with a typical Western diet, but vegetarians and especially vegans may be at risk^(9,10) . In future, this might be an issue to consider due to the increasing interest in sustainable food consumption in high-income countries. This will probably increase the intake of plant-based foods with a concurrent reduction of animal foods⁽¹¹⁾. Furthermore, the complementary feeding period with the transition from breast milk to the family diet may affect vitamin B₁₂ status. During this period, the need for especially vitamin B₁₂ is high, as the content in human breast milk and the child’s own depots are sparse^(2,12) .

Few studies have focused on the impact of vitamin B₁₂ status on psychomotor development among well-nourished children in high-income countries^(13–15) and not among representative healthy children during the first years of life. In low- and middle-income countries, the association between vitamin B₁₂ status and cognitive function has been assessed in more studies^{(4,10,16–19)} . This study aims to address the associations between vitamin B₁₂ and folate status in infancy and early childhood and cognitive and motor development in early childhood among healthy well-nourished children. We hypothesised that plasma vitamin B₁₂ and folate concentrations would be positively associated with psychomotor development during the first 3 years of life among healthy children. Furthermore, the associations between intake of animal source foods and vitamin B₁₂ and psychomotor development were examined as well.

Methods

Study design and participants

The present study used samples from the two comparable prospective observational cohorts SKOT I and SKOT-II. In SKOT-I, infants were randomly selected from the Copenhagen area using the National Danish Civil Registry⁽²⁰⁾, whereas in SKOT-II, infants were born to obese mother participating in the intervention study ‘Treatment of obese pregnant women’ (TOP) at Hvidovre Hospital⁽²¹⁾. The cohorts followed the same protocol from 9 months making it possible to pool data⁽²²⁾. Both cohorts have been described in detail previously^(20,22–25) . Briefly, inclusion criteria for both cohorts were healthy singleton full-term infants aged 9 months ± 2 weeks at the first examination and Danish speaking parents. The infants were 36 ± 3 months when they were monitored at the third and last examination. At 9 months, 311 and 166 infants participated in the SKOT-I and SKOT-II cohorts, respectively, whereas at 36 months, the numbers were 266 and 130, respectively. Data collection was carried out from 2007 to 2010 for SKOT-I and from 2011 to 2014 for SKOT-II. Written informed consent was obtained from all parents and legal guardians of the children. SKOT-I (H-KF-2007-0003) and SKOT-II (H-3-2010-122) were approved by The Committees on Biomedical Research Ethics for the Capital Region of Denmark.

Examinations at 9 and 36 months

The examinations were performed at the Department of Nutrition, Exercise and Sports, University of Copenhagen, Denmark, as described in detail previously^(22,23,26) . Length at 9 months and height at 36 months were calculated as the mean of three measurements and BMI as weight/length² or weight/height², respectively. Z-scores for weight, length, height and BMI were calculated using WHO growth standards as a reference and the WHO Anthro software⁽²⁷⁾. From background questionnaires and interviews, information regarding breast-feeding was obtained, and full breast-feeding was defined as only receiving breast milk, vitamins and water. Information regarding mothers’ education was also obtained and categorised as basic, short, medium or long.

Venous blood samples of 5 ml were taken after approximately 2 h fasting in lithium-heparin test tubes. Blood sampling was not successful in all children, so at 9 and 36 months, 401 and 331 blood samples were available, respectively. After preparation, blood samples were stored at −80°C. Vitamin B₁₂ and folate were analysed on an Immulite 2000 Xpi System analyser (Siemens Healthcare Diagnostics) and determined by competitive immunoassays using Immulite®2000 vitamin B₁₂ and Immulite®2000 Folic Acid kits with intra- and interassay CV of 4·6 and 9·6% for vitamin B₁₂ and 4·5 and 5·1 % for folate, respectively. Due to limited plasma volume, some samples had to be diluted twice for analyses, which led to a higher limit of detection for vitamin B₁₂ of 221 pmol/l for these samples. Samples below this value were coded as 111 pmol/l. Plasma ferritin was determined on an Immulite 1000 analyzer (Siemens Medical Solutions Diagnostics) using the chemiluminescent immunometric assay Immulite®Ferritin kit (Diagnostic Products Corporation) with an intra- and interassay CV of 3·4 and 3·7 %, respectively.

The definition of vitamin B₁₂ deficiency was set to <148 pmol/l⁽³⁾ and low vitamin B₁₂ status to <300 pmol/l⁽²⁸⁾. Low folate status was defined as <10 nml/l⁽²⁹⁾.

Diet

Dietary intake was recorded using a validated pre-coded food diary⁽³⁰⁾. Parents and caregivers recorded for 7 d except for SKOT-II at 36 months where 4 d recording was applied. Household measures and photo booklet for estimation of portion sizes were used as described elsewhere⁽³⁰⁾. Intake of energy and food items for each child was calculated using the software system GIES (version 1.000d; The National Food Institute). Intake of dairy products was calculated as the combined intake of milk, formula, other milk products and cheese. Human milk was not included, as the intake was not measured. Intake of meat products was calculated as the combined intake of meat, poultry and fish.

Assessment of psychomotor development

At 36 months, motor and cognitive development was assessed by the parents using the Ages and Stages Questionnaire, 3rd edition (ASQ-3) as described previously^(24,25) . Briefly, the parents were instructed by the project staff at the visit how to complete the questionnaire at home. It was recommended that the parents filled it out when the child was fed and rested. ASQ-3 is a standardised developmental checklist, which for 36 months old children consists of thirty age-appropriate questions divided into five subscale categories: communication, gross motor, fine motor, problem-solving and personal-social skills. The total ASQ-3 score was calculated by adding the scores for the subscales using only cases with completed questionnaires for all subscales. The possible score range for each subscale was 0–60 and hence the range for total ASQ-3 was 0–300.

Statistics

Characteristics are given as means and standard deviations or medians and interquartile ranges (IQR) for continuous variables, and for categorical variables as counts and percentages. Comparison between sex, cohorts or multivitamin supplementation was performed by independent t test or Mann–Whitney for parametric and non-parametric variables. Bivariate correlations between values at 9 and 36 months for vitamin B₁₂ and folate were conducted using Pearson’s correlation coefficient.

Associations between vitamin B₁₂, folate and ASQ-3 were assessed by multiple linear regression analysis using general linear models (GLM). Vitamin B₁₂ and folate at 9 and 36 months were exposures and total ASQ-3 and subscale scores were outcomes. The adjusted models included sex and age at ASQ-3 assessment. Fully adjusted models included adjustment for cohort, maternal education level and vitamin B₁₂ or folate value at 9 months for the corresponding 36 months analyses. In addition, models with vitamin B₁₂ were adjusted for ferritin and models with folate for BMI at 36 months as vitamin B₁₂ was correlated with ferritin and folate with BMI. One model also examined the interaction between sex and vitamin B₁₂ including a sex × vitamin B₁₂ term, which was removed if not significant. Standardised effect estimates were calculated to compare estimates between the different exposures and outcomes. Vitamin B₁₂ quartiles were applied in GLM models with vitamin B₁₂ as a categorical variable. As the range for ASQ variables was rather narrow and often left skewed, logistic regression models were also accomplished to assess consistency. The ASQ-3 score variables were categorised at the 25th percentile (≤ 25th percentile v. >25th percentile) and logistic regression analyses were performed using the same adjustments as for GLM.

The associations between dietary intake and vitamin B₁₂ and total ASQ-3, respectively, were investigated using GLM. Crude and models adjusted for sex, age and cohort were performed. At 9 months, many infants were still breastfed, which may have affected the milk intake. As the intake of human milk was not measured, the analyses of intake of dairy products at 9 months included a model with adjustment for current breast-feeding status as well. At 36 months, adjustment included intake of multivitamin supplements, which may affect plasma vitamin B₁₂ concentration. Difference in intake of dairy or meat products across vitamin B₁₂ quartiles was assessed by ANOVA.

Model assumptions for regression models and impact of outliers on estimates and significance were checked by residual plots and Cook’s distance, respectively. P values <0·05 were considered significant, whereas P values 0·05 to <0·10 as trends. All analyses were performed using IBM SPSS statistics (version 27.0; IBM).

Results

Of the 477 children included in the SKOT cohorts, 280 children had complete ASQ-3 and valid plasma vitamin B₁₂ measurements at 36 months. As these variables were of major interest, they defined the sample size. Of these, 278 children had folate measurements at 36 months, and at 9 months, 221 and 226 children had vitamin B₁₂ and folate measurements, respectively (online Supplementary Fig. 1). There was no significant difference in characteristics between children with or without ASQ-3 and vitamin B₁₂ measurements except that children included in the analyses were about 1 week younger at the 36 months examination than those not included (mean age (sd): 36·4 (1·02) months v. 36·7 (1·03) months, P = 0·028).

The characteristics of the children are shown in Table 1. No children had vitamin B₁₂ deficiency (<148 pmol/l) at 36 months. The number of children with vitamin B₁₂ deficiency at 9 months was not possible to evaluate due to dilution of samples with limited amount of material resulting in a higher detection limit of 221 pmol/l. Eight (4 %) children had concentrations below this value. One undiluted sample was <148 pmol (138 pmol/l). There were 37 (17 %) and 6 (2 %) children with low vitamin B₁₂ status (<300 pmol/l) at 9 and 36 months, respectively. Furthermore, no children had low folate status (<10 nmol/l) at 9 months and 2 (1 %) had at 36 months. The median (IQR) for total ASQ-3 was 270 (250–280), whereas medians for subscales ranged between 50 (40–60) and 60 (55–60). Girls had higher total ASQ-3 score compared with boys (275 (255–285) v. 260 (245–275); P ≤ 0·001), but there was no sex difference regarding vitamin B₁₂ or folate status (P ≥ 0·16). For vitamin B₁₂, there was a correlation between values at 9 and 36 months (r = 0·408, P ≤ 0·001) but not for folate (r = 0·052, P = 0·44).

Table 1. Characteristics of participants*

(Numbers and percentages; Mean values and standard deviations; median values and interquarticle range)

ASQ-3, Ages and Stages Questionnaire, 3rd edition.

* Values are expressed as mean and standard deviation (sd); median and interquartile range (IQR); n and %, as appropriate.

Vitamin B₁₂ and ASQ-3

Linear regression analyses showed that vitamin B₁₂ at 36 months was positively associated with total ASQ-3 score corresponding to an increase of 100 pmol/l vitamin B₁₂ per 1·5 increase in total ASQ-3 score (Table 2). Thus to obtain an increase of about 1 sd in total ASQ-3 score, an increase of approximately 2000 pmol/l vitamin B₁₂ would be needed. Adjustment attenuated the significance, but the associations remained significant also after adjustment for vitamin B₁₂ at 9 months, which was not significant in any of the models. Although girls had higher total ASQ-3 score than boys, there was no interaction between sex and vitamin B₁₂ (P = 0·12).

Table 2. Associations between vitamin B₁₂ and folate and ASQ-3 scores by linear regressions analyses*

(Coeffients values and 95 % confidence intervals)

ASQ-3, Ages and Stages Questionnaire, 3rd edition.

* Values are β-coefficients, 95 % CI, P-values.

† Model 1 is adjusted for sex and age at 36 months.

‡ Model 2 is adjusted for sex, age at 36 months, cohort, maternal education and value at 9 months for analyses at 36 months. Furthermore, vitamin B₁₂ analyses were adjusted for ferritin and folate analyses for BMI at 36 months.

§ n 221 and 224 for vitamin B₁₂ and folate both at 36 months, respectively.

In the ASQ-3 subscale analyses, vitamin B₁₂ at 36 months was significantly positively associated with problem-solving score and it remained significant after adjustment. Further, there was a positive trend for an association with the subscales communication and personal-social scores for vitamin B₁₂ at 36 months in the crude model, which became significant or vanished after full adjustment, respectively. Standardised effect estimates showed comparable estimates of the total ASQ-3 score and significant subscale scores (standardised estimates (CI)); total ASQ-3:0·17 (0·006; 0·33), problem-solving: 0·18 (0·011; 0·35) and communication: 0·19 (0·027; 0·35), fully adjusted models. There were no associations with the personal-social, gross or fine motor subscales.

Vitamin B₁₂ at 9 month was not associated with total ASQ-3 scores or subscale scores except for a positive association with personal-social score in the crude model.

To further investigate the impact of a low vitamin B₁₂ status at 36 months on total ASQ-3 score, vitamin B₁₂ was divided into quartiles in the GLM analyses. Overall, there was a trend for a difference between vitamin B₁₂ quartiles in relation to ASQ-scores (P = 0·072). Pairwise comparisons showed that the lowest vitamin B₁₂ quartile had lower ASQ score compared with the other quartiles ((first (lowest) v. second quartile: P = 0·036; first v. third quartile: P = 0·039 and first v. fourth (highest) quartile, P = 0·022)) (Fig. 1). When adjusting for sex and age, the overall trend for a difference between vitamin B₁₂ quartiles vanished (P = 0·160), but total ASQ-3 score in the lowest vitamin B₁₂ quartile was still lower compared with the highest quartiles (P = 0·036). Equivalent results were obtained after full adjustment.

Fig. 1. Total ASQ-3 score according to vitamin B₁₂ quartiles at 36 months (means, 95 % CI). *Significant different from other quartiles (P ≤ 0·039). ASQ-3, Ages and Stages Questionnaire, 3rd edition.

Logistic models for the total ASQ-3 scores using total ASQ-3 below or above the 25th percentile as outcome showed similar results. Per 100 pmol/l increase in vitamin B₁₂ concentration, the odds of not being in the lower quartile of total ASQ-3 score was 1·24 ((95 % CI: 1·07, 1·43), P = 0·004). Adjustment did not change the associations significantly (OR = 1·28 (95 % CI 1·05, 1·56), P = 0·016 for the full adjusted model). Contrary to the linear regression analyses, there were no associations between vitamin B₁₂ at 36 months and the subscales scores though a trend was seen for fine motor subscale in the fully adjusted model (OR = 1·21 (95 % CI 0·99, 1·47) for not being in the lower total ASQ-3 quartile per 100 pmol/l increase in vitamin B₁₂, P = 0·061). Vitamin B₁₂ at 9 months showed no associations with total ASQ-3 scales or any of the subscales.

Folate and ASQ-3

Folate was not associated with total ASQ-3 score or subscales at any time points in any of the models except for an association between folate at 9 months and communication subscale in the fully adjusted logistic regression model (OR = 1·02 (95 % CI 1·00, 1·03)) for not being in the lower total ASQ-3 quartile, per nmol/l increase in folate, P = 0·040).

Dietary intake and vitamin B₁₂

At 9 and 36 months, dietary intake was available for 272 and 262 children, respectively. The energy intake and intake of milk- and meat-related products are shown in Table 3. At 9 months, no infants were fully breastfed, while almost half of the children were still breastfed (47 %) (Table 1). Multivitamin supplements were only given to one child (0·4 %) at 9 month, whereas 226 children (81%) had received multivitamin supplements at 36 months. The vitamin B₁₂ concentration was significantly higher in the children receiving multivitamin supplement than those not receiving supplement (680 (250) pmol/l v. 540 (141) pmol/l, P ≤ 0·001.)

Table 3. Energy, dairy- and meat-related products intake in the children at 9 and 36 months

(Mean values and standard deviations; median values and interquarticle range)

At both 9 and 36 months, the intake of cheese was minimal compared with the milk intake, which included formula (Table 3). Likewise was the intake of poultry and fish low compared with the intake of meat. However, to investigate the associations between dietary intake and serum vitamin B₁₂ concentration, dairy products included all relevant milk products and meat products included all meat-related products. Intake of dairy products was positively associated with current vitamin B₁₂ concentration at 36 months. A 100 g increase in dairy products per d was associated with 30 pmol/l higher vitamin B₁₂ concentration ((95 % CI 0·11, 0·48), P = 0·002) (online Supplementary Table 1), equivalent to an increase of about 5 %. Adjustment for sex, age, intake of multivitamin supplements and cohort did not change the result significantly (P = 0·003). At 9 months, there was a trend for an association (β (95 % CI): 0·16 pmol/l per g/d (-0·02, 0·33), P = 0·082), but adjustment attenuated the association (P = 0·13) also when adjusting for breast-feeding status at 9 months (P = 0·18). Intake of meat products was not associated with vitamin B₁₂ at any time point (P ≥ 0·57). The intake of milk products across the vitamin B₁₂ quartiles was different (P = 0·011). The lowest vitamin B₁₂ quartile had lower intake of milk products than third and the fourth quartile ((mean (sd); P) 348 (141) g/d v. 421 (156) g/d, P = 0·007 and 348 (141) g/d v. 424 (140) g/d, P = 0·005, respectively). The intake of meat products did not differ across the vitamin B₁₂ quartiles (P = 0·64).

Dietary intake and ASQ

None of the diet variables were associated with total ASQ-3 at any time point (P ≥ 0·11). However, duration of full breast-feeding and breast-feeding status at 9 months were associated with total ASQ-3 score (P = 0·021, P = 0·033, respectively), but it disappeared after adjustment for sex, age, cohort and maternal education (P ≥ 0·11).

Discussion

In this study, including healthy Danish children with adequate vitamin B₁₂ concentrations, we found that plasma vitamin B₁₂ was positively associated with psychomotor development measured as total ASQ-3 at 36 month also after adjustment for essential confounders. The relation was consistent using logistic regression or vitamin B₁₂ concentrations categorised into quartiles. Furthermore, the total ASQ-3 score for children in the lowest vitamin B₁₂ quartile was about 10 scores lower compared the other quartiles. This could indicate that even though these children were not vitamin B₁₂-deficient, attention should be paid to children with low vitamin B₁₂ concentration. Though the effect size was relatively small, it is interesting that a positive association could be observed in this sample of well-nourished children which has not been reported previously. For the ASQ-3 subscales, positive associations were observed for the communication and problem-solving scores in the linear regression analyses but not in the logistic regression analyses. This may be due to the low range and variability of the responses and risk of ceiling effect in the subscales.

Although there was a highly significant correlation between vitamin B₁₂ at 9 and 36 months, vitamin B₁₂ status at 9 months was not associated with ASQ-3 at 36 months. Hence, in this study, only concurrent B₁₂ values were associated with psychomotor development though tracking from 9 months cannot be excluded.

Several studies have investigated the association between vitamin B₁₂ status and psychomotor development in children and adolescents but mainly in low- and middle-income countries^{(4,10,16–19)} or in special populations with risk of low vitamin B₁₂ status due to diet or medical issues in high-income countries^(13–15). In line with our results, a higher vitamin B₁₂ status was associated with improved cognitive and motoric development in many studies, but studies have also shown no associations^(18,31) . A Dutch observational study measured cognitive functions in adolescents who had been on a macrobiotic diet, which is close to a vegan diet and low in vitamin B₁₂, for the first 6 years of life⁽³²⁾. In early childhood, the children had reduced plasma vitamin B₁₂ concentration, impaired growth and psychomotor development compared with children on omnivorous diets. Follow-up examinations in adolescence revealed that these children underperformed in cognitive tests compared with adolescents fed on an omnivorous diet⁽¹³⁾. In a larger observational study from the USA (NHANES III), plasma vitamin B₁₂ was not associated with cognitive tests scores in children aged 6–16 years⁽³¹⁾. Compared with our study the children were older, the age range broader and tests for cognitive function were different, which may contribute to the different results.

Two Norwegian randomised studies investigated the effect of vitamin B₁₂ injection to high-risk groups with impaired vitamin B₁₂ function in infancy^(14,15) . After 1 month, infants treated with vitamin B₁₂ showed improved motor function. In our study, we did not observe any associations of motor development ASQ-3 subscales and vitamin B₁₂, but the difference in age might be important as associations between nutrition and motor development might be difficult to observe in later childhood⁽¹⁷⁾.

In a Nepalese observational study, vitamin B₁₂ status in infancy (2–12 months) was associated with cognitive development 5 years later⁽¹⁷⁾. ASQ-3 scores were assessed and in accordance with our results, the strongest association was found for total ASQ-3 score but also associations for problem-solving scores were found. In contrast, we did not find any long-term association between vitamin B₁₂ in infancy at 9 months and ASQ-3 scores at 36 months. Contrary to our study, different assessments of vitamin B₁₂ status were applied and Kvested et al. found most associations for other markers, that is, total homocysteine and methylmalonic acid, than total vitamin B₁₂ concentration. Total homocysteine and methylmalonic acid have been recognised as sensitive markers of low levels and mild deficiency reflecting status for metabolic function^(3,33) . In infancy, differences in breast-feeding might influence the vitamin B₁₂ status as some studies have showed lower vitamin B₁₂ status in breastfed infants^(7,17,34,35) . Studies from populations with a low vitamin B₁₂ status have shown associations between vitamin B₁₂ concentration and cognitive development among 12–18 months old children^(16,19) .

In addition, a few randomised trials have investigated the effect of vitamin B₁₂ on psychomotor development showing inconsistent results. A group of 6–30 months old children receiving vitamin B₁₂ supplementation for 6 months showed improved gross motor scores compared with placebo but, contrary to our study, there was no effect on total ASQ-3⁽⁴⁾. However, in the same study, where one of the four randomised groups received both folate and vitamin B₁₂, an effect was observed for gross motor and problem-solving and a trend for total ASQ-3 compared with placebo⁽⁴⁾. Contrary, no effect on psychomotor development of vitamin B₁₂ supplementation of 6–11 months old infants for 1 year was found in a Nepalese study⁽¹⁸⁾. However, the study group consisted of mildly stunted children who might suffer from insufficient levels of other nutrients and the highly selected group might reduce the generalisability of the results to populations with lower prevalence of micronutrient deficiencies.

Generally, the studies are challenging to compare as they vary in study design, motor and cognitive development tests, B₁₂ status assessment, age, sample size and setting of participants. Different specific developmental domains vary between tests and may not be directly comparable. In addition, the rapid development of the brain with different timing of regional brain growth spurts complicate comparison of results on developmental domains across ages.

We found no associations between folate and total ASQ-3 score. The association in the fully adjusted logistic model for the communication subscale might be a chance finding as no associations were noticed for the other models or for total ASQ-3, which has a broader range and variability, which increase the ability to detect associations as observed for the vitamin B₁₂ analyses. In the literature, divergent results have been reported. In the study by Kvestad et al., supplementation with folate alone did not improve total ASQ-3 and subscale scores⁽⁴⁾. However, a positive association between folate and cognitive scores in a cross-sectional study in 6–16 years old children from the USA was observed⁽³¹⁾ as well as in 12–18 months old Indian children provided that their vitamin B₁₂ concentration was above the 25th percentile⁽¹⁶⁾.

We found a positive association between the intake of dairy products and vitamin B₁₂ at 36 months but no association with meat intake. Though the intake of dairy products in the lowest vitamin B₁₂ quartile was around the recommended intake of 350 g/d⁽³⁶⁾, it was still possible to observe a difference in intake across the B₁₂ quartiles. The association was independent of intake of multivitamin supplements. This may indicate that dairy products especially milk, which constituted the main part of the dairy intake, probably is the key contributor to vitamin B₁₂ intake. This also conforms to the higher intake of dairy products compared with intake of meat products. At 9 months, many infants were still breastfed, which may underlie the missing association between milk intake and vitamin B₁₂ at this age. In many studies, animal source food and plasma vitamin B₁₂ concentrations were positively correlated^(3,37–39) , but in a study among healthy Norwegian 4–6 years old children, no associations between dietary intake and biomarkers for vitamin B₁₂ were found⁽⁶⁾. In accordance with our results, dairy products but not meat products were correlated with plasma vitamin B₁₂ concentration among 2 years old Norwegian children⁽³⁹⁾. The difference in results between dairy products and meat products may be that dairy products are the principal source of vitamin B₁₂ and that meat intake is low at this age. Furthermore, the bioavailability of vitamin B₁₂ from dairy products might also be higher as previously suggested in a study in adults⁽³⁷⁾.

A strength of the study is that the SKOT cohorts are healthy well-nourished children with detailed information on cognitive development and dietary intake of the first years of life. A limitation of the study is that only plasma vitamin B₁₂ (cobalamin) was measured as biomarker of vitamin B₁₂ status due to limited amount of material. It reflects long-term status and is not affected by recent intake. Additional biomarkers would have increased the sensitivity and specificity of vitamin B₁₂ status and increased the validity of the results. Another limitation is the reduced sample size as not all children had complete ASQ-3 assessment or blood samples. Nevertheless, as the only difference in baseline characteristics between children with or without valid B₁₂ measurements and total ASQ-3 at 36 months was a difference in age at 36 months of 7 d. It is unlikely that this has affected the results. Furthermore, we did not have measurement of intake of human milk at 9 months. There is a risk of residual confounding as this is an observational study and no causative conclusions can be made as well as associations should be interpreted with carefulness. We did not adjust for multiplicity as this was an explorative study and the risk of chance findings may occur. The results should therefore be confirmed in other studies preferentially including other biomarkers for vitamin B₁₂ status.

In summary, this study indicates that even in well-nourished children, a low vitamin B₁₂ concentration is associated with a lower score for psychomotor development. The growing concern for sustainable food production might increase the interest for plant-based diet and reduced intake of animal source foods, which may compromise the vitamin B₁₂ status. The current study indicates that special attention should be paid to the association between vitamin B₁₂ status and psychomotor development. Future studies should address how a sufficient vitamin B₁₂ status can be achieved, so optimal psychomotor development is secured.