The Journal of Nutrition

Community and International Nutrition

Vitamin B-12 Status Is Associated with

Socioeconomic Level and Adherence to an
Animal Food Dietary Pattern in Colombian
School Children1–3
Eduardo Villamor,4* Mercedes Mora-Plazas,5 Yibby Forero,6 Sandra Lopez-Arana,7 and Ana Baylin8
4
Departments of Nutrition and Epidemiology, Harvard School of Public Health, Boston, MA 02115; 5Department of Nutrition, National
University of Colombia Medical School, Bogotá, Colombia; 6Nutrition Unit, National Institute of Health of Colombia, Bogotá,
Colombia; 7Braun School of Public Health and Community Medicine, Jerusalem, Israel 91120; and 8Department of Community Health,
Warren Alpert Medical School of Brown University, Providence, RI 02903

Abstract
Vitamin B-12 is related to neurocognitive function in school-age children, yet sociodemographic and dietary correlates of
vitamin B-12 status in this age group are not well characterized. The prevalences of vitamin B-12 or folate deficiencies in
Colombia are unknown. We conducted a cross-sectional study in a representative sample of 2800 low- and middle-income
children aged 5–12 y from Bogotá’s public schools. Plasma vitamin B-12 and erythrocyte folate concentrations (mean 6
SD) were 327 6 106 pmol/L and 858 6 256 nmol/L, respectively. The prevalence of vitamin B-12 deficiency (,148 pmol/L)
was 1.6% and the prevalence of marginal status (148–221 pmol/L) was 15.0%. Only 2 children had folate deficiency (,305
nmol/L). In multivariate analysis, mean vitamin B-12 concentrations significantly decreased with age and were 15 pmol/L
higher in girls than boys (95%CI ¼ 8, 23). Vitamin B-12 was inversely related to the mother’s parity and positively
associated with the amount of money spent on food per person per day at home and the household’s neighborhood
socioeconomic status (SES) classification. Folate concentrations were lower in girls than in boys and significantly
increased with the household’s SES. We identified 4 dietary patterns with principal components analysis of a FFQ in a
random subsample (n ¼ 972). Plasma vitamin B-12 was strongly, positively associated with a pattern that included
frequent intake of beef, chicken, and dairy products in a dose-response manner (P-trend, adjusted, ¼ 0.008). Low vitamin
B-12 status is not negligible in Colombian school children and is associated with poverty and marginal intake of animal food
sources. J. Nutr. 138: 1391–1398, 2008.

Introduction
affects all population groups. Between 11 and 39% of Chinese
Vitamin B-12 deficiency is related to serious adverse health adults were recently found to be deficient (4) and up to 36% of
outcomes, including neurological deficits, anemia, congenital infants (5) and 47% of adults (6) in India were reported to have
malformations, and hyperhomocysteinemia, a risk factor for low serum vitamin B-12 concentrations. High prevalences of
atherosclerosis (1–3). The sources of vitamin B-12 include animal vitamin B-12 deficiency have also been found in population
foods, fortified foodstuffs, and supplements; thus, vitamin B-12 groups from Latin American countries, including Mexico (7),
status could be compromised in strict vegetarians and among Guatemala (8,9), Cuba (10), Venezuela (11), and Chile (12). It is
persons living in low-income settings where foods are not fortified estimated that the prevalence of vitamin B-12 deficiency in this
and animal products are not frequently consumed due to limited region may be as high as 40% (13). Additional data are needed to
purchasing power. A growing body of evidence suggests that the further document the extent of this problem in the Americas.
prevalence of vitamin B-12 deficiency worldwide is large and Recent reports indicate that vitamin B-12 could play a
significant role on the neurocognitive development of school
1 children (14,15), yet the prevalence and determinants of vitamin
Supported by the David Rockefeller Center for Latin American Studies at
Harvard University, the City of Bogotá’s Secretary of Education, the National B-12 deficiency in this age group have not been widely
University of Colombia, and the National Institute of Health of Colombia. characterized. We conducted a study to examine the prevalence
2
Author disclosures: E. Villamor, M. Mora-Plazas, Y. Forero, S. Lopez-Arana, and of vitamin B-12 deficiency in a sample of school children from
A. Baylin, no conflicts of interest. Bogotá, Colombia, who represent low- and middle-income
3
Supplemental Tables 1 and 2 are available with the online posting of this paper
at jn.nutrition.org.
families. We aimed to determine the sociodemographic and die-
* To whom correspondence should be addressed. E-mail: evillamo@hsph. tary determinants of the plasma concentration of vitamin B-12
harvard.edu. in this population.
0022-3166/08 $8.00 ª 2008 American Society for Nutrition. 1391
Manuscript received 21 January 2008. Initial review completed 9 March 2008. Revision accepted 21 April 2008.
 Colombia adopted folic acid fortification of wheat flour in per day to never. Additional characteristics of the FFQ have been
1996. The prevalence of folate deficiency in the postfortification reported (17). Preliminary results of a validation study of this FFQ
era is largely unknown. If folate concentrations have been raised indicate that it is appropriate to rank individuals according to their
substantially but vitamin B-12 status is low, the consequences of dietary vitamin B-12 intake. In a subsample of 110 children, the crude
Pearson correlation coefficient between vitamin B-12 intake estimated
vitamin B-12 deficiency could be aggravated (16). We therefore
from the FFQ and vitamin B-12 intake estimated as the mean of multiple-
examined the correlates of folic acid status with the use of day 24-h recalls was 0.47.
erythrocyte folate measurements.
Laboratory methods. An aliquot of whole blood was hemolyzed by
dilution in a hypotonic aqueous solution of 1% ascorbic acid. Another
Methods
aliquot was centrifuged at 1500 3 g for 15 min and plasma was separated.
Study population. The study was conducted as part of a project on the Erythrocyte folate was measured on the RBC lysates and vitamin B-12
health and nutritional status of school children from low- and middle- on the plasma aliquot with the use of competitive chemiluminescent
socioeconomic strata in Bogotá, Colombia. Details of the study design immunoassay in an ADVIA Centaur analyzer (Bayer Diagnostics).
have been previously published (17). Briefly, in February 2006, we
enrolled 3202 children 5–12 y old from primary schools of Bogotá in a Statistical data analyses. Outcomes were plasma vitamin B-12 and
longitudinal study to examine the impact of school health initiatives on erythrocyte folate status. Children were classified as vitamin B-12
their health and nutritional status. The sample was randomly selected deficient when concentrations were ,148 pmol/L and marginally defi-
from the list of all classes (primary school grades 1–5) for that year in all cient if concentrations were 148–221 pmol/L (9). Folate deficiency was
public schools of the city. We used a cluster sampling technique with defined as erythrocyte folate concentrations , 305 nmol/L (19).
classrooms as sampling units. The study population is representative of Sociodemographic determinants of vitamin B-12 or folate status
low- and middle-income families from Bogotá, considering that the public included child’s age and sex; maternal characteristics, including age,
school system enrolls more than one-half of all primary school children years of formal schooling, marital status, and parity; and indicators of
in the city and 89% of them are of low- and middle-socioeconomic status the household’s SES, including the amount of money spent on food per
(SES) (18). person at home every day (the total amount of money spent on food
During wk 1 of classes, we sent a questionnaire to the parents divided by the number of people in the household), home ownership, and
inquiring about their age, marital status, education level, home owner- the household socioeconomic stratum according to the city’s classifica-
ship, socioeconomic characteristics of the household (including income tion of the neighborhood’s public services fees. We assessed child food
and money spent to purchase food), and food security status. The insecurity in the household with the use of an index developed from 5
questionnaire was returned by 2466 households, representing 81% of child-specific questions that were included in a previous food security
enrolled children’s families, after accounting for siblings. During the next survey (17). The survey was a modified version of the Spanish language
3 wk, trained research assistants visited the schools to obtain a fasting USDA Household Food Security Survey Module (20) and the Commu-
blood specimen by venipuncture from enrolled children. Approximately nity Childhood Hunger Identification project (21). Child food insecurity
4 mL of blood was placed in an EDTA Vacutainer tube following a in the household was considered to be present when 3 of the 5 questions
protocol to avoid hemolysis. On the same day of collection, the aliquots had an affirmative answer.
were transported on ice and protected from sunlight to the National Dietary determinants of vitamin B-12 and folate status included 4
Institute of Health in Bogotá, where all biochemistry analyses took dietary patterns that we identified with the use of principal component
place. analysis of the 38 items in the FFQ (17). An orthogonal transformation
Between May and June, trained dieticians applied a 38-item FFQ to a was used to rotate the factors obtained to achieve a simpler structure and
random sample of 1027 mothers to obtain information on the children’s facilitate interpretability. To determine the number of factors to retain,
usual dietary intake. The reference period to determine average intake we considered eigenvalues . 1, the Scree test, and interpretability. We
was the month prior to the interview. For each item, we described multiplied the standardized frequencies of intake for each food group by
reference portion sizes in natural units or standard measures for the factor score coefficients and the sum of these products was the score
commonly consumed servings in this population and inquired about for each derived factor. The patterns identified were: animal protein (e.g.
frequency of intake with a scale comprising 9 options, from 4 to 5 times beef/pork/veal/lamb, chicken/turkey, milk, cheese), cheaper protein (e.g.

FIGURE 1 Prevalence of marginal and deficient

vitamin B-12 status in low- and middle-income
school children from Bogotá, Colombia, according
to sex and age group. Values are percentages, n ¼
2660.

1392 Villamor et al.

cow tripe/liver, spleen, chicken giblets), traditional/starch (e.g. rice, tions among siblings in the sample. We estimated adjusted differences in
potato, plantain), and snacking (e.g. candy, ice cream, packed fried vitamin B-12 or folate concentrations by categories of determinants by
snacks, soda, fruit punch). We also examined whether frequency of fitting multivariate models with the correlates that were significant in
intake of individual foods or food groups and vitamin supplements were univariate analyses at P , 0.10. We retained in the models the
significant correlates of vitamin B-12 or folate status. Analyses of dietary determinants that remained associated with each outcome at P , 0.05
determinants were restricted to the subsample of children from whom we or were considered to be biologically relevant. CI were built around the
collected the FFQ. adjusted differences with the use of robust variances, which do not rely
In univariate analyses, we estimated circulating vitamin B-12 and on the multivariate normality assumption (22). The effect of clustering
erythrocyte folate concentrations (means 6 SD) by categories of each by classroom was negligible and was excluded from the models for
determinant. We tested the significance of the associations between parsimony. All statistical tests were double-sided. A P-value of ,0.05
ordinal determinants and nutrient serostatus with the use of tests for was considered significant. Analyses were carried out with the use of the
trend. These tests were derived from linear generalized estimating Statistical Analysis Software SAS version 9.1 (SAS Institute).
equation (GEE) models in which vitamin B-12 or folate was the outcome The parents or primary care providers of all children gave written
and the determinant was a covariate representing the ordinal categories informed consent prior to enrollment. The study protocol was approved
of the correlate, introduced as a continuous variable. For dichotomous by the Ethics Committee of the National University of Colombia
determinants, we used the Wald test. We specified an exchangeable cor- Medical School and the Humans Subjects Committee at the Harvard
relation matrix in the models to account for within-household correla- School of Public Health.

TABLE 1 Plasma vitamin B-12 concentrations among school children from Bogotá, Colombia
according to sociodemographic characteristics

n1 Mean 6 SD P2 Adjusted difference (95% CI)3 P4

pmol/L
Child's sex ,0.0001 ,0.0001
Female 1328 336 6 108 15 (8, 23)
Male 1388 319 6 103 Reference
Child's age, y ,0.0001 ,0.0001
5–6 513 338 6 111 Reference
7–8 826 332 6 104 25 (217, 6)
9–10 1061 326 6 106 211 (222, 1)
11–12 260 292 6 97 236 (251, 221)
Mother's education 0.005
Incomplete primary (1–4 y) 191 317 6 105 —
Complete primary (5 y) 463 324 6 106 —
Incomplete Secondary (6–10 y) 607 328 6 109 —
Complete Secondary (11 y) 943 332 6 106 —
University ($12 y) 157 351 6 110 —
Mother's parity ,0.0001 0.0002
1 275 347 6 110 Reference
2 850 335 6 106 28 (223, 7)
3 696 329 6 107 212 (227, 3)
4 298 312 6 100 222 (240, 25)
$5 224 299 6 103 233 (253, 214)
Money spent on food per person/day at home, pesos5 ,0.0001 0.003
Q1: median 720 484 313 6 106 Reference
Q2: median 1250 421 325 6 104 5 (29, 19)
Q3: median 1880 635 329 6 106 8 (25, 21)
Q4: median 3125 505 345 6 102 21 (8, 35)
Household socioeconomic stratum6 ,0.0001 0.003
1 (lowest) 226 312 6 110 Reference
2 878 320 6 109 6 (210, 23)
3 1233 338 6 104 18 (1, 34)
4 57 349 6 95 26 (23, 55)
Child food insecurity in household 0.003
No 1939 332 6 107 —
Yes 383 314 6 101 —
1
Totals may be ,2800 due to missing values.
2
Test for trend when a variable representing the ordinal categories of each characteristic was introduced as a continuous determinant in a
univariate linear GEE model with plasma vitamin B-12 as the outcome. For child’s sex and child insecurity in the household, P-value is from
the Wald test.
3
From a multivariate linear general estimating equations model with covariates that included child’s sex and age, mother’s parity, daily
money spent on food per person, and the household socioeconomic stratum. An exchangeable correlation structure was used to account
for within-household correlation among siblings. Only estimates for the variables retained in the final model are presented.
4
Test for trend from the multivariate-adjusted model.
5
At the time of the study, the exchange rate was 1 USD ¼ 2326 Colombian pesos.
6
According to the city’s classification of neighborhoods’ public service fees.

Vitamin B-12 and folate in Bogotá, Colombia 1393

Results Sociodemographic correlates of vitamin B-12 and folate
status. The prevalence of marginal or deficient vitamin B-12
Among the 3202 children recruited, we collected a biological status was 17.2, 15.7, 15.6, and 23.5% in age groups 5–6, 7–8,
specimen in 2816 (88%). Results of vitamin B-12 or folate 9–10, and 11–12 y, respectively. Marginal or deficient status was
analyses were available in 2800 children, and this constituted more prevalent in boys than in girls between 7 and 12 y of age
the final sample for the present report. Dietary data were (Fig. 1). In multivariate analysis, plasma vitamin B-12 was
available in 972 of these children. Children who did not provide significantly higher in girls than boys and was inversely related
a blood sample were more likely to be female than those who to the child’s age and mother’s parity and positively associated
provided a sample (Supplemental Table 1). Among children who with the amount of money spent on food in the household and
provided a sample, those with dietary information were slightly the household’s socioeconomic stratum (Table 1). There was
younger and more likely to live in lower SES neighborhoods than evidence for dose-response associations between these socio-
children without dietary information. There were no other demographic determinants and vitamin B-12 concentrations.
differences between these groups. Child food insecurity was not significantly related to vitamin
Children’s age was 8.7 6 1.8; 49% were girls. Sixty-eight B-12 status after adjusting for other socioeconomic indicators.
percent of the children’s households had a daily per capita Maternal age, marital status, and home ownership were not
income ,2 USD and 75% spent ,1 USD on food per person associated with the children’s vitamin B-12 concentrations.
each day. Plasma vitamin B-12 concentration was 327 6 106 Erythrocyte folate concentrations were significantly lower in
pmol/L and erythrocyte folate was 858 6 256 nmol/L. The girls than in boys and were positively related to the household
prevalence of vitamin B-12 deficiency (,148 pmol/L) was 1.6% socioeconomic stratum (Table 2). Folate was not significantly
and marginal status (148–221 pmol/L) was 15%. Only 2 associated with child’s age. Folate and vitamin B-12 concentra-
children had erythrocyte folate concentrations ,305 nmol/L. tions were independent of each other (r ¼ 0.03).

TABLE 2 Erythrocyte folate concentrations among school children from Bogotá, Colombia
according to sociodemographic characteristics

n1 Mean 6 SD P2 Adjusted difference (95% CI)3 P4

nmol/L
Child's sex 0.008 0.004
Female 1335 844 6 235 229 (248, 29)
Male 1376 871 6 275 Reference
Child's age, y 0.58 0.39
5–6 520 852 6 316 Reference
7–8 817 858 6 246 9 (222, 41)
9–10 1058 863 6 333 15 (215, 46)
11–12 259 855 6 280 11 (232, 54)
Mother's education 0.12
Incomplete primary (1–4 y) 191 834 6 241 —
Complete primary (5 y) 463 855 6 241 —
Incomplete secondary (6–10 y) 615 860 6 300 —
Complete secondary (11 y) 934 864 6 236 —
University ($12 y) 156 870 6 232 —
Money spent on food per person/day at home, pesos5 0.19
Q1: median 720 pesos 492 852 6 235 —
Q2: median 1250 pesos 424 850 6 254 —
Q3: median 1880 pesos 628 865 6 233 —
Q4: median 3125 pesos 507 871 6 320 —
Household socioeconomic stratum6 ,0.0001 ,0.0001
1 (lowest) 226 840 6 199 Reference
2 872 833 6 237 29 (240, 21)
3 1236 878 6 275 39 (9, 70)
4 58 916 6 268 77 (4, 151)
Child food insecurity in household 0.02
No 1940 864 6 262 —
Yes 385 834 6 220 —
1
Totals may be ,2800 due to missing values.
2
Test for trend when a variable representing the ordinal categories of each characteristic was introduced as a continuous determinant in a
univariate linear GEE model with plasma vitamin B-12 as the outcome. For child’s sex and child insecurity in the household, P-value is from
the Wald test.
3
From a multivariate linear general estimating equations model with covariates that included child’s sex and age, and the household
socioeconomic stratum. An exchangeable correlation structure was used to account for within-household correlation among siblings. Only
estimates for the variables retained in the final model are presented.
4
Test for trend from the multivariate-adjusted model.
5
At the time of the study, the exchange rate was 1 USD ¼ 2326 Colombian pesos.
6
According to the city’s classification of neighborhoods’ public service fees.

1394 Villamor et al.

Dietary correlates of vitamin B-12 and folate status. We amount of money spent on food per capita was positively related
examined the associations between 4 dietary patterns and the to adherence to the animal protein pattern (adjusted score
plasma concentration of vitamin B-12. There was a strong, difference between extreme quartiles ¼ 0.19; 95% CI ¼ 0.01,
positive, linear association between the animal protein pattern 0.38; P-trend ¼ 0.003). A similar association was found with the
and vitamin B-12 concentrations (Fig. 2). Vitamin B-12 concen- household’s neighborhood socioeconomic classification (ad-
trations in the lowest and highest quartiles of adherence to this justed score difference between strata 3 or 4 and stratum
pattern were 306 and 335 pmol/L, respectively. After adjusting 1 ¼ 0.27; 95% CI ¼ 0.04, 0.50; P-trend ¼ 0.007).
for child’s age and sex, the difference in vitamin B-12 between Erythrocyte folate concentrations were not significantly
extreme quartiles was 26 pmol/L (95% CI ¼ 7, 45; P ¼ 0.008). associated with dietary patterns or with the frequency of intake
With further adjustment for SES indicators, including maternal of individual food items, including bread, breakfast cereal,
parity, money spent on food, and household’s socioeconomic legumes, or supplements.
stratum, the difference in vitamin B-12 concentrations between
extreme quartiles of animal pattern adherence was 18 pmol/L
Discussion
(P-trend ¼ 0.04); in this model, the associations between SES
indicators and vitamin B-12 were attenuated. We examined the status and correlates of plasma vitamin B-12
We next assessed the associations between the children’s and erythrocyte folate in school children from low- and middle-
frequency of intake of individual animal food sources or vitamin income families from Bogotá, Colombia. Although folate defi-
supplements and their vitamin B-12 concentrations. There were ciency was not documented, we found a nonnegligible 17%
positive, dose-response associations between the plasma con- prevalence of marginal or deficient vitamin B-12 status.
centrations of vitamin B-12 and the frequency of intake of meat, A number of sociodemographic factors were significant
dairy products, fish, and cow liver (Table 3). Vitamin B-12 determinants of vitamin B-12 status in this population. Vitamin
concentrations were also higher in children who reported B-12 concentrations were lower in boys than in girls and
regular intake of multivitamin-containing or any supplements. appeared to decrease with age. A similarly negative, cross-
Egg intake was not related to vitamin B-12 status. In a sectional association between plasma vitamin B-12 and age was
multivariate model that included child’s age, sex, and all animal reported among school children from rural Kenya (23). In older
food sources and supplements, only age, sex, and intake of meat adults, an impairment of vitamin B-12 status with age has been
and dairy products remained associated with vitamin B-12 explained through declined gastric acidity, athrophic gastritis,
concentrations. After further adjustment for the household’s and bacterial overgrowth that lead to lower absorption of
socioeconomic stratum and the amount of money spent on food protein-bound vitamin B-12 (24); however, these factors do not
per capita, the associations between meat or dairy intake and appear to play a significant role in the vitamin B-12 status of
vitamin B-12 were attenuated. The SES-adjusted differences in school-age children (25). Instead, the inverse relation between
extreme categories of meat or dairy intake were 19 pmol/L (95% male gender or age and vitamin B-12 status in our study could be
CI ¼ 25, 42; P-trend ¼ 0.10) and 28 pmol/L (95% CI ¼ 1, 54; interpreted in light of greater requirements for more rapid
P-trend ¼ 0.16), respectively. growth in boys than in girls and in older than in younger
We further examined whether associations between chil- children that are not being met with adequate intake of the
dren’s age, sex, or SES and intake of animal foods could partly vitamin. The estimated average requirements are greater for chil-
explain the observed associations of these variables with vitamin dren 9–13 y old (1.5 mg/d) than for those aged 4–8 y (1.0 mg/d)
B-12 status by fitting a linear regression model with the score of (26). Age differences in plasma vitamin B-12 could be due in part
adherence to the animal protein pattern as the outcome. to dietary intake variability in our study, because we found that
Whereas sex was not associated with intake of animal protein older children were less likely to adhere to an animal protein
(P ¼ 0.92), children 11–12 y of age were less likely to adhere to intake pattern.
this pattern compared with children 5–7 y of age (adjusted score Vitamin B-12 concentrations were strongly associated with
difference ¼ 20.36; 95% CI ¼ 20.62, 20.10). The daily indicators of SES, including maternal parity, the amount of

FIGURE 2 Plasma vitamin B-12 concen-

trations in low- and middle-income school
children from Bogotá, Colombia, according to
their adherence to 4 dietary patterns. Bars
represent means 1 SEM, n ¼ 237, 240, 235,
and 234 for quartiles 1–4, respectively.
Quartile 1 represents the lowest quartile of
pattern adherence and quartile 4 the highest
quartile of pattern adherence. The P-trend values
are derived from linear regression models in
which a term representing the quartiles of
each pattern was introduced as a continuous
variable.

Vitamin B-12 and folate in Bogotá, Colombia 1395

 TABLE 3 Plasma vitamin B-12 concentrations among school children from Bogotá, Colombia
according to dietary intake of animal source foods and supplements1

Frequency of intake2 n3 Mean 6 SD P4 Adjusted difference (95% CI)5 P6

pmol/L
Meat ,0.0001 0.04
,4 times per week 98 294 6 89 Reference
4–6 times per week 139 314 6 99 10 (215, 35)
Once per day 344 323 6 100 15 (26, 37)
$2 times per day 312 338 6 107 24 (1, 48)
Dairy ,0.0001 0.06
,4 times per week 58 279 6 83 Reference
4–6 times per week 48 304 6 102 10 (228, 48)
Once per day 273 324 6 95 30 (4, 56)
2 times per day 178 326 6 105 30 (1, 58)
$3 times per day 339 335 6 104 32 (5, 59)
Fish 0.01 0.16
,1 time per month 115 302 6 95 Reference
1–3 times per month 363 322 6 102 11 (210, 31)
Once per week 221 326 6 104 12 (211, 34)
$2 times per week 215 333 6 103 17 (27, 41)
Cow liver 0.005 0.08
Never 300 314 6 98 Reference
,1 time per month 212 316 6 102 2 (216, 20)
1–3 times per month 139 317 6 109 22 (223, 20)
Once per week 178 350 6 108 27 (8, 46)
$2 times per week 122 328 6 107 5 (217, 28)
Egg 0.54 0.12
#1 time per week 120 314 6 87 Reference
2–4 times per week 292 325 6 100 1 (219, 21)
5–6 times per week 130 323 6 122 26 (233, 21)
Once per day 331 328 6 103 24 (224, 16)
$2 times per day 97 318 6 102 225 (250, 1)
Multivitamin-containing supplement 0.05 0.31
No 595 317 6 101 Reference
Yes 182 335 6 106 9 (28, 27)
Any supplement 0.02 0.23
No 457 317 6 97 Reference
Yes 384 333 6 109 9 (26, 23)
1
In a random subsample of 972 children with information on dietary intake.
2
Frequency of intake refers to the following average portion sizes: meat including beef (30 g), chicken (30 g), cold cuts (20 g), and
hamburger or hot dog (30 g); dairy including milk (200 g), cheese (20 g), and yogurt (200 g); and fish including fresh water fish (25 g) and
canned tuna or sardines (20 g). Portion sizes for cow liver and egg were 30 g and 50 g, respectively.
3
Totals may be ,972 due to missing responses to specific food items in the FFQ.
4
Test for trend when a variable representing the ordinal categories of frequency of intake was introduced as a continuous determinant in a
univariate linear GEE model with plasma vitamin B-12 as the outcome. For multivitamin or any supplements, P-value is from the Wald test
for ‘‘yes.’’
5
From a multivariate linear GEE model with covariates that included child’s sex, age, frequency of intake of meat, dairy, fish, cow liver, and
supplements. An exchangeable correlation structure was used to account for within-household correlation among siblings.
6
Test for trend from the multivariate-adjusted model.

money spent on food at home, and a general ranking of the Meat is an expensive food item in this setting and the imposition
household’s stratum according to the neighborhood. These of cost constraints has been shown to decrease meat intake (27).
associations have not been consistently found in other settings. In support of this explanation are the strong, positive, linear
In Guatemalan school children, for example, parental income associations we found between adherence to an animal food
was lower in children with plasma vitamin B-12 , 162 pmol/L intake pattern and indicators of SES such as the daily amount of
than among children with higher concentrations, but this money spent on food per capita and the household’s neighbor-
difference was not significant (P . 0.05) (25) and there were no hood socioeconomic classification. Furthermore, the associations
associations with other SES indicators. Large household size was between SES indicators and vitamin B-12 concentrations were
inversely related to vitamin B-12 status in Guatemalan infants, attenuated when we introduced them in a model with the animal
but an index of economic status or the level of maternal education protein food pattern, suggesting that the latter is in the causal
were not (23). Among Venezuelan children, vitamin B-12 pathway between SES and vitamin B-12 status.
concentrations were not related to SES (11). The socioeconomic When we considered together the individual sources of
gap in the vitamin B-12 status of our population is likely to vitamin B-12 in this population, only meat (red meat and
represent high levels of inequity in access to animal food sources. poultry) or dairy remained associated with vitamin B-12 status,
1396 Villamor et al.
whereas multivitamin supplements or cheaper animal foods In conclusion, whereas folate deficiency is rare, marginal or
including cow liver or egg were not. One possible explanation is deficient vitamin B-12 status is moderately prevalent in low- and
that vitamin B-12 from eggs or liver has lower bioavailability middle-income school children from Bogotá, Colombia. Dietary
than vitamin B-12 from red meat or chicken (28,29). In addition, intake of animal foods, especially meat and dairy, appear to have
the frequency of intake of liver or multivitamin supplements was a major influence on the vitamin B-12 status of this population.
relatively low in this population. Our results are consistent with Inclusion of animal food sources in ongoing food assistance
the study of school children in rural Kenya, where increased programs at schools could improve the vitamin B-12 situation of
plasma vitamin B-12 was the only micronutrient response to children in Colombia. The policy of wheat flour micronutrient
supplementation with meat or dairy (30). These results suggest fortification in Colombia currently includes iron, thiamin,
that meat- or dairy-based programs may represent an effective riboflavin, niacin, and folate, but excudes vitamin B-12. Inclu-
way of improving the vitamin B-12 status of school children. sion of vitamin B-12 into the micronutrient mix for wheat flour
The associations between individual food sources and vitamin fortification needs to be considered. Future studies are
B-12 status are consistent with the linear relation we found warranted on the effects that suboptimal vitamin B-12 status
between adherence to an animal protein pattern and plasma may have on functional outcomes among school children,
vitamin B-12 concentrations. The difference in plasma vitamin including neurocognitive and school performance, anemia, and
B-12 between the highest and lowest quartiles of adherence to the morbidity.
animal protein pattern (26 pmol/L) was close to that between the
highest and lowest categories of frequency of meat intake (24
pmol/L) or dairy intake (32 pmol/L). An explanation for the
similarity is that the highest factor loadings in the animal protein Literature Cited
pattern were for dairy and red meat intake (Supplemental Table 1. Lindenbaum J, Healton EB, Savage DG, Brust JC, Garrett TJ, Podell
2). Both the dietary patterns and individual foods analytic ap- ER, Marcell PD, Stabler SP, Allen RH. Neuropsychiatric disorders
proaches offer advantages and are complementary. The dietary caused by cobalamin deficiency in the absence of anemia or macrocy-
patterns analysis accounts for interactive effects of nutrients in tosis. N Engl J Med. 1988;318:1720–8.
different foods (31) that may affect the bioavailability of vitamin 2. Ray JG, Wyatt PR, Thompson MD, Vermeulen MJ, Meier C, Wong PY,
Farrell SA, Cole DE. Vitamin B12 and the risk of neural tube defects in a
B-12. In addition, analyses of dietary patterns overcome the
folic-acid-fortified population. Epidemiology. 2007;18:362–6.
possibility of finding significant associations by chance alone
3. Quinlivan EP, McPartlin J, McNulty H, Ward M, Strain JJ, Weir DG,
when several foods or food groups are examined individually in Scott JM. Importance of both folic acid and vitamin B12 in reduction of
relation to nutrient status. On the other hand, analyzing the risk of vascular disease. Lancet. 2002;359:227–8.
relations between individual food sources and vitamin B-12 4. Hao L, Ma J, Zhu J, Stampfer MJ, Tian Y, Willett WC, Li Z. Vitamin
concentrations in multivariate models allows the identification of B-12 deficiency is prevalent in 35- to 64-year-old Chinese adults. J Nutr.
specific foodstuffs that strongly and independently contribute to 2007;137:1278–85.
vitamin B-12 status and which may be specifically considered for 5. Taneja S, Bhandari N, Strand TA, Sommerfelt H, Refsum H, Ueland
PM, Schneede J, Bahl R, Bhan MK. Cobalamin and folate status in
supplementation programs. infants and young children in a low-to-middle income community in
We did not find folate deficiency in this population. Folate India. Am J Clin Nutr. 2007;86:1302–9.
fortification of wheat flour (1.54 mg/kg) was mandated in 6. Refsum H, Yajnik CS, Gadkari M, Schneede J, Vollset SE, Orning L,
Colombia in 1996. Although accessible published data on folate Guttormsen AB, Joglekar A, Sayyad MG, et al. Hyperhomocysteinemia
status in Colombia prior to fortification are unavailable, it is and elevated methylmalonic acid indicate a high prevalence of cobal-
amin deficiency in Asian Indians. Am J Clin Nutr. 2001;74:233–41.
possible to speculate that the lack of folate deficiency in Bogotá
7. Allen LH, Rosado JL, Casterline JE, Martinez H, Lopez P, Munoz E,
school children might be linked to the fortification policy. It may
Black AK. Vitamin B-12 deficiency and malabsorption are highly
also be due to high intake of folate-rich foods, including beans, prevalent in rural Mexican communities. Am J Clin Nutr. 1995;62:
but we did not find specific associations between dietary intake of 1013–9.
these foodstuffs and folate concentrations, perhaps because of 8. Casterline JE, Allen LH, Ruel MT. Vitamin B-12 deficiency is very
limited variability in the frequency of intake. Interestingly, despite prevalent in lactating Guatemalan women and their infants at three
the lack of deficiency, we did observe a gradient in folate status by months postpartum. J Nutr. 1997;127:1966–72.
household SES. This association could be indicative of unequal 9. Jones KM, Ramirez-Zea M, Zuleta C, Allen LH. Prevalent vitamin B-12
deficiency in twelve-month-old Guatemalan infants is predicted by
distribution of fortified products within the city and suggests that maternal B-12 deficiency and infant diet. J Nutr. 2007;137:1307–13.
folate deficiency could be a problem in other areas of the country 10. Arnaud J, Fleites-Mestre P, Chassagne M, Verdura T, Garcia Garcia I,
with limited access to quality-controlled, fortified wheat flour. Hernandez-Fernandez T, Gautier H, Favier A, Perez-Cristia R, et al.
Our study has some limitations. First, its cross-sectional Vitamin B intake and status in healthy Havanan men, 2 years after the
nature prevents us from establishing the temporal sequence of the Cuban neuropathy epidemic. Br J Nutr. 2001;85:741–8.
associations found and limits the possibility of drawing causal 11. Garcia-Casal MN, Osorio C, Landaeta M, Leets I, Matus P, Fazzino F,
Marcos E. High prevalence of folic acid and vitamin B12 deficiencies in
inferences. Second, we lacked functional indicators of vitamin
infants, children, adolescents and pregnant women in Venezuela. Eur J
B-12 deficiency such as elevated serum methyl-malonic acid; Clin Nutr. 2005;59:1064–70.
however, low plasma vitamin B-12 concentrations have been 12. Hirsch S, de la Maza P, Barrera G, Gattas V, Petermann M, Bunout D.
related to adverse hematological and neurocognitive outcomes in The Chilean flour folic acid fortification program reduces serum
children (15) and have been shown to increase in response to homocysteine levels and masks vitamin B-12 deficiency in elderly
supplementation (30). Thus, plasma concentrations of vitamin people. J Nutr. 2002;132:289–91.
B-12 are commonly used as an indicator of vitamin B-12 status in 13. Allen LH. Folate and vitamin B12 status in the Americas. Nutr Rev.
2004;62:S29–33.
population studies. Third, although dietary intake appears to
14. Allen LH, Penland JG, Boy E, deBaessa Y, Rogers LM. Cognitive and
play a major role in the vitamin B-12 status of this population, we neuromotor performance of Guatemalan schoolers with deficient,
did not have indicators to determine whether and to what extent marginal, and normal plasma vitamin B-12. FASEB J. 1999;13:A544.
malabsorption could be a risk factor for marginal or deficient 15. Louwman MW, van Dusseldorp M, van de Vijver FJ, Thomas CM,
status. Schneede J, Ueland PM, Refsum H, van Staveren WA. Signs of impaired

Vitamin B-12 and folate in Bogotá, Colombia 1397

 cognitive function in adolescents with marginal cobalamin status. Am J Kenyan school children is highly prevalent and improved by supple-
Clin Nutr. 2000;72:762–9. mental animal source foods. J Nutr. 2007;137:676–82.
16. Johnson MA. If high folic acid aggravates vitamin B12 deficiency what 24. Baik HW, Russell RM. Vitamin B12 deficiency in the elderly. Annu Rev
should be done about it? Nutr Rev. 2007;65:451–8. Nutr. 1999;19:357–77.
17. Isanaka S, Mora-Plazas M, Lopez-Arana S, Baylin A, Villamor E. Food 25. Rogers LM, Boy E, Miller JW, Green R, Rodriguez M, Chew F, Allen
insecurity is highly prevalent and predicts underweight but not over- LH. Predictors of cobalamin deficiency in Guatemalan school children:
weight in adults and school children from Bogota, Colombia. J Nutr. diet, Helicobacter pylori, or bacterial overgrowth? J Pediatr Gastro-
2007;137:2747–55. enterol Nutr. 2003;36:27–36.
18. Alcaldia Mayor de Bogota. Secretarı́a de Educacion. Estadisticas del 26. Institute of Medicine. Vitamin B12. Dietary reference intakes for
sector educativo de Bogota 2005 y avances 2006. [Accessed 2007 Oct 9] thiamin, riboflavin, niacin, vitamin B6, folate, vitamin B12, pantothenic
Available from: http://wwwsedbogotaeduco/secretaria/export/SED/ acid, biotin, and choline. Washington, DC: National Academy Press;
seducativo/compendio_estadistico/M_STADISTICASPDF. 2006. 1998. p. 306.
19. Institute of Medicine. Folate. Dietary reference intakes for thiamin, 27. Darmon N, Ferguson EL, Briend A. Impact of a cost constraint on
riboflavin, niacin, vitamin B6, folate, vitamin B12, pantothenic acid, nutritionally adequate food choices for French women: an analysis by
biotin, and choline. Washington, DC: National Academy Press; 1998. linear programming. J Nutr Educ Behav. 2006;38:82–90.
p. 201. 28. Doscherholmen A. Inhibition by raw eggs of vitamin B12 absorption.
20. Harrison GG, Stormer A, Herman DR, Winham DM. Development of a JAMA. 1978;240:2045.
spanish-language version of the U.S. household food security survey 29. Heyssel RM, Bozian RC, Darby WJ, Bell MC. Vitamin B12 turnover in
module. J Nutr. 2003;133:1192–7. man. The assimilation of vitamin B12 from natural foodstuff by man
21. Wehler CA, Scott RI, Anderson JJ. The Community Childhood and estimates of minimal daily dietary requirements. Am J Clin Nutr.
Hunger Identification Project: a model of domestic hunger: demon- 1966;18:176–84.
stration project in Seattle, Washington. J Nutr Educ. 1992;24: 30. Siekmann JH, Allen LH, Bwibo NO, Demment MW, Murphy SP,
S29–35. Neumann CG. Kenyan school children have multiple micronutrient
22. White H. A heteroskedasticity-consistent covariance matrix estima- deficiencies, but increased plasma vitamin B-12 is the only detectable
tor and a direct test for heteroskedasticity. Econometrica. 1980;48: micronutrient response to meat or milk supplementation. J Nutr.
817–30. 2003;133:S3972–80.
23. McLean ED, Allen LH, Neumann CG, Peerson JM, Siekmann JH, 31. Hu FB. Dietary pattern analysis: a new direction in nutritional
Murphy SP, Bwibo NO, Demment MW. Low plasma vitamin B-12 in epidemiology. Curr Opin Lipidol. 2002;13:3–9.

1398 Villamor et al.