Provision of a School Snack Is Associated with Vitamin B-12 Status, Linear Growth, and Morbidity in Children from Bogotá, Colombia

Abstract

In 2004, Bogotá's Secretary of Education (SED) initiated a snack program in public primary schools. A midmorning food ration was provided free of charge to children to supplement 30 and 50% of their daily requirements of energy and iron, respectively. The purpose of this study, an observational investigation of 3202 children ages 5–12 y, was to examine whether the snack program improved children's nutritional and health status. We measured micronutrient levels (plasma ferritin and vitamin B-12, and erythrocyte folate), anthropometry, and reported morbidity during the first semester of the 2006 school year. After adjusting for socioeconomic status and other school interventions, children at schools receiving the snack (n = 1803) had greater increases in plasma vitamin B-12 (42 pmol/L; P < 0.0001) from baseline to 3 mo of follow-up than children at schools not receiving the snack (n = 1399). They also experienced a smaller decrease in height-for-age Z-scores than children who did not receive the snack (P = 0.001). Provision of the SED snack was associated with significantly fewer reported days with morbidity symptoms (e.g. cough with fever, diarrhea with vomiting), 44% fewer doctor visits (P = 0.02), and 23% fewer days of school absenteeism (P = 0.03). The snack was not related to ferritin or folate levels. In conclusion, provision of a school-administered snack was related to improved vitamin B-12 status and linear growth and decreased reported morbidity. Although provision of the snack was not related to BMI changes over a 4-mo period, snack components such as candy and sugar-sweetened beverages should be replaced with healthier options, as the rates of child overweight in Colombia are not negligible.

Introduction

Nutritional status is a strong determinant of health and neurocognitive performance of school-age children (1). Deficiencies of nutrients from animal source foods, such as iron and vitamin B-12, have been associated with impaired cognitive performance, increased susceptibility to infections, and stunted growth in this age group (2–4). In many developing countries, these deficiencies are highly prevalent, because animal source food intake is low due to limited purchasing power. In Bogotá, Colombia, e.g., up to 17% of school children have marginal vitamin B-12 status (5). School-based interventions that incorporate animal-source foods are effective at improving micronutrient status and health outcomes in such settings (6,7).

Several of these countries, however, are undergoing a nutrition transition through which they face a dual burden of malnutrition, with high rates of both undernutrition and overweight (8). Food-based interventions in such populations need to be evaluated both for their efficacy in improving nutritional status and for any potential adverse effects they might have on overweight or obesity (9).

In 2004, the city of Bogotá's Secretary of Education (SED)¹⁰ initiated a school snack program in public primary schools. A mid-morning food ration was provided free of charge to children to supplement 30 and 50% of their daily requirements of energy and iron, respectively. To examine whether the snack program improved children's nutritional and health status, we conducted an observational longitudinal study during the first semester of 2006 among 3202 children ages 5–12 y. We hypothesized that provision of the snack would be related to improved micronutrient status and growth and to decreased school absenteeism.

Methods

The school snack program

In 2004, the city of Bogotá's SED initiated a school snack program with the purpose of improving the nutritional status of children in public primary schools in the city. The program consisted of the daily provision, free of charge, of a mid-morning food ration to children enrolled in grades 0–5 (ages ∼5–12 y) at public primary schools in Bogotá. The snack was designed to provide 30, 50, and 40% of the recommended daily intake of energy (1600 kcal/d¹¹ for children <8 y and 1800 kcal/d for children ≥8 y), iron (9 mg/d for <8y and 13 mg/d for ≥8 y), and calcium (450 mg/d for < 8y and 525 mg/d for ≥ 8 y), respectively, according to the Colombian Institute of Family Welfare [Instituto Colombiano de Bienestar Familiar (ICBF)] (10,11). The snack consisted of a beverage, a cereal and/or protein component, and a “sweet” component (which also included peanuts and Petit Suisse cheese) arranged into 9 different daily menus; fruit was provided in 5 of the 9 menu days. Examples of the menu components and mean nutrient composition of the snack are provided in Supplemental Table 1. Nutrient values for the snack components were compiled using laboratory-analyzed data and food composition tables from Colombia (12) and the US (13,14).

The SED snack was administered daily during school days (Monday through Friday). It was distributed by the teacher ∼30 min before the mid-morning break and children were asked to consume it in the classroom under direct observation of the teachers. By the end of 2005, the school snack coverage reached ∼70% of the ∼470,000 children who were enrolled in the public primary school system (10). Primary schooling coverage in Bogotá was 88% in 2005. The public school system enrolled 57% of all primary school children in the city at the time, the majority of whom (89%) belonged to low-and middle-socioeconomic strata (15).

Study design and population

We conducted an observational longitudinal study to examine whether participation in the school snack program was related to children's nutritional and health status. In February 2006, a random sample of 4000 children enrolled in the public primary school system was selected using a cluster sampling strategy. Clusters were defined as the classes (primary school grades 0–5) of all 361 public schools in the city at the time. The sampling units were the classrooms (n = 8500) and 166 were randomly selected to reach the target sample size. During the first week of class in February 2006, consent was obtained from the parents to participate in the study and responses were obtained for 3207 children. The parents of 5 children declined participation; thus, the final sample size was 3202 children from 3032 households (after accounting for siblings). The total number of schools with children participating in the study was 38 and 25 (66%) of these schools were covered by the SED snack program. The program was first introduced in poorer school districts. We compared the outcomes of children who received the SED snack with those of children who did not receive it during the first academic semester of 2006 (about February to June). A longer term comparison was not possible, because after July, coverage was extended to the schools that were not receiving the snack by the start of the year. The research protocol was approved by the Ethics Committee of the National University of Colombia Medical School. The Institutional Review Board at the Harvard School of Public Health approved the use of data from the study.

Field procedures

At the time of enrollment, information on the parents' sociodemographic characteristics, including age, marital status, education level, and indicators of the household socioeconomic status, was obtained using a self-administered questionnaire that was completed and returned by 2466 households, representing 81% of enrolled children's families (2637 children, after accounting for siblings).

Baseline biochemical and anthropometric measurements were collected at the beginning of the school year in February and follow-up measurements ∼4 mo thereafter before the mid-year break. Teams of trained research assistants visited the schools on previously scheduled days. They confirmed assent for participation from the children and proceeded to collect anthropometric measurements with the use of standardized techniques (16). Weight was measured to the nearest 0.1 kg on Tanita HS301 electronic scales, while height was measured to the nearest 1 mm using wall-mounted portable Seca 202 stadiometers. Phlebotomists obtained a fasting blood specimen by venipuncture drawing an aliquot of ∼4 mL into an EDTA tube under a protocol to avoid hemolysis. The tubes were protected from sunlight and transported on ice the same day to the National Institute of Health in Bogotá, where all biochemistry analyses took place. Due to logistic constraints, baseline and follow-up anthropometric measurements were obtained in a group of 1570 participants as the other children could only have their first measure collected after February. Blood samples at both time points were collected in 2326 children.

Information was collected on the incidence of morbidity symptoms using pictorial diaries that were given to the parents each week. The diaries had cartoons depicting symptoms of common illnesses. The parents were instructed to check each day the child had conditions including fever, cough, diarrhea, or vomiting and whether the child visited the doctor or did not attend school. Between February and June, we obtained a total of 27,065 weekly diaries from 2768 children, with a median 11 diaries per child.

Compliance with consumption of the SED snack was assessed through direct observation by the teachers and recorded semiquantitatively for each snack component as consumed: all, one-half or more, less than one-half, or nothing.

Information was collected from the teachers on whether their classes received other school interventions. Some schools that were not covered by the SED snack in 2006 received another snack by the ICBF. The ICBF snack consisted of milk and a biscuit that provided ∼250 kcal and 6 mg of iron (17). Also, an iron supplementation program was implemented in some schools by the Secretary of Health consisting of the oral administration of 2 mg iron/kg body weight for 15 consecutive school days every 4–6 mo.

Laboratory procedures

Hemoglobin concentrations were determined with the use of a HemoCue photometer. Plasma ferritin and vitamin B-12 concentrations were measured using a competitive chemiluminescent immunoassay in an ADVIA Centaur analyzer (Bayer Diagnostics). C-reactive protein (CRP) plasma was measured with the use of a turbidimetric immunoassay on an ACS180 analyzer (Bayer Diagnostics). The packed red cell volume was hemolyzed by dilution in a hypotonic aqueous solution of 1% ascorbic acid. Erythrocyte folate was measured on the RBC lysates with the use of chemiluminescent immunoassay.

Data analyses

Definition of exposure.

Children exposed to the snack were those enrolled in schools that participated in the SED snack program from the beginning of the 2006 academic year (n = 1803). Unexposed children were those enrolled in schools that were not yet covered by the SED snack program during the first semester of the 2006 school year (n = 1399).

Definition of outcomes.

Micronutrient status outcomes were primarily assessed using hemoglobin, plasma ferritin, and vitamin B-12 and erythrocyte folate concentrations as continuous variables. Anemia was defined as hemoglobin <115 g/L after subtracting 12 g/L from the individual values as an adjustment for an altitude of 2500 m (18); iron deficiency was defined as ferritin <15 μg/L if CRP ≤ 10 mg/L or <30 μg/L if CRP >10 mg/L (19); vitamin B-12 status was defined as normal (>221 pmol/L) or marginal (≤221 pmol/L) (20); and folate deficiency was defined as erythrocyte folate <305 nmol/L (21).

Growth outcomes were primarily assessed using age- and sex-specific Z-scores as continuous variables. Height-for-age Z-scores (HAZ) and BMI for age Z-scores (BAZ) were calculated using 2007 WHO reference data (22). Stunting was defined as HAZ < −2 SD (23); thinness was defined based on the BMI-for-age and sex-specific cutoff points corresponding to <18.5 in adults according to Cole et al. (24); overweight or obesity was defined using BMI cutoff points for age and sex corresponding to ≥25 in adults as recommended by the International Obesity Task Force (25).

We calculated rates of morbidity (fever, cough with fever, diarrhea, and diarrhea with vomiting) as the number of days with each symptom or combination of symptoms divided by the number of days under observation. Rates were calculated for doctor visits and school absenteeism in the same manner. Weekends and vacation days were excluded from the days under observation for the school absenteeism endpoint.

Statistical analyses

The primary analytic strategy consisted of comparing changes in micronutrient status or growth parameters, and morbidity rates between children who received the SED snack and children whose schools were not yet participating in the SED snack program during the first semester of the 2006 school year. For each continuous micronutrient status (hemoglobin, plasma ferritin and vitamin B-12, and erythrocyte folate concentrations) and anthropometric outcomes (HAZ and BAZ), we estimated the difference in change from baseline to follow-up between the snack exposure groups using linear regression models for repeated measurements (PROC MIXED, SAS Institute). For morbidity and school absenteeism outcomes, rate ratios and 95% CI were estimated from Poisson regression models with the log-link function. In all models, an exchangeable correlation structure was specified to account for potential within-household correlations among siblings and for the clustering effect of the sampling strategy. All models were adjusted for child, maternal, or sociodemographic characteristics that significantly differed between groups at baseline. Values in the text are mean ± SD.

Results

Of the 3202 children enrolled, 56% (n = 1803) were in schools receiving the SED snack and 44% (n = 1399) were in schools that did not receive the SED snack during the first semester of 2006. The age of the children at baseline was 8.8 ± 1.8 y. Compared with children in schools that did not receive the SED snack, children receiving the SED snack were more likely to be male, have lower vitamin B-12 status, have received iron supplements at school, and have mothers who were less educated, less likely to be single, with higher parity, and from lower SES households (Table 1). These differences are expected, because the program was first introduced in poorer school districts.

TABLE 1.

Child, maternal, and socioeconomic characteristics of school children from Bogotá, Colombia, by snack group¹

	No SED snack	SED snack	P2
n	1399	1803
Child characteristics
Child age, y	8.7 ± 1.8	8.8 ± 1.8	0.63
Male, %	46.3	51.1	0.006
Anthropometric status, %
Stunted	8.1	10.4	0.14
Thin	10.3	8.9	0.36
Overweight or obese	10.5	10.0	0.73
Anemia, %	7.6	6.3	0.27
Iron deficiency, %	2.9	3.3	0.58
Marginal vitamin B-12 status (≤221 pmol/L), %	12.3	21.9	<0.0001
Folate deficiency (erythrocyte folate <305 nmol/L), %	0.1	0.1	0.93
Received iron supplements in 2006, %	34.1	45.9	<0.0001
Received ICBF snack in 2006, %	80.5	0	<0.0001
Maternal characteristics
Mother's age, y	35.1 ± 6.8	35.4 ± 6.7	0.23
Maternal education, y	8.9 ± 3.3	8.4 ± 3.3	0.0008
Mother is single parent, %	31.0	24.1	0.0001
Parity	2.6 ± 1.1	2.8 ± 1.1	0.0001
Household socioeconomic status
Household socioeconomic stratum, %			<0.0001
1 (lowest)	4.6	12.4
2	22.6	47.9
3	68.5	38.8
4	4.3	1.0

¹Values are mean ± SD or %.

²P-values are from univariate regression models in which each characteristic was introduced as the outcome and snack group as the predictor. An exchangeable correlation matrix was used to account for potential correlations within households among siblings and clustering due to the sampling strategy.

Children with measurements of micronutrient status at both time points did not differ from those without both measurements regarding exposure to the SED snack or socioeconomic status; however, children with both measurements were older and more likely to be male than children without both measurements. The median time between the baseline and follow-up measurements was 103 d. In both the SED snack and no SED snack groups, hemoglobin, plasma ferritin, and vitamin B-12 concentrations increased over time, whereas erythrocyte folate concentrations decreased (Table 2). After adjusting for socioeconomic status and other school interventions, children who received the SED snack had a 42 pmol/L greater increase in plasma vitamin B-12 (P < 0.0001) than children in schools that did not receive the SED snack. The prevalence of marginal vitamin B-12 status changed from 21.9% at baseline to 2.7% at follow-up in children who received the SED snack. The corresponding change in children who did not receive the SED snack was from 12.3 to 4.4%. The SED snack was not associated with significant changes in hemoglobin, ferritin, or folate.

TABLE 2.

Micronutrient status of school children from Bogotá, Colombia at baseline and 3-mo follow-up by snack group¹

			Difference2
	No SED snack	SED snack	Unadjusted (95% CI)	Adjusted (95% CI)
Hemoglobin, g/L
n	1068	1258
Baseline	140 ± 10	140 ± 9
3-mo follow-up	141 ± 9	142 ± 10
Change	1 (1, 2)	2 (1, 3)	1 (0, 2)	1 (0, 2)
Plasma ferritin, μg/L
n	1073	1243
Baseline	43.8 ± 23.5	41.0 ± 22.9
3-mo follow-up	45.8 ± 25.3	44.9 ± 24.8
Change	2.0 (0.6, 3.4)	3.8 (2.6, 5.1)	1.8 (−0.1, 3.7)	0.6 (−2.5, 3.8)
Plasma vitamin B-12, pmol/L
n	1033	1165
Baseline	336 ± 100	315 ± 109
3-mo follow-up	384 ± 106	380 ± 108
Change	47 (42, 53)	65 (59, 70)	17 (9, 25)	42 (29, 55)
Erythrocyte folate, nmol/L
n	1038	1180
Baseline	860 ± 276	847 ± 237
3-mo follow-up	755 ± 173	740 ± 200
Change	−105 (−123, −87)	−106 (−122, −90)	−1 (−26, 23)	−21 (−53, 12)

¹Baseline and 3-mo follow-up values are mean ± SD; change values are means and 95% CI.

²Estimates are from linear regression models for repeated measurements and represent the difference in the change over time between the SED snack and no SED snack groups. Adjusted differences are from models including the covariates child age and sex, maternal education and marital status, household socioeconomic stratum, and other nutritional interventions (iron supplementation at school, and whether child received ICBF snack). The ferritin model also included a variable for elevated CRP (>10 mg/L). An exchangeable correlation matrix was specified to account for potential correlations within households among siblings and clustering due to the sampling strategy.

The group of children with anthropometric measurements at both baseline and follow-up did not differ from that of children without the 2 measurements regarding exposure to the SED snack. Nevertheless, children with both anthropometric measures were older, more likely to be female, less likely to receive the ICBF snack, and in lower household socioeconomic strata than children with the 2 measurements. The median time from baseline to the follow-up anthropometric measurements was 119 d. HAZ decreased in both groups, but children in the SED snack group had a smaller decrease in HAZ than children not receiving the SED snack (P = 0.001) (Table 3). BAZ change did not differ between the 2 groups.

TABLE 3.

Anthropometric status of school children from Bogotá, Colombia at baseline and 4-mo follow-up by snack group¹

			Difference2
	No SED snack	SED snack	Unadjusted (95% CI)	Adjusted (95% CI)
HAZ
n	605	965
Baseline	−0.71 ± 0.97	−0.83 ± 0.97
4 mo follow-up	−0.75 ± 0.96	−0.84 ± 0.97
Change	−0.04 (−0.05, −0.03)	−0.01 (−0.02, 0.00)	0.04 (0.02, 0.05)	0.04 (0.02, 0.06)
BAZ
n	600	963
Baseline	0.10 ± 0.97	0.11 ± 1.00
4 mo follow-up	0.27 ± 0.95	0.29 ± 0.97
Change	0.17 (0.14, 0.19)	0.18 (0.16, 0.20)	0.02 (−0.01, 0.05)	0.03 (−0.03, 0.09)

¹Baseline and 4-mo follow-up values are means ± SD; change values are means and 95% CI.

²Estimates are from linear regression models for repeated measurements and represent the difference in the change over time between the SED snack and no-SED snack groups. Adjusted differences are from models including the covariates child age and sex, maternal education and marital status, household socioeconomic stratum, and other nutritional interventions (iron supplementation at school, and whether child received ICBF snack). In all models, an exchangeable correlation structure was specified to account for potential within-household correlations among siblings and for the clustering effect of the sampling strategy.

Children who received the SED snack had lower rates of reported morbidity than children who did not receive the SED snack (Table 4). Specifically, children receiving the SED snack reported 40% fewer days with fever (P = 0.0003), 57% fewer days with cough and fever (P < 0.0001), 30% fewer days with diarrhea (P = 0.03), 55% fewer days with diarrhea and vomiting (P = 0.0007), and 44% fewer doctor visits (P = 0.02) after adjusting for potential confounders. Children receiving the SED snack were absent from school on 23% fewer days (P = 0.03) than children from schools that did not receive the SED snack.

TABLE 4.

Morbidity and school absenteeism of children from Bogotá, Colombia by snack group

	No SED snack	SED snack
n	1241	1527
Fever
Days/child-years	1656/231	1302/288
Rate of days/child-year	7.2	4.5
Unadjusted rate ratio (95% CI)	1.0	0.63 (0.59, 0.68)
Adjusted rate ratio (95% CI)1	1.0	0.60 (0.46, 0.79)
Cough with fever
Days/child-years	809/231	564/288
Rate of days/child-year	3.5	2.0
Unadjusted rate ratio (95% CI)	1.0	0.56 (0.50, 0.62)
Adjusted rate ratio (95% CI)	1.0	0.43 (0.31, 0.58)
Diarrhea
Days/child-years	1368/231	1154/288
Rate of days/child-year	5.9	4.0
Unadjusted rate ratio (95% CI)	1.0	0.68 (0.63, 0.73)
Adjusted rate ratio (95% CI)	1.0	0.70 (0.51, 0.96)
Diarrhea with vomiting
Days/child-years	255/231	199/288
Rate of days/child-year	1.1	0.7
Unadjusted rate ratio (95% CI)	1.0	0.63 (0.52, 0.75)
Adjusted rate ratio (95% CI)	1.0	0.45 (0.29, 0.71)
Visited doctor
Days/child-years	652/231	545/288
Rate of days/child-year	2.8	1.9
Unadjusted rate ratio (95% CI)	1.0	0.67 (0.60, 0.75)
Adjusted rate ratio (95% CI)	1.0	0.56 (0.35, 0.90)
Absent from school
Days/child-years	1379/165	1417/206
Rate of days/child-year	8.4	6.9
Unadjusted rate ratio (95% CI)	1.0	0.83 (0.77, 0.89)
Adjusted rate ratio (95% CI)	1.0	0.77 (0.60, 0.98)

¹Estimates are from Poisson regression models with the log link function and offset by the log person-time of follow-up, adjusted for child age and sex, maternal education and marital status, household socioeconomic stratum, and other nutritional interventions received (iron supplements in school, child received ICBF snack). In all models, an exchangeable correlation structure was specified to account for potential within-household correlations among siblings and for the clustering effect of the sampling strategy.

Compliance with snack consumption was obtained from 1543 children on a median of 49 d per child. Compliance was high, with one-half or more of the snack consumed on 94% of the days. The percent of total days when the child consumed all of the snack components was 94% for the beverage, 87% for the cereal/protein component, 94% for the “sweet,” and 96% for the fruit.

Discussion

The provision of a midmorning snack at school was related to improved vitamin B-12 status and linear growth and to decreased reported morbidity in children from Bogotá, Colombia. These findings need to be interpreted with caution for a number of reasons. First, it is uncertain that these associations represent causal effects, because the study was not a randomized trial. Randomization was not an option, because the snack program was an official policy of the local city government, with a predetermined roll-out plan. Therefore, there may be potential residual confounding for measured or unmeasured factors despite adjusting for key differences between treatment groups. Second, follow-up was relatively short; it is not known whether the potential effects of the snack could be sustained through longer periods. We could not extend the duration of the study, because the snack program was introduced in all participating schools by the second semester of the school year. Also, within such a short period, we were unable to evaluate potential adverse effects of the snack on child adiposity. Third, information on anthropometry and micronutrient status could not be collected in all children and characteristics related to the selection of these subgroups could have been related to the response to the snack intervention. Fourth, the validity of the self-reported morbidity information is uncertain in this setting, although morbidity diaries have been used successfully in other studies (26,27).

Despite its limitations, our observational study offers interesting insights into the potential effectiveness of school-based nutritional interventions in this setting, especially to improve micronutrient status. Milk was the primary source of vitamin B-12 in the snack and the increased plasma vitamin B-12 among children receiving the snack was similar in magnitude to that in a study of school children in Kenya who were given 200–250 mL/d of milk for 1 y (6). Other studies have shown significant relations between dairy consumption and plasma vitamin B-12 concentrations (28,29). In the Kenya study, milk was also associated with increased height but only among the stunted children (7). The apparent effect on HAZ in our study, although significant, was very small, possibly due to the short duration of follow-up (4 mo). It will be important to determine whether this type of school-based intervention could have sustained effects on growth over longer periods.

The SED snack was not related to changes in ferritin, a proxy for iron stores, after adjusting for other nutritional interventions, including iron supplementation, at the schools. Some of the snack components were fortified with iron, and iron fortification has been shown to improve iron status of school children in other studies (30–33). However, absorption of iron from supplements is greater than that from fortified foods; thus, the potential benefits of iron from the snack may have been attenuated in this population, because both children in the SED and no-SED snack schools received iron supplements and iron supplementation of school children has been associated with increases in serum ferritin for comparable periods (e.g. 4 mo) (34). Overall, the prevalence of iron deficiency was relatively low at baseline and we may have lacked the statistical power to detect small improvements in iron status.

The SED snack was not related to changes in erythrocyte folate. However, we noted a decrease in mean erythrocyte folate concentrations from baseline to follow-up in both groups. The reasons for this decline are speculative. It has been suggested that treatment with iron could result in decreased erythrocyte folate (35) and these children may be having relatively high iron intakes during the school year, both from the snack interventions and from iron supplementation programs. Whether increased iron intake could be related to lower folate status in this population warrants further investigation.

Children consuming the SED snack reported fewer days with morbidity symptoms than children in the no-SED snack group. Vitamin B-12 deficiency has been associated with impaired immune function (36), but associations with infectious morbidity have not been clearly demonstrated. Some trials of multi-micronutrient–fortified food found decreased diarrhea and respiratory-related morbidity in school children consuming the fortified food (30,37). Children in Thailand fed a multi-micronutrient–fortified seasoning powder had lower incidence but not shorter duration of respiratory-related symptoms and diarrhea, which the authors attributed to zinc (37). South African children consuming a fortified biscuit had fewer school days missed due to respiratory- and diarrhea-related illnesses than children consuming an unfortified biscuit (30). In our study, it is unclear if the fewer reported days with morbidity symptoms in the snack group may have been due to potential effects of micronutrients or energy from the snack. Because the study was not randomized, it is not possible to rule out confounding by unmeasured factors.

Provision of the snack was not related to the children's BMI. The SED snack provided about one-third of the children's daily energy needs, the amount expected from a meal. Therefore, the SED snack might have displaced food intake from other sources. In Kenyan children receiving either a milk or high-energy snack at school, energy intake from home foods decreased from preintervention levels and the total energy intake of children who received the energy snack did not increase (38). It is a limitation of our study that we were unable to measure change in usual dietary intake before and after the intervention; we do not know if the snack may have affected food intake from other meals. Effects of an energy-based snack on ponderal growth need to be monitored longitudinally. It was not possible to assess whether the snack could affect adiposity, because follow-up was relatively short. The need to evaluate the potential effect of nutritional interventions on child overweight is critical (9), as the rates of child overweight and obesity are rapidly increasing in many developing countries such as Colombia. The SED snack derives a substantial proportion of its caloric content from high-energy, low-nutrient–dense foods, including sugar-sweetened beverages and candy. Some of the SED snack daily menus alone could provide at least 150 kcal/d from sugar-sweetened beverages and candy; this amount is very close to the 10% maximum of total energy/d from free sugar that is recommended by WHO (39). These foods should be replaced in the SED snack with healthier options such as daily fresh fruit.

In conclusion, the administration of a midmorning snack at school was related to improved vitamin B-12 status and linear growth and to reduced morbidity in children in Bogotá. Future studies should address the impact of school snacks on educational outcomes, such as school performance and neurocognitive function.