Impact factor (WEB OF SCIENCE - Clarivate)

2 year: 7.2 | 5 year: 6.6

Articles

Impacts of school feeding on educational and health outcomes of school-age children and adolescents in low- and middle-income countries: A systematic review and meta-analysis

Dongqing Wang1, Sachin Shinde1, Tara Young1, Wafaie W Fawzi1,2,3

1 Department of Global Health and Population, Harvard T.H. Chan School of Public Health, Harvard University, Boston, Massachusetts, USA
2 Department of Epidemiology, Harvard T.H. Chan School of Public Health, Harvard University, Boston, Massachusetts, USA
3 Department of Nutrition, Harvard T.H. Chan School of Public Health, Harvard University, Boston, Massachusetts, USA

Share:

Facebook
Twitter
LinkedIn
Abstract

Background

School feeding programs are ubiquitous in low- and middle-income countries (LMICs) and may have critical implications for the health and education of school-age children and adolescents. This systematic review aimed to assess the impacts of school feeding on educational and health outcomes of children and adolescents in LMICs.

Methods

Interventional studies on the effects of school feeding on nutritional and health outcomes of children and adolescents receiving primary or secondary education in LMICs were included. MEDLINE, EMBASE, CINAHL, the Cochrane Library, and grey literature were searched (through December 2019) to identify eligible studies. We included randomized controlled trials and controlled before-after studies on school feeding conducted in LMICs among children and adolescents aged 6 to 19 who received primary or secondary education. Two reviewers independently conducted study selection, data extraction, and risk of bias assessment. Meta-analyses were performed for outcomes available in three or more independent studies. Subgroup analyses were conducted by study design and school feeding modality whenever possible.

Results

Fifty-seven articles met the inclusion criteria for the review, including 44 randomized controlled trials and 13 controlled before-after studies; 19 articles were included in the meta-analysis. School feeding resulted in a significant increase in height (mean difference = 0.32 cm; confidence interval (CI) = 0.03, 0.61; P = 0.032) and weight (mean difference: 0.58 kg; 95% 95% CI = 0.22, 0.93; P = 0.001) over 12 months, compared to those in the control groups. School feeding also resulted in a significant increase in the percentage of school days attended (2.6%; 95% CI = 1.2%, 3.9%; P < 0.001).

Conclusions

School feeding is an important approach to improving the health and education outcomes of children and adolescents living in LMICs. More well-designed research is needed to establish further the effectiveness of school feeding for nutritional outcomes and academic achievement.

Registration

PROSPERO ID: CRD42020159003.

Print Friendly, PDF & Email

Across the developing world, millions of school-age children go to school every day hungry [1]. It is estimated that 73 million children of primary school age living in extreme poverty in 60 countries do not have access to national school feeding programs [1]. Attending classes while hungry severely impacts children’s abilities to learn, thrive, and realize their full educational and health potential as adults [2]. School feeding programs, which provide food to school-going children and adolescents, are ubiquitous worldwide. Most countries have some forms of school meal programs, though in varying formats and scales [3], and it is estimated that 368 million school-aged children and adolescents globally were reached by school feeding programs in 2017 [3]. In low- and middle-income countries (LMICs), it is estimated that some 305 million schoolchildren, approximately half of all those enrolled, eat a daily meal at school; for many of these children, this represents the only regular meal of the day [1]. Therefore, school feeding programs are imperative for the second Sustainable Development Goal (SDG), which entails ending hunger and achieving food security [4].

Besides alleviating hunger, school feeding programs may also increase school attendance, improve academic performance, and generate health, nutrition, and social protection benefits [3,57]. Nonetheless, school feeding programs tend to have incomplete coverage in LMICs, where the need for such interventions is greatest [6]. The coronavirus disease 2019 (COVID-19) pandemic has had devastating impacts on people’s livelihoods [8] and has propelled 265 million people worldwide into acute food insecurity [9]. With potentially less food available to children and adolescents at their homes during the pandemic, the role of school meal programs in safeguarding nutrition and education by providing regular and nutritious food will become even more critical. In the meantime, existing determinants of food insecurity remain at regional and household levels, such as household income, household size, household educational level, food production and availability, and food price [10]. School feeding programs play pivotal roles in SDG 3 (ensuring healthy lives and promoting well-being for all at all ages) and SDG 4 (ensuring inclusive and equitable quality education and promoting lifelong learning opportunities for all); it is also highly relevant for SDG 5 (achieving gender equality and empowering all women and girls) since girls dropped out of school disproportionally compared to their male counterparts [4].

Despite being one of the most long-lasting and well-established social protection programs in LMICs, continuous evidence synthesis is needed on the impacts of school feeding on educational and health outcomes in the ever-changing landscape of child and adolescent development. The only Cochrane systematic review on school feeding reported that participants in lower-income countries who were fed at school gained more weight, attended school more frequently, and had better performances in math and short-term cognitive tasks compared to those not fed at school in control groups [11]. However, this Cochrane review was published more than a decade ago and does not include all of the studies currently available [11]. More recent reviews on the benefits of school feeding exist, yet are limited in scope regarding the age range, school level, and the types of outcomes included [12,13]. Therefore, an updated synthesis of the evidence of the impacts of school feeding on a wide array of educational and health outcomes among children and adolescents is imperative.

In this systematic review and meta-analysis, we aimed to assess the impacts of school feeding programs on educational and health outcomes of children and adolescents in LMICs. We focused on educational and health outcomes due to their critical implications to long-term development and well-being and their high relevance to the SDGs. Through this work, our goal was to provide an evidence base that will be informative to the design and implementation of future programs that seek to provide food to school-going children and adolescents in the developing world.

METHODS

Eligibility criteria

Inclusion criteria

We included studies that met all of the inclusion criteria detailed below. We did not place restrictions on the year, sample size, or duration of the intervention.

  1. Randomized controlled trials (RCTs) and controlled before-after studies (CBAs) were eligible. The intervention could be allocated individually or in clusters (classes or schools).
  2. Published articles or grey literature were eligible.
  3. Studies conducted in low-income, lower-middle-income, or upper-middle-income countries, as defined by the World Bank for the 2020 fiscal year, were eligible.
  4. Studies among children and adolescents (girls and boys) aged 6 to 19 years who were receiving primary or secondary education (i.e., primary, middle, or high school) were eligible.
  5. Studies that examined the impacts of the provision of food or beverages in the school, including formal meals (breakfast, lunch, or dinner) or snacks, were eligible. We also included food distributed to households conditional on school enrollment and consumed outside the school setting (ie, take-home ration) [6].
  6. The comparison (control) group in each study did not receive the school feeding intervention; comparisons among school feeding menus with different food compositions were also eligible.
  7. Studies that included at least one outcome of children and adolescents pertinent to education, nutrition, anthropometry, cognition, or morbidity, were eligible. Eligible education outcomes included school enrollment, school attendance, school dropout, grade repetition, and academic performances (math, reading, spelling, vocabulary). Eligible nutrition outcomes included micronutrient status (eg, hemoglobin concentrations and anemia, ferritin concentrations, retinol concentrations) and macronutrient status (eg, blood lipid levels and blood albumin levels). Eligible anthropometry outcomes included height, weight, body mass index, mid-upper arm circumference, skinfold thickness, and bone development (including their z-scores). Eligible cognition outcomes included intelligence quotient, on-task behavior, and memory. Eligible morbidity outcomes included infectious disease diagnoses (eg, malaria and helicobacter pylori infection) or symptoms (eg, fever, cough, diarrhea, and vomiting).

Exclusion criteria

We excluded studies with any of the following characteristics.

  1. Interventional studies without a proper control group, such as uncontrolled before-after studies, were ineligible.
  2. Observational studies were ineligible.
  3. Editorials, commentaries, opinions, or review articles were ineligible; these articles, however, were reviewed to identify eligible studies.
  4. Studies conducted among preschool children only were ineligible. Feeding interventions among preschool children are important and of great interest but are beyond the scope of this work, which will focus on primary and secondary school settings.
  5. Studies that examined the impacts of micronutrient fortification, micronutrient supplementation, or nutrition education were ineligible unless such interventions were complementary to eligible school feeding interventions.
  6. Clinical treatment programs targeting patients with medical conditions or exclusively toward individuals with underweight, overweight, or obesity were ineligible.
  7. Descriptive studies of school feeding programs without the inclusion of specific outcomes were ineligible.

Information sources and search strategy

We searched four databases to identify eligible studies, from the inception of each database through December 2019: MEDLINE (via PubMed), EMBASE, CINAHL, and the Cochrane Library. The selection of the electronic databases was made in consultation with a health science librarian with expertise in systematic searching and covered all databases recommended by the Cochrane Handbook for Systematic Reviews of Interventions [14]. A search strategy including a combination of keywords, indexing (eg, MeSH terms), and free-text terms, was used. The specific search terms for each database are provided in Table S1 in the Online Supplementary Document. We also searched ClinicalTrials.gov and other governmental or organizational websites (World Food Programme, World Health Organization, Food and Agriculture Organization, and World Bank) for studies not identified from the database searching. We conducted a manual search of references included in key articles and previous reviews. We included articles published in English or Chinese. The six articles excluded were in Persian (two articles), Portuguese (one article), or Spanish (three articles). We used EndNote X9 (Clarivate Analytics, Pennsylvania, United States) to store the records retrieved from searches of electronic databases. We imported the records into Covidence (Veritas Health Innovation, Melbourne, Australia) to facilitate streamlined management of the records. Duplicate records were detected and removed by Covidence.

Study selection

To improve the accuracy of study selection, two reviewers (DW and SS, or DW and TY) independently screened titles and abstracts to assess the initial eligibility of the articles, and irrelevant studies were removed. The full texts of potentially eligible studies were then independently assessed by two reviewers (DW and SS) to confirm eligibility. Disagreements between reviewers were resolved by discussion. Specific reasons for study exclusions were recorded and summarized using the flow diagram for the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) [15].

Data extraction

To improve the accuracy of the extracted data, two reviewers (DW and SS) independently extracted data from the retained studies and entered data into an extraction form, which was pilot-tested using five randomly selected studies. Discrepancies in data extraction were resolved by discussion. We extracted the following information from each article: title, corresponding authors (name and contact information), journal (or source of information), year of publication, year of intervention, country and setting, source of funding, study design, sample size (number of participants and number of clusters when relevant), age and sex distributions of the sample, intervention (including timing, duration, food and nutritional content, and co-interventions), comparator/control, outcomes assessed, and main findings with point estimates and measures of variance (standard errors, 95% confidence intervals, or p-values). We collated multiple reports of a single study as additional results might be available in different articles.

Risk of bias assessment

Two reviewers (DW and SS) independently evaluated the risk of bias, with any disagreement resolved by discussion. For RCTs, we used version 2 of the Cochrane risk-of-bias tool (RoB 2), which includes five domains for individually randomized RCT [16] and an additional domain for cluster RCT [17]. We judged each domain as “low risk of bias,” “high risk of bias,” or “some concerns.” We considered an RCT to be of low risk of bias if it is judged to have a low risk of bias for all domains; we considered an RCT to be of a high risk of bias if it is judged to have a high risk of bias in at least one domain; we considered an RCT to have some concerns if it raises some concerns in at least one domain but is not of a high risk of bias for any domain [16].

For CBAs, we used the Risk of Bias in Non-randomized Studies of Interventions (ROBINS-I) tool, which considers biases from seven domains [18]. We judged each domain as “low risk of bias,” “moderate risk of bias,” “serious risk of bias,” “critical risk of bias,” or “no information.” We considered a CBA to be of low risk of bias if it is judged to have a low or moderate risk of bias for all domains; we considered a CBA to be of a high risk of bias if it is judged to have a serious or critical risk of bias in one or more domains; we considered a CBA to have some concerns if the assessment has no information for one or more domains but is of low or moderate risk of bias for all other domains.

Data synthesis

We summarized the characteristics and key findings of all included studies in the form of a summary table. Random-effects, inverse-variance-weighted meta-analyses were conducted for outcomes included in three or more independent studies with a sufficiently consistent definition of the outcome and the timing of the outcome assessment. As a result, the meta-analyses included nine outcomes, including 1) height; 2) height-for-age Z-score (HAZ); 3) weight; 4) weight-for-age Z-score (WAZ); 5) body mass index-for-age Z-score (BAZ); 6) hemoglobin concentrations; 7) plasma/serum ferritin concentrations; 8) mathematical or arithmetic skills; and 9) school attendance. We used the random-effects method because the effect of school feeding is presumed to be heterogeneous across time and populations, and also because the menus of previous school feeding interventions have been quite heterogeneous.

We treated school feeding as a dichotomous exposure. Effect estimates for continuous outcomes were expressed as mean differences with 95% confidence intervals (CIs) comparing the intervention group with the control group. To account for the different durations of the intervention across studies, we converted the estimates to the differences between the two groups over a period of 12 months using the monthly difference between arms. School attendance was standardized across studies as the percentage of possible school days attended in each study. When two or more relevant school-feeding arms (eg, in-school feeding and take-home ration, or different school feeding menus) were present, the average effects across arms were used. When cluster-based studies did not account for the clustered design in the analysis, we used the extracted average cluster size and an intracluster correlation coefficient (ICC) to correct the variance estimate using the approach recommended in the Cochrane Handbook for Systematic Reviews of Interventions [14]. Based on values recommended or commonly used in previous literature [11,19,20], the ICC values used to correct the variance estimate were 0.016 for HAZ, 0.025 for weight, WAZ, BAZ, hemoglobin concentrations, and plasma/serum ferritin concentrations, and 0.15 for attendance and mathematical/arithmetic skills.

We assessed the robustness of the results by removing each study from the meta-analyses (ie, the “one-study-removed” procedure). Whenever possible (ie, when at least two studies were available in each subgroup), we conducted subgroup analyses by study design (RCT or CBA) and modality of intervention (formal meals or snacks) to assess the sources of heterogeneity of the effects. We did not conduct subgroup analyses by risk of bias as most of the included studies had a high risk of bias. Other subgroup analyses we considered that were not ultimately feasible due to the small number of studies or lack of subgroup-specific results included the presence of co-interventions (by itself or combined with complementary interventions), year of study, country or region, level of food insecurity of the region, age group of participants (primary or secondary education), sex of participants, and type of report (published or unpublished). We used funnel plots to detect publication bias for outcomes available in at least five studies. We conducted meta-analyses using Comprehensive Meta-Analysis Software Version 3 [21] using a two-sided level of 0.05.

Assessment of certainty of evidence

The overall certainty of evidence for each outcome included in the meta-analyses was assessed using the Grading of Recommendation, Assessment, Development, and Evaluation (GRADE) approach that considers risk of bias, publication bias, imprecision, inconsistency, and indirectness [2227]. The evidence for each outcome was accordingly judged as high, moderate, low, or very low [22].

Registration and reporting

The full protocol of this work was published elsewhere [28]. This work was registered with the International Prospective Register of Systematic Reviews (PROSPERO ID: CRD42020159003). We conducted and reported this study in accordance with the Preferred Reporting Items for Systematic Review and Meta-Analysis Protocols (PRISMA) [15].

RESULTS

Basic characteristics of the included studies

We identified 7854 articles from four electronic databases and a manual search of references of key articles, previous reviews, and grey literature. Among the retrieved articles, 7613 were judged as irrelevant and excluded after the title and abstract screening. Among the 241 articles included in the full-text review, 184 were excluded due to various exclusion criteria, resulting in 57 articles included in the data synthesis (Figure 1).

Figure 1.  PRISMA flow diagram.

Among the 57 retained articles, 44 and 13 were based on RCTs [2972] and CBAs [7385], respectively. The 57 articles arose from 39 independent studies, of which 26 were RCTs and 13 were CBAs, and most (12 RCTs and 10 CBAs) allocated the intervention by school. The majority of the studies included a separate control group that did not receive school feeding, with four studies [30,33,38,48] employing a crossover design where the same individuals received both the intervention and the control. Five studies each were conducted in China [38,43,61,62,82] and India [48,54,58,72,76], three in Jamaica [30,36,74], two each in Ghana [69,71], Indonesia [35,73], Iran [57,81], Kenya [39,80], Peru [32,33], Philippines [37,65], South Africa [68,77], Uganda [50,83], and Vietnam [53,78], and one each in Bangladesh [84], Benin [79], Burkina Faso [52], Colombia [75], Lebanon [85], Papua New Guinea [29], Senegal [67], and Thailand [55].

Seventeen studies focused on the provision of in-school snacks [39,43,53,54,57,58,62,65,69,73,78,8085], eight on the provision or modification of in-school breakfast [30,32,33,36,38,61,74,77], seven on the provision or modification of in-school lunch [55,67,68,71,72,75,76], and seven included more than one modality of school feeding [29,35,37,48,50,52,79]. In-school meals were prepared either centrally in one or a handful of kitchens or a decentralized manner in each school. The distribution of take-home rations was involved in two studies [50,52]. For in-school meals, the compliance was often monitored through direct supervision by teachers or study staff when students consumed the food. Details on the costs of food or the costs of operation were rarely provided.

Most of the studies used school feeding as the sole intervention without any complementary interventions (to all participants or those in the intervention arms), with the exception of three studies that provided deworming [61,63,68], four that provided nutrition or health education [72,80,81,85], and one that provided both deworming and nutrition education [78]. The duration of intervention ranged from one week to 2 years. Two studies were conducted among secondary school students [74,79], three studies among both primary and secondary school students [54,72,81], and the rest 34 studies were among primary school students. The sample size for each study ranged from 39 to 4236 students. Four studies were conducted among girls only [43,54,75,79], one study among boys only [73], and the remaining 34 studies among girls and boys. Ten studies were conducted among children of middle childhood (6 to 9 years) [30,38,48,53,57,69,76,78,80,83], 10 studies among adolescents (10 to 19 years) [43,54,55,62,67,68,72,74,79,85], and the remaining 18 studies including both children and adolescents; the age distribution for one study was not reported [37]. Most of the studies were conducted in governmental/public schools. The main characteristics of the included studies are provided in Table 1.

Table 1.  Characteristics of the included studies

WordPress Data Table

BAP – bone alkaline phosphatase, BAZ – body mass index for age z-scores, BMC – bone mineral content, BMD – bone mineral density, BMI – body mass index, CBA – controlled before-after study, CRP – C-reactive protein, GPA – grade point average, HAZ – height-for-age Z-score, Hb – hemoglobin, HDL-C – high-density lipoprotein cholesterol, IGF-1 – insulin-like growth factor 1, IQ – intelligence quotient, LDL-C – low-density lipoprotein cholesterol, MAFA – mid-upper-arm fat area, MAMA – mid-upper-arm muscle area, MUAC – mid-upper arm circumference, PTH – parathayroid hormone, RBC – red blood cell, RCT – randomized controlled trial, RDA – recommended dietary allowance, SD – standard deviation, TC:HD – ratio of total cholesterol to HDL cholesterol, THR – take-home ration, TIBC – total iron binding capacity, TSF – triceps skinfold, WAZ – weight-for-age Z-score, WHZ – weight-for-height Z-score, WISC – Wechsler Intelligence Scale for children, d – day(s), mo – month(s), y – year(s)

Risk of bias

Among the 44 articles based on RCTs, 40 were judged to have a high risk of bias, and four were judged to have some concerns (Table S2 in the Online Supplementary Document). The primary sources of bias for RCTs included bias in selection of the reported results (32 had a high risk of bias), bias arising from the randomization process (20 had a high risk of bias), and bias due to missing outcome data (18 had a high risk of bias). Among the 13 articles based on CBAs, 11 were judged to have a high risk of bias, one was judged to have a low risk of bias [84], and one was judged to have some concerns [85]. The primary sources of bias for CBAs included bias due to confounding (10 had a high risk of bias), bias in selection of the reported results (seven had a high risk of bias), bias due to missing data (five had a high risk of bias), and bias in selection of participants into the study (five had a high risk of bias).

Meta-analyses

Nine outcomes were included in three or more independent studies with sufficiently consistent outcome definitions, allowing for meta-analyses. These outcomes were: 1) height; 2) HAZ; 3) weight; 4) WAZ; 5) BAZ; 6) hemoglobin concentrations; 7) plasma/serum ferritin concentrations; 8) mathematical or arithmetic skills; and 9) school attendance. Nineteen studies were included in any of the meta-analyses [29,35,36,52,53,56,57,60,62,65,66,68,69,7173,76,78,85]. The main reasons for outcomes being excluded from meta-analyses were failure to report quantitative results, lack of variance estimates, reported estimates not able to be synthesized with other studies, or outcomes being included in only one or two studies.

Height

The meta-analysis for height included seven studies [29,36,53,57,73,76,78]. In the random-effects meta-analysis including all seven studies (Figure 2), school feeding resulted in a statistically significant increase in height; participants who received school feeding gained, on average, 0.318 (95% CI = 0.027, 0.609; P = 0.032) centimeters greater heights over 12 months compared to the those in the control groups. No publication bias was detected using the funnel plot (Figure S1 in the Online Supplementary Document). The positive effect on height was robust to the removal of any particular study; the estimates with one study removed ranged from 0.141 (95% CI = 0.054, 0.228; P = 0.002) centimeters with Lampl 1978 removed to 0.463 (95% CI = 0.043, 0.883; P = 0.031) centimeters with Bailey 1962 removed. The effect on height was stronger in RCTs (mean difference = 0.858; 95% CI = 0.312, 1.403; P = 0.002) than in CBAs (mean difference = 0.050; 95% CI = -0.359, 0.460; P = 0.81) (Figure S2 in the Online Supplementary Document), and stronger when formal meals (breakfast, lunch, or dinner) were provided (mean difference = 0.586; 95% CI = 0.109, 1.063; P = 0.016) than when snacks were provided (mean difference = 0.111; 95% CI = -0.362, 0.584; P = 0.65) (Figure S3 in the Online Supplementary Document). The certainty of the evidence for height was rated as moderate.

Figure 2.  Random-effects meta-analysis of school feeding interventions on 12-month height gain (in centimeters), comparing the group receiving school feeding intervention to the control group. I2 = 82.02%.

Height-for-age Z-score

The meta-analysis for HAZ included six studies [53,69,71,72,78,85]. In the random-effects meta-analysis including all six studies (Figure 3), school feeding did not have a statistically significant effect on HAZ (mean difference over 12 months = -0.018; 95% CI = -0.048, 0.012; P = 0.23), and the results were robust to the removal of any particular study. No publication bias was detected using the funnel plot (Figure S4 in the Online Supplementary Document). The certainty of the evidence for HAZ was rated as moderate.

Figure 3.  Random-effects meta-analysis of school feeding interventions on 12-month change in height-for-age Z-score, comparing the group receiving school feeding intervention to the control group. I2 = 4.85%.

Weight

The meta-analysis for weight included seven studies [29,36,53,57,73,76,78]. In the random-effects meta-analysis including all seven studies (Figure 4), school feeding resulted in a statistically significant increase in weight; participants who received school feeding gained, on average, 0.576 (95% CI = 0.221, 0.932; P = 0.001) kilograms more weights over 12 months compared to the those in the control groups. No publication bias was detected using the funnel plot (Figure S5 in the Online Supplementary Document). The positive effect on weight was robust to the removal of any particular study; the estimates with one study removed ranged from 0.327 (95% CI = 0.080, 0.573; P = 0.009) kilograms with Lampl 1978 removed to 0.721 (95% CI = 0.265, 1.177; P = 0.002) kilograms with Bailey 1962 removed. The effect on weight was stronger in RCTs (mean difference = 0.993; 95% CI = 0.433, 1.554; P = 0.001) than in CBAs (mean difference = 0.308; 95% CI = -0.140, 0.756; P = 0.18) (Figure S6 in the Online Supplementary Document), and stronger when formal meals were provided (mean difference = 0.904; 95% CI = 0.521, 1.287; P < 0.001) than when snacks were provided (mean difference = 0.123; 95% CI = -0.303, 0.550; P = 0.57) (Figure S7 in the Online Supplementary Document). The certainty of the evidence for weight was rated as moderate.

Figure 4.  Random-effects meta-analysis of school feeding interventions on 12-month weight gain (in kilograms), comparing the group receiving school feeding intervention to the control group. I2 = 93.20%.

Weight-for-age Z-score

The meta-analysis for WAZ included five studies [52,53,69,78,85]. In the random-effects meta-analysis including all five studies (Figure 5), school feeding did not have a statistically significant effect on WAZ (mean difference over 12 months = 0.043; 95% CI = -0.022, 0.108; P = 0.19). The lack of statistical significance appeared to be due to the non-significant estimates from Hall 2007, which had a large sample size and a greater contribution to the pooled estimate. The effect with Hall 2007 removed was positive and statistically significant (mean difference over 12 months = 0.100; 95% CI = 0.012, 0.189; P = 0.027). The asymmetry in the funnel plot suggests a potentially slight selective publication of studies with positive findings (Figure S8 in the Online Supplementary Document). The certainty of the evidence for WAZ was rated as low.

Figure 5.  Random-effects meta-analysis of school feeding interventions on 12-month change in weight-for-age Z-score, comparing the group receiving school feeding intervention to the control group. I2 = 25.38%.

Body mass index-for-age Z-score

The meta-analysis for BAZ included four studies [52,71,72,85]. In the random-effects meta-analysis including all four studies (Figure 6), school feeding did not have a statistically significant effect on BAZ (mean difference over 12 months = 0.074; 95% CI = -0.088, 0.235; P = 0.37), and the results were robust to the removal of any particular study. The asymmetry in the funnel plot suggests a potential selective publication of studies with positive findings (Figure S9 in the Online Supplementary Document). The certainty of the evidence for BAZ was rated as low.

Figure 6.  Random-effects meta-analysis of school feeding interventions on 12-month change in body mass index-for-age Z-score, comparing the group receiving school feeding intervention to the control group. I2 = 33.92%.

Hemoglobin concentrations

The meta-analysis for hemoglobin concentrations included six studies [35,52,53,62,65,68]. In the random-effects meta-analysis including all six studies (Figure 7), school feeding did not have a statistically significant effect on hemoglobin concentrations (mean difference over 12 months = 0.065 g/L; 95% CI = -1.676, 1.807; P = 0.94), and the results were robust to the removal of any particular study. No publication bias was detected using the funnel plot (Figure S10 in the Online Supplementary Document). The certainty of the evidence for hemoglobin concentrations was rated as low.

Figure 7.  Random-effects meta-analysis of school feeding interventions on 12-month change in hemoglobin concentrations (g/L), comparing the group receiving school feeding intervention to the control group. I2 = 23.87%.

Plasma/serum ferritin concentrations

The meta-analysis for plasma or serum ferritin concentrations included three studies [35,53,68]. In the random-effects meta-analysis including all three studies (Figure 8), school feeding did not have a statistically significant effect on hemoglobin concentrations (mean difference over 12 months = 0.612 μg/L; 95% CI = -0.722, 1.947; P = 0.37), and the results were robust to the removal of any particular study. The certainty of the evidence for plasma or serum ferritin concentrations was rated as low.

Figure 8.  Random-effects meta-analysis of school feeding interventions on 12-month change in plasma or serum ferritin concentrations (μL), comparing the group receiving school feeding intervention to the control group. I2 = 57.62%.

Mathematical/arithmetic skills

The meta-analysis for mathematical or arithmetic skills included five studies [36,60,62,66,76]. In the random-effects meta-analysis including all five studies (Figure 9), school feeding did not have a statistically significant effect on mathematical/arithmetic skills (mean difference in scores = 0.320; 95% CI = -0.148, 0.789; P = 0.18), and the results remained largely robust to the removal of any particular study. No publication bias was detected using the funnel plot (Figure S11 in the Online Supplementary Document). The certainty of the evidence for mathematical or arithmetic skills was rated as moderate.

Figure 9.  Random-effects meta-analysis of school feeding interventions on 12-month change in mathematical or arithmetic skills, comparing the group receiving school feeding intervention to the control group. I2 = 32.37%.

School attendance

The meta-analysis for school attendance included three studies [36,56,60]. In the random-effects meta-analysis including all three studies (Figure 10), school feeding resulted in a statistically significant increase in the percentage of school days attended; participants who received school feeding attended, on average, 2.56 (95% CI = 1.20, 3.92; P < 0.001) percent more school days compared to the those in the control groups. The positive effect on attendance was robust to the removal of any particular study; the estimates with one study removed ranged from 2.30 (95% CI = 0.82, 3.78; P = 0.002) percent with Kazianga 2012 removed to 3.48 (95% CI = 0.51, 6.46; P = 0.022) percent with Powell 1998 removed. The certainty of the evidence for school attendance was rated as moderate.

Figure 10.  Random-effects meta-analysis of school feeding interventions on 12-month change in school attendance (percentage of possible school days attended), comparing the group receiving school feeding intervention to the control group. I2 = 0.00%.

DISCUSSION

We present a systematic review and meta-analysis on the impacts of school feeding interventions on a wide range of health and education outcomes among children and adolescents living in LMICs. We report that school feeding programs confer benefits on height, weight, and school attendance. This work presents an evidence base that will be informative to the design and implementation of school feeding programs targeted toward children and adolescents in resource-constrained settings.

A few systematic or non-systematic reviews have previously synthesized the impacts of school feeding interventions on the health and education of children and adolescents [1113,86,87]. The only Cochrane systematic review on the impacts of school feeding interventions was published in 2007 and included 18 studies, of which nine were from LMICs [11]. Comprehensive at the time of publication, this Cochrane systematic review included five RCTs conducted in LMICs. More recent reviews on the benefits of school feeding interventions are available but were limited in scope. A qualitative review assessed the impacts of school feeding on the education and health of children in primary schools in LMICs [12]. However, studies conducted at the secondary school level or published before 1990 were not included in this review. Another two reviews [86,87] evaluated interventions for improving education outcomes among children and adolescents in LMICs. Nonetheless, they did not specifically focus on school feeding, nor did they include outcomes related to nutrition and health. The impacts of school feeding on the anthropometry and nutritional status of primary-school-age children and preschool and adolescent girls in LMICs have been reviewed [13], but this work did not specifically examine the impacts on educational outcomes. As more studies become available, updated evidence synthesis on a wide range of outcomes based on systematic literature review is critical.

School meals may provide nutritional benefits for children and adolescents. The benefits of school feeding on anthropometry and school attendance in our review are consistent with those reported in previous work. The estimated benefit of school feeding on weight gain (0.58 kg over 12 months) is similar to the estimates from the Cochrane systematic review [11], which reported that children and adolescents in LMICs receiving school feeding gained an average of 0.25 and 0.75 kg more weight over 12 months in RCTs and CBAs, respectively, compared to those in the control groups. We found a significant effect on height gain (0.32 cm over 12 months) comparing the school feeding group to the control group; this is also generally in line with the slightly mixed results shown in the Cochrane systematic review, which reports non-significant gain in RCTs and a significant gain of 1.43 cm in CBAs [11].

We did not find significant benefits of school feeding on indicators of undernutrition such as HAZ, WAZ, and BAZ. The lack of observed benefits on the anthropometry Z-scores may be due to the small magnitude of the effects (in absolute terms) that might have been insufficient to materially alter the underlying nutritional status or the relative standings in the population. However, most studies were of short or moderate durations, with few studies exceeding 12 months, so they might not have accumulated adequate time for the nutritional benefits of the school feeding interventions to fully manifest. Similarly, the non-significant effects on hemoglobin concentration and ferritin concentrations may be explained by the failure to include iron-rich, animal-source food. When optimally designed considering all the nutritional needs of children and adolescents (beyond merely providing total energy) and provided for an extended period of time, school feeding interventions may have the potential to prevent undernutrition. Future studies should preferably provide formal school meals (as opposed to snacks only) with animal-source food for years and conduct longitudinal follow-up of the participants to better evaluate school feeding programs’ long-term benefits.

In addition to the direct nutritional benefits, school feeding has long been used to confer educational benefits. The Cochrane systematic review estimates that children and adolescents receiving school meals attended 4 to 6 more school days per year (in a school year of 172 school days) than those without school meals [11]. Assuming also a school year of 172 days, our estimate of 2.6% more school days attended is equivalent to 4.5 more school days attended, in line with the estimates from the Cochrane systematic review. This finding confirms that school meals can motivate the parents to send their children to school and keep the students in attendance.

The high level of heterogeneity of the school feeding menus and modality indicates that the estimates from the meta-analyses should be interpreted as the average effects; different school feeding programs may well confer quite different levels of benefits. This observation also highlights a historical lack of clear consensus and guidance on the design of school meals in resource-limited settings. It was beyond the scope of the current study to provide definitive recommendations on the design of school feeding programs. However, recent years have seen significant developments in the innovative tools for the development and standardization of school feeding programs in LMICs. For example, the Partnership for Child Development School Meals Planner can be used to create nutritionally balanced and market-costed school meals using daily reference nutrient intake [88]. Similarly, the upcoming PLUS School Menus software will be able to optimize school meals [89]. Based on mathematical optimization algorithms, the PLUS School Menus software calculates the most cost-effective menus, covering the nutritional needs of children from different age groups and optimizing the use of locally sourced food [89]. With global scale-up and capacity building at the country and school levels, tools like these will prove vital for maximizing the potential benefits of school meals in resource-constrained settings. Even with such tools, school feeding interventions nevertheless allow a great deal of flexibility in the design (eg, menu, quantity, timing) based on the specific context; the key elements for successful programs need to be better understood.

A challenge that complicates the interpretation of our results is the generally high risks of bias in previous studies. The primary concerns for both RCT and CBA study designs were the selected reporting of the results and missing data of the outcomes. While school feeding is one of the most long-lasting and well-established nutrition interventions worldwide, it was noted that many studies on school feeding were lacking in the rigor of the methodology or the quality of the reporting. Future studies to evaluate the effectiveness and implementation of school feeding programs should consider the transparent reporting of all pre-specified outcomes, preferably supplemented by a registered protocol. Also, the number of participants with missing data for each outcome (and the reasons for missingness), the potential of selection bias, and measures taken to control for confounding should all be carefully considered and clearly described. Randomized studies should additionally report the process of randomization clearly, so there is no ambiguity on how the intervention received by each study unit (eg, each school) was determined.

Future research priorities may not be whether school feeding impacts nutrition or education in general, but instead on assessing and improving the quality of school feeding programs. Some specific knowledge gaps include identifying appropriate and measurable nutrition indicators, examining school feeding programs by finer age strata, improving the quality and diversity of school meal menus, and strategies for scaling up. The consensus among stakeholders regarding such issues is yet to be reached, and additional studies will contribute valuable insights to the broader research agenda of establishing school feeding guidelines.

The main strengths of this systematic review included the coverage of a wide array of health and education outcomes and the inclusion of all modalities of school feeding. This study has a few potential limitations. First, as most of the included studies had relatively short follow-up periods, we could not make strong inferences regarding the long-term impacts of the school feeding interventions. Second, most of the studies compared school feeding intervention with no intervention at all, which limited our abilities to evaluate whether different contents of school feeding interventions may have different impacts. Third, as a limited number of studies reported process details of the intervention such as the fidelity of the intervention [90] or provided stratified estimates by age or sex, our abilities to conduct subgroup analyses by those characteristics were constrained. Fourth, combining estimates of anthropometry measures across studies conducted among different age groups of children and adolescents was not ideal. However, we also included more age-standardized metrics whenever possible (eg, HAZ, WAZ, and BAZ), which lessened this issue.

In conclusion, this study provides a much-needed synthesis on the impacts of school feeding interventions on the health and education outcomes of children and adolescents living in LMICs. Findings from this study are informative to various stakeholders, including researchers, policymakers, non-governmental organizations, and individual schools, in the design and implementation of school feeding programs. We highlight a need for rigorously designed school feeding studies that have long-term follow-up periods, that are conducted among adolescents in secondary school settings, and that evaluate the impacts on micronutrient deficiencies and academic achievements. Future work is needed to develop clear guidelines and scale up emerging tools for school feeding among children and adolescents in LMICs who are urgently in need of this critical nutrition intervention, especially during the ongoing COVID-19 pandemic.

Additional material

Online Supplementary Document

Acknowledgments

We thank Jacqueline Cellini at Francis A. Countway Library of Medicine at Harvard University for conceptual consultation and assistance with developing the PubMed search strategy.

[1] Funding: This work received no financial support.

[2] Authorship contributions: DW and WWF designed research. DW, SS, and TY conducted title and abstract screening. DW and SS conducted full-text screening, data extraction, and risk of bias assessment. DW performed statistical analysis. DW wrote the paper. All authors read and critically edited the paper. All authors approved the final paper.

[3] Competing interests: The authors completed the ICMJE Declaration of Interest Form (available upon request from the corresponding author), and declare no conflicts of interest.

references

[1] LJ Drake, N Lazrak, M Fernandes, K Chu, S Singh, and D Ryckembusch. Establishing global school feeding programme targets: how many poor children globally should be prioritized, and what would be the cost of implementation? Front Public Health. 2020;8:530176. DOI: 10.3389/fpubh.2020.530176. [PMID:33344395]

[2] Plaut D, Thomas M, Hill T, Worthington J, Fernandes M, Burnett N. Getting to Education Outcomes: Reviewing Evidence from Health and Education Interventions. Child and Adolescent Health and Development 3rd edition: The International Bank for Reconstruction and Development/The World Bank; 2017.

[3] Drake L, Fernandes M, Aurino E, Kiamba J, Giyose B, Burbano C, et al. School Feeding Programs in Middle Childhood and Adolescence. Child and Adolescent Health and Development 3rd edition. Washington, DC: The International Bank for Reconstruction and Development/The World Bank; 2017.

[4] BX Lee, F Kjaerulf, S Turner, L Cohen, PD Donnelly, and R Muggah. Transforming our world: implementing the 2030 agenda through sustainable development goal indicators. J Public Health Policy. 2016;37:13-31. DOI: 10.1057/s41271-016-0002-7. [PMID:27638240]

[5] Adelman S, Gilligan D, Lehrer K. How effective are food for education programs?: A critical assessment of the evidence from developing countries. Intl Food Policy Res Inst; 2008.

[6] Bundy D, Burbano C, Grosh ME, Gelli A, Juke M, Drake L. Rethinking school feeding: social safety nets, child development, and the education sector. Washington, DC: The World Bank; 2009.

[7] R Aliyar, A Gelli, and SH Hamdani. A review of nutritional guidelines and menu compositions for school feeding programs in 12 countries. Front Public Health. 2015;3:148 DOI: 10.3389/fpubh.2015.00148. [PMID:26301209]

[8] K Govender, RG Cowden, P Nyamaruze, RM Armstrong, and L Hatane. Beyond the disease: Contextualized implications of the COVID-19 pandemic for children and young people living in Eastern and Southern Africa. Front Public Health. 2020;8:504 DOI: 10.3389/fpubh.2020.00504. [PMID:33194933]

[9] D Headey, R Heidkamp, S Osendarp, M Ruel, N Scott, and R Black. Impacts of COVID-19 on childhood malnutrition and nutrition-related mortality. Lancet. 2020;396:519-21. DOI: 10.1016/S0140-6736(20)31647-0. [PMID:32730743]

[10] W Drammeh, NA Hamid, and A Rohana. Determinants of household food insecurity and its association with child malnutrition in Sub-Saharan Africa: A review of the literature. Current Research in Nutrition and Food Science Journal. 2019;7:610-23. DOI: 10.12944/CRNFSJ.7.3.02

[11] EA Kristjansson, M Petticrew, B MacDonald, J Krasevec, L Janzen, and T Greenhalgh. School feeding for improving the physical and psychosocial health of disadvantaged students. Cochrane Database Syst Rev. 2007;CD004676. DOI: 10.1002/14651858.CD004676.pub2. [PMID:17253518]

[12] LH Jomaa, E McDonnell, and C Probart. School feeding programs in developing countries: impacts on children’s health and educational outcomes. Nutr Rev. 2011;69:83-98. DOI: 10.1111/j.1753-4887.2010.00369.x. [PMID:21294742]

[13] Watkins K, Gelli A, Hamdani S, Masset E, Mersch C, Nadazdin N, et al. Sensitive to Nutrition? A Literature Review of School Feeding Effects in the Child Development Lifecycle. London: Working Paper Series, 2015.

[14] Higgins JPT, Thomas J, Chandler J, Cumpston M, Li T, Page MJ, et al. Cochrane Handbook for Systematic Reviews of Interventions. 2 ed. Newark: John Wiley & Sons, Inc.; 2019.

[15] MJ Page, JE McKenzie, PM Bossuyt, I Boutron, TC Hoffmann, and CD Mulrow. The PRISMA 2020 statement: an updated guideline for reporting systematic reviews. BMJ. 2021;372:n71 DOI: 10.1136/bmj.n71. [PMID:33782057]

[16] JA Sterne, J Savović, MJ Page, RG Elbers, NS Blencowe, and I Boutron. RoB 2: a revised tool for assessing risk of bias in randomised trials. BMJ. 2019;366:l4898 DOI: 10.1136/bmj.l4898. [PMID:31462531]

[17] Eldridge S, Campbell M, Campbell M, Dahota A, Giraudeau B, Higgins J, et al. Revised Cochrane risk of bias tool for randomized trials (RoB 2.0): Additional considerations for cluster-randomized trials. Available: https://www.semanticscholar.org/paper/Revised-Cochrane-risk-of-bias-tool-for-randomized-Eldridge-Campbell/d0b0ef43d768e6d3a574cc84d75e91722a2f7d4e. Accessed: 1 July 2021.

[18] JA Sterne, MA Hernán, BC Reeves, J Savović, ND Berkman, and M Viswanathan. ROBINS-I: a tool for assessing risk of bias in non-randomised studies of interventions. BMJ. 2016;355:i4919 DOI: 10.1136/bmj.i4919. [PMID:27733354]

[19] X Du, K Zhu, A Trube, DR Fraser, H Greenfield, and Q Zhang. Effects of school-milk intervention on growth and bone mineral accretion in Chinese girls aged 10–12 years: accounting for cluster randomisation. Br J Nutr. 2005;94:1038-9. DOI: 10.1079/BJN20051584. [PMID:16351784]

[20] PZ Schochet. Statistical Power for Random Assignment Evaluations of Education Programs. J Educ Behav Stat. 2008;33:62-87. DOI: 10.3102/1076998607302714

[21] Borenstein M, Hedges L, Higgins J, Rothstein H. Comprehensive Meta-Analysis Version 3. Biostat, Englewood, NJ. 2013.

[22] H Balshem, M Helfand, HJ Schünemann, AD Oxman, R Kunz, and J Brozek. GRADE guidelines: 3. Rating the quality of evidence. J Clin Epidemiol. 2011;64:401-6. DOI: 10.1016/j.jclinepi.2010.07.015. [PMID:21208779]

[23] GH Guyatt, AD Oxman, G Vist, R Kunz, J Brozek, and P Alonso-Coello. GRADE guidelines: 4. Rating the quality of evidence—study limitations (risk of bias). J Clin Epidemiol. 2011;64:407-15. DOI: 10.1016/j.jclinepi.2010.07.017. [PMID:21247734]

[24] GH Guyatt, AD Oxman, V Montori, G Vist, R Kunz, and J Brozek. GRADE guidelines: 5. Rating the quality of evidence—publication bias. J Clin Epidemiol. 2011;64:1277-82. DOI: 10.1016/j.jclinepi.2011.01.011. [PMID:21802904]

[25] GH Guyatt, AD Oxman, R Kunz, J Brozek, P Alonso-Coello, and D Rind. GRADE guidelines 6. Rating the quality of evidence—imprecision. J Clin Epidemiol. 2011;64:1283-93. DOI: 10.1016/j.jclinepi.2011.01.012. [PMID:21839614]

[26] GH Guyatt, AD Oxman, R Kunz, J Woodcock, J Brozek, and M Helfand. GRADE guidelines: 7. Rating the quality of evidence—inconsistency. J Clin Epidemiol. 2011;64:1294-302. DOI: 10.1016/j.jclinepi.2011.03.017. [PMID:21803546]

[27] GH Guyatt, AD Oxman, R Kunz, J Woodcock, J Brozek, and M Helfand. GRADE guidelines: 8. Rating the quality of evidence—indirectness. J Clin Epidemiol. 2011;64:1303-10. DOI: 10.1016/j.jclinepi.2011.04.014. [PMID:21802903]

[28] D Wang and WW Fawzi. Impacts of school feeding on educational and health outcomes of school-age children and adolescents in low-and middle-income countries: protocol for a systematic review and meta-analysis. Syst Rev. 2020;9:55 DOI: 10.1186/s13643-020-01317-6. [PMID:32178734]

[29] M Lampl, FE Johnston, and LA Malcolm. The effects of protein supplementation on the growth and skeletal maturation of New Guinean school children. Ann Hum Biol. 1978;5:219-27. DOI: 10.1080/03014467800002841. [PMID:686663]

[30] AM Chandler, SP Walker, K Connolly, and SM Grantham-McGregor. School breakfast improves verbal fluency in undernourished Jamaican children. J Nutr. 1995;125:894-900. [PMID:7722692]

[31] SM Chang, SP Walker, J Himes, and SM Grantham-McGregor. Effects of breakfast on classroom behaviour in rural Jamaican schoolchildren. Food Nutr Bull. 1996;17:1-10. DOI: 10.1177/156482659601700307

[32] E Jacoby, S Cueto, and E Pollitt. Benefits of a school breakfast programme among Andean children in Huaraz, Peru. Food Nutr Bull. 1996;17:1-11. DOI: 10.1177/156482659601700111

[33] E Pollitt, E Jacoby, and S Cueto. School breakfast and cognition among nutritionally at-risk children in the Peruvian Andes. Nutr Rev. 1996;54:785S-789S. DOI: 10.1111/j.1753-4887.1996.tb03894.x. [PMID:8700448]

[34] SM Grantham-McGregor, S Chang, and SP Walker. Evaluation of school feeding programs: some Jamaican examples. Am J Clin Nutr. 1998;67:785s-9s. DOI: 10.1093/ajcn/67.4.785S. [PMID:9537629]

[35] S de Pee, CE West, D Permaesih, and S Martuti. Muhilal, Hautvast JG. Orange fruit is more effective than are dark-green, leafy vegetables in increasing serum concentrations of retinol and beta-carotene in schoolchildren in Indonesia. Am J Clin Nutr. 1998;68:1058-67. DOI: 10.1093/ajcn/68.5.1058. [PMID:9808223]

[36] CA Powell, SP Walker, SM Chang, and SM Grantham-McGregor. Nutrition and education: a randomized trial of the effects of breakfast in rural primary school children. Am J Clin Nutr. 1998;68:873-9. DOI: 10.1093/ajcn/68.4.873. [PMID:9771865]

[37] J-P Tan, J Lane, and G Lassibille. Student outcomes in Philippine elementary schools: An evaluation of four experiments. World Bank Econ Rev. 1999;13:493-508. DOI: 10.1093/wber/13.3.493

[38] G Ma, X Hu, S Gao, and D Bai. Effect of energy intake at breakfast on school performance. Wei Sheng Yan Jiu. 1999;28:286-8. [PMID:12712699]

[39] JH Siekmann, LH Allen, NO Bwibo, MW Demment, SP Murphy, and CG Neumann. Kenyan school children have multiple micronutrient deficiencies, but increased plasma vitamin B-12 is the only detectable micronutrient response to meat or milk supplementation. J Nutr. 2003;133:3972S-3980S. DOI: 10.1093/jn/133.11.3972S. [PMID:14672298]

[40] JH Siekmann, LH Allen, MR Watnik, P Nestel, CG Neumann, and Y Shoenfeld. Titers of antibody to common pathogens: relation to food-based interventions in rural Kenyan schoolchildren. Am J Clin Nutr. 2003;77:242-9. DOI: 10.1093/ajcn/77.1.242. [PMID:12499348]

[41] SE Whaley, M Sigman, C Neumann, N Bwibo, D Guthrie, and RE Weiss. The impact of dietary intervention on the cognitive development of Kenyan school children. J Nutr. 2003;133:3965S-3971S. DOI: 10.1093/jn/133.11.3965S. [PMID:14672297]

[42] M Grillenberger, CG Neumann, SP Murphy, NO Bwibo, P van’t Veer, and JG Hautvast. Food supplements have a positive impact on weight gain and the addition of animal source foods increases lean body mass of Kenyan schoolchildren. J Nutr. 2003;133:3957S-3964S. DOI: 10.1093/jn/133.11.3957S. [PMID:14672296]

[43] X Du, K Zhu, A Trube, Q Zhang, G Ma, and X Hu. School-milk intervention trial enhances growth and bone mineral accretion in Chinese girls aged 10-12 years in Beijing. Br J Nutr. 2004;92:159-68. DOI: 10.1079/BJN20041118. [PMID:15230999]

[44] M Sigman, SE Whaley, CG Neumann, N Bwibo, D Guthrie, and RE Weiss. Diet quality affects the playground activities of Kenyan children. Food Nutr Bull. 2005;26:S202-12. DOI: 10.1177/15648265050262S211. [PMID:16075570]

[45] K Zhu, X Du, CT Cowell, H Greenfield, B Blades, and TA Dobbins. Effects of school milk intervention on cortical bone accretion and indicators relevant to bone metabolism in Chinese girls aged 10-12 y in Beijing. Am J Clin Nutr. 2005;81:1168-75. DOI: 10.1093/ajcn/81.5.1168. [PMID:15883444]

[46] K Zhu, Q Zhang, LH Foo, A Trube, G Ma, and X Hu. Growth, bone mass, and vitamin D status of Chinese adolescent girls 3 y after withdrawal of milk supplementation. Am J Clin Nutr. 2006;83:714-21. DOI: 10.1093/ajcn.83.3.714. [PMID:16522922]

[47] CG Neumann, SP Murphy, C Gewa, M Grillenberger, and NO Bwibo. Meat supplementation improves growth, cognitive, and behavioral outcomes in Kenyan children. J Nutr. 2007;137:1119-23. DOI: 10.1093/jn/137.4.1119. [PMID:17374691]

[48] S Muthayya, T Thomas, K Srinivasan, K Rao, AV Kurpad, and JW van Klinken. Consumption of a mid-morning snack improves memory but not attention in school children. Physiol Behav. 2007;90:142-50. DOI: 10.1016/j.physbeh.2006.09.025. [PMID:17081574]

[49] K Zhu, H Greenfield, X Du, Q Zhang, G Ma, and X Hu. Effects of two years’ milk supplementation on size-corrected bone mineral density of Chinese girls. Asia Pac J Clin Nutr. 2008;17:Suppl 1147-50. [PMID:18296324]

[50] Alderman H, Gilligan DO, Lehrer K. The impact of alternative food for education programs on School Participation and Education Attainment in Northern Uganda. Draft, World Bank, IFPRI, and University of British Columbia. 2008.

[51] Adelman S, Alderman H, Gilligan DO, Lehrer K. The impact of alternative food for education programs on learning achievement and cognitive development in Northern Uganda. Unpublished manuscript. 2008.

[52] Kazianga H, De Walque D, Alderman H. Educational and health impacts of two school feeding schemes: Evidence from a randomized trial in rural Burkina Faso. Washington, D.C.: The World Bank, 2009.

[53] TK Lien, BT Nhung, NC Khan, T Hop, NT Nga, and NT Hung. Impact of milk consumption on performance and health of primary school children in rural Vietnam. Asia Pac J Clin Nutr. 2009;18:326-34. [PMID:19786380]

[54] RP Tupe and SA Chiplonkar. Zinc supplementation improved cognitive performance and taste acuity in Indian adolescent girls. J Am Coll Nutr. 2009;28:388-96. DOI: 10.1080/07315724.2009.10718101. [PMID:20368377]

[55] K Mayurasakorn, P Sitphahul, and PO Hongto. Supplement of three eggs a week improves protein malnutrition in Thai children from rural areas. J Med Assoc Thai. 2010;93:301-9. [PMID:20420104]

[56] EM Omwami, C Neumann, and NO Bwibo. Effects of a school feeding intervention on school attendance rates among elementary schoolchildren in rural Kenya. Nutrition. 2011;27:188-93. DOI: 10.1016/j.nut.2010.01.009. [PMID:20471807]

[57] K Rahmani, A Djazayery, MI Habibi, H Heidari, AR Dorosti-Motlagh, and M Pourshahriari. Effects of daily milk supplementation on improving the physical and mental function as well as school performance among children: results from a school feeding program. J Res Med Sci. 2011;16:469-76. [PMID:22091261]

[58] M Vaz, M Pauline, US Unni, P Parikh, T Thomas, and AV Bharathi. Micronutrient supplementation improves physical performance measures in Asian Indian school-age children. J Nutr. 2011;141:2017-23. DOI: 10.3945/jn.110.135012. [PMID:21918063]

[59] H Alderman, DO Gilligan, and K Lehrer. The impact of food for education programs on school participation in northern Uganda. Econ Dev Cult Change. 2012;61:187-218. DOI: 10.1086/666949

[60] H Kazianga, D De Walque, and H Alderman. Educational and child labour impacts of two food-for-education schemes: Evidence from a randomised trial in rural Burkina Faso. J Afr Econ. 2012;21:723-60. DOI: 10.1093/jae/ejs010

[61] L Li, X Zhao, J Wang, T Muzhingi, PM Suter, and G Tang. Spirulina can increase total-body vitamin A stores of Chinese school-age children as determined by a paired isotope dilution technique. J Nutr Sci. 2012;1:e19. DOI: 10.1017/jns.2012.21. [PMID:25191548]

[62] M Kleiman-Weiner, R Luo, L Zhang, Y Shi, A Medina, and S Rozelle. Eggs versus chewable vitamins: Which intervention can increase nutrition and test scores in rural China? China Econ Rev. 2013;24:165-76. DOI: 10.1016/j.chieco.2012.12.005

[63] CG Neumann, NO Bwibo, L Jiang, and RE Weiss. School snacks decrease morbidity in Kenyan schoolchildren: a cluster randomized, controlled feeding intervention trial. Public Health Nutr. 2013;16:1593-604. DOI: 10.1017/S1368980013000876. [PMID:23537728]

[64] CG Neumann, L Jiang, RE Weiss, M Grillenberger, CA Gewa, and JH Siekmann. Meat supplementation increases arm muscle area in Kenyan schoolchildren. Br J Nutr. 2013;109:1230-40. DOI: 10.1017/S0007114512003121. [PMID:22856533]

[65] MM Cervo, LO Llido, EB Barrios, and LN Panlasigui. Effects of canned pineapple consumption on nutritional status, immunomodulation, and physical health of selected school children. J Nutr Metab. 2014;2014:861659. DOI: 10.1155/2014/861659. [PMID:25505983]

[66] JL Hulett, RE Weiss, NO Bwibo, OM Galal, N Drorbaugh, and CG Neumann. Animal source foods have a positive impact on the primary school test scores of Kenyan schoolchildren in a cluster-randomised, controlled feeding intervention trial. Br J Nutr. 2014;111:875-86. DOI: 10.1017/S0007114513003310. [PMID:24168874]

[67] Diagne A, Moustapha Lo M, Sokhna O, Diallo FL. Evaluation of the impact of school canteen programs on internal efficiency of schools, cognitive acquisitions and learning capacities of students in rural primary schools in Senegal. PEP Working Paper Series 2013-14. 2014.

[68] M van der Hoeven, M Faber, J Osei, A Kruger, and CM Smuts. Effect of African leafy vegetables on the micronutrient status of mildly deficient farm-school children in South Africa: a randomized controlled study. Public Health Nutr. 2016;19:935-45. DOI: 10.1017/S1368980015002037. [PMID:26146039]

[69] R Lee, L Singh, D van Liefde, M Callaghan-Gillespie, M Steiner-Asiedu, and K Saalia. Milk Powder Added to a School Meal Increases Cognitive Test Scores in Ghanaian Children. J Nutr. 2018;148:1177-84. DOI: 10.1093/jn/nxy083. [PMID:29905824]

[70] S Adelman, DO Gilligan, J Konde-Lule, and H Alderman. School Feeding Reduces Anemia Prevalence in Adolescent Girls and Other Vulnerable Household Members in a Cluster Randomized Controlled Trial in Uganda. J Nutr. 2019;149:659-66. DOI: 10.1093/jn/nxy305. [PMID:30926996]

[71] A Gelli, E Aurino, G Folson, D Arhinful, C Adamba, and I Osei-Akoto. A School Meals Program Implemented at Scale in Ghana Increases Height-for-Age during Midchildhood in Girls and in Children from Poor Households: A Cluster Randomized Trial. J Nutr. 2019;149:1434-42. DOI: 10.1093/jn/nxz079. [PMID:31100125]

[72] S Anitha, J Kane-Potaka, TW Tsusaka, D Tripathi, S Upadhyay, and A Kavishwar. Acceptance and Impact of Millet-Based Mid-Day Meal on the Nutritional Status of Adolescent School Going Children in a Peri Urban Region of Karnataka State in India. Nutrients. 2019;11:2077 DOI: 10.3390/nu11092077. [PMID:31484377]

[73] KV Bailey. Rural nutrition studies in Indonesia. IX. Feeding trial on schoolboys. Trop Geogr Med. 1962;14:129-39. [PMID:13864022]

[74] C Powell, S Grantham-McGregor, and M Elston. An evaluation of giving the Jamaican government school meal to a class of children. Hum Nutr Clin Nutr. 1983;37:381-8. [PMID:6654697]

[75] GB Spurr and JC Reina. Marginal malnutrition in school-aged Colombian girls: dietary intervention and daily energy expenditure. Hum Nutr Clin Nutr. 1987;41:93-104. [PMID:3570874]

[76] DK Agarwal, KN Agarwal, and SK Upadhyay. Effect of mid-day meal programme on physical growth & mental function. Indian J Med Res. 1989;90:163-74. [PMID:2767740]

[77] LM Richter, C Rose, and RD Griesel. Cognitive and behavioural effects of a school breakfast. S Afr Med J. 1997;87:93-100. [PMID:9180808]

[78] A Hall, TTM Hanh, K Farley, TPN Quynh, and F Valdivia. An evaluation of the impact of a school nutrition programme in Vietnam. Public Health Nutr. 2007;10:819-26. DOI: 10.1017/S1368980007382530. [PMID:17381906]

[79] H Alaofè, J Zee, R Dossa, and HT O’Brien. Education and improved iron intakes for treatment of mild iron-deficiency anemia in adolescent girls in southern Benin. Food Nutr Bull. 2009;30:24-36. DOI: 10.1177/156482650903000103. [PMID:19445257]

[80] O Ohiokpehai, DM David, and J Kamau. The effects of soybean (Glycine max) and pigeon pea (Cajanus cajan) food mixtures on the nutritional status of school children in Suba District, Kenya. J Food Agric Environ. 2009;7:59-63.

[81] H Joulaei, E Nwagwu, A Nasihatkon, L Azadbahkt, R Shenavar, and V Keshtkar. To assess the effects of nutritional intervention based on advocacy approach on malnutrition status among school-aged children in Shiraz. J Res Med Sci. 2013;18:739-45. [PMID:24381614]

[82] S Lin, X Hu, F Zhang, Q Ruan, W Tang, and L Tao. Effects of egg and milk supplementation on growth and development among children in poor rural area. Zhonghua Yu Fang Yi Xue Za Zhi. 2015;49:733-7. [PMID:26733034]

[83] JI Baum, JD Miller, and BL Gaines. The effect of egg supplementation on growth parameters in children participating in a school feeding program in rural Uganda: a pilot study. Food Nutr Res. 2017;61:1330097. DOI: 10.1080/16546628.2017.1330097. [PMID:28659739]

[84] AM Adams, R Ahmed, AH Latif, S Rasheed, SK Das, and E Hasib. Impact of fortified biscuits on micronutrient deficiencies among primary school children in Bangladesh. PLoS One. 2017;12:e0174673. DOI: 10.1371/journal.pone.0174673. [PMID:28380009]

[85] MD El Harake, S Kharroubi, SK Hamadeh, and L Jomaa. Impact of a Pilot School-Based Nutrition Intervention on Dietary Knowledge, Attitudes, Behavior and Nutritional Status of Syrian Refugee Children in the Bekaa, Lebanon. Nutrients. 2018;10:913 DOI: 10.3390/nu10070913. [PMID:30018221]

[86] Krishnaratne S, White H, Carpenter E. Quality education for all children? What works in education in developing countries. New Delhi: International Initiative for Impact Evaluation (3ie), Working Paper. 2013;20:155.

[87] B Snilstveit, E Gallagher, D Phillips, M Vojtkova, J Eyers, and D Skaldiou. PROTOCOL: Interventions for improving learning outcomes and access to education in low- and middle-income countries: a systematic review. Campbell Syst Rev. 2017;13:1-82. DOI: 10.1002/CL2.176

[88] M Fernandes, R Galloway, A Gelli, D Mumuni, S Hamdani, and J Kiamba. Enhancing linkages between healthy diets, local agriculture, and sustainable food systems: the school meals planner package in Ghana. Food Nutr Bull. 2016;37:571-84. DOI: 10.1177/0379572116659156. [PMID:27435223]

[89] de Inestrillas R, Kauffmann D, Piccini S. Mastro Cd, Momcilovic P. PLUS School Menus: the future of school menu design. UNSCN News. 2020:75-82.

[90] T Greenhalgh, E Kristjansson, and V Robinson. Realist review to understand the efficacy of school feeding programmes. BMJ. 2007;335:858-61. DOI: 10.1136/bmj.39359.525174.AD. [PMID:17954518]

Correspondence to:
Dongqing Wang, Ph.D., M.P.H.
Research Associate
Department of Global Health and Population
Harvard T.H. Chan School of Public Health
90 Smith Street, 310 WS-5
Boston, Massachusetts
USA 02120
[email protected]