RATIONALE: School feeding aims to alleviate hunger and enhance child outcomes. Since the first Cochrane systematic review of school meals, there has been a marked increase in studies and reviews of school feeding programs. However, most systematic reviews are geographically limited and use qualitative analysis. We reviewed worldwide papers and performed several meta-analyses, providing a more comprehensive picture of the effectiveness of school feeding.
OBJECTIVES: 1. To assess effectiveness of school feeding programs for improving the physical and psychological health of children experiencing socioeconomic disadvantage worldwide. 2. To assess effectiveness of school feeding programs for improving the health of children experiencing socioeconomic disadvantage compared with children who are more advantaged.
SEARCH METHODS: We searched 17 subject-specific and multidisciplinary databases and registries up to November 2023. In November 2024, two Information Specialists ran a top-up search for randomized controlled trials (RCTs). We handsearched references of included studies and relevant reviews.
ELIGIBILITY CRITERIA: We included individually randomized, cluster-randomized, and cross-over trials, as well as longitudinal non-randomized studies of interventions (NRSIs). Studies had to compare the provision of free or reduced-price food in schools versus no school feeding, focusing on socioeconomically disadvantaged children. The food had to contain at least 3% of the daily energy requirement and at least 10% of the daily protein requirement for the specified age group(s). Eligible participants were primary or secondary school students aged five to 19 years.
OUTCOMES: Our critical outcomes were change in: math achievement, reading achievement, attendance, enrollment, height-for-age z-score (HAZ), weight-for-age z-score (WAZ), and overweight/obesity. Our important outcomes were change in: overall academic achievement, fluid intelligence, working memory, behavioral/emotional outcomes, height, weight, and anemia. We planned to study changes between baseline and final outcomes. In one study with extreme contamination, we used the first follow-up.
RISK OF BIAS: We assessed the risk of bias for RCTs by outcome using the appropriate version of the Cochrane risk of bias tool (RoB 2): RoB 2 for individually randomized trials, for cluster-RCTs, and for cross-over trials. We evaluated the quality of NRSIs using the School-based Measurement & Assessment of Results Tool (SMART), which is adapted from the Newcastle Ottawa Scale.
SYNTHESIS METHODS: We used standardized mean differences (SMDs) with 95% confidence intervals (CIs) for educational and cognitive outcomes. We used mean differences (MDs), odds ratios (ORs), or incidence rate ratios (IRRs) for others. All meta-analyses used random-effects generic inverse variance. We conducted equity subgroup analyses by sex and socioeconomic status. We used GRADE to assess our confidence in the evidence for critical outcomes reported in RCTs.
INCLUDED STUDIES: We included 40 studies with 83 reports. There were 13 RCTs (12 cluster-randomized trials and one individually randomized trial) and 27 NRSIs. Most studies (34) were from low- and middle-income countries (LMICs). In total, there were more than 91,885 students (four studies didn't report sample size). One study involved 59,613 students, while the others had between 60 and 6038 students. The studies included 48 outcomes; we meta-analyzed or reported 14.
SYNTHESIS OF RESULTS: Low- and middle-income countries (LMICs) School feeding compared to no school feeding improves math achievement slightly (SMD 0.14, 95% CI 0.06 to 0.23; P = 0.001; I² = 68%; 6 cluster-RCTs, 5587 participants; high-certainty evidence), but may have little to no effect on reading achievement (SMD 0.02, 95% CI -0.06 to 0.11; P = 0.61; I= 54%; 3 cluster-RCTs, 3417 participants; low-certainty evidence). School feeding programs lead to a slight increase in enrollment rates (MD 3.44% increase, 95% CI 0.83 to 6.04; P = 0.01; I² = 0%; 2 cluster-RCTs, 5200 participants, high-certainty evidence) but may have little to no effect on attendance (MD 0.17%, 95% CI -2.64% to 2.97%; P = 0.91; I= 81%; 3 cluster-RCTs, 3566 participants; low-certainty evidence). School feeding programs likely lead to a slight increase in HAZ (MD 0.06, 95% CI 0.03 to 0.09; P < 0.001; I= 0%; 2 cluster-RCTs, 3678 participants; moderate-certainty evidence) and WAZ (MD 0.08, 95% CI 0.05 to 0.12; P < 0.001; I= 0%; 3 cluster-RCTs, 2132 participants; moderate-certainty evidence). Two cluster-RCTs assessed change in obesity/overweight. One 10-month study found that the odds of being overweight/obese were 53% lower among adolescents receiving school meals compared to control adolescents (OR 0.47, 95% CI 0.30 to 0.72). Another study found no cases of overweight/obesity before or after the intervention. These findings are of very low certainty. Some researchers encountered implementation problems beyond their control, including conflicts, delays, and bureaucratic decisions. Heterogeneity in contexts, outcomes, child populations, and statistics was a limitation of this review. High-income countries (HICs) One NRSI found very uncertain evidence that children assigned to a breakfast club increased their attendance by more than 1.6% on average compared to children in the control group. Equity analyses The equity (subgroup) analyses by sex and by socio-economic status were all non-significant. There were only two studies in each subgroup analysis; they were likely underpowered. 2 2 2 2
AUTHORS' CONCLUSIONS: In LMICs, school feeding programs lead to a slight improvement in math achievement, but may have little to no effect on reading achievement. School feeding programs lead to a slight increase in enrollment, but may have little to no impact on attendance. They likely lead to slight gains in HAZ and WAZ. There may be little to no association between school feeding and overweight/obesity, but the evidence is very uncertain. We recommend that researchers and policymakers view research as an integral part of the implementation process. To reduce heterogeneity in outcomes, we recommend even greater co-ordination of research, and that researchers and interest holders work together to identify a core set of outcomes.
FUNDING: The authors would like to thank the following donors for their generous support in making this review possible: Dubai Cares, the World Food Programme's School Meals and Social Protection Service, and the Research Consortium for School Health and Nutrition.
REGISTRATION: Protocol (2022): https://doi.org/10.1002/14651858.CD014794 Original review (2007): https://doi.org/10.1002/14651858.CD004676.pub2 Original Campbell protocol (2006): doi.org/10.1002/CL2.12.
