Table 2.
Characteristics of studies conducted in other (non-U.S.) OECD countries included in the systematic review.
| Author, Year | Location; Participant Characteristics | Study Design | Year(s) | Universal Meal Provision | Outcome Measure(s) | Results | Risk of Bias 1 |
|---|---|---|---|---|---|---|---|
| Andersen et al. 2014 [82] | Denmark; 9 schools (3–4th grade students) assigned to free lunch (3 months) and packed lunch from home (3 months); n = 834 students | QE: Pre/post (with control) | 2011–2012 SY | Free school lunch (+ snacks) | Diet: Foods and nutrients consumed (measured using a validated 7-day food record tool [self-administered] | Free school lunches was associated with improved diets, including higher intakes of, vegetables (16% higher intake; p < 0·0001) and fish (48% higher; p < 0·0001, which resulted in higher intakes of vitamin D (42% higher; p < 0·0001) and iodine (11% higher; p < 0·0001). Additionally, students consumed significantly less saturated fat (30% lower; p < 0·0001). There were no significant differences in total calories. | Low |
| Asakura et al. 2017 [34] | Japan; 14 elementary schools (n = 629 students) and 13 junior high schools (n = 281 students) | CS | 2014 | Universal school lunches | Diet: Diet records completed by parents/guardians on three non-consecutive days (two school days and one weekend day) + plate waste at school | School lunches were positively associated with total diet quality (the prevalence of inadequate nutrient intakes was higher on weekend days compared with school days for almost all of the nutrients assessed). | High |
| Ask et al. 2006 [70] | Norway; 10th grade students in one school with 1 intervention classroom with free breakfasts for 4 months (n = 26 students) and 1 control classroom (n = 28 students) | QE: Pre/post (with control) | 2005 | Pilot free breakfast intervention | (1) Diet: Diet quality (measured using a non-validated FFQ, which was used to calculate overall HEI scores) (2) BMI: objective measurements by school nurse |
(1) Free breakfast was positively associated with overall HEI scores among male students (16 pt increase in HEI score; p < 0.05) (2) No changes in BMI were observed among students with free breakfast, but was significantly higher among control students who did not receive a school breakfast |
Very High |
| Ask et al. 2010 [74] | Norway; 9th grade students (1 intervention school with free lunches for 4 months [n = 58 students] and 1 control school [n = 92 students]) | QE: Pre/post (with control) | 2007 | Pilot free lunch intervention | (1) Diet: Healthy food scores (measured using a non-validated FFQ,) (2) BMI: objective measurements by research team |
(1) No association with food scores (2) No association with BMI |
Very High |
| Bartelink et al. 2019 [98] | Netherlands; 4 intervention and 4 control schools (n = 1676 children age 4–12 years). | QE: Pre/post (with control) | 2015–2017 | Free school lunch (+ structured PA after lunch) | BMI: objective measurements by research team | Free school lunch ( + PA) was associated with ↓ BMI z-score after two years of follow-up (standardized effect size = −0.083, p = 0.01) | Low |
| Dalma et al. 2020 [85] | Greece; 28 intervention (n= 1442 students) and 23 control primary schools (n= 986 students) | Cluster RCT | 2014–2015 SY | Free lunch (+ nutrition education) | Diet: Food insecurity (measured using the Food Security Survey Module [FSSM]; parent report) | Free school lunch was associated with ↓ food insecurity (average FSSM score decrease of 0.31 points; p = 0.045), with the greatest reduction observed among food insecure households with hunger (average decrease of 1.04 points; p = 0.023). | Low |
| Gatenby 2011 [79] | England; two primary schools (one higher- and one lower-income [147 students ages 8–11, and a subsample of n = 20 students with food diaries]) | CS | 2004 | Universal free lunch | Diet: Plate waste + Food diaries (5 days) and photos taken by students | Students who received school meals consumed significantly less at lunch on average compared with students who brought meals from home. However, due to compensation outside of lunch, there were no differences in overall nutrients consumed throughout the day. | Very High |
| Holford 2015 [63] | Scotland; all primary schools (students ages 4–11 years) | QE: Pre/post (with control) | 2003–2013 | Universal free school lunch | Participation | Universal free school lunch was associated with an ↑ on participation among students previously eligible for free school meals (3.3% pt; p < 0.05) as well as among students not previously eligible (14.4% pt; p < 0.001) | Low |
| IlløKken et al. 2017 [75] | Norway; one intervention elementary school with students receiving free school lunch for six months (n = 55 students) and one control school (n = 109 students); students ages 10–12 years | QE: Pre/post (with control) | 2014–2015 SY | Free school lunch | Diet: Healthy food scores (measured using an FFQ) | Free school lunch was associated with ↑ in healthy food scores (change in total healthy food score of 1.7 vs. 0.5; p < 0.01), primarily due to an increased frequency of consuming fruits (p < 0.01), vegetables (p < 0.01), and fish spread (p = 0.02). | High |
| Jenkins et al. 2015 [73] | Wales; 111 primary schools randomly assigned to free school breakfast (n = 55 schools) or control (delayed intervention [n = 56 schools]); students ages 9–11 years | Cluster RCT | 2004–2005 to 2006–2007 SY | Primary School Free Breakfast Initiative | Diet: 24-h recalls | There were no differences in the nutritional quality of breakfasts consumed at school or at home, except significantly higher levels of selenium (5.1 μg vs. 3.2 μg; p < 0.01 and carbohydrates (59.8 g vs. 48.7 g; p < 0.01) in school meals. Among students who ate a school breakfast, 49% had already consumed a breakfast at home that morning, although there were no significant differences in caloric intake over a 24-h period. | High |
| Laursen et al. 2015 [90] | Denmark; 9 schools (3–4th grade students) assigned to free lunch (3 months) and packed lunch from home (3 months); n = 797 students | Cluster RCT | 2011–2012 SY | Free school lunch (+ snacks) | Attendance | No association with attendance rates | Low |
| MacLardie et al. 2008 (Scottish Govt report) [62] | Scotland; 5 local authorities | QE: Pre/post (no control) | 2007–2008 SY | Free school meals trial for P1-P3 pupils (universal free lunch) | (1) Participation (2) Finances |
(1) An ↑ in participation of 22% pts was observed in schools with free school lunches, with the greatest increases among students not previously registered for free school meals (28% pts). An increase in participation was also observed among students previously eligible for free school meals (4% pt increase). (2) The cost of implemented school meals varied from £1.79 to £4.65 per additional meal. Costs tended to be lower in areas with a greater number of additional meals served. |
Very High |
| Mhurchu et al. 2012 [84] | New Zealand; 14 primary schools with staggered implementation of free school breakfasts (n = 424 students ages 5–13 years) | Cluster RCT | 2010 | Free school breakfast program | (1) Diet: Short-term hunger (measured using satiety scale for children [self-report]), student breakfast habits (parent-report), and child/household food security (measured using the CCHIP Scale [parent-report]). (2) Attendance (3) Academic Performance: standardized tests of math and literacy |
(1) Free school breakfast was associated with a ↓ in children’s self-reported short-term hunger (increase of 8.6 units on the satiety scale; p = 0.001). No association with child or household food security, or breakfast frequency. (2) No association with overall attendance. In a sub-analysis examining students who frequently attended the SBP (≥50% of the time), free school breakfast was associated with a 1.6% increase in attendance (p = 0.016). (3) No association with academic performance |
Low |
| Moore et al. 2014 [72] | Wales; 111 primary schools randomly assigned to free school breakfast (n = 55 schools) or control (delayed intervention [n = 56 schools]), students ages 9–11 years | Cluster RCT | 2004–2005 to 2006–2007 SY | Primary School Free Breakfast Initiative | (1) Diet: healthy food consumption and breakfast skipping (measured using a validated dietary recall questionnaire) (2) Academic Performance: Cognitive tests administered in classrooms |
(1) Free breakfast was associated with an ↑ in the number of healthy items at breakfast (0.25 more servings of healthy foods [p < 0.01]), with greater improvements observed in lower-income schools. While there was no overall association with breakfast skipping, there was a significant reduction in breakfast skipping among children from lower-income schools and households (p < 0.05). (2) No association with cognitive tests. |
High |
| Munday et al. 2017 [81] | New Zealand; one kindergarten class (n = 17 students); 2.5 months of exposure to intervention | QE: Pre/post (no control) | 2014 | Free lunches + educational component | (1) Diet: Foods and nutrients consumed (measure using a 24-h modified dietary recall questionnaire data and a vegetable- and fruit-specific FFQ [teacher and parent-report]) (2) Attendance |
(1) Free lunch was associated with ↓ in snack food consumption at school (p = 0.015). No association with overall nutrients. (2) No association with attendance. |
Very High |
| Murphy et al. 2011 [71] | Wales; 111 primary schools randomly assigned to free school breakfast (n = 55 schools) or control (delayed intervention [n = 56 schools]), students ages 9–11 years | Cluster RCT | 2004–2005 to 2006–2007 SY | Primary School Free Breakfast Initiative | (1) Diet: healthy food consumption and breakfast skipping (measured using a validated dietary recall questionnaire) (2) Academic Performance: Cognitive tests administered in classrooms |
(1) Free breakfast was associated with an ↑ in the number of healthy items at breakfast (0.23 more servings of healthy foods [p < 0.01]), with greater improvements observed in lower-income schools. No association with breakfast skipping. (2) No association with cognitive tests. |
High |
| Sabinsky et al. 2018 [83] | Denmark; 4 intervention school and 4 control schools (n = 984 students in grades 2–6; students ages 7–13 years | QE: Pre/post (with control) | 2008 | Free school lunches | Diet: Diet quality (measured using digital photography on 3 consecutive days + a validated Meal Index of dietary Quality (Meal IQ) | Free meals were associated with ↑ dietary quality of the lunch eaten compared with packed lunches (p = 0.004), due in part to reductions in saturated fat and snacks and increases in vegetables and fish. When the school meals were not provided for free, selection of these meals was limited and no difference in dietary quality was observed. | Low |
| Spence et al. 2020 [78] | England; Two primary schools (students age 4–7 years) before and after implementation of UIFSM | QE: Pre/post (no control) | 2008–2009 SY and 2017–2018 SY | UIFSM | Diet: Foods and nutrients consumed (measuring using a validated 24-hr food diary on four consecutive days) | UIFSM was associated with ↓ consumption of non-milk extrinsic sugar (i.e., added sugar [mean change −4.6%, p < 0.001]) and biscuits (i.e., shortbread cookies [−0.4, p < 0.001]) at lunch. The reductions in added sugar were observed in students’ overall diets as well (−3.8%, p < 0.001). However, an ↑ in cakes/sweet puddings were observed which were offered with school lunches (after the main meal) daily. | High |
| Petralia et al. 2016 [86] | Greece; 162 schools provided with free lunches (primary and secondary schools) | QE: Pre/post (no control) | 2012–2013 SY | Free lunch (+ nutrition education) | Diet: Food insecurity (measured using the Food Security Survey Module [FSSM]; parent report) | Free school lunch was associated with ↓ food insecurity (decrease from 64.2% of households with food insecurity to 59.1%; p < 0.001). The greatest reductions were observed among food insecure households with hunger; each additional month of free school meals was associated with a 13% increase in the odds of not reporting hunger problems (OR 1.13, 95% CI 1.02–1.25). | High |
| Vik et al. 2019 (BMC Public Health) [76] | Norway; one intervention elementary school with students receiving free school lunch for one year (n = 55 students) and one control school (n = 109 students); students ages 10–12 years | QE: Pre/post (with control) | 2014–2015 SY | Free School Lunches | (1) Diet: Diet quality (measured using validated FFQs, with results used to calculate healthy food scores) (2) BMI: objectively measured at school |
(1) Free school lunches were associated with ↑ healthy food scores (F = 10.941, p =0.001) after one year of exposure, with the greatest increases among lower-SES students. (2) Free school lunches were associated with ↑ BMI z-scores (F = 10.007, p = 0.002) after one year of exposure. |
Low |
| Vik et al. 2019 (BMC Res Notes) [77] | Norway; one intervention elementary school with students receiving free school lunch for one year (n = 55 students) and one control school (n = 109 students); students ages 10–12 years | QE: Pre/post (with control) | 2014–2015 SY | Free School Lunches | Diet: Frequency of consuming meals (measured using a validated questionnaire) | There was no association between free school meals and meal frequency after 1 year of exposure. | Low |
| Yamaguchi et al. 2018 [80] | Japan; Four municipalities, n = 719 elementary school children (ages 6–12 years). | CS | 2013 | Universal school lunch | Diet: dietary habits (measured using the validated brief diet history questionnaire-10 years old [BDHQ-10]; self-administered) | Universal school lunches were associated with a reduction in SES-related disparities in children’s diets (a reduction in the inequality of vegetable intake by 9.9% and fruit intake by 3.4%) | Low |
CS: Cross-Sectional; FFQ: Food Frequency Questionnaire; HEI: Healthy Eating Index; QE: Quasi-Experimental; RCT: Randomized controlled trial. 1 Risk of Bias was based on adapted Newcastle–Ottawa Scales (NOS) for cross-sectional and cohort studies (Supplemental Tables S1 and S2).