Table 1.
Characteristics of studies conducted in the United States included in the systematic review.
| Author, Year | Location; Participant Characteristics | Study Design | Year(s) | Universal Meal Provision | Outcome Measure(s) | Results | Risk of Bias 2 |
|---|---|---|---|---|---|---|---|
| Adams et al. 2020 [68] | Virginia; 6 Title I elementary schools (grades 1–5), n = 1155 plate waste measurements | CS | 2016 | CEP | Diet: Added sugar consumption at lunch (measured using plate waste determined using digital photography) | In CEP schools, foods selected had on average 11.2 g of added sugar and beverages had on average 11.0 g of added sugar. Students consumed on average 6.6 g of added sugar from foods and 3.6 g of added sugar from beverages (~10% of calories consumed from foods and ~35% calories consumed from beverages; ~2.5% of added sugars consumed out of the 10% recommend by the Dietary Guidelines for Americans) | Low |
| Bartfeld et al. 2019 [87] | Wisconsin; elementary schools throughout the state with varying breakfast models (including universal free and mean-tested) | QE: Pre/post (with control) | 2009–2010 to 2013–2014 SY | USBP | (1) Attendance: percent of school days attended and low attendance (i.e., attending fewer than 95% of available days) measured among student in grades 1–5 (2) Academic Performance: test scores in reading and math measured among student in grades 3–5 |
(1) USBP was not associated with attendance overall, but in sub-analyses was associated with increased attendance among low-income students; USBP was associated with a 0.24% pt ↑ in the % of days attended (p = 0.023) and a 3.5% pt ↓ in the percent of students with low attendance (p <0.001) (2) USBP was not associated with academic performance overall, but in sub-analyses was associated with 0.07 SD high math scores (p = 0.001) and 0.04 SD higher reading scores (p = 0.035) among higher-income students |
Low |
| Bartfeld et al. 2020 [89] | Wisconsin; 37 CEP elementary schools and 108 comparison (i.e., eligible non-CEP) elementary schools (grade 1–5) | QE: Pre/post (with control) | 2013–2014 to 2015–2016 SY | CEP | Attendance: percent of school days attended and low attendance (i.e., attending fewer than 95% of available days) | After two years of exposure, CEP was associated in a 3.5% pt ↓ in low attendance (p = 0.045) compared with control schools. In sub-analyses, CEP was associated with a 4.2% pt ↓ in the probability of low attendance (p = 0.035) among lower-income students. | Low |
| Bernstein et al. 2004 [51] (USDA Report) |
USA; Six school districts (in six states); elementary schools within each district randomly assigned to USBP (n = 79 schools) or control (maintain means-tested SBP; n = 74 schools). | QE: Pre/post (with control) | 1999–2000 to 2002–2003 SY | USBP | (1) Participation
(2) Attendance (3) Academic Performance: test scores in reading and math |
(1) Offering free school breakfasts was associated with an ↑ in breakfast participation that was maintained for three years (a 15% pt gain after three years; p = 0.01). (2) No association with attendance (3) No association with test scores |
Low |
| Brown 2009 [57] (GAO Report) | USA; 5 states and 14 districts in other states that implemented ERP | QE: Pre/post (no control) | 2007–2008 SY | ERP | (1) Participation
(2) Finances |
(1) ERP was associated with an ↑ in participation in the SBP (9% average increase) and NSLP (11% average increase) among students who were eligible for reduced-price meals (2) Federal reimbursements only partially offset programs costs for the participation states/school districts |
Very High |
| Crepinsek et al. 2006 [66] | USA; national sample of elementary schools (153 matched schools in six school districts with USBP or means-tested breakfast [n = 4358 students, grades 2–6]) | Cluster RCT | 1999–2000 to 2002–2003 |
USBP | Diet: food and nutrient intakes (measured using one 24-h recall) | USBP was positively associated with the consumption of a nutritionally substantive breakfast (80% vs. 76%; p < 0.01), including increased servings of fruit and dairy. There was no association with overall breakfast skipping or overall dietary intakes over a 24-h period. | Low |
| Dykstra et al. 2016 [54] | Philadelphia, PA; 16 schools (students grade 4–6; n = 821 student/parent dyads) with USBP | CS | 2013 | USBP | (1) Participation
(2) Diet: Breakfast skipping (measured using the Breakfast Patterns Survey [student self-report]) |
(1) On the day of data collection, 38.8% of students reported consuming a school breakfast and participating in the SBP on 32.1% of possible days (with 87.0% of students participating in the SBP at least 1 day during the fall semester). There was significantly higher participation among minority students (Black students participated on 36.5% of days, Hispanic students participated on 25.0% of days, and white students participated on 18.7% of days (p < 0.001). No differences in SBP participation by free or reduced-price eligibility or by food insecurity levels. (2) 16.9% of students reported skipping breakfast on the morning of data collection. Rates of skipping breakfast did not differ between students from food-insecure households and students from food-secure households. |
High |
| Gordanier et al. 2020 [88] | South Carolina; elementary and middle schools throughout the state that adopted CEP vs. non-CEP schools (both eligible and non-eligible schools), students grade 3–8 | QE: Pre/post (with control) | 2013–2014 to 2015–2016 SY | CEP | (1) Attendance
(2) Academic Performance: State standardized test scores (Math and English Language Acquisition] |
(1) CEP was associated with a ↓ in absences among elementary students (−0.231 days per year; p < 0.05). No significant associations were observed with absences among middle school students. (2) CEP was associated with an ↑ in math test scores among elementary students (0.06 SD; p < 0.01). No significant associations were observed for math scores among middle school students nor for reading scores for elementary or middle school students. |
Low |
| Gross et al. 2019 [19] | Maryland; One district with 5 CEP schools and one matched control district with 3 schools (CEP-eligible but not participating), n = 427 household surveys | CS | 2017 | CEP | Diet: Food insecurity (measured using the USDA Six-Item Short Form of the Food Security Survey Module [parent report] | CEP was associated reduced odds of household food insecurity (i.e., students had twice the odds of being in a food-insecure household if they attended a school that was CEP-eligible but not participating [OR 2.85, 95% CI 1.67, 4.88]). | High |
| Khan et al. 2011 [53] | Vermont; one middle school (grades 6–8) with USBP, n = 373 students | CS | 2005 | USBP | Diet: Food insecurity (measured using a 9-item validated survey [student self-report]) | Food insecure children were significantly less likely to eat breakfast at home compared with food secure children (32.9% vs. 18.6% of students did not eat breakfast at home; p = 0.007), and 91.3% of food insecure students reported eating breakfast at school. | High |
| Kleinman et al. 2002 [67] | Boston, Massachusetts; three schools before and after implementing USBP (n = 97 students in grades 4–6) | QE: Pre/post (no control) | 1998–1999 to 1999–2000 SY | USBP | (1) Diet: Nutrient intakes and hunger (measured using 24-h dietary recalls and 5-item version of the Child Hunger Index Child Report survey [student self-report] + an 8-item hunger/food insufficiency questionnaire [parent report]) (2) Attendance (3) Academic Performance: test scores in math, reading, science, and social studies (based on school records) |
(1) USBP was not associated with differences in the percent of students who were nutritionally at risk overall, but children who had improvements in nutritional status had significant ↑ in USBP participation (p < 0.001) and ↓ in self-reported hunger (mean change in hunger score of −1.4; p < 0.0001). (2) Among students with improved nutrient intakes, there was a significant ↓ in absences (−4.4 days absent; p < 0.01). (3) Among students with improved nutrient intakes, there was a positive association with math grades (mean change 0.6; p < 0.05). No other significant associations were observed with grades. |
High |
| Leos-Urbel et al. 2013 [48] | New York City, New York; elementary and middle schools before and after implementing USBP (n = 723,843 students in grades 3–8) | QE: Pre/post (no control) | 2002–2003 to 2003–2004 SY | USBP | (1) Participation
(2) Attendance (3) Academic performance: scores in statewide English and math tests |
(1) USBP was associated with ↑ breakfast participation (an increase of 5% among students previously eligible for free meals [p < 0.05])., 21% among students previously eligible for reduced price meals [p < 0.01], and 36% among students previously eligible for full-price meals [p < 0.01]). (2) There was no association with overall attendance rates. In sub-analyses, universal free breakfast was associated with a small ↑ in attendance among low-income black students (0.37%; p < 0.01) and higher-income Asian students (0.25%; p < 0.05). (3) There was no association with academic performance. |
Low |
| Logan et al. 2014 [37] (Report to USDA) | National; 7 states (285 participating LEAs 1 and 528 matched non-participating LEAs) | QE: Pre/post (with control) | 2009–2010 to 2012–2013 SY | CEP | (1) Participation (2) Finances |
(1) CEP was associated with ↑ participation (5.2% increase in NSLP participation, p < 0.01; 9.4% increase in SBP participation, p < 0.01) (2) CEP was associated with ↑ federal reimbursement (5.6% for NSLP [3.5% pts, p < 0.01]; 1.9% for SBP [3.5% pts, p < 0.01]). CEP was also associated with ↓ in time spent by staff distributing/processing applications for FRP meals (p < 0.01) and verifying eligibility of students (p < 0.01), resulting in a combined savings of on average 68 min/student annually, which represents a labor saving cost of approximately USD 29/student annually (partially offset by increases in staff time to claim reimbursable meals [increase of 30 min/student annually, p < 0.01]. Federal funding (reimbursements) per student ↑ by 13.5% (USD 5.33/student annually p < 0.01), and there was no impact on non-Federal finances (e.g., state reimbursement or student payments for non-reimbursable meals) resulting in potential net gains for LEAs participating in CEP. |
Low |
| McLaughlin et al. 2002 [50] (USDA Report) |
USA; Six schools districts (in six states); elementary schools within each district randomly assigned to USBP (n = 79 schools) or control (maintain means-tested SBP; n = 74 schools). | Cluster RCT | 1999–2000 to 2000–2001 SY | USBP | (1) Participation
(2) Diet: Frequency of breakfast and nutrients consumed (measured by 24 h recall [with parent assistance]) (3) Attendance (4) Academic Performance: test scores for math and reading and cognitive tests (5) BMI: objective measurements at school (6) Finances |
(1) USBP was associated with 16% pt ↑ in participation (p = 0.01), with the largest increases among students not previously eligible for free or reduced-price breakfasts. (2) Students in schools with USBP were more likely to eat breakfast on all five school days (p < 0.01) but there was no difference observed in most nutrients consumed over the course of a day. (3) There were no differences observed in attendance. (4) There were no differences observed in math or reading score gains nor cognitive functioning. (5) There were no differences observed in the prevalence of overweight. (6) The increases in breakfast participation resulted in lower per-meal labor costs in schools with USBP, with the increases in labor costs offset by the increase in meals served (average cost per breakfast served was USD 0.11 lower). |
High/ Low2 |
| Poblacion et al. 2017 [69] | USA; national dataset of households with children and school meal participation rates to model potential impact of CEP | SM | 2014 | CEP | Diet: food insecurity (measured using simulation modeling based on change in income-to-poverty ratios of food-insecure people in households with children using prevalence estimates from national data) | Free lunches from CEP was associated with an estimated increase of 3.73% of students becoming food secure (due to families increasing their food purchasing power). When examining the combined impact of USBP and NSLP with CEP, the estimated increase in purchasing power was associated with 3.23% of food insecure households with children becoming food secure. | Low |
| Pokorney et al. 2019 [58] | Pennsylvania; all CEP schools (n = 654) and eligible non-CEP schools (n = 1221) | QE: Pre/post (with control) | 2013–2014 to 2014–2015 SY | CEP | Participation | CEP was associated with an 8% ↑ in lunch served (RR = 1.08, 95% CI 1.03, 1.12). In sub-analyses, CEP was associated with an 69% ↑ in lunches served among higher-income students (RR = 1.69, 95% CI 1.11, 2.56), but also a slight decrease among students previously eligible for free or reduced-price meals (RR = 0.91, 95% CI 0.86, 0.96). | Low |
| Ribar et al. 2013 [49] | North Carolina; elementary schools that changed between USBP and mean-tested SBP (n = 4 schools) and matched schools with no change (n = 6) | QE: Pre/post (with control) | 2007–2008 to 2008–2009 SY | USBP | (1) Participation: Grades 1–5 (2) Attendance: Grades 1–5 (3) Academic Performance: state standardized test scores in math and reading (grades 3–5) and science (grade 5), |
(1) USBP was associated with a 16.4% ↑ in breakfast participation overall (p <0.05), with the greatest increases among higher-income students (27.5%; p < 0.05). (2) No association with attendance. (3) No association with test scores. |
Low |
| Rivas 1994 [55] | Brownsville, Texas; one school district before and after implementing Provision 2 | QE: Pre/post (no control) | 1993–1994 SY | Provision 2 | (1) Participation
(2) Finances |
(1) Provision 2 was associated with a 16% ↑ in overall school meal participation (2) Provision 2 was associated with a 5% ↑ in district food service revenue |
Very High |
| Robinson 1994 [56] (GAO Report) | USA; 3 states with four school districts implementing the USDA No-Fee School Meal Pilot Program | QE: Pre/post (no control) | 1990–1991 to 1992–1993 SY | (1) Participation
(2) Finances |
(1) Universal free meals was associated with ↑ participation (10% pt for NSLP). (2) Federal reimbursement increased by 33% overall due to increased student meal participation; districts incurred reduced administrative costs. |
Very High | |
| Schwartz et al. 2020 [39] | New York City, New York; middle schools with universal free lunch through Provision 2 (free breakfast was available in all schools prior to the start of the study) | QE: Pre/post (no control) | 2010–2013 | Provision 2 | (1) Participation
(2) Attendance (3) Academic Performance: standardized test scores (English Language Arts [ELA] and math) (4) BMI: objective measurements by schools |
(1) Provision 2 was associated with ↑ school lunch participation (5.39% among lower-income students [p < 0.01] and 10.97% among higher income students [p < 0.05]). (2) No association with attendance (3) Provision 2 was associated with ↑ in math scores (0.036 SD; p < 0.01) and ELA scores (0.030 SD; p < 0.01), with the greatest increases among higher-income students (0.083 SD [p < 0.01] in math and 0.059 SD [p < 0.01] for ELA). (4) No association with BMI. In sub-analyses, Provision 2 was associated with a 2.5% reduced probability (p < 0.01) of obesity among higher-income students. |
Low |
| Soldavini et al. 2019 [52] | North Carolina; 2285 public schools (elementary, middle, and high schools) with varying SBP models (including USBP) | CS | 2017 | USBP | Participation | USBP was positively associated with the odds of student participation at breakfast for all school levels, except high school students who were previously eligible for free or reduced-price meals. | Low |
| Tan et al. 2020 [60] | USA; national data from K-8 schools (80 CEP schools [n = 842 students] and 118 non-CEP schools [n = 1463 students]) | QE: post-only (with control) | 2013–2015 | CEP | Participation | CEP was associated with ↑ NSLP participation among students near the cutoff for free or reduced-price meals (11.7% higher likelihood of participation, p = 0.023) and higher-income students (eligible for full- price) [18.5% higher likelihood of participation, p < 0.001] compared with students at schools not participating in CEP. CEP was also associated with ↑ USBP participation among higher-income students (19.6%; p < 0.05). | Low |
| Taylor et al. 2020 [95] | Vermont; 116 school staff members (e.g., principals, food service staff, business managers, and nurses) from K-12 schools throughout the state with CEP | CS | 2017 | CEP | (1) Academic performance: staff perceptions (measured using an online survey) (2) Finances: staff perceptions (measured using an online survey) |
(1) Within CEP schools, a higher percentage of school staff perceived that free meals were associated with improved academic performance (64.4% agreed vs. 34.5% disagreed) and students were more ready to learn (83.0% agreed vs. 14.8% disagreed). (2) Approximately half of participants (52.4%) perceived that school finances had improved with CEP, but only 44% perceived that the school meal program deficit was reduced. |
Very High |
| Turner et al. 2019 [59] | California; Public schools throughout the state with varying school meal provisions | QE: Pre/post (with control) | 2013–2014 to 2016–2017 SY | CEP or Provision 1, 2, or 3 | Participation | Universal free school meals was associated with ↑ lunch participation (5.79% pt increase) and ↑ breakfast participation (3.48% pt increase) | Low |
| Wahlstrom et al. 1999 [22] | Minnesota; 6 elementary schools piloting USBP and 3 control schools | QE: Pre/post (no control) | 1993–1994 to 1996–1997 | USBP | (1) Participation: students in grades K-8 (varying by school) (2) Academic Performance: standardized achievement test scores (grades 3–6) |
(1) High school breakfast participation rates were observed and maintained in schools with USBP (no statistical analyses conducted). (2) A general increase in standardized achievement test scores for math and reading were observed (no statistical analyses conducted). |
Very High |
CS: Cross-sectional study; GAO: Government Accountability Office; LEA: Local Education Agency; QE: Quasi-experimental; RCT: Randomized controlled trial; SM: Simulation modeling; SFA: School Food Authority; USBP: Universal School Breakfast Program. 1 LEAs include traditional school districts as well as public and non-public nonprofit local entities (e.g., charter schools, non-public schools, archdiocese running multiple non-public schools, etc.) that enter into agree-ments with State agencies to operate the NSLP and SBP. 2 Risk of Bias varied by outcome (see Supplemental Table S3).