Table 3. GRADE Assessment for Certainty of Evidence by Outcome.
| Studies, No. | Design | Sources | Risk of biasa | Inconsistencyb | Indirectnessb | Imprecisionb | Publication biasc | Magnitude of effectd | Influence of all plausible confoundinge | Dose-response gradientf | Summary of findings | Certaintyg |
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Lunch participation rates | ||||||||||||
| 3 | NRSI | Gordanier et al,9 2019; Pokorney et al,10 2019; Schneider et al,11 2021 | Not serious | Not serious | Serious; US population not fully represented as data are only from 3 states | Not serious | Not detected | No | No | No | Implementation of universal free meals at lunch was associated with increased lunch participation | Moderate |
| Breakfast participation rates | ||||||||||||
| 1 | NRSI | Schneider et al,11 2021 | Not serious | Very serious; only 1 study | Serious; US population not fully represented as data are only from 1 state | Not serious | Not detected | No | No | No | Implementation of universal free meals at breakfast was associated with increased breakfast participation | Very low |
| Attendance | ||||||||||||
| 2 | NRSI | Bartfeld et al,8 2019; Gordanier et al,9 2019 | Not serious | Serious; results were consistent in direction, but only 2 studies | Serious; US population not fully represented as data are only from 2 states | Not serious | Not detected | No | No | No | Attendance rates did not change or were modestly improved in schools with universal free meals compared to schools without universal meals | Low |
| Anthropometrics | ||||||||||||
| 1 | NRSI | Localio et al,12 2024 | Not serious | Very serious; only 1 study | Serious; US population not fully represented as data are only from 1 state | Not serious | Not detected | No | No | No | Implementation of universal free school meals was associated with reduced prevalence of obese students, and increased prevalence of normal students | Very low |
| Disciplinary Action | ||||||||||||
| 1 | NRSI | Domina et al,13 2024 | Not serious | Very serious; only 1 study | Serious; US population not fully represented as data are only from 1 state | Not serious | Not detected | No | No | No | Implementation of universal free school meals was associated with reduced number of suspensions | Very low |
Abbreviations: GRADE, Grading of Recommendations, Assessment, Development, and Evaluation; NRSI, nonrandomized study of interventions.
a
Downgrading domain. Response options: not serious, serious, very serious, or extremely serious.
b
Downgrading domain. Response options: not serious, serious, or very serious.
c
Downgrading domain. Response options: undetected or strongly detected.
d
Upgrading domain. Response options: no, large, or very large.
e
Upgrading domain. Response options: no, would reduce demonstrated effect, or would suggest spurious effect.
f
Upgrading domain. Response options: no or yes.
g
GRADE rating options: high, moderate, low, very low.