Fernandez‐Jimenez 2019.

Study characteristics
Methods	Study design: cluster‐RCT Length of follow‐up from baseline: 5 months Unit of allocation: school Unit of analysis: child
Participants	Service type: centre‐based (Head Start) Operation: public Country (region): USA (Harlem, New York City) Country income classification: high Low‐SES sample: yes Population description: minority community (underserved, urban, multi‐ethnic). Harlem is predominantly a mix of immigrant Hispanic and African‐American populations. Inclusion criteria: the preschools must be located in Harlem, New York; public; have children between ages 3 and 5 years of age; provide meals for the children; and, must make available use of their applicable programme operation space. Preschools were also asked to commit to not to participate in any other major structured health intervention programme aside from the usual curriculum during the study. Exclusion criteria: not reported Number of services randomised: Number of children randomised: 585 (413 intervention, 172 control) Characteristics Children Age: 4.1 years (SD 0.6) Gender (% female): 51.10% Ethnicity: non‐Hispanic black: 37.4%; Hispanic/Latino: 53.9%; other/multiracial: 8.7% Parents Age (years): not reported Gender (% female): not reported Ethnicity: not reported Parent/family SES: not reported Method of recruitment: not reported Missing data/dropout: approximately 20% of children were lost to follow‐up or had incomplete data. No school withdrew from the study. Reasons for dropout: mainly because preschoolers aged out and moved to other schools and/or residencies Characteristics of dropouts: not reported
Interventions	Programme name: the FAMILIA (Family‐Based Approach in a Minority Community Integrating Systems‐Biology for Promotion of Health) trial Number of conditions: 1 intervention, 1 control Intervention duration: 4 months Intervention setting: ECEC Intervention strategies: Health curriculum Children Education: 37 h of educational programme over 4 months Partnerships Families Education: invitations to informational and educational meetings, called FAMILIA days, and a minimum of 11 family health activities (12 h) that the teacher regularly provided to parents during the duration of the child educational intervention Resources: teachers regularly sent educational material with the students to work on with their families at home. Intensity of intervention: 37‐h educational programme for children over 4 months; 12 h of educational meetings for parents/caregivers Intervention delivered by: research team, ECEC staff Modality: face‐to‐face Theoretical basis: not reported Description of control: wait‐list control
Outcomes	Outcomes relating to child dietary intake: not reported Outcomes relating to child physical measures: BMI z‐score Number of participants analysed: Intervention baseline: 304 Intervention follow‐up: 304 Control baseline: 144 Control follow‐up: 144 Data collection measure: objectively measured (CDC) Data collector: not reported Validity of measures used: not reported Outcome relating to child language and cognitive performance: not reported Outcome relating to child social/emotional measures: emotional comprehension score Number of participants analysed: Intervention baseline: 304 Intervention follow‐up: 304 Control baseline: 144 Control follow‐up: 144 Data collection measure: Test of Emotion Comprehension (TEC) questionnaire Data collector: not reported Validity of measures used: not reported Outcome relating to child quality of life: not reported Outcome relating to cost: not reported Outcome relating to adverse consequences: Study authors state only, "No adverse events were reported" Number of participants analysed: not reported Data collection measure: not reported Data collector: not reported Validity of measures used: not reported
Notes	Funding source: American Heart Association under grant No. 14SFRN20490315
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	The random sequence generation procedure was not described.
Allocation concealment (selection bias)	Unclear risk	No information on the method of allocation concealment reported
Blinding of participants and personnel (performance bias) Physical outcomes	Low risk	We assumed that due to the nature of the intervention, ECEC service staff and study personnel delivering the intervention were not blind to the study allocation. However, the outcome is not likely to be influenced by lack of blinding.
Blinding of participants and personnel (performance bias) Social/emotional outcomes	Low risk	We assumed that due to the nature of the intervention, ECEC service staff and study personnel delivering the intervention were not blind to the study allocation. However, the outcome is not likely to be influenced by lack of blinding.
Blinding of outcome assessment (detection bias) Physical outcomes	Low risk	Blinding not reported, however children's height and weight were measured and were not likely to be influenced by lack of blinding.
Blinding of outcome assessment (detection bias) Social/emotional outcomes	Unclear risk	No information provided on whether the individuals conducting the Test of Emotion Comprehension were blind to group allocation. It is unclear whether the outcome is likely to be influenced by lack of blinding.
Incomplete outcome data (attrition bias) Physical outcomes	Low risk	Study authors stated that "After a median follow‐up of 5.4 months, ~20% of children were lost to follow‐up or had incomplete data; therefore, 448 children (304 in the intervention group and 144 in the control group) were included in the main analysis (complete‐case intention‐to‐treat analysis) of the primary outcome of the study." Similar reasons for loss to follow‐up were noted between groups (i.e. unable to contact, unable to schedule, moved, withdrew consent). Study authors stated that "The primary analysis was supplemented by sensitivity analyses conducted using multiple imputation procedures, which explored different assumptions for the distribution of the missing data. Similar results were obtained across these sensitivity analyses. Therefore, we might reasonably assume that the missing data did not significantly affect the validity of our findings." Therefore, risk of attrition bias was considered to be low.
Incomplete outcome data (attrition bias) Social/emotional outcomes	Low risk	Study authors stated that "After a median follow‐up of 5.4 months, ~20% of children were lost to follow‐up or had incomplete data; therefore, 448 children (304 in the intervention group and 144 in the control group) were included in the main analysis (complete‐case intention‐to‐treat analysis) of the primary outcome of the study." Similar reasons for loss to follow‐up were noted between groups (i.e. unable to contact, unable to schedule, moved, withdrew consent). Study authors stated that "The primary analysis was supplemented by sensitivity analyses conducted using multiple imputation procedures, which explored different assumptions for the distribution of the missing data. Similar results were obtained across these sensitivity analyses. Therefore, we might reasonably assume that the missing data did not significantly affect the validity of our findings." Therefore, risk of attrition bias was considered to be low.
Selective reporting (reporting bias)	Low risk	The outcomes reported in the paper were pre‐specified in the protocol paper.
Recruitment bias	High risk	Individual recruitment occurred after randomisation of clusters.
Baseline imbalance	Low risk	No significant differences were found between groups at baseline, except that there was a greater proportion of Hispanic/Latino compared with non‐Hispanic Black children in the intervention group as compared with the control group. Study authors state that "Interaction models were also fitted to identify possible baseline score‐, age‐, sex‐, race/ ethnicity‐, and socioeconomic‐by‐treatment effects for the main outcome variable."
Loss of clusters	Low risk	No loss of clusters
Incorrect analysis	Low risk	Study authors stated that “Multilevel linear mixed‐effects models that account for the hierarchical cluster randomized design were used to test for the adjusted intervention effect.”
Contamination	Unclear risk	All schools were from the same community, however there is not enough information provided to assess whether contamination exists.
Other bias	Unclear risk	Protocol states that school recruitment would occur in a 2:1 ratio (intervention: control), however the study states that recruitment occurred in a 3:2 ratio. No explanation provided for deviation from protocol. There is insufficient information to assess whether an important risk of bias exists.