Ward 2020a.

Study characteristics
Methods	Study design: cluster‐RCT Length of follow‐up from baseline: 9 months Unit of allocation: family child‐care homes (i.e. family day care) Unit of analysis: child
Participants	Service type: family child‐care homes Operation: not reported Country (region): USA (North Carolina) Country income classification: high Low‐SES sample: yes Population description: counties across central North Carolina; to help target family child‐care homes in high‐need areas, recruitment efforts focus primarily on counties with a higher‐than‐average prevalence of childhood overweight and obesity among children aged 2‐4 years and below average median household income. Inclusion criteria: eligible family child‐care homes must have had at least 2 children currently enrolled who are between the ages of 18 months and 4 years (but not entering kindergarten before follow‐up measures), serve at least 1 meal and 1 snack to children, and have been in business for 2 years (as a demonstration of business stability) with no plans to close in the coming year. Exclusion criteria: not reported Number of services randomised: 166 (83 intervention, 83 control) Number of children randomised: 496 (242 intervention, 254 control) Characteristics Children Age: 35.7 months (SD 11.4) Gender (% female): 50.4% Ethnicity: black or African American: 63.3%; white: 27.2%; other: 9.5%; Hispanic or Latino: 4.1% Parents Age (years): not reported Gender (% female): not reported Ethnicity: not reported Parent/family SES not reported ECEC staff (family child‐care home provider) Age (years): 49.3 (SD 9.1) Gender (% female): Ethnicity: black or African American: 74.1%; white: 18.1%; other: 7.8%; Hispanic or Latino: 4.8% Service/ECEC staff SES: education high school diploma or GED: 24.7%; Associate's degree or 60 h college credit: 49.4%; Bachelor's degree or greater: 25.3% Method of recruitment: community partners shared information about the project with local family child‐care homes. Then, study staff followed up with family child‐care home providers via mail, email, and telephone to invite study participation. During telephone follow‐ups, family child‐care homes were screened for eligibility (i.e. enrolling at least 2 children aged 1.5–4 years, providing at least 1 meal and snack/day, being open year‐round, and having been in business for 2 years with no plans to close in the coming year). Study staff then visited eligible family child‐care home providers to explain study details and obtain written informed consent. Study staff worked through the family child‐care home provider to share study information with parents (including project contact information for questions) and collect informed consent. Missing data/dropout: at postintervention 38% were lost to follow‐up. Reasons for dropout: children no longer being enrolled in the family child‐care home, or family child‐care home providers refusing to participate in measures. Characteristics of dropouts: not reported
Interventions	Programme name: Keys to Healthy Family Child Care Homes Number of conditions: 1 intervention, 1 control Intervention duration: 9 months Intervention setting: family child‐care home Intervention strategies: Ethos and environment ECEC staff Education: module 1 focuses on the provider's own health behaviours and encourages comparison against current recommendations; module 2 focuses on comparing current practices of family child‐care homes against recommendations and identifying areas of improvement; module 3 targets financial, record keeping for tax purposes, creating and enforcing policies and contracts, communicating with parents, professionalism, and marketing. Workshops: 3 x 3‐h face‐to‐face group workshops (1 on each module: Healthy You, Healthy Home, Healthy Business) delivered by health coaches. Resources: 3 x set of educational materials and toolkit resources (per module). Resources include pedometers, re‐useable water cups, child‐sized divided plates, poly spots, diet and physical activity‐related book. Support: 12 x one‐on‐one coaching contacts (lasting approximately 30‐45 min) via face‐to‐face and phone/email (4/module) employing motivational interviewing techniques. During each contact, the health behaviour coach works with the provider to review current goals, assess progress toward goals, problem‐solve around any barriers that might be encountered, and revise action plans as needed. Participants are provided with tracking sheets and encouraged to self‐monitor on a daily or weekly basis to help them stay on track. Partnerships Healthcare Delivery: health coaches delivered staff workshops and provided support. Intensity of intervention: 3 x workshops; 3 x home visits; 9 x phone calls Intervention delivered by: research team, ECEC staff, healthcare staff Modality: face‐to‐face, telephone, online, written Theoretical basis: the Socio‐ecologic Framework, Social Cognitive Theory and Self Determination Theory Description of control: alternative intervention control (a business‐focused intervention)
Outcomes	Outcomes relating to child dietary intake: Diet quality (total, total fruit intake, whole fruit intake, total vegetables intake, greens and beans intake, whole grains intake, dairy intake, total protein intake, seafood and plant protein intake, fatty acids intake, sodium intake, empty calories intake) Number of participants analysed: Intervention baseline: 242 Intervention follow‐up: 149 Control baseline: 253 Control follow‐up: 142 Data collection measure: diet observation in child‐care protocol Data collector: data collectors Validity of measures used: not reported Outcomes relating to child physical measures: BMI, BMI percentile Number of participants analysed: Intervention baseline: 242 Intervention follow‐up: 149 Control baseline: 253 Control follow‐up: 142 Data collection measure: objectively measured (CDC) Data collector: data collectors Validity of measures used: not reported Outcome relating to child language and cognitive performance: not reported Outcome relating to child social/emotional measures: not reported Outcome relating to child quality of life: not reported Outcome relating to cost: not reported Outcome relating to adverse consequences: not reported
Notes	Funding source: National Heart, Lung, and Blood Institute (HL108390), the Centers for Disease Control and Prevention (U48‐DP005017), and the National Institute of Diabetes and Digestive and Kidney Diseases (DK056350)
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	The study statistician used computerised block randomisation to assign family child‐care homes into either the intervention or control arm (1:1) (SAS 9.3, Cary, NC).
Allocation concealment (selection bias)	Low risk	Baseline data collected before cluster randomisation
Blinding of participants and personnel (performance bias) Diet outcomes	Unclear risk	We assumed that due to the nature of the intervention, teachers were not blind to the study allocation. It is unclear whether the outcome could be influenced by lack of blinding.
Blinding of participants and personnel (performance bias) Physical outcomes	Low risk	We assumed that due to the nature of the intervention, teachers 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) Diet outcomes	Low risk	Children's diet quality was estimated from observed intakes of food and beverages collected via the Diet Observation in Child Care protocol by data collectors who were blinded to arm assignment.
Blinding of outcome assessment (detection bias) Physical outcomes	Low risk	Children's height, weight, and waist circumference were measured by data collectors who were blinded to arm assignment.
Incomplete outcome data (attrition bias) Diet outcomes	Low risk	Data were available for 149 (62%) children in the intervention group and 142 (56%) children in the control group. Most (84%) of the missing data are due to children no longer being enrolled in the service. Study authors stated that "To address missing child‐level data, multiple imputation was used, models were re‐run, and results were compared against the original. One hundred samples were imputed and analyzed in SAS (Proc MI) using available physical activity, HEI [healthy eating index], and covariate data. Comparison of baseline data from completers and non‐completers suggests that data are missing at random." Therefore, risk of attrition bias was considered to be low.
Incomplete outcome data (attrition bias) Physical outcomes	Low risk	Data were available for 149 (62%) children in the intervention group and 142 (56%) children in the control group. Most (84%) of the missing data are due to children no longer being enrolled in the service. Study authors stated that "To address missing child‐level data, multiple imputation was used, models were re‐run, and results were compared against the original. One hundred samples were imputed and analyzed in SAS (Proc MI) using available physical activity, HEI [healthy eating index], and covariate data. Comparison of baseline data from completers and non‐completers suggests that data are missing at random." Therefore, risk of attrition bias was considered to be low.
Selective reporting (reporting bias)	Unclear risk	Waist circumference mentioned in protocol and methods of study, but findings not reported
Recruitment bias	Unclear risk	Study authors stated that "Participants included a convenience sample of family child care home providers in central North Carolina and children aged 1.5‐4 years enrolled in these family child care homes, recruited in five cohorts over 2 years." Although randomisation occurred after baseline data collection, it is unclear if additional children were recruited from the same cluster.
Baseline imbalance	Unclear risk	Baseline differences between groups were not reported. Models of child‐level outcomes included child age, sex, and BMI as covariates.
Loss of clusters	Unclear risk	Loss of 3 intervention clusters (8 children) and 4 control clusters (11 children). Loss was a small percentage (< 5%) of the overall sample
Incorrect analysis	Low risk	Study authors stated that "Models of child‐level outcomes accounted for clustering; included child age, sex, and BMI as covariates; and for primary outcomes (HEI [healthy eating index score], MVPA [moderate‐to‐vigorous physical activity/hour]) used p values < 0.025."
Contamination	Unclear risk	No evidence to make assessment
Other bias	Low risk	No clear other source of bias