Kim 2018.

Methods	Study design: Cluster‐randomised controlled trial Funding: “there has been no significant financial support for this work that could have influenced its outcome.”
Participants	Description: Children aged 2‐5.5 years with confirmed autism spectrum disorder (ASD) attending applied behaviour analysis (ABA) early intervention agencies for children N (Randomised): 5 agencies, 35 children Age: Child (mean): intervention = 4.4 years, control = 4.0 years % Female: Child: intervention = 15%, control = 7% SES and ethnicity: Not reported Inclusion/exclusion criteria: “inclusion criteria: 1) aged between two and five and a half years, 2) confirmed ASD diagnosis via parental report, 3) reported no extreme food restrictions or medical conditions impeding any kind of food consumption, and 4) received no additional feeding‐related interventions” Recruitment: “Participants were recruited via five ABA early intervention agencies for children with ASD located in the metropolitan area of Seoul, Korea.” Recruitment rate: Unknown Region: Seoul (Korea)
Interventions	Number of experimental conditions: 2 Number of participants (analysed): Intervention = 13, control = 14 Description of intervention: “The exposure program was administered as one of the ABA curriculum activities. The therapists and assistants (hereafter referred to as ‘staff’) were instructed to conduct a 5–10 min activity designed for a one‐week basis, one activity a day, for four days a week. The one‐week activity set—comprising four different activities—was repeated for four weeks in a month until a new one‐week activity set was started the next month.” “The final program consisted of 24 play activities, grouped into three levels by the degree of exposure, which was determined based on the time of contact, as well as the size and number of the stimuli. Each activity was repeated four times with three different vegetable and the expected time of contact, as well as the number of vegetables, had increased along with the level process.” Duration: 6 months Number of contacts: 96 sessions Setting: Early intervention agencies Modality: Face‐to‐face Interventionist: Therapists and assistants Integrity: “The first author and an undergraduate research assistant checked treatment fidelity using a 7‐item checklist on a regular basis (twice a week) during agency visits. Interrater agreement for treatment fidelity ranged from 85% to 100%.” Date of study: Unknown Description of control: “In this study, the control group received their usual treatment. The training manual was provided to the control group after the completion of this study.”
Outcomes	Outcome relating to children's fruit and vegetable consumption: Child’s consumption of vegetables (pieces) assessed by staff counting the number of pieces consumed Outcome relating to absolute costs/cost‐effectiveness of interventions: Not reported Outcome relating to reported adverse events: Not reported Length of follow‐up from baseline: 6 months Length of follow‐up post‐intervention: Immediate Subgroup analyses: None Loss to follow‐up (at immediate) Overall: 23% (8/35) Analysis: Unknown if adjusted for clustering Sample size calculation not performed “The biggest limitation of this study is the small sample size and the selection of participants using convenient sampling method.”
Notes
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Agencies were randomly assigned to either the exposure or control No further information provided
Allocation concealment (selection bias)	Unclear risk	No information provided
Blinding of participants and personnel (performance bias) All outcomes	Low risk	Objective measure of child’s vegetable intake and unlikely to be influenced by performance bias
Blinding of outcome assessment (detection bias) All outcomes	Low risk	The experimenter recorded the number of pieces of each food item taken by the child and it is unlikely that this would be influenced by detection bias
Incomplete outcome data (attrition bias) All outcomes	High risk	Overall loss to follow‐up is: 8/35 (23%), no ITT reported
Selective reporting (reporting bias)	Unclear risk	Unclear, no protocol, trial registration
Other bias	High risk	Recruitment bias (low risk): the parents were blind to the group assignment to avoid confounding variables Baseline imbalance (high risk): from demographic table appears to be no imbalance between groups. However, in table 3 the consumption of vegetable in the exposure group looks significantly higher which trial authors don’t report accounting for Loss of clusters (low risk): all clusters analysed Incorrect analysis (low risk): no clustering adjustment reported. The review authors adjusted for in the meta‐analysis Contamination bias (low risk): agencies at different locations