Spill 2010

Methods	Study design: Cluster‐randomised controlled trial – cross‐over Funding: "Supported by the National Institute of Diabetes and Digestive and Kidney Diseases (R01 DK082580) and the Robert Wood Johnson Foundation"
Participants	Description: Children aged 3 to 6 years enrolled in daycare at the Bennett Family Center on campus at The Pennsylvania State University N (Randomised): 5 classrooms, 51 children Age: Mean = 4.4 years % Female: 57% SES and ethnicity: “Of the 51 children in the study, 46 parents provided demographic information for their children. Of these 46 children, 28 (61%) were white, 14 (30%) were Asian, 3 (7%) were black or African American, and 1 (2%) was American Indian or Alaska Native. Parents of the children had above‐average educational levels and household incomes; 90% of mothers and 85% of fathers had a college degree, and 79% of households had an annual income >$50,000.” Inclusion/exclusion criteria: Provided by study author: "Children with an allergy to the foods being served were not eligible to participate in the study. Parents and guardians provided informed written consent for both their own participation and that of their child." Recruitment: “Recruitment began in April 2008 by distributing letters to parents who had children aged 3–6 years enrolled in daycare at the Bennett Family Center at the University Park campus of The Pennsylvania State University.” Recruitment rate: Provided by study author: "100% of children whose parents signed consent form were included in the study" Region: Pennsylvania (USA)
Interventions	Number of experimental conditions: 4 Number of participants (analysed): Overall = 51 Description of intervention: One day a week for 4 weeks, children were provided with a first course and main course at lunch. Across the weeks the portion size of raw carrots and dip served as the first course of lunch was varied (30 g, 60 g, or 90 g) and during 1 week no first course was provided. Cooked broccoli was served as the vegetable with the main lunch course Duration: 4 weeks Number of contacts: 4 (1 day a week) Setting: Preschool Modality: Face‐to‐face Interventionist Preschool teacher Integrity: Provided by study author: "All children were served the food assigned in the experimental condition. There was no deviation from study protocol. No unplanned or unintended interventions." Date of study: Recruitment began in April 2008 Description of control: N/A
Outcomes	Outcome relating to children's fruit and vegetable consumption: Child’s consumption of vegetables for different first course portion sizes (grams). “Uneaten items were removed, and weights were recorded to the nearest 0.1 g with digital scales. Consumption of the foods and milk was determined by subtracting postmeal weights from premeal weights.” 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: Unclear Length of follow‐up post‐intervention: Immediately Subgroup analyses: Provided by study author: "Differences between girls and boys in age, body weight, height, BMI percentile, and BMI z score were analyzed by using t tests. Analysis of covariance was used to assess the influence of continuous variables (age, body weight, height, BMI percentile, and BMI z score) on the relation between carrot portion size and the main study outcomes. Children who consumed all of the carrots (95% of the weight served) at any meal were identified, and data were analyzed both with and without these children to determine whether they influenced the results. The effect of individual children who were influential on the main study outcomes was assessed." Loss to follow‐up: There was no loss to follow‐up Analysis: Unclear if adjusted for clustering Sample size calculations performed.
Notes	Sensitivity analysis ‐ primary outcome: Vegetable intake listed as primary outcome in trial registry.
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	“Children were enrolled from 5 classrooms; the order of the experimental conditions across study weeks was assigned to classrooms by using a Latin square design.” Provided by study authors: "The orders of the experimental conditions across study weeks were created using Latin squares and then assigned to classrooms using a random number generator."
Allocation concealment (selection bias)	Unclear risk	It is not clear who undertook randomisation of classrooms. Provided by study authors: "Classrooms (and the associated condition order) were assigned a color coding so that participants and teachers were uninformed of the experimental condition."
Blinding of participants and personnel (performance bias) All outcomes	Low risk	“Incidents of food and drink spillage were recorded by researchers. Teachers were instructed to redirect conversations pertaining to food to nonfood‐related topics to minimize the influence on lunch intake.” Objective outcome measurement. Children were not blinded and it seems unlikely that this would influence their intake. Staff present during the meal and staff who served the food to children were not blinded and it seems unlikely this would influence child intake
Blinding of outcome assessment (detection bias) All outcomes	High risk	“Uneaten items were removed, and weights were recorded to the nearest 0.1 g with digital scales”. “Incidents of food and drink spillage were recorded by researchers.” Appears that researchers who weighed the food were the same researchers who recorded incidents of food and drink spillage. Researchers were not blinded and this may have had an impact on how the outcome was recorded in different classrooms
Incomplete outcome data (attrition bias) All outcomes	Low risk	“A total of 51 children were enrolled, and all of them completed the study” There were no children who dropped out over the study
Selective reporting (reporting bias)	Low risk	There is no study protocol and unable to determine if all prespecified outcomes have been reported as described Provided by study authors: "All outcomes collected were reported in the paper (vegetable and food intake)"
Other bias	Low risk	There are no other sources of potential bias