Roset‐Salla 2016

Methods	Study design: Cluster‐randomised controlled trial Funding: "This work was supported by a grant for investigation in nursing from Collegi Oficial d’ Infermeria de Barcelona, 2009 (grant number PR‐5001/09); Primer Premio Nacional de Investigación en Enfermería, 2009, from Hospital Universitario Marqués de Valdecilla; and a grant for investigation in nursing from Acadèmia de Ciències Mèdiques de Catalunya i Balears, filial Maresme, 2010. The funders had no role in the design, analysis or writing of this article."
Participants	Description: Children aged 1 to 2 years attending 12 daycare centres and their parent N (Randomised): 12 day‐care centres, 206 children, 195 parents Age: Child (mean): Intervention = 1.3 years, Control = 1.4 years Parent (mean): Intervention = 35 years, Control = 35 years % Female: Child: Intervention = 37%, Control = 49% Parent: Intervention = 93%, Control = 85% SES and ethnicity: Educational level: Primary = 10%, Secondary = 35%, University = 55% Inclusion/exclusion criteria: No explicit inclusion criteria stated for this trial Exclusion criteria: “Children still exclusively breast‐feeding at the time of the study, children whose parents were not responsible for their alimentation, children with special diets due to chronic diseases (such as coeliac disease, food intolerances or allergies, inflammatory bowel disease), parents with language difficulties, parents unable to attend the educational workshops and those who did not sign the informed consent.” Recruitment: “At the beginning of the school term, all parents of the children attending the participating day‐care centres were invited to informative meetings regarding the study with the use of pamphlets and posters.” Recruitment rate: Child: 35% (206/581) Region: The city of Mataró (north of Barcelona), Spain
Interventions	Number of experimental conditions: 2 Number of participants (analysed): Child: Intervention = 75, Control = 67 Parent: Intervention = 74, Control 72 Description of intervention: “All parents from the day‐care centres in the intervention group (IG) were invited to attend four educational workshops on alimentation at the beginning of the study and one reminder at 4 months. A model of participatory‐active education was used, in order to achieve practical skills in addition to nutritional knowledge. Cognitive (teaching how to improve diet), emotional (addressing beliefs and attitudes of the participants through discussion and analysis techniques) and skill areas (developing dietary skills) were included. The aim was to incorporate new and better dietary knowledge and to change the habits of the participants.” Duration: 6 months (workshops in October ‐ November and a reminder in March) Number of contacts: 5 workshops Setting: Preschool Modality: Face‐to‐face Interventionist: Nurses trained in nutrition Integrity: No information provided Date of study: October 2010 to May 2011 Description of control: “The parents included in the control group (CG) did not receive any education related to nutrition. In order to avoid drop outs, the participants of the CG were invited to a workshop on a subject not related to the study or nutritional education (manipulation and conservation.”
Outcomes	Outcome relating to children's fruit and vegetable consumption: Child’s consumption of fruits and vegetables (servings per day) assessed using a 78‐item food frequency questionnaire (FFQ) completed by parents 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: 8 months Length of follow‐up post‐intervention: 2 months Subgroup analyses: None Loss to follow‐up: Child: Intervention = 32%, Control = 35% Parent: Intervention = 9%, Control = 8% Analysis: Did not adjust for clustering. Unknown if sample size calculation performed.
Notes	First reported outcome (changes in vegetable and garden produce servings per day) was extracted for inclusion in the meta‐analysis. To enable inclusion in meta‐analysis, we calculated post‐intervention means by group by summing baseline and change from baseline means, assuming baseline SDs for post‐intervention SDs, and we calculated an effective sample size using ICC of 0.014 to account for clustering Sensitivity analysis ‐ primary outcome: Primary outcome not stated, fruit or vegetable intake 2nd listed outcome after adherence to Mediterranean diet
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Randomly allocated to experimental group but the random sequence generation procedure is not described
Allocation concealment (selection bias)	Unclear risk	There is no information provided about allocation concealment and therefore it is unclear if allocation was concealed
Blinding of participants and personnel (performance bias) All outcomes	High risk	Dietary intake (self‐reported): There is no blinding to group allocation of participants and this is likely to influence performance
Blinding of outcome assessment (detection bias) All outcomes	High risk	Dietary intake (self‐reported): There is no blinding to group allocation of participants and because this is a self‐reported measure this is likely to introduce detection bias
Incomplete outcome data (attrition bias) All outcomes	High risk	“Of the parents randomized to the IG only sixty‐seven (65 %) attended three or more workshops, with the remaining parents considered drop outs. The reasons for not attending the workshops were mainly difficulties in family timetables and illness of the children”. 35% of the intervention group did not attend the minimum of 3 workshops and were considered dropouts. Therefore analysis was not undertaken according to intention‐to‐treat principles and risk of attrition bias is high
Selective reporting (reporting bias)	Unclear risk	There is no study protocol therefore it is unclear if there was selective outcome reporting
Other bias	Unclear risk	There were baseline imbalances for certain characteristics between the conditions (e.g. servings of legumes), although adjusted for in the analysis and so the impact of this is unclear. Analysis did not accounted for effect of clustering, but we calculated an effective sample size prior to pooling in meta‐analysis to account for this