Beets 2009.

Methods	Study name: Positive Action (Hawaii) Study design (e.g. RCT, cluster RCT): cluster RCT Intervention arm(s): Positive Action intervention Comparator arm (s): standard education (as usual) Sample size calculation performed: not reported Subgroups prespecified: yes Subgroup analyses: yes, by gender (boys vs girls) Start date: 2001‐2002 Duration of follow‐up: immediately post intervention only Number of follow‐ups: 1 Follow‐up time points: immediately post intervention, at the end of grade 5 ICC (if reported): student violent behaviours (0.06), substance use (0.05), and sexual activity (0.28)
Participants	Number of schools randomised: 20 overall (intervention: 10, control: 10) Number of participants randomised (total and by arm): N/S Age (range or mean (SD)) or grade at the start: N/S; range from 6 to 8 years at baseline Gender: 50:50 among responders Ethnicity: Hawaiian or part Hawaiian (26.1%), multiple ethnic backgrounds (22.6%), non‐Hispanic White (8.6%), African American (1.6%), American Indian (1.7%), other Pacific Islander (4.7%), Japanese (4.6%), other Asian (20.6%), other (7.8%), unknown (1.6%) SES: not reported Inclusion criteria: schools meeting criteria (1) 25% of students received free or reduced‐price lunch; (2) Hawaiian schools were in the lower 3 quartiles of SAT scores and (3) were located on Oahu, Maui, or Molokai; (4) public schools from kindergarten to fifth or sixth grade (i.e. not specialised schools); (5) annual school stability rate > 80% Exclusion criteria: students who left the study schools during the study period
Interventions	Randomisation before or after baseline survey: before Duration of the intervention (excluding follow‐up): 4 to 5 years Description of the intervention: programme is a school curriculum with school‐wide, family, and community components designed to improve behaviour, character, and academic performance. It involves children in kindergarten to 12th grade. The programme has a detailed curriculum with almost daily lessons and a school‐wide climate programme undertaken by the principal and a Positive Action co‐ordinator or committee, with family and community involvement components. Sequenced elementary school curriculum of 140 lessons per grade per academic year is delivered in 15‐ to 20‐minute periods. During the academic year of 35 weeks, total programme time is approximately 35 hours. Teachers delivered the intervention; teachers, administrators, and support staff attended training at the start of the academic year. Booster sessions of 30 to 50 minutes were provided for each school at least once over the academic year. Content of lessons was grouped over 6 units: self‐concept; mind and body positive action, social and emotional actions; getting along with others; being honest; and self‐improvement. Involved structured discussions and activities with teachers and structured or semi‐structured small group activities including games, role‐play, and skill practice between students Brief description of the theoretical model: theory of self‐concept, consistent with theories of triadic influence (loosely stated) Description of the comparator: students in the control group received their usual school curriculum
Outcomes	Primary outcomes: substance use for alcohol, drugs, tobacco; violent behaviours (carrying a knife or razor to cause harm, threatening to cut or stab someone, cut or stabbed someone, carried a gun, shot someone); sexual activity (voluntary heterosexual sex) Secondary outcomes: teacher reports of substance use and violent behaviours
Setting	Country: USA; State: Hawaii (schools on 3 islands: Oahu, Maui, Molokai) Setting: school Focus: universal
Process measures	Process data reported: not reported Method (qualitative or quantitative): N/A Description: N/A
Statistics	Sample size: N = 1993; 86% (n = 1714) of children responded to the questionnaire at fifth grade; teachers reported on 1225 children. Intervention: 10 schools (n = 976 in analysis); control: 10 schools (n = 738 in analysis) Unit of randomisation: school Unit of analysis: individual Method to promote equivalence between groups: Stratification: eligible schools were stratified based on year 2000 school report card data by an index including percentage of students receiving free or reduced‐price school lunch, school size, % student stability, and student ethnic distribution. Additional characteristics of the school (i.e. student‐teacher ratios and expenditure/student) and characteristics of student populations (i.e. proportion of gifted, special education, and English as a second language students) were also used, as were indicators of student behavioural and school performance outcomes (i.e. disciplinary referrals, suspension rates, and standardised achievement scores) Statistical models: 2‐level over‐dispersion random‐effects Poisson models; generalised linear latent and mixed models Baseline differences adjustment: no significant differences on the index tests or teacher reports of negative student behaviour between intervention and control groups Repeated measures methods in analysis: N/A; only baseline and 1 follow‐up reported
Notes	Equity: descriptive data at follow‐up: ethnicity primarily Hawaiian or part Hawaiian (26.1%), multiple ethnic backgrounds (22.6%), non‐Hispanic White (8.6%), African American (1.6%), American Indian (1.7%), other Pacific Islander (4.7%), Japanese (4.6%), other Asian (20.6%), other (7.8%), and unknown (1.6%) Funding National Institute on Drug Abuse (grants DA13474 and DA018760) Randomisation method, e.g. block, stratification, computer: stratification of the 111 schools was based on an index, resulting in 19 strata with at least 3 similar schools in each; 1 intervention and 1 control school were then randomly assigned in each stratum Clustering accounted for in sample size calculation (if relevant): not reported Cluster randomisation methods to account for clustering in analysis: 2‐level random‐effects models (binary and Poisson) were implemented to account for the heirarchical nature of the study design and the clustering
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Not reported
Allocation concealment (selection bias)	Unclear risk	Not reported
Blinding of participants and personnel (performance bias) All outcomes	High risk	Participants, children, teachers, and project staff not blinded to intervention allocation and receipt
Blinding of outcome assessment (detection bias) All outcomes	High risk	Self‐reported outcome measures
Incomplete outcome data (attrition bias) All outcomes	Unclear risk	Insufficient information provided
Selective reporting (reporting bias)	Unclear risk	No published protocol
Other bias	Unclear risk	Impact of addition of new students and loss of students from schools unclear