Stanton 2004.

Methods	Cluster randomized trial: communities assigned to study groups; 35 sites Location: Baltimore, MD (USA) Time frame: recruitment 1999 to 2000 Sample size calculation (and outcome of focus): no mention Report included intraclass correlation coefficients and number of clusters for use in calculating design effect
Participants	General with N: 817 youth; African American; 13 to 16 years old Source: low‐income housing developments, tenant associations and local recreation center staff Inclusion criteria: youth living in low‐income communities Exclusion criteria: no mention
Interventions	Study focus: multiple risk reduction, including sexual risk Theory or model: Protection Motivation Theory Focus on Kids (FOK or youth program): 8 weekly group meetings at 1.5 hours each on risk reduction Youth + parent program (i.e. ImPACT) on monitoring and communicating (20‐minute video, role‐play, and discussion) Youth + parent program + booster session for FOK (90‐minutes each at 7, 10, 13, 16 months after intervention) Follow‐up: 6, 12, 18, and 24 months after intervention (only 24‐month data reported)
Outcomes	Primary: in past 6 months been pregnant or gotten girl pregnant (self‐report), used birth control during last sex (specify a method), or used condom during last sex Secondary: NA
Notes
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Use of a random numbers table
Allocation concealment (selection bias)	Unclear risk	Randomization of sites occurred after all youths in selected sites identified. All youths meeting inclusion criteria were eligible.
Blinding of participants and personnel (performance bias) All outcomes	High risk	Presume no blinding of participants; not feasible due to type of intervention
Blinding of outcome assessment (detection bias)	Unclear risk	No mention
Outcome measures	High risk	Contraceptive use and pregnancy from self‐report
Incomplete outcome data (attrition bias) All outcomes	High risk	Loss to follow‐up: 40%; groups ranged from 38% to 41%
Other bias	Low risk	Analysis for cluster randomized trial: reported intraclass correlation coefficients for each outcome and number of clusters. We calculated design effects, and then effective sample sizes, according to recommended methods (Higgins 2011).