Levine 2003.

Methods	Study design: controlled before‐after study Sampling frame: clinic patients with hypertension Sampling method: random assignment Collection method: review of records Description of the community coalition: Sandtown‐Winchester High Blood Pressure Control Program. Johns Hopkins University and Hospital led a coalition composed of city health department, city schools, district mayor’s office, recreation, social welfare sectors, and community churches in East Baltimore, MD, USA. A needs assessment was conducted using Maryland Department of Health data, hospital discharge data, community household survey, and interviews of community leaders to establish goals and objectives
Participants	Communities: urban, low income, 82% African American with a median age of 30, East Baltimore, MD Country: USA Ages included in assessment: adults Reason provided for selection of intervention community: high risk, African American population Intervention community (population size): 120,000 Comparison community (population size): 120,000
Interventions	Name of intervention: Sandtown‐Winchester High Blood Pressure Control Program Theory: Precede‐Proceed Model     Aim: to control hypertension Description of costs and resources: unknown Components of the intervention: (1) information provided to the patient  to clarify and reinforce components of treatment (15 min); (2) family and peer education to enhance social support (2 hours); and (3) a small group approach to enhance motivation and commitment (three 2‐hour sessions) Start date: unknown Duration: 48 months
Outcomes	Outcomes and measures: review of health records, % with high blood pressure under control (BP < 140/90) Time points: baseline and 60 months
Notes	Source of funding: government
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Study authors state that patients were randomly assigned to treatment and control groups but do not explain the method of randomization
Allocation concealment (selection bias)	High risk	Allocation not concealed
Baseline outcome measurement similar	Low risk	At baseline, % in control of high blood pressure was similar for treatment and control groups (38% vs 41%)
Baseline characteristics similar	Unclear risk	Study authors do not report characteristics separately for the 2 groups
Blinding of outcome assessment (detection bias) All outcomes	High risk	No blinding reported
Incomplete outcome data (attrition bias) All outcomes	Unclear risk	Treatment group attrition 16% and control group attrition 30%
Blinding of participants and personnel (performance bias) All outcomes	High risk	No blinding reported
Protection against contamination	Unclear risk	None reported
Selective reporting (reporting bias)	Low risk	Relevant outcomes reported