Holder 2000.

Methods	Study design: controlled before‐after study Sampling frame: adult population Sampling method: random‐digit dialing telephone survey Collection method: phone interviews, police incidence reports, and emergency services records Description of the community coalition: Coalition included local law enforcement officers, medical service providers, alcohol‐beverage control agents, CBO staff, community activists, parents, and youth
Participants	Communities: 2 California communities and 1 South Carolina community, compared with similar communities that did not receive the intervention Country: USA Ages included in assessment: general public Reasons provided for selection of intervention community: Intervention sites had existing community coalitions through which the intervention could be implemented Intervention community (population size): approximately 100,000 at each site Comparison community (population size): approximately 100,000 at each site
Interventions	Name of intervention: Community Trials Project Theory: not reported Aim: to reduce excessive drinking and related injury and violence problems Description of costs and resources: not reported Components of the intervention: (1) community mobilization; (2) community awareness; (3) responsible beverage service; (4) underage‐access law enforcement; and (5) intoxicated‐patron law enforcement Start date: 1992 Duration: 48 months
Outcomes	Outcomes and measures: alcohol‐related traffic accidents and assaults with emergency services events Time points: time series accident data from 1988 to 1997, hospital discharge data from 1991 to 1997
Notes	Outcomes measured at population level Funding source: government
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	High risk	Not randomly assigned communities
Allocation concealment (selection bias)	High risk	No allocation concealment
Baseline outcome measurement similar	Unclear risk	Matched comparison communities
Baseline characteristics similar	Unclear risk	Matched comparison communities
Blinding of outcome assessment (detection bias) All outcomes	Low risk	Data collected from state records and hospital emergency services records
Incomplete outcome data (attrition bias) All outcomes	Low risk	Data collected from state records and hospital emergency services records
Blinding of participants and personnel (performance bias) All outcomes	Low risk	Change in outcome assessed by population survey and public records
Protection against contamination	Unclear risk	Do not describe geographic proximity of control communities
Selective reporting (reporting bias)	Low risk	Relevant outcome data were reported