Smart 1976.

Methods	STUDY TYPE: Interrupted Time Series COUNTRY: Canada SETTING: Provinces of British Columbia and Ontario DURATION OF STUDY PERIOD: 1962 to 1972 ANALYSIS: Simple mean comparisons using t test on de‐trended data
Participants	Adult population purchasing alcohol
Interventions	INTERVENTION: Type: Complete ban on alcohol (beer, wine and spirits) and tobacco advertising Media: Newspaper Radio Television Billboards Notice‐boards Duration of intervention: 1 September 1971 to 31 October 1972 CONTROL: Type: No ban Duration of control: Pre‐ban before 1 September 1971 Variable depending on data type (monthly or yearly) and type of alcohol Monthly data: Beer: 1968 to 1 September 1971 Wine: 1968 to 1 September 1971 Spirits: October 1970 to 1 September 1971 Post‐ban after 31 October 1972 Variable depending on data type (monthly or yearly) and type of alcohol Monthly data: Beer: 31 October 1972 to August 1972 (note no monthly data for post‐ban period) Wine: 31 October 1972 to 1974 Spirits: 31 October 1972 to December 1973 COMPARISON: The consumption rates were compared to those in the province of Ontario where no ban had been in place during the same period
Outcomes	PRIMARY OUTCOME: Per capita alcohol consumption: Measured by sales data for alcohol beverages from Statistics Canada for British Columbia and Ontario. Using population estimates from the dicennial censuses (1961 to 1971) per capita consumption estimates were made for beer, wine and spirits SECONDARY OUTCOMES: None reported
Notes	ETHICS: Not applicable as nationally aggregated data. FUNDING: Addiction Research Foundation, Canada and Alcoholism Foundation of British Columbia
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	High risk	Not a RCT
Allocation concealment (selection bias)	High risk	Not a RCT
Was knowledge of the allocated interventions adequately prevented during the study	Low risk	The outcome of consumption was objectively measured by routine data collection and was thus unlikely to have been influenced by knowledge of the intervention
Were incomplete outcome data adequately addressed	Unclear risk	Data were not available for all alcohol types across all the same periods. The author states that he was unable to obtain the data despite requests
Was the study free from selective outcome reporting bias	Low risk	There is no indication that other outcomes would be of interest
Was the intervention unlikely to affect data collection (ITS)	Low risk	The data were collected from routine sources before and after the ban
Was the intervention independent of other changes (ITS)	High risk	The ban was initiated by a unanimous political vote, but the ban was stopped after elections when there was a change in political power. There is a likelihood that other political or social changes may have coincided with the period of the ban
Was the shape of the intervention effect pre‐specified (ITS)	Low risk	An increase in consumption was predicted after the ban was removed. This was tested and the point was dated
Was the study free from other risks of bias	High risk	There is an acknowledged possibility that advertising from other states would not have been stopped by the ban, causing a dilution effect. Seasonality may have affected results and this is addressed in the analysis. Mediators of alcohol use, other than advertising, are not discussed

RCT: randomised controlled trial

SD: standard deviation