Table 1

Results from controlled trials

Studies	Objectives	Outcomes
Monterey: Altman et al 24Community intervention with retailer education	Illegal sales	Proportion of successful purchases fell from 75% to 0% in intervention and from 64% to 39% in comparison communities. The difference was significant towards the end of the assessment period. Sales rate in intervention communities < 20% at final four assessments. Proportion of clerks asking for ID rose more in intervention communities.
	Smoking behaviour	Cross sectional analyses of smoking behaviour (30 day tobacco use) found an effect on 7th graders which was not sustained to the end of study, but no significant effect in other grades. A longitudinal analysis of grade cohorts using generalised estimating equations found a significant effect favouring intervention communities only for the 7th grade. Females in intervention communities less likely to use tobacco post intervention than females in comparison communities.
	Perceived access	Intervention community 7th graders were less likely to report tobacco purchases at survey points 2 and 3, and 9th graders less likely to do so time 4
Minnesota (TPOP): Forster et al 21-23 Comprehensive community intervention	Illegal sales	There was a decline in over the counter purchase success in all communities, from 36.7% to 3.1% in intervention, 41% to 8.8% for control. The net difference did not reach significance.
	Smoking behaviour	Baseline prevalence marginally lower in intervention than control communities. Prevalence of all levels of smoking climbed sharply in control communities over the course of the study, the increase in the intervention communities was less pronounced. The difference in prevalence was only significant for daily smoking (−4.9%, 95% CI −9.0% to −0.7%). Effects were homogenous across sex and grade. For daily smokers there was a non-significant trend towards greater effectiveness among younger students.
	Perceived access	Perceived availability from commercial sources was high in all communities but there was a small decline in intervention areas (79.8% to 77.2% rating as high). There was also a decline in proportions citing a commercial source for most recent cigarette and making a purchase attempt in last month in intervention versus an increase in control areas.
Massachusetts: Rigotti et al 25 Enforcement compared to information alone	Illegal sales	There were no significant differences in baseline intervention and control compliance rates (35% and 28%). Compliance increased among the merchants in all communities but more rapidly in intervention than control according to a mixed effects model with adjustment for time point, type of store and type of sale. Most effects in first 6 months, with rise to > 70% compliance in intervention areas. Final compliance rates 82% in intervention and 45% in control areas.
	Smoking behaviour	There were significant differences between intervention and control communities at baseline and both follow ups, even after adjustment for age, sex, and ethnic group. The rate of current tobacco use rose slightly in the intervention communities but remained stable in controls, with borderline significance. No other measures of use changed differentially.
	Perceived access	Young people's self reported difficulty in buying cigarettes increased over time but there were no differences between the intervention and control groups. There were similar shifts in the sources of tobacco in the two groups with a reduction in the proportion buying their own in their community.
Gateshead, UK: Bagott et al 19 20 Enforcement (intended)	Illegal sales	Only assessed as part of intervention—no sales made. No checks in control area.
	Smoking behaviour	There were no significant changes in smoking prevalence. Regular smoking was more prevalent in the intervention school before and after.
	Perceived access	Few children reported being refused sales and there was no change over time. Over half of regular smokers bought cigarettes every day.
Sydney Australia 1995: Staff et al 28 Retailer education and community awareness	Smoking behaviour	In some school year/sex subgroups there were significant changes in prevalence from baseline. In the intervention group, 2 were decreases and 3 increases. In the control group there were 2 increases. When logistic models were used there was only an indication of the effect of the intervention in year 7, the youngest students.
	Perceived access	Ease of purchase was rated as being greatest from vending machines (93% rated as easy or very easy) and lowest in supermarkets (60%). The proportion of males in the intervention area who rated purchasing from petrol stations as easy or very easy was significantly lower post intervention but no other significant changes were noted.
Chicago: Jasonet al 12 Active enforcement with different scheduled frequency	Illegal sales	Baseline sales rates were high and similar across all interventions and communities: 86–89% over 5 months. Sales decreased in all conditions following warning. For last 6 months of intervention, average sales rates were 19% in 2 month, 34% in 4 month, and 42% in 6 month condition. Publicity about the project caused a decrease in sales in all conditions in months 11/12, and this contributes to the low average rate in 2 month condition.
Erie County 1987: Skretnyet al 46 Mailed education package	Illegal sales	There were no significant differences in illegal sales between intervention and control groups (77%v 86%). 40% of intervention stores posted the warning signs provided, no control stores had warning signs.
Erie County 1995: Cummings et al 13 Enforcement (with different numbers of checks) compared to warning only	Illegal sales	There were 44 violations/385 checks. Only one store fined twice. The baseline compliance rates were similar in enforcement (36%) and non-enforcement (35%) communities. Rates at follow up were 74% and 72% (difference NS). Rates were similar in stores that had and had not been fined. Significant change in asking for age ID from baseline. Lack of differential effect for enforcement may be due to the prior warning and widespread publicity about the sting operations in all communities. Also coincided with FDA's proposed regulations, and tobacco industry mailings.
Harlem: Gemson et al 14 Enforcement or education	Illegal sales	Baseline sales were high; 70% sold loose and 98% sold loose or packet. No significant group differences at baseline. Control group showed no significant change over time. Both education and enforcement group sales declined from baseline to 6 and 12 months, but there was a clear benefit of the enforcement strategy (from 100% to 47% for any sale at 1 year).
New South Wales, Australia: Schofieldet al 15 Education, or education and threat of enforcement	Illegal sales	There were no intervention effects on compliance; there was a pre- to post-test increase across all groups in the number of retailers requiring proof of age (17–43%) but no differences between intervention and control.
Project TRUST San Diego: Keay, Wildey et al 26 27 Retailer education and community awareness	Illegal sales	Baseline sales rates were 70% intervention/65% control. Intervention stores showed a significant decrease to 32% after intervention. Control stores showed a non-significant reduction to 59%. Results were similar 6 months later.
Santa Clara: Altman et al 16 17 Retailer education compared to community education alone	Illegal sales	There was no differential effect on sales as a function of the type of contact project staff had with merchants. Baseline purchase rate was 74%; at 6 months 39%; at 1 year 56%. There was no change in vending machine sales.
Sydney, Australia 1992: Chapman et al 18 Threat of enforcement	Illegal sales	The combined effects of publicity about undercover buying operations and warning letters threatening prosecution reduced sales of cigarettes to minors by 29% (95% CI 8% to 50%). 31% of retailers receiving warning letter sold at T2 compared to 60% of those not sent letter. However among 146 retailers who had not sold at T1 and were resurveyed, 24% did sell at T2.