Is it reasonable to expect that enacting a law prohibiting the sale of tobacco to minors will reduce youth smoking?

Since 1987 it has been known that the mere adoption of a law prohibiting the sale of tobacco to minors is insufficient to prevent the sale of tobacco to children.1 This fact has since been confirmed by dozens of studies from around the world.8–10 Many studies have used cross-sectional surveys to determine if features of the law affect the likelihood that a youth will use tobacco. Commonly considered features include bans on vending machines, self-service and single cigarette sales. While several studies found weak associations between the strength of a law and youth smoking,11–16 others did not.11 ,17 ,18 A problem with cross-sectional studies is that tougher laws may be enacted where anti-smoking sentiment has already reduced youth smoking, or conversely in response to high rates of youth tobacco use. Cross-sectional analyses often lack data on whether the sale of tobacco to minors has been disrupted by law enforcement. An analysis of US states failed to demonstrate any association between the strength of a law and the strength of enforcement.19 In summary, it cannot be expected that enacting a law, no matter how strong, will have any impact on adolescent smoking in the absence of effective enforcement.

Opponents of restrictions on the sale of tobacco to children claim that these laws encourage adolescent tobacco use by making tobacco a forbidden fruit.20 ,21 This claim has been refuted by dozens of studies that show no increase in smoking after the institution of such restrictions.

Are merchant education programmes sufficient to reduce youth tobacco use?

Staff et al conducted a controlled study involving over 6000 English youth to evaluate the impact of a merchant education campaign.22 The only indication of an impact on commercial distribution was that male students rated it harder to purchase from petrol stations. Logistic models indicated a 54% reduction of smoking uptake and a 39% reduction in daily smoking among 7th graders in the intervention group. No effect was seen among older youths.

A 3-year randomised controlled study of community mobilisation and merchant education without enforcement was conducted in four Arizona communities.23 The intervention produced a temporary reduction in cigarette purchases by 7th graders but not by 9th and 11th graders. During the period when 7th graders made fewer purchases, smoking was reduced by a relative 23% in this age group. Neither purchases nor smoking were reduced among older youth.

There is no evidence that merchant education programmes, in the absence of law enforcement, have any impact on smoking by youth older than about 12 years of age.

Can enforcement disrupt the commercial distribution of tobacco to minors?

Opponents of efforts to disrupt the sale of tobacco to minors claim that such efforts are futile because social sources will fill the void making tobacco more available.24–28 In reality, adolescents who purchase cigarettes are the primary social sources for other youth.29–31 The disruption of commercial distribution creates supply shortages, driving up the cost of tobacco on the street and discouraging sharing among peers as smokers protect their supply.32 ,33 In Texas it was demonstrated that enforcement resulted in a marked decline in smoking among middle school students and a simultaneous decrease in the number of commercial and social sources of tobacco.34 ,35

The disruption of the sale of tobacco to minors has resulted in fewer purchases made by youth,23 ,33 ,36–38 fewer youth reporting that commercial sources are their usual source,33 ,39 ,40 fewer youth reporting a commercial source for their last cigarette,36 ,41 ,42 and more reports that cigarettes are harder to purchase.22 ,43 There is ample proof that enforcement can disrupt the commercial distribution of tobacco to minors, and this is a crucial test of the quality of an enforcement programme. Therefore, we sorted enforcement programmes into those that did, or did not show evidence that the sale of tobacco to minors had been disrupted.

Can enforcement that is too weak to disrupt the commercial distribution of tobacco to minors be expected to impact youth tobacco use?

Two studies were conducted in England on the impact of a particularly ineffective approach to enforcement.44 ,45 Enforcement was conducted as is customary through test purchases conducted by underage decoys.19 However, the decoys were exceptionally young (under 13). The decoys were refused in 100% of their attempts to purchase tobacco. Yet 95% of young adolescent smokers in the same community reported that they purchased tobacco from shops at least once per week, and 55% purchased on a daily basis. Thus, cigarette sales to children remained vibrant in these communities despite the publicised enforcement programme. No impact on youth smoking was observed.

Rigotti et al conducted a six-community randomised controlled trial of enforcement in greater Boston.46 Legal challenges from tobacco retailers temporarily derailed enforcement efforts5 and the intervention failed to disrupt sales to minors as the percentage of youth who purchased tobacco did not change. There was no effect on tobacco use among students in grades 9–12.

There is no evidence that enforcement that is too weak to disrupt the commercial distribution of tobacco to minors has any effect, positive or negative, on youth smoking.

Can enforcement that disrupts the commercial distribution of tobacco to minors be expected to impact youth tobacco use?

Table 1 displays the 19 interventions in which the sale of tobacco to minors was disrupted. In each case the intervention was followed by a decline in youth tobacco use. Some of these studies will be described in more detail here.

Forster conducted a randomised controlled study involving seven pairs of communities in Minnesota. Local laws were enacted and enforced in all communities assigned to the intervention group. The intervention reduced the number of purchase attempts and the number of youth relying on commercial sources of tobacco.36 Among 8th and 10th graders, daily smoking in the intervention group was reduced by 28% relative to the controls. Substantial activity in the control communities during the long-term follow-up period prevented a long-term effect specific to the intervention communities from being seen. This randomised controlled trial provides strong evidence that the disruption of tobacco sales to minors reduced adolescent smoking rates.

Cummings examined youth smoking rates and purchasing behaviours in a longitudinal analysis of 12 communities.39 Test purchases were conducted to measure merchant compliance with the law. Frequent smoking increased by 28% in the communities with higher violation rates and decreased by 16% in the communities with low violation rates for a differential decrease of 44%.

Jason demonstrated that the number of youth purchasing tobacco in communities with high levels of enforcement was only about one-quarter that in the communities with less effective enforcement.52 Regular tobacco use among high school students in communities with high levels of enforcement was about half that seen in neighbouring low-enforcement communities.

The Health Promotion Unit of the Central Coast region of New South Wales, Australia initiated an effort to curtail the sale of tobacco to children by conducting test purchases and issuing $1000 fines to violators.37 ,38 Subsequent rounds of compliance tests and prosecutions revealed violation rates of 10%, 8% and by 1997, 0%. Over the ensuing decade, the Unit maintained violation rates consistently below 5%. Surveys conducted every 3 years from 1993 to 2002, revealed that the proportion of all youth who had recently attempted to purchase cigarettes fell from 21.6% to 5.7%, a reduction of 73.6%. The prevalence of current smoking among youth was reduced by 50% (from 26% in 1993 to 13% in 2002).38 Initially, a decrease in smoking was evident only in the youngest adolescents, with declines among the older adolescents being evident only as a birth cohort effect as youths who did not initiate smoking during early adolescence did not take it up later. With fewer older adolescents smoking, subsequent birth cohorts had fewer smoking role models and social sources of tobacco. The prevalence of smoking among youth on the Central Coast showed an uninterrupted decline over 9 years. Only after enforcement had spread to the rest of New South Wales did youth smoking rates begin to decline on the state level, and only after enforcement had spread across the country did youth smoking begin to decline in national surveys. This study demonstrated that (1) with tough enforcement very low violation rates can be maintained with minimal effort; (2) the number of purchase attempts goes down (by 73.6% in this case); (3) the youngest youth are impacted first; and (4) 9 years or more of follow-up are needed to observe the full impact on older adolescents.

Jason et al conducted a randomised controlled study involving four pairs of communities, comparing the effects of strict enforcement of sales and possession laws (merchant violation rate of 4%) to moderate enforcement of both laws (merchant violation rate of 17%).53 Compared to the control communities, smoking initiation and daily use were reduced by more than 70% among white adolescents. There was a fourfold reduction in the number of cigarettes smoked by white smokers.

Texas reported a statewide merchant violation rate of 13% for 1998 and 15% for 1999.59 Between 1998 and 1999, the proportion of middle school students who reported commercial access to tobacco dropped by two-thirds while the prevalence of tobacco use in this age group dropped by 31%.34 At the same time, commercial access to tobacco did not change among high school students and there was no reduction in the prevalence of tobacco use in this age group. The lack of change in the older adolescents makes it unlikely that the decline in smoking among younger youths was due to factors that would affect all youths such as media or pricing.

Pokorny et al included the density of non-compliant retailers as a factor in a cross sectional multivariate analysis of the determinants of smoking using data from 5234 6th–8th grade students from 11 Midwestern communities.55 Retail tobacco availability (RTA) was measured as the density of non-compliant merchants per 1000 youth aged 10–17 years. Increased RTA was associated with increased odds that a youth initiated smoking. RTA ranged from 0.7 to 5.4 non-compliant retailers per 1000 youths. Each additional non-compliant retailer per 1000 youths increased the odds of initiation by 49%, or from our perspective, the elimination of each additional non-compliant retailer decreased the odds of initiation.

Reductions in national adolescent smoking rates of 50% or more have been reported in the USA, Canada, Australia and The Netherlands following the implementation of enforcement programmes.38 ,50 ,58 ,60 In The Netherlands, the proportion of youth that smoked and bought their own tobacco fell by 53%, from 13.5% to 6.4%, while the prevalence of smoking among youth 13–15 years old fell by 64%, from 20.3% to 7.4%.58

From 1997 to 2008 smoking among US 8th graders fell by 65%.61 In 1992, the average merchant violation rate for 93 US cities was 77%.9 Since 2006, all states have had merchant violation rates under 20%.62 Between 1997 and 2001 reliance on store purchases by adolescent smokers decreased from 38.7% to 23.5% in national surveys, indicating that improved violation rates had reduced commercial access.26 From 1997 to 2003 the prevalence of daily smoking among US 10th graders fell by 51% nationwide.61 State-level merchant violation rates over this 7-year period were a significant predictor of youth smoking when controlled for prices, anti-tobacco media and clean indoor air policies. Smoking among 10th graders was reduced by an average of 1% for every 2% decrease in the state violation rate.50

Although enforcement started later in Canada than the US, Canada achieved very low violation rates more quickly than the US. Between 1999 and 2009, the prevalence of current smoking among 15–19-year-old Canadian youth dropped by 53%.60 While other factors may have contributed to the declines in youth smoking in the US, Canada, Australia and The Netherlands, the fact remains that youth smoking has fallen by at least 50% in all four countries that instituted enforcement programmes, and no other countries have reported a similar phenomenon.

Nine additional studies of various designs documented reductions in youth smoking in conjunction with efforts to prevent the distribution of tobacco to minors, but none of these studies clearly documented a disruption of commercial access.11–14 ,16 ,63–67 Among these is a study that demonstrated the adolescent smoking had declined in Sweden only in the two regions with the lowest merchant violation rates.64

In summary, there are many examples of interventions in which the disruption of the sale of tobacco to minors was followed by substantial declines in youth smoking. Some of the studies documenting this effect had strong designs with long follow-up periods and these studies had the strongest measured outcomes.36 ,38 ,42 The positive studies include non-controlled studies,33 ,41 ,49 non-randomised controlled studies,39 ,42 and randomised controlled studies.36 ,51 ,53 Positive effects were seen in rural communities,36 ,39 suburban communities,41 ,42 ,53 ,55 across large regions or states,37 ,38 ,40 and countrywide.50 ,58 Positive effects remained in all studies in which it was possible to adjust for other tobacco control initiatives.47 ,50 ,56 Against the many studies documenting enforcement programmes that resulted in a decline in youth smoking, there were zero publications in the literature that documented that an intervention had disrupted the commercial distribution of tobacco to minors without it having had a positive impact on youth smoking. Therefore, all available evidence indicates that interventions that successfully disrupt the sale of tobacco to minors can be expected to reduce the rate of tobacco use among adolescents. This conclusion contradicts those of prior reviews that concluded that youth access interventions are ineffective.

Starting in 19871 public health officials proceeded by trial and error through a series of supply-side initiatives of increasing aggressiveness to determine what was required to impact youth smoking. Initial efforts consisted of enacting local ordinances alone.41 When this was seen to be ineffective, aggressive merchant education programmes were tried.68 When these too failed to produce the desired result, enforcement programmes of different types were implemented. Persistence eventually paid off as the current review indicates that all enforcement programmes that disrupted the sale of tobacco to minors have had a favourable impact on youth smoking.

From the beginning there were individuals within the tobacco control community who were opposed to a shift of focus from clean indoor air to youth access.69 While public health workers struggled to develop successful interventions, these critics continuously called for them to abandon their efforts.7 ,20 ,27 ,28 ‘Rather than continue to devote scare (sic) resources to youth access effort (sic), it is time to accept the fact—now well documented and explained—that youth access is a failed policy and abandon it so that resources can be focused on policies that work’.27 ‘Given the limited resources available for tobacco control, as well as the expense of conducting youth access programmes, tobacco control advocates should abandon this strategy and devote the limited resources that are available for tobacco control towards other interventions with proven effectiveness’.7 A perceived battle over resources offers a motivation for creating a false controversy about the effectiveness of youth access.21 ,27 In reality, enforcement of tobacco sales laws is inexpensive, is typically paid for by license fees and is very efficient in terms of cost per year of life saved, and could be fully funded with a 1 cent per pack tax on cigarettes with no need to divert resources from other programmes.70

Since the disruption of the commercial distribution of tobacco to minors defines a supply-side intervention, this is a critical factor to assess in evaluating the suitability of studies for inclusion in a systematic review. However, no prior review even considered whether interventions had any impact on commercial distribution.7 ,21 ,24 ,71 By mixing ineffective interventions that had no impact on commercial distribution to minors with interventions that effectively disrupted commercial distribution, a false controversy about the effectiveness of the supply side strategy has been sustained. Systematic reviews that mixed failed and successful interventions are not valid since signatories of the FCTC are not constrained to adopt a mixture of failed and successful interventions as they address the issue of tobacco sales to minors.