The Scottish Government

« Previous | Contents | Next »

Listen

4. Evidence for the effectiveness of smoking prevention programmes

Key points

• Several major long-term comprehensive American state-wide smoking prevention programmes appear to have contributed to declines in teenage smoking rates.
• A small number of media campaigns have been shown to contribute to reducing youth smoking rates, as part of a wider smoking control strategy. They were intensive and long-lasting and used strong, carefully designed messages on TV but supported by other media.
• Some multi-stranded campaigns involving community action have been effective but the contribution of community action and its possibly effective elements have not been clearly identified. A Scottish community-based smoking prevention initiative was not successful.
• Reviews of the evaluations of large numbers of school-based smoking and drug prevention programmes show that most are ineffective in reducing smoking rates. Although there is no clear "best-buy," a number of peer-led programmes reported positive results and a peer-supported preventive programme in South Wales is showing early promise. Some programmes using social influence methods reported lower short-term effects on smoking rates but why these were effective and other similar programmes were not is unclear.
• Few programmes appear to have addressed the underlying factors associated with higher rates of smoking and other drug use by young people.
• Although rates of smoking and other drug use continue to rise after leaving school, few preventive programmes have focused on this older age group.

4.1 In this chapter, we consider the available published evidence for the effectiveness of interventions that have been specifically designed to discourage young people from smoking. Most of the evaluations have been conducted in the United States and included media campaigns, community-based programmes and schools-based programmes. Several comprehensive programmes involving a wide range of concurrent elements have also been assessed.

Comprehensive programmes

4.2 Wakefield and colleagues described the effects on teenage smoking of five state-wide programmes in the USA⁷⁷. There were no control groups or states. Instead, the smoking prevalence among teenagers and young people was measured and compared with the US prevalence. The programmes were implemented in California (1989-96), Massachusetts(1993-97), Arizona (1994), Oregon (1996) and Florida (1997). The funding of the programmes varied; some had lower than anticipated budgets which is thought to have strongly influenced their effectiveness. Furthermore, the tobacco industry employed lobbying against a number of the programmes. In general, the programmes consisted of mass media campaigns, grants for smoking cessation services, school based programmes, environmental tobacco smoke ( ETS) legislation, enforcement and youth and community programmes.

4.2 The apparent effects of the campaigns on reported smoking prevalence among teenagers varied:

• In California there was no change in 12-17 year old prevalence from 1990-93, an increase from 1993-96, and an increase in non-smoker susceptibility. Among 8 ^th graders and 10 ^th graders, relative increases in smoking prevalence from 1993-1996 were less than in other states.

• In Massachusetts the relative increase in 30 day prevalence was less than for the rest of the US for 8th and 10 ^th graders from 1993-1996, and for 9 ^th and 10 ^th graders from 1993-97. There was a relative decline in lifetime use for 8 ^th graders compared to an increase in the rest of the USA.
• In Oregon smoking prevalence was the same as national trends for 8 ^th and 11 ^th graders.
• In Florida the relative declines in 30 day prevalence for middle and high school students were greater than national trends.

4.3 The Californian tobacco programme has been the subject of further evaluation of its long-term impact, covering its first 13 years, from 1989-2002 ⁷⁸. Downward trends in adolescent ever-smoking rates were first observed among 12-13 year olds in 1993 and carried through in subsequent years among 14-15 year olds and then 16-17 year olds. The prevalence of current smokers among 12-17 year olds actually increased significantly from 1994-1996 and a significant fall was not seen until 1999, ten years after the programme began. Compared with the rest of the US, current smoking rates among 12-17 year olds were substantially lower in California at the outset (18% v 27% in 1991) and the gap had widened by 1999 (20% v 34%), even though an overall increase in smoking rates had occurred across the country. It is thus plausible that the California programme successfully discouraged early teenagers from starting to smoke and prevented the overall rise in teenage smoking across the US during that period from being as large as it might otherwise have been.

4.4 Thus, overall, there appeared to be some slowing or reversal of the upward trend in smoking in states where the interventions took place. However, as with the studies reviewed by the Cochrane Collaboration, the heterogeneity of the programmes makes it very difficult to assess what were the components of the programmes, or indeed other factors, which contributed to their success.

Mass media campaigns

Sources of evidence

4.5 The main sources of evidence we used were systematic reviews by the Cochrane Collaboration ⁷⁹ and the US Task Force on Community Preventive Services ⁸⁰. A recent Health Development Agency review was based on the Cochrane and US Task Force reviews ⁸¹. The US Task Force review identified 12 studies of sufficient quality. The Cochrane review found only six studies that met their criteria, all of which were included in the US Task Force review. They had the following characteristics:

• All were mass media campaigns carried out in the late 1980s to 1990s either alone or in combination with other primary prevention interventions such as school programmes.
• They targeted children from late primary school to young adults aged 18 in the US (5) or Norway (1).

4.6 The US Task Force found that mass media campaigns either alone or when combined with other interventions such as school-based and community-wide educational activities can be effective in preventing the uptake of smoking. Among the five studies which measured self-reported smoking before and after the campaigns, the decrease in rates varied from 0.02% to 9.5%. It is not known whether these decreases were maintained.

4.7 The Cochrane review found that only two of the six studies were associated with reduced smoking behaviour among children and young people. The review's authors concluded there was moderate evidence that mass media interventions could be effective.

Characteristics of effective campaigns

4.8 The reviews found that the effective campaigns were sustained for at least two years and were relatively intense. They used a variety of media with brief repeated messages that motivated young people to remain tobacco free. The effective messages were designed to provoke emotional reactions among young people. They were of two types - exposing the strategies of the tobacco industry, and providing information and support to help young people to remain non-smokers. The primary prevention message was either the sole component of the campaign or part of a broader anti-tobacco intervention. Most of the successful interventions and the messages they used were based on initial market research or built on successful components of previous campaigns. The content was adjusted for the age of the target audience. The type of media and times of exposure were adjusted to suit those used by young people. The media used in the campaigns included broadcasts on television and radio and adverts in printed media and on billboards.

Additional non-systematic review level evidence

4.9 A recent evaluation of the " TRUTH" campaign in the US was not included in the review as it was published in 2005 ⁸². This was a before-and-after evaluation of a national campaign to inform young people of the tobacco industry's tactics. Changes in estimated smoking prevalence among young people attributed to the campaign were calculated using national annual survey data and correlating this with the degree of exposure to the campaign. Between 1999 and 2002, youth smoking rates declined from 25.3% to 18%. It was estimated that 22% of this decline (a 1.6% fall in the rate) was due to the campaign.

4.10 Further useful information on the nature of effective mass media campaigns for smoking prevention is provided in a literature review and a qualitative study by Devlin and colleagues ⁸³ which identified the following additional requirements for successful media interventions:

• Segmenting "consumers" into target markets and making the campaign relevant to them by tailoring the message to their culture, motivations, attitudes and behaviours. This should recognise that young people's perceived benefits of smoking vary according to their personal and social circumstances. Three types of smoker were identified: resigned smokers, (older and of lower socio-economic groups), contented experimenters (younger smokers who do not consider the dangers of smoking) and reluctant experimenters (who view smoking as a "social tool", and are from higher socio-economic groups)

• Repeating, refreshing and updating the messages throughout the campaign
• Using "branding" to increase recall of key messages.

Summary of main findings

4.11 These reviews suggest that media campaigns should:

• Be combined with other interventions
• Be based on successful components of previous campaigns or initial market research, including identifying and understanding the target markets
• Use the full range of media which young people are currently using
• Include adverts which provoke an emotional reaction and/or expose the tobacco industry and are regularly refreshed
• Be intense and sustained over a number of years.

Community-wide interventions

4.12 The rationale for community-wide interventions is based on the observation that smoking is a behaviour which is embedded in a social context. Accordingly, manipulation of the wider social environment, to discourage the uptake of smoking, may help young people remain smoke free. In this context, community-wide interventions are either developed and/or supported by the community. The community has "ownership" of the programme which usually involves a number of diverse agencies working in partnership.

4.13 Community-wide interventions have been the subject of a Cochrane Collaboration systematic review ⁷⁹ upon which a review by the Health Development Agency was recently based ⁸¹. The Cochrane review identified 17 studies of sufficient quality, based in the USA, UK and Finland. The target populations were aged from 8-24 years; some were focused on deprived areas.

Interventions

4.14 The interventions were in general multi-component programmes - often including school based interventions as well as mass media campaigns, parental involvement and community action. Some were part of wider campaigns to reduce cardiovascular or cancer risk in the community, others targeted young people, either focussing on drug use in general or specifically on smoking prevention. The extent of community involvement varied, as did the duration and intensity of the intervention. Further, the duration of follow-up varied markedly.

4.15 Nine studies compared community-based interventions with no intervention. Only two (the Minnesota Heart Health Programme and the North Karelia Project) showed reduced smoking prevalence in the intervention group. Both had been primarily designed as a heart disease prevention campaign. The remainder found no significant difference between the groups. The studies in which the interventions were effective used the social learning theory or the social influences approach. Six studies compared community-wide programmes with single component controls. The single component was most often a school-based programme, but others used, for example, media only. Two studies found a statistically significant reduction in the community group, while three found no difference, and one found no difference but did identify a decline in smoking in both the intervention and control groups.

4.16 The heterogeneity of the interventions, populations, methods and outcomes measured in community interventions, as well as difficulties in selecting appropriate comparison populations make the assessment of whether community interventions are effective problematic. The review authors concluded that there was some limited evidence of effectiveness of community wide interventions for smoking prevention. They identified the following features which may increase the likelihood of programme success:

• Programmes should be built upon the effective elements of existing campaigns
• Programmes need to be flexible to address variability between communities
• Developmental work should be carried out with representative samples of the target audience to implement appropriate messages and activities
• Programme messages and activities should be guided by theoretical constructs
• Community activities need to reach the intended audience.

4.17 Breathing Space is the only major anti-smoking community based initiative in Scotland that has been evaluated ⁸⁴. Breathing Space was an experimental health promotion initiative designed to assess the potential for shifting community attitudes to smoking in a low-income area in Edinburgh (Wester Hailes). The initiative was based on community development principles and practice combined with health education theories and methods. It was thought that a shift towards less tolerance towards smoking in the community would contribute significantly towards reducing smoking in the medium to long term. It focused on four settings - community, primary care, young people (including school) and the workplace. A survey was conducted before the start of the intervention in 1999, in Wester Hailes and three other comparable housing estates and repeated in the same areas about 2.5 years later. It was found that there was little awareness in Wester Hailes of the Breathing Space campaign materials and no difference between Wester Hailes and the control areas in awareness of any health promotion activities that were smoking related. Overall, there was no evidence that the initiative had achieved its intended outcome. The study raised questions about the appropriateness of using community development approaches in this context, given the varying and competing understandings of community development methods among multi-agency teams and the structural issues that prevented partnership collaboration from achieving successful outcomes ⁸⁵.

School-based smoking preventive interventions

4.18 The main sources of evidence were

• A meta-analysis of smoking prevention programmes ⁸⁶. A recent HDA review of reviews considered only this analysis ⁸¹.
• A systematic review of schools-based programmes for preventing smoking ⁸⁷.
• A systematic review of school based smoking prevention trials with long term follow up ⁸⁸.

4.19 Rooney and Murray ⁸⁶ identified 90, mostly US based, studies undertaken from 1974-91. The study populations aged from 11-18 years old. The interventions, classified as "social" or "peer-led type" included information on the short health effects of smoking, the factors which encourage young people to start smoking, the training in techniques to resist the offer of cigarettes and pro-smoking influences in the media, as well as making a public declaration of not smoking. The length of follow up varied from 2-20 months. Smoking prevalence was mostly self- reported.

4.20 There was some evidence of effectiveness, with an estimated 5% relative reduction in smoking. Programmes which employed peer-led preventive activities as part of a multifaceted programme delivered to late primary school children with yearly "booster" sessions were recommended. Information on whether the reductions were sustained was not available.

4.21 The authors concluded that the following characteristics would increase the likelihood of success:

• They are delivered early in the school career
• They are delivered by peer leaders
• They are part of a multi-component campaign
• Boosters sessions occur
• Peers are not over-trained.

4.22 Thomas and colleagues ⁸⁷ identified all randomised controlled trials evaluating behavioural interventions in schools to prevent children (5-12) and adolescents (13-18) from starting smoking. They identified 16 studies of sufficient quality; 15 described interventions based on social influences, eight of which showed reduced uptake of smoking in the intervention compared with the control group; however, the largest and most methodologically sound trial found no long term effect. It is unclear which factors contributed to the apparent success of some interventions but not others.

4.23 Wiehe and colleagues ⁸⁸ reviewed all school-based randomized controlled trials with follow-up to age 18 and one year post-intervention which measured smoking prevalence. They identified eight out of 177 studies which met the inclusion criteria, only one of which showed long term reduced smoking prevalence in the intervention group. This was an American school-based drug abuse prevention trial which included smoking prevention in a white middle-class population. The review authors could not ascertain whether the success of this intervention was based on its content, intensity, implementation or other factors.

Other evidence

4.24 The ASSIST trial is a large scale evaluation of a schools-based smoking prevention initiative, currently being conducted in schools in South Wales and the Bristol area of England ⁸⁹. Fifty-nine schools involving 10,730 pupils were randomly allocated either to continue with their normal smoking education programme, or to do so with additional "peer supporter" training. Peer-nominated students in Year 8 (aged 12-13) were recruited as peer supporters and given intensive training off the school premises by professional health promotion staff. The peer supporters were trained to intervene informally with other pupils in their year in everyday situations to discourage them from smoking. Students have then been followed up for two years to see whether smoking prevalence in the intervention schools was lower than that in the schools which did not receive the training. Outcome data at one year indicated that the risk of students who were occasional or experimental smokers at baseline going on to report weekly smoking at one year follow up was 18.2% lower than in intervention schools. This finding was supported by analysis of salivary cotinine. Qualitative data from the process evaluation indicate that the majority of peer supporters adopted a pragmatic approach, concentrating their attentions on friends and peers whom they felt could be persuaded not to take up smoking, rather than those they considered to be already 'addicted' or who were members of smoking 'cliques' ⁹⁰.

Summary of findings

4.25 There is some limited evidence of short-term effectiveness for school-based smoking prevention programmes which use social influences and other approaches based on the social context of smoking. However, only about half of the adequately evaluated programmes using these approaches were effective. There was minimal evidence for the long-term effectiveness of school-based programmes. As most of the published studies were conducted in the US and do not provide sufficient information to be able to reproduce the programme, they do not provide a clear indication of the best way forward. The ASSIST programme has reported promising early results, indicating that the peer supporter approach may be worth pursuing.

School-based drug preventive programmes

4.26 Drug prevention interventions often aim to prevent the use of tobacco and alcohol as well as illegal drugs. The interventions designed to prevent the uptake of illegal drugs can also inform the development of smoking prevention interventions. We thus considered two further systematic reviews .

4.27 A Cochrane Collaboration review of school-based drug preventive programmes was published in 2005 ⁹¹. It identified 32 studies which met the inclusion criteria ( RCTs, case-controlled trials or controlled prospective studies), 29 of which were RCTs which compared the intervention with usual curricula. Twenty-eight of the 29 studies were undertaken in the USA, mostly during the 80s and 90s. The interventions (often aimed at preventing use of tobacco and alcohol, as well as illicit drugs) were skills based programmes (designed to improve individual pupils' social skills), programmes designed to influence drug knowledge or those designed to modify psychological factors (affective programmes). The target populations were largely pre-teenage children.

Findings

4.28 Despite drug knowledge improving among the intervention groups exposed to the knowledge-based and affective programmes, there was no evidence of these interventions affecting drug use. In addition to improving drug knowledge, decision making, self esteem and peer pressure resistance, skills based approaches appeared to reduce reported drug use (heroin and cannabis, and in those assessing smoking, tobacco). However, no difference in drug use was found when skills programmes were compared "head to head" with knowledge or affective approaches.

4.29 A Cochrane Collaboration review of non school-based drug prevention interventions identified 17 studies all published within the last ten years ⁹². The interventions considered included multi-component community studies, family intervention studies, education and skills training and brief intervention.

4.30 No conclusive evidence of effectiveness was obtained. There was limited evidence of effectiveness of some family interventions and some brief interventions (in primary care or further education college settings).

Conclusions

4.31 Of the various types of intervention studied, the comprehensive multi-stranded state-wide American campaigns appear to have been the most successful but took several years to achieve positive results. There are a small number of apparently effective media campaigns, characterised by careful design, targeting, high intensity and long duration. Most community-wide interventions did not lead to reduced smoking rates even when they were large, well-resourced and well-conducted. Of the two that achieved reduced smoking rates, it was unclear which elements made the difference. The results of school-based smoking prevention and drugs education programmes were also generally disappointing. While some studies found social skills-based approaches had short-term effects, particularly those involving peer education, in other comparative evaluations they did not appear to be any more effective than knowledge or affect-based approaches. Studies with follow-up for longer than one year have generally not shown lasting effects. The ASSIST peer support programme in Wales and the Bristol area appears to be showing some promise with results from two years of follow-up awaited.

4.32 There are two further important considerations. First, most if not all the community-based and school-based programmes were run as research studies. As a general rule, such programmes are conducted with a higher degree of quality control, better resourcing and more training and commitment of staff than would be the case in programmes that have been rolled out across the country in the "real world". Consequently, the impact of the real world interventions is invariably less than that of those conducted as a pilot or research study.

4.33 Second, the findings in chapter 2 show clearly that smoking and other drug use is more common among young people with particular family and social circumstances and especially among those who have shown signs of disturbed or anti-social behaviour. Teenage smoking is also more common among girls than boys. None of the interventions reviewed above appeared to take any of these factors into account. If these underlying issues are not being addressed, it is perhaps not surprising that the interventions are ineffective.

4.34 These conclusions have important implications for future smoking prevention interventions in Scotland which will be considered in Chapter 6.

« Previous | Contents | Next »