You are here

Article Text

Article menu
Review
Anti-tobacco policy in schools: upcoming preventive strategy or prevention myth? A review of 31 studies
  1. Maria Rosaria Galanti1,
  2. Alessandro Coppo2,
  3. Elin Jonsson3,
  4. Sven Bremberg3,
  5. Fabrizio Faggiano2
  1. 1Department of Public Health Sciences, Karolinska Institutet, Centre for Epidemiology and Community Medicine, Health Care District, Stockholm, Sweden
  2. 2Department of Translational Medicine, Avogadro University, Novara, Italy
  3. 3Department of Public Health Sciences, Karolinska Institutet, Stockholm, Sweden
  1. Correspondence to Dr Maria Rosaria Galanti, Department of Public Health Sciences, Karolinska Institutet, Centre for Epidemiology and Community Medicine, Health Care District, Stockholm 17177, Sweden; rosaria.galanti{at}ki.se

Abstract

Objective To summarise the evidence on effectiveness of school anti-tobacco policies (exposure) in preventing tobacco use (outcome) among high school students.

Data sources The search was conducted between 1 September and 30 November  2011 on six electronic databases with keywords: ‘policy’, ‘ban’, ‘restriction’ and ‘environment’ in combination with ‘adolescent’ or ‘student’, ‘school’ and ‘smoking’ in titles, abstracts or keywords. Restrictions were made to articles published in English.

Study selection Studies were included if they targeted the relevant grades/age; reported at least one outcome measure of students’ ever or current tobacco use; reported on the effects of exposure to policy separately from other interventions. Inclusion criteria were assessed independently by two of the coauthors. Of 2723 articles initially identified, 31 articles met the inclusion criteria (1.1%).

Data extraction Independent multiple observers extracted the data following the GRADE system guidelines to classify the level of evidence in relation to the review objective.

Data synthesis Studies were very heterogeneous in the definitions of exposure to school anti-tobacco policy and of tobacco use, adjustment for potential confounders and reporting of results, therefore summary quantitative measures of effect were not calculated. Qualitative summary statements were derived by reviewing the results reported in text and tables for distinct policy constructs. Evidence could be classified as low or very low, resting on cross-sectional studies with high risk of bias. Studies were rather consistent in indicating that comprehensive smoking bans, clear rules, strict policy enforcement, availability of education and prevention were associated with decreased smoking prevalence. Formally adopted and written policies, surveillance of students’ behaviour and presence/severity of sanctions were not consistently associated to students’ tobacco use.

Conclusions The evidence concerning the effectiveness of a school policy alone in preventing youth tobacco use is weak and inconclusive. Experimental studies or observational studies with longitudinal design are warranted, employing clear definitions of policy components and careful control for confounding.

  • Public policy
  • Prevention
  • Environment

https://doi.org/10.1136/tobaccocontrol-2012-050846

Statistics from Altmetric.com

    Request Permissions

    If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

    Introduction

    The importance of schools addressing health promotion and risk factor prevention at the environmental level has become conspicuous in recent years.1 The prevention of tobacco use among youths is a paradigmatic example in this domain. There are several lines of reasoning supporting the environmental perspective in school-based tobacco control. First and foremost, the effectiveness of pedagogic curricula alone in determining changes at the level of the student population is weak.2 This can partly be explained by the fact that young people are often exposed to social contexts where smoking is considered a normative behaviour and tobacco is easily available. Therefore, an environment denormalising smoking may represent a key strategy in prevention.3 Furthermore, classical environmental measures (eg, increasing retail prices, prohibition to sell) provide favourable cost-effectiveness comparisons in real-life situations,4 ,5 an argument that cannot be ignored in times of rethinking resource allocation in the public sector. In addition, recent developments indicate the importance of schools as physical environments and therefore implicated in the health protection of students and employees, besides pedagogic and educational tasks.6

    The introduction and enforcement of anti-tobacco policies in schools could in principle be regarded as a very promising preventive strategy. However, there is no consensus or a prevailing view on the definition of a school tobacco policy (STP) or on the effectiveness of such policies. To begin with, the very concept and content of STPs may be challenging. A formal definition could be taken from the Cambridge Advanced Learner’s Dictionary & Thesaurus (Cambridge University Press, Cambridge, UK) as a ‘set of ideas or a plan of what to do in particular situations that has been agreed officially by a group of people, a business organisation, a government or a political party’. In a public health framework, policy has been defined as ‘a guide to action to change what would otherwise occur, … a statement of commitment to certain areas of concern’.7 The common key points of these definitions seem to refer to problem-solving and change strategies. However, the peculiar feature of the public health definition above is the emphasis on the role of guide that a policy should have in indicating priorities in areas of concern for the health of the public. It may sound obvious that this general statement would go beyond the healthcare sector, and include other public institutions such as schools that have an indirect function in promoting health (preventing ill health), but schools may face several challenges along this path.8

    Second, the purpose of a STP may not be straightforward: should it be to deter youths from using tobacco in the first place, to be protected from exposure to second-hand smoke or only to secure refrainment from tobacco use on the school premises? Should the concern be for smoking only or for all forms of tobacco use? How far can locally developed rules go from an existing legislation?

    Empirical evidence on the effects of introducing or enforcing school anti-tobacco policies has not been systematically reviewed and summarised.

    To fill this knowledge gap we conducted a review of international literature, broadly relying on the principles highlighted by the Cochrane Collaboration9 in order to summarise the evidence of effectiveness of anti-tobacco school policies in preventing tobacco use by students. Specifically, we addressed the following question: is there evidence that formal and strongly enforced school policies against tobacco are associated with decreased likelihood of smoking or use of other tobacco compared to no/informal/weakly enforced policies, among high school students?

    Methods

    Inclusion criteria

    We included in this review articles: (1) published in English in peer reviewed journals; (2) comparing schools with or without an anti-tobacco policy with any study design; (3) reporting at least one measure of effect on tobacco use by students (smoking and/or smokeless tobacco); (4) evaluating the effects of an anti-tobacco policy alone or evaluating policy effects within a multicomponent intervention, provided that it was possible to disentangle the unique policy effect; (5) regarding high schools. Any definition of STP was accepted, and no time constraints were posed. We also included articles where policy status was derived from the students’ perception and not from objective external sources, but these were analysed separately.

    Search history

    The search was conducted independently from 1 September to 30 November 2011 by two of the coauthors (EJ and AC). The following databases were searched: Pub Med, PsychInfo, Eric, Google scholar, Global Health and Web of Science, with keywords: ‘policy’, ‘ban’, ‘restriction’ and ‘environment’ in combination with ‘adolescent’ or ‘student’, ‘school’ and ‘smoking’ in titles, abstracts or keywords.

    After each search, double entries were cleared and abstracts reviewed to check for inclusion criteria. When in doubt, two other coauthors (FF and MRG) were requested to read the article in order to assess the presence of inclusion criteria. The flowchart of the search history is displayed in figure 1.

    Figure 1
    Figure 1

    Flowchart of inclusion criteria for selected articles.

    The data to be extracted and the framework for classification and comparison of policy characteristics were determined collaboratively. Two authors independently read each study in detail and checked on agreement.

    Results

    In total, 31 original studies published between 1989 and 2011 were included in the review.10–40 A synopsis of the included studies with size and age of the student samples, policy definition, and primary and secondary outcomes related to tobacco use, is shown in online supplementary table S1, whereas online supplementary table S2 shows a summary of study-specific effects.

    The majority of the studies were based in North America (11 in Canada and 7 in the USA), followed by European countries (n=7) and Australia and New Zealand (n=3). Two studies were based in Asian countries, while one included a comparison between US and Australian data.

    Study design

    Without exceptions, the included studies were cross-sectional, comparing concurrent variations in the prevalence or individual probability of tobacco use according to existing school-level policies. Three articles26–28 referred to the same study, employing repeated cross-sectional survey data in order to model changes in smoking prevalence connected to scaled introduction of policy measures. The most recent of these articles reported on a group of students eligible to be followed-up for 3 years (ie, between 10th and 12th grade), but without individual linkage. A few studies reported information on the time the anti-tobacco policy had been in place, or the time of exposure of the study population.26–29 The majority of the included studies explored the association between school policy and tobacco use by students as exclusive or primary aim. Six studies included other anti-smoking measures, other outcomes, mediation or contextual effects.14 ,16 ,20 ,30 ,31 ,36

    Study populations

    The student populations were recruited from all kinds of schools in a range of ages from 10 to 21 years, with a preponderance of studies enrolling students between 13 and 16 years.

    Dimensions and components of STPs

    In all studies, specific STPs were analysed according to a predefined set of components or characteristics, but these differed greatly. Some studies analysed the effect of single STP component, such as the presence of smoking ban,35 of STP rules as such,26 ,32 or of specific sanctions.19 However, the majority of the studies adopted a complex descriptive approach of the exposure under study, with policy components differently aggregated into larger dimensions. The most frequently encountered dimensions were: comprehensiveness or strength,10 ,12 ,13 ,15 ,25 ,29 enforcement,10 ,14–16 19–25 ,31 ,33 ,34 ,36 ,39 dissemination, communication and participation,14 ,22 ,31 ,34 ,38 degree of formality,16–18 ,25 ,34 ,38 emphasis or orientation.13–15 ,29

    Studies used different operational definitions for these complex dimensions as well as different measurement scales for the underlying policy components. For instance, in a US study10 policy comprehensiveness was defined according to a multiple-items tool addressing applicability, restrictions, repercussions, programmes, notification and evaluation. In another US study, comprehensiveness was defined as number of components applied in the policy out of four investigated.29 In addition, enforcement could be variably defined through consequences for violation,14 subjective rating by staff,15 ,16 or subjective rating by students.20 ,36 Conversely, the same policy component (for instance sanctions in cases of rule breaking) could be used as an empirical definition of different policy dimensions, such as enforcement or emphasis.13 ,17 ,29 ,33

    Concerning the modality of assessment of STPs, studies could be divided into 2 groups: 9 studies where the assessment was based exclusively on students’ subjective perception of rules and of their enforcement in school,20 ,21 ,28 ,31–33 ,36 ,37 ,40 and the remaining 22 where the assessment was performed exclusively or also through interviews or surveys of the school administrators.

    Outcome definitions

    All studies included some measure of smoking by students, and two studies included measures of smokeless tobacco use.35 ,40 Broadly speaking, outcomes related to tobacco use were conceptualised either with reference to lifetime experience (eg, ever smoking) or to current/recent use (eg, current smoking). However, substantial variations could be observed between operational definitions of these two timeframes as well as of behavioural frequency (see online supplementary table S1). For instance, of 25 studies examining current/recent tobacco use as main or secondary outcome about half referred to any use in the past 30 day for the definition of current behaviour, but in 12 studies other definitions were adopted, such as weekly smoking,28 having smoked from a cigarette during 2 of the past 30 days,22 or various combinations of self-reported daily or occasional smoking investigated with different questions.13 ,37 ,38 ,40

    Three studies did not allow the direct exploration of smoking outcomes among students. In one study14 smoking was assessed in combination with other substances. In another study, the only outcome was perception of smoking by students or teachers in the school area.16 A third study assessed the locations where smoking occurs (on or off school properties).37 The three studies were retained because of potentially interesting tobacco-specific secondary outcomes.

    Policy effects on tobacco use by students

    A summary of policy effects on tobacco use by students by primary or secondary outcomes is reported in online supplementary table S2.

    Smoking bans, permissions and restrictions

    The presence or self-report of different levels of smoking bans was investigated in 13 studies, with mixed results. Five of these studies16 ,24 ,30 ,35 ,37 suggested a 20% to 60% decreased probability of tobacco use among students in schools with strict bans, supported by three studies that presented an increased risk with more liberal attitudes, especially concerning smoking by teachers.11 ,19 ,40 However, other studies failed to detect clear relationships between smoking bans per se and students’ behaviour,12 ,18 ,32 ,38 while one study suggested an association in the opposite direction (higher likelihood of smoking progression) if the students perceived the presence of strict bans.36

    Degree of formality

    Eight studies investigated the effect of the degree of formality of the adopted policy in relation to students’ behaviour, for instance whether the policy was written and/or clearly stated.16–18 23–25 ,34 ,38 The presence of a written policy with clearly stated rules and goals was associated with lower probability of students smoking or lower perceived smoking in some studies.16 ,23 ,24 One study25 investigated the degree of formality together with level of restrictions or bans, and found that strong policy (written, with universal restriction) was associated with lower probability of daily and weekly smoking than a policy that was not written and/or not universal. In one study, written policy dissemination to students was analysed separately from the presence of a formal policy itself.38 While written dissemination was linked to lower probability of smoking, the opposite association was found between formal policy and students smoking, but this was no longer observed after adjustment for individual-level variables. However, some studies failed to detect any association between formal or clearly stated school policies and students’ behaviour.17 ,18 ,34 Similarly, studies where the presence of clearly stated anti-smoking rules was self-reported by students yielded mixed results, as some found associations in the hypothesised direction26 ,31 ,37 while others did not.28 ,33

    Policy comprehensiveness or strength

    Comprehensiveness or strength of the policy was analysed in six studies,10 ,12 ,13 ,15 ,25 ,29 conceptualised in different ways from complex multidimensional indexes10 to simple statements on target groups included in the bans or restrictions.15 Studies were largely inconsistent, with some not observing any association between policy comprehensiveness and students’ behaviour.10 ,15 Among the studies that found comprehensiveness being associated with lower probability of smoking by students25 ,29 one investigated degree of formality together with the extent of the smoking restrictions.25

    Policy enforcement

    Definitions of enforcement employed in the studies were as heterogeneous as those of comprehensiveness, frequently including the perception of smoking in the environment self-reported by staff or students. Other definitions included systems to monitor students’ behaviour, sanctions and perceived compliance with the rules. Associations of strict policy enforcement with smoking by students or tobacco use in the anticipated direction of lower smoking rates overall or on school premises were reported in eight studies.10 ,16 ,19 ,21 ,24 ,25 ,33 ,36 In these studies, the association of smoking with enforcement (indicated by ORs) ranged from 0.39 (0.34 to 0.43)33 to 0.89 (0.85 to 0.99).36 Other studies also indirectly suggested associations in the same direction, that is, exposure to staff or students smoking and to non-compliance with policy rules was associated with increased likelihood of own smoking.28 ,31 ,34 However, these associations were not always consistent across age groups19 ,33 or types of exposure (eg, to smoking by staff or students).25 ,28 In one study, the proportion of students perceiving strict enforcement predicted smoking prevalence at the general and at the school area level in the hypothesised direction, but individual behaviour in the opposite direction.21 In another study, field observations of smokers in the school area was associated with higher smoking prevalence, and multiple involvement of school staff in the enforcement with lower smoking prevalence, but strength of enforcement was not associated.34 In a study addressing individual mediators of school policy effects, it was found that policy enforcement was linked to proximal predictors of adolescent smoking, such as perceived availability, perceived peer norms and perception of risks, in the anticipated direction.20 Nevertheless, a few studies showed no association between components of policy enforcement and tobacco use by students;15 ,30 ,39 one study showed associations opposite to the expectations (that is, higher probability of students smoking with stricter enforcement),23 while one study22 showed different associations depending on whether enforcement was defined through staff reports (associations mostly in the predicted direction) or student perceptions (inconsistent associations).

    A total of 12 studies specifically investigated the presence, the perception and/or the content of sanctions or punishment emphasis in case of rule breaking of the anti-tobacco policy in relation to students’ behaviour. Findings were quite mixed, with a majority of either null13 ,15 ,29 ,30 ,38 or counterintuitive reported associations, that is, risk increasing with sanctions.6 ,19 ,26 ,37 In one study17 schools using only a disciplinary approach had a higher probability of students smoking compared to schools using education and counselling. Of the two studies that reported decreased use with heavier sanctions, one did not present results for smoking separately from those of other substances.14 The other study found that sanctions that were put in place at school were associated with lower smoking probability, while informing parents showed the opposite association.39

    Communication and participation

    Six studies examined indicators of policy development, diffusion and communication in relation to students’ behaviour. Despite heterogeneity of definitions, findings were in general negative, that is, no associations were detected with the majority of these indicators.14 ,22 ,24 ,31 ,34 A few studies have suggested, however, that some dimensions of policy communication may be important. In one study, developing/overseeing and communicating the school policy was not associated with lower probability of current smoking, but clearly stated purpose and goals was.24 In another study, a written communication of policy to students and staff and the consistency of anti-smoking messages in the school environment were associated with lower prevalence of smoking and of smoking at school.38

    Policy emphasis/orientation

    Four studies explored the relation between an overall indicator of policy inspiring principles (emphasis or orientation) and smoking prevalence, in particular whether this indicator was connected to prevention, harm minimisation, cessation, or discipline.13–15 ,29 Results were not consistent, as in two studies policy emphasis did not make any difference on youth smoking,13 ,15 while two studies indicated that emphasis on prevention rather than on cessation, and on total abstinence rather than on harm minimisation was associated with lower smoking prevalence.14 ,29

    Availability of education or cessation support

    Five studies indicated that policies including prevention and education components were associated with lower prevalence of smoking,17 ,18 ,23 ,30 ,35 but two studies did not detect an association.22 ,34 Two studies compared schools including anti-smoking educational or cessation components programmes, schools introducing policy measures only, or a combination of the two.26 ,27 Educational or cessation components were associated with a decreased risk of being an occasional smoker rather than a non-smoker but not of being a regular smoker. However, there was no association with policy alone or a combination of policy and educational programmes. One study found that perceived support from teachers in general was associated with lower probability of students’ daily smoking.39 With regard to cessation, available or mandatory cessation was not associated with smoking in some studies,30 ,31 associated with a lower risk of smoking in one study,34 and with higher risk of smoking in another study.24

    Objective assessment of policy versus students’ perception

    STP variables were exclusively based on students’ perception in nine studies.20 ,21 ,28 ,31–33 ,36 ,37 ,40 In 5 studies14 ,22 ,26 ,34 ,39 policy information was collected from students and from staff, while in the remaining 17 studies the policy variables were assessed through interviews with staff only. The proportion of studies reporting at least 1 association in the hypothesised direction (ie, STP components associated with lower smoking prevalence) was higher in studies based on student self-reports (7 out of 9) than in studies based on staff reports (11 out of 17). However, the presence of counterintuitive results (eg, the perception of strict rules and sanctions associated with higher smoking prevalence) was more common in studies where policy assessment relied exclusively on student perception. In fact, 6 out of 9 such studies reported at least 1 ‘counterintuitive’ association, compared to 9 of the 17 studies where the policy assessment rested exclusively on staff reports.

    Policy effects on other endpoints

    The most commonly investigated secondary outcome concerned students’ own tobacco use on the school grounds or while at school, as opposed to overall use.13 ,21 ,22 ,37 ,38 To these should be added a few studies that investigated smoking by students in school as observations performed by others, including the students themselves.10 ,14 ,15 Although referring to the same events these studies are cited separately, because reporting on an undesirable behaviour as involving others implies quite different cognitive and evaluation processes than reporting on the same behaviour as one's own, and the gap due to unreliable reports is likely to increase with policy strength. Both these groups of studies were rather consistent in indicating that clear rules,37 a comprehensive ban, consistency of rules,38 strict enforcement,10 ,14 ,15 ,21 and availability of education and prevention22 were associated with lower likelihood of students smoking on the school area. However, one study conducted only among smokers suggested that the policy characteristics effective in decreasing smoking at school may actually increase smoking outside the school area.37

    In other studies, a variety of outcomes indirectly linked to tobacco use were explored. These included purchasing tobacco,13 knowledge of consequences of smoking,13 ,17 ,20 positive or negative expectations and attitudes about smoking,17 ,19–21 perceived availability of cigarettes in school and frequency of peer smoking,17 ,20 recalling of being taught anti-tobacco curricula35 and academic performance in general.32 Apart from one negative study13 there was a tendency for the remaining studies to report associations in the anticipated direction, that is, aspects of policies were associated with increased awareness of risks,17 ,20 with negative attitudes and intentions,17 ,20 ,21 and with decreased perceived availability and frequency of peer smoking.17 ,20 Recalling anti-tobacco curricula was more common in schools with a policy,35 and poorer academic performance was predicted by perceiving absence of school anti-tobacco rules.32

    Discussion

    Research on effectiveness of anti-STP seems not to have progressed beyond an initial stage, with a distinctive lack of experimental, quasiexperimental and longitudinal studies. This broad search of the scientific literature resulted in 31 cross-sectional studies. Therefore, the evidence of effectiveness of STP can be classified as ‘very low quality’ with high risk of bias.41

    This study is the first attempt to systematically review and summarise the state of the art concerning specifically STP effects on youth behaviour. However, some previous related works deserve mention, because their conclusions are generally in line with those of this review or may convey useful insights for future studies. A review conducted with the broader scope to identify contextual factors explaining differences in smoking prevalence between schools42 reached the conclusion that policy comprehensiveness, strength and harsher sanctions, but not other aspects of tobacco control, could explain school variation in smoking. A narrative review of the effectiveness of school anti-drug policies (not specifically smoking) concluded that a more comprehensive anti-smoking policy may have some effect in decreasing smoking prevalence, but may also displace the behaviour from school grounds to off school.43 A non-systematic review of school contextual effects on pupil behavioural outcomes presented results in the hypothesised direction from three studies, two of which did not include any specific evaluation of STP.44 Finally, a more recent systematic review analysed whole-school programmes aiming at changing school organisation, practices and ethos.45 This analysis suggested that such programmes may lead to decreased use of substances.

    Given the heterogeneity of the studies included in the present review and the methodological limitations discussed below, questions whether and to what extent an anti-tobacco policy in school can deter youths from taking up or from progressing into tobacco use cannot be answered from the existing literature.

    Some components of the investigated policies may be regarded as more promising than others, as they showed rather consistent expected associations with primary or secondary outcomes related to tobacco use by students. For instance, universal tobacco bans or restrictions, clear rules against tobacco use and consistent enforcement towards students and adults in school were most often associated with decreased likelihood of smoking or decreased smoking prevalence at the school level. However, whether a policy was written and/or disseminated as a written document did not consistently affect students’ behaviour beyond its content or strength. Sanctions and strict surveillance measures deserve a particular attention, because their presence and strength were often either not associated at all with student behaviour or associated with an increased likelihood of smoking. In the absence of longitudinal observations or intervention studies it would be erroneous to dismiss these components as ineffective or countereffective, owing to the possible sources of bias highlighted below. In addition, there may be different effects of sanctions depending on contextual characteristics. For instance, countermeasures with emphasis on education seemed to be most often associated with lower smoking prevalence than countermeasures with emphasis on punishment or cessation. In addition, sanctions applied at school seemed to be more effective in deterring smoking than delayed consequences, such as informing parents.39

    Finally, some of the studies suggested that while STPs may be effective in curbing smoking on the school premises, they do little in preventing smoking in general, and may even contribute to increasing the frequency of the behaviour out of school.37 Indirectly, this conclusion is supported by the only study conducted in a western country where smokeless tobacco use was analysed in relation to school policy. Use was not affected by prohibitions at the school level, but was increased where there was no limitation of use during school time.40

    Methodological considerations

    Despite the search limitations, for instance concerning the exclusion of articles published in languages other than English or of unpublished works, we are confident that the studies included in our review fairly represent the state of art of the research in this domain.

    All of the studies included in this review adopted a cross-sectional design, which precluded causal inference but also made them vulnerable to various risks of bias. Inverse causality is a possible explanation for associations of higher tobacco use with policies that are more comprehensive or harsh prohibitions (ie, high tobacco use rates prompt more restrictive measures). In studies in which the exposure to policy was assessed through student self-reports there is the additional possibility that perception of stricter enforcement or other policy characteristics may be influenced by actual behaviour, for instance smokers become more aware than non-smokers of the existence and content of policy rules.

    A second important limitation precluding conclusions is the heterogeneity of exposure definition in the included studies. STPs had very different conceptualisation, components and operational definitions even when similar terminology was employed (eg, enforcement, comprehensiveness). This would make direct comparisons, and in the future, any data pooling very difficult.

    The age range of the study populations also varied considerably between studies. Since age is a major determinant of the prevalence of tobacco use, with a doubling of the initiation rates between early and middle adolescence,46 ,47 discrepancies between studies should be always interpreted in the light of age distribution.

    Heterogeneity of outcomes and limited descriptions of the instruments used for assessment are additional factors hampering the interpretation of results and the comparison between studies. None of the studies employed any validation of student self-reported behaviours. All studies apart from two35 ,40 considered only cigarette smoking as the target behaviour, therefore policy effects on smokeless tobacco use remain to be investigated.

    In a few studies, there was no mention at all of adjustment for potential confounders.13 ,16 ,35 ,39 In the other studies there was variability in the factors considered for adjustment, an unmeasured confounders at either individual level or school level pose problems for studies without randomisation. The risks of bias connected with improper control for confounding when the exposure is measured at the group level have been highlighted.42

    Studies also differed as to the statistical methods employed to examine the relation between policy and the proposed outcomes. Even if the hierarchical nature of the data was taken into account in a set of studies, simple comparisons between prevalence, ordinary regression modelling or other analytical methods were often used. Data relative to individuals grouped in clusters, such as students in schools, tends to be correlated. Ignoring such dependence, for instance by using ordinary linear or logistic regression, may lead to underestimation of SEs, with consequent spurious statistical significance and risk of incorrect inference.48

    Finally, contextual effects due to social norms concerning tobacco use may interact with the effects of school policies and also account for conflicting results between studies. Indeed, there was a greater tendency to report favourable effects of a school policy in studies based on surveys conducted up to the year 2000, compared to studies conducted later, suggesting that a possible preventive effect of local policies may be concealed by increasing social disapproval of tobacco use.

    Areas of primary importance for future research

    Prospective studies with focus on intervention evaluation should be conducted, preferably with an experimental or quasiexperimental design. Primary and secondary outcomes should be clearly identified, by stating: type of tobacco use involved in the policy; timeframe and definition of use; topography of the target behaviour (eg, in school or elsewhere); follow-up time. Policy definition should rest on a small number of well-defined components, allowing the possibility to reproduce and test their effects. At this time, when there is little evidence that the effectiveness of a policy is a function of some particular values of these components, it would be premature to recommend a complete policy template. However, theoretical and empirical considerations suggest that a well-defined policy should target all areas and all subjects in a school, should report a clear description of consequences for violations, be regularly enforced and widely communicated. Table 1 is an attempt to operationalise the above components. Finally, the context in which a given policy is going to be evaluated should be explicit, in order to allow the study of potential interactions (eg, with campaigns, preventive programmes, legislation).

    View this table:
    Table 1

    Policy dimensions and components suggested for formal evaluation

    Conclusions

    Some components of an anti-tobacco policy in schools may be effective in deterring young people from initiating or progressing in tobacco use. Conclusions about evidence of effectiveness should be cautious, owing to the low methodological quality of the studies addressing this question, which remains to be investigated in large, possibly multicentric studies, employing an experimental or a quasiexperimental design.

    What this paper adds

    • This is the first systematic review addressing the effects of an anti-tobacco policy on young people's use of tobacco. In general, there is high expectation on the efficacy of such measure of environmental prevention, because it is inexpensive and theoretically easily implementable.

    • The research in this field is dramatically poor. There is a lack of shared definition of what is a school tobacco policy (STP), the studies are mainly cross-sectional in type and the components of the policy poorly described.

    • The overall evidence of effectiveness is weak, due to lack of experimental, quasiexperimental and longitudinal studies. Notwithstanding, this review highlights promising preventive components of an anti-tobacco policy, to be included in future evaluations, such as comprehensiveness, consistency and enforcement.

    Acknowledgments

    We acknowledge the contribution of Ewa Andersson, Department of Public Health Sciences, Karolinska Institutet, for administrative assistance.

    References

    1. ↵
      1. Bonell CP,
      2. Sorhaindo AM,
      3. Allen EE,
      4. et al
      . Pilot multimethod trial of a school-ethos intervention to reduce substance use: building hypotheses about upstream pathways to prevention. J Adolesc Health 2010;47:555–63.
      OpenUrlCrossRefPubMedWeb of Science
    2. ↵
      1. Thomas R,
      2. Perera R
      . School-based programmes for preventing smoking. Cochrane Database Syst Rev 2006;(3):CD001293.
    3. ↵
      1. Wilson LM,
      2. Avila Tang E,
      3. Chander G,
      4. et al
      . Impact of tobacco control interventions on smoking initiation, cessation, and prevalence: a systematic review. J Environ Public Health 2012;2012:961724.
      OpenUrlPubMed
    4. ↵
      1. DiFranza JR,
      2. Carlson RP,
      3. Caisse RE Jr.
      . Reducing youth access to tobacco. Tob Control 1992;1:58.
      OpenUrl
    5. ↵
      1. Ross H,
      2. Chaloupka FJ
      . The effect of cigarette prices on youth smoking. Health Econ 2003;12:217–30.
      OpenUrlCrossRefPubMedWeb of Science
    6. ↵
      1. Oliver LC
      . Protecting our children's health: the ultimate government ‘stimulus package’. J Public Health Policy 2009;30:158–62.
      OpenUrlCrossRefPubMedWeb of Science
    7. ↵
      1. Milio N
      . Glossary: healthy public policy. J Epidemiol Community Health 2001;55:622–3.
      OpenUrlFREE Full Text
    8. ↵
      1. Hallfors D,
      2. Van Dorn RA
      . Strengthening the role of two key institutions in the prevention of adolescent substance abuse. J Adolesc Health 2002;30:17–28.
      OpenUrlCrossRefPubMedWeb of Science
    9. ↵
      1. Higgins JPT,
      2. Green S
      . Cochrane Handbook for Systematic Reviews of Interventions Version 5.1.0 2011.
    10. ↵
      1. Adams ML,
      2. Jason LA,
      3. Pokorny S,
      4. et al
      . The relationship between school policies and youth tobacco use. J Sch Health 2009;79:17–23.
      OpenUrlCrossRefPubMedWeb of Science
    11. ↵
      1. Barnett TA,
      2. Gauvin L,
      3. Lambert M,
      4. et al
      . The influence of school smoking policies on student tobacco use. Arch Pediatr Adolesc Med 2007;161:842–8.
      OpenUrlCrossRefPubMedWeb of Science
    12. ↵
      1. Boris NW,
      2. Johnson CC,
      3. Huang S,
      4. et al
      . Targeting school tobacco policy: lessons from the Acadiana Coalition of Teens Against Tobacco (ACTT). Health Promot Pract 2009;10:549–56.
      OpenUrlAbstract/FREE Full Text
    13. ↵
      1. Darling H,
      2. Reeder AI,
      3. Williams S,
      4. et al
      . Is there a relation between school smoking policies and youth cigarette smoking knowledge and behaviors? Health Educ Res 2006;21:108–15.
      OpenUrlAbstract/FREE Full Text
    14. ↵
      1. Evans-Whipp TJ,
      2. Bond L,
      3. Toumbourou JW,
      4. et al
      . School, parent, and student perspectives of school drug policies. J Sch Health 2007;77:138–46.
      OpenUrlCrossRefPubMedWeb of Science
    15. ↵
      1. Evans-Whipp TJ,
      2. Bond L,
      3. Ukoumunne OC,
      4. et al
      . The impact of school tobacco policies on student smoking in Washington State, United States and Victoria, Australia. Int J Environ Res Public Health 2010;7:698–710.
      OpenUrlCrossRefPubMed
    16. ↵
      1. Griesbach D,
      2. Inchley J,
      3. Currie C
      . More than words? The status and impact of smoking policies in Scottish schools. Health Promot Int 2002;17:31–41.
      OpenUrlAbstract/FREE Full Text
    17. ↵
      1. Hamilton G,
      2. Cross D,
      3. Lower T,
      4. et al
      . School policy: what helps to reduce teenage smoking? Nicotine Tob Res 2003;5:507–13.
      OpenUrlAbstract/FREE Full Text
    18. ↵
      1. Huang HL,
      2. Chen FL,
      3. Hsu CC,
      4. et al
      . A multilevel-based study of school policy for tobacco control in relation to cigarette smoking among children in elementary schools: gender differences. Health Educ Res 2010;25:451–63.
      OpenUrlAbstract/FREE Full Text
    19. ↵
      1. Kumar R,
      2. O'Malley PM,
      3. Johnston LD
      . School tobacco control policies related to students’ smoking and attitudes toward smoking: national survey results, 1999–2000. Health Educ Behav 2005;32:780–94.
      OpenUrlAbstract/FREE Full Text
    20. ↵
      1. Lipperman-Kreda S,
      2. Grube JW
      . Students’ perception of community disapproval, perceived enforcement of school antismoking policies, personal beliefs, and their cigarette smoking behaviors: results from a structural equation modeling analysis. Nicotine Tob Res 2009;11:531–9.
      OpenUrlAbstract/FREE Full Text
    21. ↵
      1. Lipperman-Kreda S,
      2. Paschall MJ,
      3. Grube JW
      . Perceived enforcement of school tobacco policy and adolescents’ cigarette smoking. Prev Med 2009;48:562–6.
      OpenUrlCrossRefPubMedWeb of Science
    22. ↵
      1. Lovato CY,
      2. Sabiston CM,
      3. Hadd V,
      4. et al
      . The impact of school smoking policies and student perceptions of enforcement on school smoking prevalence and location of smoking. Health Educ Res 2007;22:782–93.
      OpenUrlAbstract/FREE Full Text
    23. ↵
      1. Lovato CY,
      2. Zeisser C,
      3. Campbell HS,
      4. et al
      . Adolescent smoking: effect of school and community characteristics. Am J Prev Med 2010;39:507–14.
      OpenUrlCrossRefPubMed
    24. ↵
      1. Lovato CY,
      2. Pullman AW,
      3. Halpin P,
      4. et al
      . The influence of school policies on smoking prevalence among students in grades 5–9, Canada, 2004–2005. Prev Chronic Dis 2010;7:A129.
      OpenUrlPubMed
    25. ↵
      1. Moore L,
      2. Roberts C,
      3. Tudor-Smith C
      . School smoking policies and smoking prevalence among adolescents: multilevel analysis of cross-sectional data from Wales. Tob Control 2001;10:117–23.
      OpenUrlAbstract/FREE Full Text
    26. ↵
      1. Murnaghan DA,
      2. Sihvonen M,
      3. Leatherdale ST,
      4. et al
      . The relationship between school-based smoking policies and prevention programs on smoking behavior among grade 12 students in Prince Edward Island: a multilevel analysis. Prev Med 2007;44:317–22.
      OpenUrlCrossRefPubMedWeb of Science
    27. ↵
      1. Murnaghan DA,
      2. Leatherdale ST,
      3. Sihvonen M,
      4. et al
      . A multilevel analysis examining the association between school-based smoking policies, prevention programs and youth smoking behavior: evaluating a provincial tobacco control strategy. Health Educ Res 2008;23:1016–28.
      OpenUrlAbstract/FREE Full Text
    28. ↵
      1. Murnaghan DA,
      2. Leatherdale ST,
      3. Sihvonen M,
      4. et al
      . School-based tobacco-control programming and student smoking behaviour. Chronic Dis Can 2009; 29:169–77.
      OpenUrlPubMedWeb of Science
    29. ↵
      1. Pentz MA,
      2. Brannon BR,
      3. Charlin VL,
      4. et al
      . The power of policy: the relationship of smoking policy to adolescent smoking. Am J Public Health 1989;79:857–62.
      OpenUrlCrossRefPubMedWeb of Science
    30. ↵
      1. Piontek D,
      2. Buehler A,
      3. Donath C,
      4. et al
      . School context variables and students’ smoking. Testing a mediation model through multilevel analysis. Eur Addict Res 2008;14:53–60.
      OpenUrlCrossRefPubMed
    31. ↵
      1. Piontek D,
      2. Buehler A,
      3. Rudolph U,
      4. et al
      . Social contexts in adolescent smoking: does school policy matter? Health Educ Res 2008;23:1029–38.
      OpenUrlAbstract/FREE Full Text
    32. ↵
      1. Poulin CC
      . School smoking bans: do they help/do they harm? Drug Alcohol Rev 2007;26:615–24.
      OpenUrlCrossRefPubMedWeb of Science
    33. ↵
      1. Reitsma AH,
      2. Manske S
      . Smoking in Ontario schools: does policy make a difference? Can J Public Health 2004;95:214–18.
      OpenUrlPubMedWeb of Science
    34. ↵
      1. Sabiston CM,
      2. Lovato CY,
      3. Ahmed R,
      4. et al
      . School smoking policy characteristics and individual perceptions of the school tobacco context: are they linked to students’ smoking status? J Youth Adolesc 2009;38:1374–87.
      OpenUrlCrossRefPubMedWeb of Science
    35. ↵
      1. Sinha DN,
      2. Gupta PC,
      3. Warren CW,
      4. et al
      . School policy and tobacco use by students in Bihar, India. Indian J Public Health 2004;48:118–22.
      OpenUrlPubMed
    36. ↵
      1. Wakefield MA,
      2. Chaloupka FJ,
      3. Kaufman NJ,
      4. et al
      . Effect of restrictions on smoking at home, at school, and in public places on teenage smoking: cross sectional study. BMJ 2000;321:333–7.
      OpenUrlAbstract/FREE Full Text
    37. ↵
      1. Watts AW,
      2. Lovato CY,
      3. Card A,
      4. et al
      . Do students’ perceptions of school smoking policies influence where students smoke?: Canada's Youth Smoking Survey. Cancer Causes Control 2010;21:2085–92.
      OpenUrlCrossRefPubMedWeb of Science
    38. ↵
      1. Wiium N,
      2. Burgess S,
      3. Moore L
      . Brief report: multilevel analysis of school smoking policy and pupil smoking behaviour in Wales. J Adolesc 2011;34:385–9.
      OpenUrlCrossRefPubMedWeb of Science
    39. ↵
      1. Wiium N,
      2. Wold B
      . Actions taken by schools when tobacco policies are violated: associations with adolescent smoking prevalence. Scand J Psychol 2011;52:494–501.
      OpenUrlCrossRefPubMed
    40. ↵
      1. Øverland S,
      2. Aaro LE,
      3. Lindbak RL
      . Associations between schools’ tobacco restrictions and adolescents’ use of tobacco. Health Educ Res 2010;25:748–56.
      OpenUrlAbstract/FREE Full Text
    41. ↵
      1. Guyatt G,
      2. Oxman AD,
      3. Akl EA,
      4. et al
      . GRADE guidelines: 1. Introduction-GRADE evidence profiles and summary of findings tables. J Clin Epidemiol 2011;64:383–94.
      OpenUrlCrossRefPubMed
    42. ↵
      1. Aveyard P,
      2. Markham WA,
      3. Cheng KK
      . A methodological and substantive review of the evidence that schools cause pupils to smoke. Soc Sci Med 2004; 58:2253–65.
      OpenUrlCrossRefPubMedWeb of Science
    43. ↵
      1. Evans-Whipp T,
      2. Beyers JM,
      3. Lloyd S,
      4. et al
      . A review of school drug policies and their impact on youth substance use. Health Promot Int 2004;19:227–34.
      OpenUrlAbstract/FREE Full Text
    44. ↵
      1. Sellstrom E,
      2. Bremberg S
      . Is there a "school effect" on pupil outcomes? A review of multilevel studies. J Epidemiol Community Health 2006;60:149–55.
      OpenUrlAbstract/FREE Full Text
    45. ↵
      1. Fletcher A,
      2. Bonell C,
      3. Hargreaves J
      . School effects on young people's drug use: a systematic review of intervention and observational studies. J Adolesc Health 2008;42:209–20.
      OpenUrlCrossRefPubMedWeb of Science
    46. ↵
      1. DiFranza JR,
      2. Savageau JA,
      3. Fletcher K,
      4. et al
      . Symptoms of tobacco dependence after brief intermittent use: the Development and Assessment of Nicotine Dependence in Youth-2 study. Arch Pediatr Adolesc Med 2007;161:704–10.
      OpenUrlCrossRefPubMedWeb of Science
    47. ↵
      1. Galanti MR,
      2. Rosendahl I,
      3. Wickholm S
      . The development of tobacco use in adolescence among "snus starters" and "cigarette starters": an analysis of the Swedish "BROMS" cohort. Nicotine Tob Res 2008;10:315–23.
      OpenUrlAbstract/FREE Full Text
    48. ↵
      1. Peugh JL
      . A practical guide to multilevel modeling. J Sch Psychol 2010;48:85–112.
      OpenUrlCrossRefPubMed

    Supplementary materials

    • Supplementary Data

      This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.

      Files in this Data Supplement:

    Footnotes

    • Contributors EJ and AC contributed to the search. MRG, AC, EJ, SB and FF contributed in data analysis and interpretation. MRG is responsible for the overall content of the review.

    • Funding The work of MRG and EJ was partially funded by the Swedish National Institute of Public Health, grant HFÅ 2011-174. The work of AC was partially funded by the Edo Tempia Foundation. The funders did not have any influence on the content and conclusions of the paper.

    • Competing interests None of the authors declare any competing interests. SB is presently employed at the Swedish National Institute of Public Health as a technical expert in the area of children's health.

    • Provenance and peer review Not commissioned; externally peer reviewed.