Interventions characteristics

The included trials were conducted in the following LMICs: three in China, two each in India, and South Africa and one each in Romania, Thailand, Indonesia and Mexico. Each trial comprised 1-intervention arm with the exception of two that comprised 2-intervention arms and one included 3-intervention arms, so in total the review included 15-intervention arms (see table 2 for trials characteristics). All trials included at least one control group who received a regular school curriculum. In one trial39 the control group received the same but delayed intervention after intervention completion. Contents of school curriculum in relation to smoking prevention were discussed only in four trials.39–42 These trials indicated absence of smoking prevention contents or activities in school curriculum, of these two trials39 42 also reported absence of anti-smoking school policy. All trials focused primarily on preventing smoking initiation but one.39 Four trials38 40 43 44 also involved smoking cessation and other substances-use, respectively. All interventions focused on face-to-face activities inside schools. Four trials39 40 42 45 also implemented activities outside schools. Booster-sessions were delivered in four trials only.

Primary-outcomes

To assess the effectiveness of each intervention, adolescents’ smoking behaviour were compared in the intervention and control arms before the intervention and after a minimum of 6 months following intervention completion. Self-administered questionnaires filled confidentially at schools was the only method used for assessing outcomes of all trials. Due to financial constraints, none of the trials used biological measures to check the validity of self-reported smoking status. Only five of the included intervention arms were effective, compared with controls, in preventing adolescents at schools from smoking initiation (table 1).

Most trials defined never-smokers as those who never tried smoking in their life, even a puff or two based on the Global Youth Tobacco Survey definition.46 However, those who smoked a puff or two were considered never-smokers in one study.44 Changes in never-smoking rates in the past one43 or two38 months before the survey were used in assessing the outcome of two trials. Two trials38 40 did not separate the findings on cigarette smoking from other tobacco use. Some trials presented findings as changes in ever-smoking prevalence among those who never smoked before and after the intervention.38 39 47 Whereas others40–42 44 48 49 calculated OR of ever-smoking rates or measured difference in number of never-smokers between intervention and controls.

Accordingly, pooling findings in a meta-analysis was not appropriate due to the inconsistency in defining, measuring, assessing and reporting outcome measures across the included trials. Consequently, narrative data synthesis was used in this review.

Risk of bias

Attrition and selection were identified as the most common sources of bias. Risk of bias was appraised as considerable across the included studies as half of the included trials included high risk of bias in at least one of the five Cochrane risk of bias criteria32 and no trial has low risk of bias in all these criteria. Ineffective trials have a higher risk of bias ratio than effective trials (table 3). Only one effective trial44 had high risk of bias caused by selecting intervention schools purposively based on being cooperative before starting the intervention. The assessment could not identify low risk of selection and detection bias in most trials due to insufficient evidence of blinding participants, deliverers or outcome-assessors. Although blinding is difficult in behaviour change intervention, findings might have been influenced by these biases, as self-reporting of smoking inside schools was the only method used for assessing interventions’ outcomes.

Cultural appropriateness

Most (10/15) intervention arms were based on effective interventions originally developed in high-income countries (see table 4). The effectiveness in originating countries was stated by intervention designers in the published articles as justifications for using these interventions, which we also checked in this review. In three39 43 48 of these imported interventions, developers of original interventions trained local public-health-specialists and researchers to adapt the intervention to the targeted context, to ensure balancing fidelity and cultural appropriateness. Local people were involved in designing interventions in all arms. Most (11/15) arms used both top-down and bottom-up approaches by involving both experts and community members with an understanding of what was feasible and acceptable. However, only three arms collected quantitative data on feasibility and acceptability of the contents before implementation.

Table 4 summarises approaches, dimensions and strategies used for cultural appropriateness of each intervention arms. Contents of all interventions were delivered by people (mainly teachers) who share culture with the targeted-population (constituent-involving cultural appropriateness strategy), using dominant local languages (linguistic cultural appropriateness). Two-thirds (10/15) of interventions considered using a culturally appropriate package of contents and materials such as images, colours, clothes and pictures of community members (peripheral cultural appropriateness). Cultural values and beliefs of targeted communities were considered when designing nine arms (sociocultural adaptation strategy). However, only three arms demonstrated relevance of interventions to the targeted population (evidential cultural appropriateness). Only in two intervention arms, both were effective, were cultural tailoring which is defined as using all the above cultural appropriateness strategies included.

Deep cultural adaptation is defined as going beyond changing intervention contents and delivery methods to match the targeted population’s characteristics (such as using local people, clothes, music and language to develop and deliver interventions) to involve through incorporating cultural, environmental, psychological and social factors that influence smoking in the targeted population into the proposed intervention.29 50 Measures for deep cultural adaptation were mostly recognised in the effective interventions. Whereas all the ineffective interventions, except one,47 involved either unclear or surface cultural adaptation of the imported interventions. These adaptations were limited to altering the language and appearance of contents to suit the targeted populations (peripheral and linguistic cultural appropriateness) with some (in three arms only) including weak consideration of local sociocultural predictors of smoking. Involving adolescents in designing interventions, by exploring their perspectives on why and why-not their peers smoke and how schools could prevent them from smoking, was considered only in two interventions,44 51 both were effective.

Theoretical constructs

The design of all effective interventions, except one,40 was informed by at least one theory but insufficient details were available on how theory was incorporated. Table 5 maps the presence and absence of the 17 TDF theoretical domains.

All interventions provided information on smoking harms (knowledge domain). The information delivered in the ineffective arms was only about smoking-related illnesses, except in two arms41 47 as social consequences were added. Only effective interventions explained the consequences of secondary smoking. Interventions that combined explaining the health, environmental, social and emotional consequences of smoking were effective.39 44 None of the interventions that explained national smoking rates to correct pupils’ overestimation of smoking rates (normative education) was effective.

All interventions aimed at either enhancing pupils’ social influence skills (by making them aware of social pressure to smoke and training them to refuse smoking offers by friends, relatives or tobacco companies), or social competence skills (by providing training on at least one of the followings: self-awareness, self-esteem, self-control, stress-coping techniques, problem solving and decision-making), or both. Training on social influence skills was emphasised in all effective interventions while combining both skills was effective only in one arm44 (skills domain).

Nine intervention arms used role-plays, group discussion and activities or videos in raising awareness of smoking consequences to make the contents attractive and memorable after the intervention (memory and attention domain). Only effective interventions44 45 combined these methods with encouraging pupils to discuss their views on advantages and disadvantages of smoking before deciding to smoke or not (decision process domain).

Six arms aimed to increase barriers and minimise facilitators of smoking in pupils’ environment (environmental-context and resources domain). All these arms involved enhancing accessibility to information on smoking consequences inside schools. While pupils’ exposure to this information was high only in the effective interventions, this exposure was either low49 or not evaluated38 42 in the ineffective arms. Additionally five arms included policies that prevent pupils, teachers, parents and visitors from smoking inside schools. In the effective arms, improving information accessibility and anti-smoking policy went beyond schools to include home40 or the wider community.39 This included motivating the community to advocate for national anti-smoking policy39 or banning tobacco promotion.40

Activities to reduce social influencers of adolescents’ smoking (social influence domain) were considered in all the effective interventions, except one.44 Only effective interventions used peer-pressure to create a positive attitude toward non-smoking, or introduced pupils (alone or with teachers or parents) who never smoked as role-models by announcing their names in school newsletters. These role-models contributed in supporting other pupils to avoid smoking; informally discussed their beliefs about smoking harms and shared their experience of maintaining non-smoking behaviour and refusing smoking offers by friends. The intervention42 that aimed to change influencers of smoking at home through parents’ education on smoking harms without using pupils as role-models or peer-pressure were ineffective even when parents signed contracts not-to-smoke at home. Some ineffective arms aimed to change social norms only through explaining the social refusal of smoking by adolescents or obtaining written commitments from teachers or parents or verbal public commitments from pupils in front of their classmates not-to-smoke.

All arms considered correcting pupils’ beliefs about smoking consequences, at least on health. Although pupils’ beliefs that smoking is harmful had improved after interventions in seven arms, only five arms39 40 45 52 showed translating these beliefs into action by avoiding smoking. All arms39 44 that involved correcting beliefs about the emotional, addictive, environmental and social consequences of smoking in addition to harms on health were effective.

Besides increasing awareness and beliefs that smoking is harmful, three arms aimed to enhance pupils’ beliefs about their capability to avoid smoking. The effective arms45 52 enhanced participants’ self-confidence in their ability to avoid smoking and supporting their relatives and peers to avoid or quit smoking, trained them in this and allowed them to practice the acquired skills in role-plays and in the presence of professionals such as teachers or health professionals . One effective arm39 established school-based support groups for the trained participants even after the intervention completion. The ineffective arm43 trained pupils, using filmed real-life scenarios, to refuse smoking offers after explaining smoking harms and encouraged them to leave smokers when they smoke. However, authors acknowledged that condensing these activities over a short period due to time and resources constraints may have contributed to its ineffectiveness.

All arms involved enabling adolescents to make an informed and conscious decision to remain non-smokers (intention theoretical domain). Although the intention to smoke markedly reduced in 10 arms, adolescents’ ability to translate this good intention to action by staying never-smokers at the end of the follow-up was demonstrated only in 5 effective arms.39 40 45 52 One effective44 and three ineffective38 43 47 interventions did not report changes in adolescents’ intention to smoke.

The identity and social role domain was coded in seven arms, four were effective. In the ineffective arms,42 48 49 all participants were required to make public commitments inside school to avoid smoking and discuss smoking harms with peers, but it was unclear if this commitment was obligatory by school or voluntary. The effective arms39 40 44 45 allowed pupils who never smoked to make a self-conscious voluntary intention to be identified as non-smokers, be role-models and take active roles in persuading their relatives or peers to avoid smoking.

Training on coping strategies with anxiety and depression and stress-management (emotion management domain) was provided in six arms, only one44 was effective. Only this effective arm allowed participants to practice the acquired skills and burn out techniques like physical activities and hobbies through enhancing adolescents’ accessibility to some relevant facilities in the community.

The domains of personal-goals, behavioural regulation or optimism were only used in one intervention, which was effective. This intervention encourage pupils to set proximal and distal goals for themselves, then educated them on how smoking hinders achieving their goals and how a better quality of life could be obtained without smoking (personal-goals setting domain). It also enhanced them to monitor their usage of free time and emotional reaction, trained them on anxiety and anger management, encouraged them to use their free time to practice hobbies and exercises to beat boredom; enabled them to overcome accessibility constraints to leisure facilities (behavioural regulation domain). The same intervention also stimulated pupils’ self-confidence that they will win sports competitions and have a healthy and bright future by avoiding smoking (optimism domain).

The reinforcement domain was used in three arms, through social rewards for never-smokers. The effective interventions39 45 rewarded pupils (as well as teachers and parents in one arm)39 who maintained non-smoking behaviour until the end of follow-up by announcing their names in school newsletters and posters, to encourage others to imitate them. The ineffective arm42 rewarded winners of schools’ competition for the best anti-smoking presentations and essays, without publishing their smoking status, by giving them schools’ smoking-control-committee membership.

Intervention functions

Table 6 illustrates the interventions effectiveness in relation to the involved intervention functions. All trials used education and training functions to deliver the above-explained theoretical domains of knowledge and skills. Besides explaining smoking-related illness, effective interventions discussed other (addictive, emotional and environmental) consequences of primary and secondary smoking, using memorable educational methods such as group discussion, role-plays and videos.

All effective arms involved the persuasion function, through illuminating disadvantages and advantages of smoking using real-life scenarios in role-plays or videos followed by debate or group discussion on that; then training pupils to balance the disadvantages and advantages of smoking before deciding to smoke or not. The persuasion function in ineffective arms was limited to explaining biological hazards of smoking using animal experiments,42 showing pictures of smoking-related illnesses49 or discussing reasons for refusing smoking offers only from pupils’ perspective.43

The incentivisation function was under-represented in the included interventions but used more in the effective arms. Only social incentives were used, as no financial incentives were offered in any included intervention. The discussion above about the reinforcement domain explains the difference between the used incentives in the effective and ineffective arms.

No intervention used the coercion function. No trial reported using or creating an expectation of punishments of smokers, even when smoking inside schools.

The included interventions showed limited use of the restrictive function but this was used more in effective arms; through preventing pupils, teachers, visitors and parents from smoking inside or around school premises. In the effective arms, the anti-smoking policy went beyond school boundaries to the wider environment through disseminating the national anti-smoking policy and enhancing its implementation by community members,39 or advocating banning tobacco advertisements through pupil-signed petitions directed to the government.40 The two ineffective arms49 53 established smoking control committees aiming to support anti-smoking activities and banning smoking inside schools, but it was unclear if the anti-smoking policy was enforced or not.

The included arms showed limited use of the modelling function, which was only included in effective interventions, through declaring pupils who never smoked as role-models, then training them to discuss smoking harms with their peers. One effective arm39 declared teachers and parents plus pupils who never smoked as role-models. The effective interventions used peer-educators who never smoked to: chair, stimulate, summarise and present outcomes of group activities and run formal peer educational sessions inside classes. Their role in the effective arms also included: (1) sharing personal experience on making friendships without smoking; (2) illustrating (through role-plays and videos) positive attitude toward non-smoking and ways to resist peers and social pressure to smoke; (3) leading informal discussion outside classes with smoker and non-smoker pupils about various smoking consequences. Whereas peer-educators, with unreported smoking status, were used only in two ineffective arms, mainly to assist49 or deliver42 formal educational sessions on smoking harms inside classes or to speak to smoker pupils outside classes.

The environmental reconstructing function was identified in four arms, only one was effective.39 52 The effective arm39 52 encouraged social norms against smoking through establishing smoke-free initiatives run by a smoking-prevention-committee, which consists of pupils who never smoked and formally promised to support their peers to avoid smoking. Their activities went beyond schools to include pupils’ home and neighbourhood environments. Whereas this function in ineffective arms, when existed, was limited to school celebration of the World No-Smoking-day42 or producing school-posters discouraging smoking.48 49 Two ineffective arms42 49 established a school smoking-related committee but the role of this committee was unexplained. One ineffective intervention48 reported taking additional efforts to prevent pupils’ exposure to smoking at home without explaining how.

Only two arms, one of which39 was effective, considered the enablement intervention function, through improving pupils’ capability (beyond training and education) and opportunity to remain never-smokers. The effective arm39 offered smoking prevention, quitting and counselling services at schools for smoker and non-smoker parents, teachers and pupils. Whereas in the ineffective arm54 this function was limited to the provision of school-based quitting services for smoker pupils only.