Background Comprehensive tobacco advertising/promotion bans are effective against adolescent smoking but many developing countries have implemented only partial bans. This study examines the association between advertising/promotions exposure and adolescent cigarette smoking risk in North Africa, and possible mediation of this association by parent and peer smoking.
Methods Adolescent data (n=12 329) from the Global Youth Tobacco Survey were analyzed (Libya, 2007; Egypt, 2005; Morocco, 2006; Tunisia 2007; and Sudan, 2005). Current smoking (any cigarette use in the past 30 days) and never-smokers' initiation susceptibility (composite of openness to accepting a cigarette from a friend and intention to start smoking in the next year) outcomes were examined. Advertising/promotion exposures included media and in-person contacts. Weighted univariate, bivariate and multivariable analyses were conducted.
Results Current smoking prevalence ranged from 5.6% (Egypt) to 15.3% (Tunisia) among boys, and 1.1% (Libya and Egypt) to 2.0% (Morocco and Sudan) among girls. Initiation susceptibility ranged from 14.1% (Sudan) to 25.0% (Tunisia) among boys, and from 13.3% (Sudan) to 15.0% (Libya) among girls. Ninety-eight percent of adolescents reported exposure to at least one type of advertising/promotion. In multivariable analyses adjusting for demographics, each type of advertising/promotion was significantly and positively associated with boys' current smoking status; most advertising/promotion exposure types were also positively associated with initiation susceptibility among boys and girls. Peer smoking only partially mediated these associations.
Conclusions Tobacco advertising/promotion exposure was highly prevalent and associated with adolescents' smoking risk in these countries. The comprehensiveness and enforcement of advertising/promotion bans needs to be enhanced.
- Advertising and Promotion
- Global health
- Low/Middle income country
Statistics from Altmetric.com
Tobacco companies have been increasingly shifting their focus to low- and middle-income countries.1 It is estimated that by 2030, 80% of smoking-related deaths will be in the developing world.2 One developing region where tobacco use is prevalent and growing is in Northern Africa.3 In 2010, the smoking prevalence among adult male subjects in this region ranged from 29% to 53%, and it is increasing rapidly.2 ,4 In the Eastern Mediterranean Region, which includes North African countries, approximately 15% of the total adult male mortality in 2000 was attributed to smoking.5 Cancer incidence in this region was estimated to increase between 100% and 180% in the next 15 years.4
Despite these statistics, little is known about contributors to adolescent smoking in this region. A recognised risk factor for adolescent smoking in other locations is exposure to tobacco promotions and advertising. Tobacco companies use both indirect marketing strategies, such as sponsorship of athletic events, concerts and movies, as well as direct strategies, such as billboards, commercials and magazine advertisements.6 Tobacco advertising/promotions affect youth smoking through shaping youths’ curiosity about smoking, perceived norms, perceived prevalence of smoking, and perceptions about the benefits and risks of smoking.7 ,8 Advertising and promotions also indirectly influence adolescents’ smoking risk by affecting the smoking habits of influential others in youths’ lives such as parents and peers.7 In one recent meta-analysis, exposure to pro-tobacco marketing and media was estimated to double the odds of tobacco use initiation.9 Even brief exposures to tobacco advertisements can be influential on adolescents’ intentions to smoke.10 Advertising and promotional items are specifically designed to appeal to youth 11 and to establish brand identity.12
A report by WHO draws attention to the special influence of tobacco in movies which has escaped many tobacco control efforts.13 Adolescents attend movies more frequently than other age groups.8 More than half of all Hollywood films feature tobacco imagery.8 As with other media/promotions exposures, seeing smoking in movies shifts adolescents’ attitudes toward smoking favourably, and influences their perceptions of the normality of smoking.14 Research suggests a dose–response relationship with on-screen smoking: the more adolescents view tobacco imagery, the more likely they are to smoke.8 A meta-analysis of US studies suggested that 44% of 11–15-year-old adolescents’ risk of smoking initiation is attributable to exposure to smoking in movies.15
Using data from the Global Youth Tobacco Survey (GYTS), this paper examines the report of youths’ exposure to tobacco advertising and promotions in North African countries and the relationship between these types of exposures and adolescents’ cigarette smoking risk. All countries analysed were signatories of the Framework Convention of Tobacco Control before fielding the GYTS, indicating a pledge to reduce tobacco consumption among adolescents in their countries. Results can inform the evaluation of the implementation of the Framework Convention of Tobacco Control in these countries.
Data were derived from the GYTS, a multinational survey aimed at assessing tobacco use, self-reported exposure to environmental tobacco smoke, knowledge and attitudes, and the factors that make youth susceptible to tobacco use. Youth in grades associated with ages 13–15 years were targeted in the GYTS. For the period 1999–2008, GYTS was conducted in 168 sites around the world. GYTS uses a standardised methodology for constructing sampling frames, selecting schools and classes, preparing questionnaires, conducting field procedures, and processing data. Coordinated internationally by WHO and Centers for Disease Control and Prevention, in each country data collection was coordinated either by a Ministry of Health or Ministry of Education. Probability samples of youth were generated using a two stage cluster sampling design. The first sampling stage involved selection of schools proportional to enrolment size. In the second stage, classrooms were randomly chosen within selected schools; all students in selected classrooms were eligible to participate. Students were presented with a self-administered questionnaire which was completed anonymously; questionnaires took 30–40 min to complete. A standardised questionnaire was used in all countries, but countries were allowed to add customised questions.
The present analysis draws on the most recent data from five North African countries (survey year in parentheses): Libya (2007), Egypt (2005), Morocco (2006), Tunisia (2007) and Sudan (2005) (n=15 155). The analyses were limited to respondents who had non-missing values for all included variables (boys n=6301, girls n=6028). Analyses of smoking initiation susceptibility were further restricted to those who reported never smoking (boys n=3638, girls n=4250). Compared with those excluded from the analysis, those included were significantly older (mean age 14.1 vs 13.7 years), less likely to be current smokers (5.1% vs 8.7%), less likely to be vulnerable to smoking initiation (15.9% vs 22.1%), less likely to have friends who smoke (33.1% vs 40.4%), but similarly likely to have parents who smoked (34.1% vs 35.5%).
Both current cigarette smoking and cigarette smoking initiation susceptibility were examined as outcome variables. Adolescents were coded as current smokers if they answered one or more days to the following question: ‘During the past 30 days (1 month), on how many of them did you smoke cigarettes?’ Adolescents who reported never trying a cigarette were considered non-susceptible to initiation if they answered ‘definitely not’ to both the following questions: (1) ‘If one of your best friends offered you a cigarette, would you smoke it?’ and (2) ‘Do you think you might start smoking over the next year?’ All other never-smokers were coded as susceptible to initiation. Because Morocco did not include the second question, adolescents from this country were not included in initiation susceptibility analyses.
A number of indicators of exposure to cigarette advertising and promotions were examined as predictors of smoking outcomes. Exposure was self-reported by respondents; question wording, response options and coding are presented in table 1. Responses to all questions were dichotomised to reflect any exposure because preliminary analyses indicated associations with smoking outcomes did not vary significantly by level of exposure for the majority of variables.
Friends’ and parents’ smoking were examined as potential mediators of the relationship between advertising/promotions exposure and smoking outcomes. Respondents were asked how many of their closest friends smoked with the response options none, some, most or all. We dichotomised responses to reflect having any close friends who smoked. Respondents were also asked if their parents smoked, with the response options none, both, father only or mother only, except for Egypt, where response options were no, my father and mother smoke, my father only smokes, my mother only smokes, both were ex-smokers, my father was an ex-smoker, my mother was an ex-smoker. For consistency across countries, and because of very low rates of maternal smoking, we dichotomised responses to reflect ANY current smoking by father or mother.
Three variables were included as controls. First, year of survey (2006 or 2007, 2005 as referent) was included to account for any time trends. Second, age at survey was included to account for increases in risk of smoking with age. Third, grade level (lowest, middle, highest) was included to capture social age as persons in higher grade levels are likely exposed to different smoking norms regardless of age.
Analyses were conducted in Stata 10.1 using svy commands to account for the complex sampling design and incorporating population weights. Prevalence estimates of current cigarette smoking, cigarette initiation susceptibility, exposure to the various forms of tobacco advertising and promotions, and friends’ and parents’ smoking were generated separately by country and gender. Bivariate logistic regression analyses were conducted to assess the crude relationship between advertising and promotions exposure and the outcome variables. We also explored whether advertising/promotions were differentially related to outcomes across countries and survey years by including interaction terms (country * exposure, survey year * exposure) in logistic regression models. In multivariable logistic regression models, we further tested the relationship between each advertising/promotions exposure and smoking outcomes net of demographic characteristics (age and grade in school) and survey year. Finally, we assessed whether the relationship between advertising/promotions exposure was mediated by parent or friends’ smoking by adding these variables to the models and assessing changes in the magnitude and significance of the exposure effect estimates.
The prevalence of current cigarette smoking and never smokers’ initiation susceptibility are displayed in table 2. Significant differences between countries were observed only in current smoking. Among school-attending boys, current smoking prevalence ranged from 5.6% (Egypt) to 15.3% (Tunisia); initiation susceptibility among never-smokers ranged from 14.1% (Sudan) to 25.0% (Tunisia). Among school-attending girls, current smoking prevalence ranged from 1.1% (Libya and Egypt) to 2.0% (Morocco and Sudan); initiation susceptibility among never-smokers ranged from 13.3% (Sudan) to 15.0% (Libya). Boys were significantly more likely than girls both to be a current smoker and to be vulnerable to smoking initiation.
The prevalence of exposure to tobacco advertisements and promotions among boys and girls by country is also presented in table 2. Substantial proportions of adolescents reported seeing each type of advertising and promotions queried. Nearly 90% of all adolescents reported seeing an actor on TV or in a movie smoking. About 40% of all participants had seen a smoking advertisement at live cultural or sporting event, and about 55% had seen a smoking advertisement in a magazine. Over half the participants reported seeing a billboard advertising tobacco. There were significant differences across countries in the prevalence of advertising exposure for both boys and girls. Particularly noteworthy were the especially high rates of seeing actors smoke on television or in movies in Tunisia, and the high rates of having seen an advertisement at a live sporting or cultural event in Sudan. Boys were significantly more likely than girls to report certain exposures, namely, owning something with a cigarette brand logo, seeing an advertisement at a live sporting or other cultural event, and being offered a free cigarette by a tobacco company representative.
Exposure to parents and friends who smoke is also presented in table 2. Parental smoking was the highest among boys and girls in Egypt (48.3% and 48.5%) and Tunisia (46.2% and 46.8%), while exposure to friends who smoke was the highest among boys and girls in Tunisia (61.9% and 41.7%) and Sudan (44.7% and 32.9%). Although boys and girls reported similar levels of parental smoking, boys were significantly more likely to report having friends who smoked.
Associations between advertising and promotions exposure and boys’ current cigarette use are presented in table 3. In analyses not shown, associations between advertising/promotions and boys’ current smoking did not vary significantly according to country or year of survey; thus data were pooled across countries. All advertising/promotions exposures were significantly associated with boys’ current smoking (OR range 1.53–2.24). Adding demographic controls and year of survey in Model 2 slightly diminished associations for two variables (actors smoking on TV/movies and seeing a televised advertisement), while for other exposures associations became stronger (owning something with a cigarette logo and being offered a free cigarette). In Model 3, parents’ smoking was added, and was found to be significantly positively associated with boys’ current smoking (adjusted OR (AOR) range 1.53–1.59). Notably, addition of this variable did not substantially attenuate associations between advertising/promotions exposure and boys’ current smoking. In Model 4, friends’ smoking was added and was found to be strongly positively associated with boys’ current smoking (AOR range 7.22–7.41). While associations between seeing actors smoking on TV/movies and current smoking became non-significant with this addition, all other advertising/promotions variables remained significantly positively associated with boys’ current smoking.
The relationship between advertising/promotions exposure and never-smoking boys’ susceptibility to smoking initiation is presented in table 4. In analyses not shown, associations between advertising/promotions and boys’ susceptibility to smoking initiation did not vary significantly according to country or year of survey; thus data were pooled across countries. Only four of the seven exposures were significantly positively associated with initiation susceptibility in crude analyses: owning an object with a cigarette brand logo (OR=1.68), seeing an advertisement on a televised sporting or cultural event (OR=1.48), seeing an advertisement in a magazine (OR=1.52), and seeing an advertisement at a live sporting or cultural event. After adding controls in Model 2, these associations remained largely the same. Adding parental smoking in Model 3 did not alter associations, although parental smoking was significantly positively associated with boys’ vulnerability to smoking initiation (AOR range 1.30–1.35). Adding friends’ smoking in Model 4 slightly attenuated associations, but in all cases advertising/promotions exposure remained significantly positively associated with boys’ smoking initiation vulnerability. Friends’ smoking was also significantly positively related to this outcome (AOR range 2.11–2.15).
The relationship between advertising/promotions exposure and never-smoking girls’ susceptibility to smoking initiation is presented in table 5. In analyses not shown, associations between advertising/promotions and girls’ susceptibility to smoking initiation did not vary significantly according to country or year of survey except for seeing actors smoking on TV/movies; for all other exposures, data were pooled across countries. When examined separately by country, seeing actors smoking on TV/movies was only significantly related to smoking initiation susceptibility among girls in Sudan (OR 0.47, 95% CI 0.28 to 0.77). Adding controls, parental smoking and friends’ smoking only somewhat diminished this association (Model 4 AOR=0.60, 95% CI 0.37 to 0.96). Among the other exposures, four were significantly positively associated with initiation susceptibility in crude analyses: owning an object with a cigarette brand logo (OR=1.85), seeing an advertisement on a televised sporting or cultural event (OR=1.33), seeing an advertisement at a live sporting or cultural event (OR=1.22) and being offered a free cigarette by a tobacco company representative (OR=1.46). After adding controls in Model 2, these associations remained largely the same. Adding parental smoking in Model 3 did not alter associations, although parental smoking was significantly positively associated with girls’ vulnerability to smoking initiation (AOR range 1.24–1.29). Adding friends’ smoking in Model 4 slightly attenuated associations; advertising/promotions exposure remained significantly positively associated with girls’ smoking initiation vulnerability, except for seeing an advertisement at a live event, which became non-significant. Friends’ smoking was also significantly positively related to girls’ initiation susceptibility (AOR range 2.79–2.82).
Cigarette use is prevalent in North Africa with significant impacts on public health. Understanding contributors to adolescent smoking initiation in this region is important for public health intervention, given most smokers report initiation during adolescence. This paper examines one policy-targetable contributor to adolescent smoking—exposure to tobacco advertising and promotions—and its relationship to adolescents’ smoking risk in North African countries. There are four major findings.
First, a substantial proportion of boys were either current smokers (5.6–15.3%) or susceptible to cigarette use initiation (14.1–25.0%) in each of the countries studied. Girls’ rates of current smoking were extremely low (≤2%), although initiation susceptibility rates were comparable with boys (13.3%–15.0%). The overall global rate of current smoking in the GYTS 2000–2007 was similar to the overall rate in the current study at 9.5%.16 Current smoking rates are somewhat comparable with those in the African region, which varied widely from a low of 1.6% in Eritrea to a high of 21.5% in Seychelles.17 The gender differences in current smoking in North Africa despite similar levels of susceptibility may suggest the religious and cultural taboos against smoking in this region may be especially salient for girls.18 However, equality in susceptibility between the genders could signal a potential increase in smoking among girls in the future, especially given the tobacco industry's stated goal of aggressively targeting women with marketing.19 Policies and programmes are warranted to prevent further encroachment into this population.
Second, self-reported exposure to tobacco advertising and promotions was substantial across all the surveyed nations. While the specific exposure types varied in prevalence between countries, 98% of adolescents reported exposure to one or more types of tobacco advertising or promotions. We were unable to identify another GYTS that examined this cumulative exposure, although a summary of GYTS surveys in 1999 also found advertising exposure to be highly variable across countries and regions of the world.16 As of the 2008 WHO Report on the Global Tobacco Epidemic (the report most closely following the GYTS surveys included in this study), only Sudan had implemented a total advertising ban according to the requirements at that time.20 As of 2011, none of the countries had comprehensive bans.17 For example, four of the countries did not ban advertising in international magazines and newspapers; three did not ban advertising on international TV/radio and on the internet. Given that one requirement of the Framework Convention on Tobacco Control is a total ban on tobacco advertising and promotions, clearly there is a need for these countries to pass adequate legislation. Research indicates partial advertising bans are not effective because tobacco manufacturers will find other communication avenues.21 Even with complete bans, however, adolescents may be exposed to advertising/promotions if enforcement is weak. For example, laws preventing billboard advertisements were passed in 2002 in Egypt, yet the 2005 GYTS survey found that 59.5% of boys and 60.5% of girls sampled were exposed to advertisements on billboards.22 Also, in Sudan, the same year that a complete ban on indirect and direct advertising went into effect, over half of the boys and almost half of the girls reported seeing a cigarette advertisement in a newspaper or magazine, a sporting event, or on a billboard.23 A Moroccan survey conducted in 2006 indicated a majority of adults had no knowledge of tobacco control laws that had been passed by the legislature, including prohibitions on tobacco advertising and selling cigarettes to youth under age 16.24 Enforcement of laws that are passed will be essential for the effectiveness of advertising and promotions bans in this region.
Third, tobacco advertising and promotions exposures were significantly and positively related to both current smoking among boys, and to initiation susceptibility among never smokers for both boys and girls. For the most part, these associations were consistent across countries and years of the survey. Positive associations between advertising/promotions and smoking risk are consistent with past studies conducted in other settings.9 Although the broad-based exposure of girls to tobacco advertising/promotions has not yet translated to high current smoking rates, given girls’ susceptibility to initiation is on par with boys, girls’ smoking rates may increase in the future. These findings underline the importance of advertising and promotions bans for tobacco control. As noted above, progress has been made in passing legislation banning tobacco advertising and promotions in these countries, although loopholes do remain; for example, for most countries, bans do not cover international media outlets. Further research on the adequacy of such laws in protecting adolescents from exposure is warranted.
Fourth, we found that although parent and peer smoking were strongly related to current smoking among boys and to initiation susceptibility among boys and girls, they did not completely explain the associations between advertising/promotions exposures and smoking outcomes. These associations remained largely unchanged after accounting for parental smoking, and were only somewhat attenuated after accounting for friend smoking. Given mediation was partially supported only for peer smoking, this may indicate youth-centric advertising and promotions in these countries. Research in other contexts suggests that tobacco advertising and promotions affect adolescents’ smoking risk through perceptions of smoking acceptability, curiosity about smoking, and perceived benefits and risks of smoking.6 ,7 Future research that establishes these mediational pathways in these countries may be helpful in designing campaigns to counteract tobacco industry promotions and advertising if total bans are not immediately feasible or enforceable.
Although this study has numerous strengths, including the use of large, population-based probability samples and consistent survey questionnaires across countries, findings should be interpreted with knowledge of study limitations. The GYTS is fielded among school-attending youth, and thus misses out-of-school youth. According to the World Bank, secondary school enrolment rates around the time of the last GYTS in these countries were 85% and above for Egypt, Libya and Tunisia, but were 32% for Sudan and 53% for Morocco.25 Given these youth often have riskier behavioural profiles, the GYTS may underestimate tobacco use in the broader population of youth, especially in Morocco and Sudan. We also limited our analyses to respondents with complete data, which may result in underestimates of current smoking, smoking susceptibility and peer smoking. This could potentially bias study results. The data analysed were cross-sectional, which leaves open the possibility of reverse-causality. It is possible that current smoking, for example, predisposes youth to pay more attention to tobacco advertisements and promotion, which could explain part of the association between self-reported advertising exposure and current smoking status. Data are self-reported, and thus vulnerable to recall and reporting biases. Finally, we were unable to adjust for important potential confounders, such as socioeconomic status, because the GYTS only asks a very narrow set of demographic questions (eg, age, grade in school and gender).
In conclusion, exposure to tobacco advertising and promotions was prevalent among Northern African adolescents, and such exposure was significantly related to adolescents’ risk for smoking cigarettes. Such findings can serve as a baseline against which to measure progress of the implementation of the Framework Convention on Tobacco Control in this region, in particular comprehensive advertising bans. Research is warranted to examine trends in adolescents’ exposures to tobacco advertising/promotions and smoking risk since the previous GYTS surveys were fielded, especially given recent social changes in many of these countries.
A substantial proportion of adolescents in North African countries reported being exposed to tobacco advertising and promotions at the time of the last Global Youth Tobacco Survey: 98% reported exposure to one or more types of tobacco advertising and promotions.
Exposure to advertising and promotions was linked with increased odds of current smoking among adolescent boys, and increased risk of susceptibility to smoking initiation among boys and girls.
Although parental and peer smoking were associated with increased odds of current smoking and susceptibility to smoking initiation, only peer smoking partially mediated the association between advertising/promotion exposure and smoking risk.
An earlier version of this manuscript was presented at the American Public Health Association Annual meeting 2011 in Washington, DC.
Contributors ASM led the conceptualisation of the research questions, conducted the data analyses, led the writing of the manuscript and provided final approval of the version to be published. ECL and LA contributed to the interpretation of results, the writing of the manuscript and provided final approval of the version to be published. CCJ contributed to the study's conceptualisation, the interpretation of the analysis results, provided feedback on drafts of the manuscript and provided final approval of the version to be published.
Funding Dr ASM's time on this project was supported in part by HRSA MCHB T76MC04927.
Competing interests None.
Ethics approval The Social Behavioural IRB at Tulane University deemed the study to be exempt from review due to its use of de-identified, publicly available data.
Provenance and peer review Not commissioned; externally peer reviewed.
Data sharing statement Data are publicly available for download from the CDC website.
Patient consent For the current study, no consent was sought, since it was a secondary analysis of existing, publicly available data. Consent procedures for the GYTS surveys followed each country's research ethics board requirements.
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.