OBJECTIVE: To test the specificity of the association between cigarette advertising and adolescent smoking initiation.
METHODS: A longitudinal survey of 2102 adolescents, aged 10 to 17 years at baseline, who never smoked was conducted by using masked images of 6 cigarette advertisements and 8 other commercial products with all brand information digitally removed. The exposure variable was a combination of contact frequency and cued recall of brands for cigarette and other advertisements. Multilevel mixed-effects Poisson regressions were used to assess smoking initiation 9 months after the baseline assessment as a function of cigarette-advertisement exposure, other advertisement exposure, and baseline covariates.
RESULTS: Thirteen percent (n = 277) of students initiated smoking during the observation period. Although the incidence of trying smoking was associated with increased exposure to cigarette advertisements (10% in the low, 12% in the medium, and 19% in the high cigarette-advertisement exposure tertile initiated smoking), exposure to other advertisements did not predict smoking initiation. Compared with low exposure to cigarette advertisements, high exposure remained a significant predictor of adolescent smoking initiation after controlling for baseline covariates (adjusted relative risk: 1.46 [95% confidence interval: 1.08–1.97]; P < .05).
CONCLUSIONS: Our results support the notion of a content-related effect of cigarette advertisements and underlines the specificity of the relationship between tobacco marketing and teen smoking; exposure to cigarette advertisements, but not other advertisements, is associated with smoking initiation.
WHAT'S KNOWN ON THIS SUBJECT:
It has been well documented that exposure to tobacco marketing is a risk factor for smoking initiation among youth. However, few studies have tested the specificity of this association.
WHAT THIS STUDY ADDS:
This study extends findings from other studies and shows (from a longitudinal design) that exposure to cigarette advertising is significantly associated with youth smoking initiation, whereas exposure to advertising for other commercial products is not.
In recent years, increasing efforts have been made to reduce the prevalence of tobacco use and the exposure to environmental tobacco smoke worldwide.1 One strategy for combating the smoking epidemic is to prevent smoking uptake during adolescence.2 The argument for smoking prevention among youth is based on the observational finding that if smoking does not start during adolescence, then it is unlikely ever to occur and on data indicating that the probability of continuing smoking is related to early age at initiation.3 Even infrequent experimental smoking in youth may increase the risk of smoking in adulthood.4 Nicotine is a potent psychoactive drug, and once smoking has begun, cessation is difficult and smoking is likely to become a long-term addiction.5 There are a number of reasons why adolescents start to experiment with smoking; the most important reasons are psychosocial in nature, such as peer and parental smoking.6
One risk factor for smoking initiation is exposure to tobacco advertisements. Many studies document an association between tobacco marketing and smoking experimentation in young people.7 Data from a recent meta-analysis suggest that tobacco marketing is influential for adolescents who have never smoked.8 According to this meta-analysis, exposure to protobacco marketing increases the odds of adolescents holding positive attitudes toward smoking and more than doubles the odds of them initiating smoking.
Most of this research is observational in nature. Sir Austin Bradford Hill9 has defined criteria that, if fulfilled, strengthen the interpretation of an association between a given risk factor and a disease as causal. These criteria include, among others, (1) strength, (2) consistency, (3) specificity, (4) temporality, and (5) scientific plausibility. A recent systematic review10 applied the Hill criteria to determine whether there is evidence of a causal link between exposure to tobacco promotion and tobacco use in children and summarized the results of studies conducted in 5 continents with more than 300 000 subjects. DiFranza and colleagues10 showed that (1) children are exposed to tobacco promotion before the initiation of tobacco use (criterion temporality), (2) exposure increases the risk for initiation (criterion strength), (3) there is a dose-response relationship (greater exposure results in higher risk) (criterion strength), (4) the increased risk is robust (criterion consistency), and (5) the risk is scientifically plausible. Although all of these findings are in line with the Hill criteria, 1 criterion was not mentioned in the review: specificity. In summary, the association between tobacco advertising and youth smoking has been the subject of a substantial number of studies, but few have tested the effect of tobacco advertising compared with advertisements for other commercial products.11,12 Thus, the relationship, although considered causal,13 lacks evidence for specificity. The question addressed in this study is, “Does exposure to tobacco advertisements, rather than exposure to advertising in general, contribute to youth smoking?” It is conceivable that some adolescents have enhanced attention and memory for all forms of advertising, along with the propensity to act on that information. It is this proclivity that accounts for much of the association between exposure to cigarette advertising and youth smoking, rather than the specific messages transmitted by the advertisements.
Although several countries all over the world (eg, Finland, Italy, and New Zealand) have strong tobacco-marketing regulations in place, other countries, such as the United States and Germany, have implemented considerably weaker tobacco-marketing policies. For example, Germany has banned tobacco advertisements in television, radio, newspapers, and magazines, but tobacco marketing is still allowed at point of sale, on billboards, and in cinemas before movies that show after 6:00 pm. In addition, brand stretching is also legally possible.
In a previous cross-sectional study14 we emphasized the specificity of the finding of an association between cigarette advertising and teen smoking, showing a multivariate association with exposure to cigarette and not control advertisements. A major limitation of this study was the cross-sectional design, which made it impossible to determine temporal sequence. The current longitudinal design enables us to follow-up those students that were never smokers at the time of exposure, weakening the argument that the advertising–smoking association simply reflects higher attention and memorization among smokers for cigarette advertising.
We invited 120 randomly selected schools from 3 states in Germany (Brandenburg, Hamburg, and Schleswig-Holstein) to participate in a school-based survey.14 We distributed a self-administrated written survey in 2008 to adolescents (aged 10–17 years) enrolled in the 29 schools that agreed to participate and repeated the survey 9 months later. Trained research staff administered both confidential surveys during class time with parental written permission and student assent. To permit a linking of individual information on subsequent surveys, each questionnaire was labeled with a 7-digit individual code generated by the student, a procedure that had been tested in previous studies15 but slightly modified for this trial. Study implementation was approved by the ministries of cultural affairs of the 3 states involved. Ethical approval was obtained from the ethical committee of the medical faculty of the University of Kiel (reference D 417/08).
A total of 3415 students were surveyed at baseline (response rate: 81.4%), of whom 51.6% were girls. The mean age was 12.5 years (SD: 1.06). The prevalence of those who ever tried smoking was 31.1%. A total of 3029 students (88.7% of the baseline sample) completed follow-up surveys, including 2102 of 2346 (89.6%) students who were not smokers at baseline.
Student self-reports included (1) demographic data (age and gender), (2) advertising exposure measures, (3) outcome measures (smoking initiation), and (4) potential covariates.
The theoretical basis of the current research is a model of the effects of advertising exposure derived from theories of memory and attention, developed in the psychological literature.16,17 According to these theories, the effects of exposure to advertising are mediated by attentional mechanisms that determine what the subject perceives and memory mechanisms that determine the amount of information retained from each advertisement. The underlying theories are described in detail elsewhere.18 According to Klitzner et al18 it can be assumed that “differing amounts and types of information will be retained from advertisements depending on the amount and manner of attention paid by subjects to ad stimuli and their efficiency in transferring this information from short-term to long-term memory stores.”18(p288)
Students were provided with masked colored images of different advertisements (fixed images of billboard images for cigarette advertisements and mostly television commercials for the other advertisements) with all brand information digitally removed (Fig 1). The images included 6 cigarette brands and 8 “control” advertisements for products that included sweets, clothes, mobile telephones, and cars. The following cigarette brands were included in the survey (with advertisement theme or cue in parentheses): (1) Marlboro (cowboy, horses); (2) F6 (sunrise); (3) Gauloises (couple); (4) Pall Mall (Empire State Building); (5) L&M (couple); and (6) Lucky Strike (cigarette packs). These 6 cigarette brands are among the 8 most popular cigarette brands in Germany.19 For other commercial products, the following advertisements were included in the survey (with product type and advertisement theme or cue in parentheses): (1) Jack Wolfskin (trekking clothing; climber); (2) Volkswagen (car; the performer Seal); (3) Tic Tac (candy; elevator); (4) Dr Best (tooth brush; tomato); (5) Kinder Pingui (sweet; penguins); (6) T-Mobile (mobile telephone; dog); (7) Spee (detergent; fox); and (8) Toyota (car).
Advertisement contact frequency was measured by asking the students to rate, on a 4-point scale, how often they had seen the advertisement extract (0, never; 1, 1–4 times; 2, 5–10 times; 3, >10 times). Cued brand recall performance was measured by asking the students which brand was advertised (open format). Correct brand names were post coded as 1 and all other answers as 0 (misspellings of brands were counted as correct). Because contact frequency and cued brand recall were highly internally consistent, the 2 measures were combined into a single scale (1 for cigarette-advertisement exposure and another for other advertisement exposure) using Stata's “α” and “generate” commands (Stata Corp, College Station, TX). Cronbach's α was 0.74 for both scales.
We assessed lifetime smoking experience by asking, “How many cigarettes have you smoked in your life?” (never smoked, just a few puffs, 1–19 cigarettes, 20–100 cigarettes, or >100 cigarettes).20 The cohort consisted of those who had never smoked at baseline; any smoking by the follow-up survey, even just a few puffs, was considered initiation of smoking.
Covariates were included to assess the independent association between exposure to tobacco advertising and adolescent smoking. The following factors were included as covariates because they were, on the basis of previous research,14,21 associated with exposure to tobacco advertising and adolescent smoking and not considered to be part of the causal pathway between the exposure and the behavioral outcome. Sociodemographic factors included age, gender, and socioeconomic status (SES). The SES of the students was approximated with a combination of student and class teacher ratings. Students answered 3 items of the Programme for International Student Assessment,22 which is a cultural and social capital assessment that asks for the number of books in the household (5-point scale, from 0 [none] to 4 [>100]) and parenting characteristics (“My parents always know where I am” and “My parents know other parents from my school”)23; class teachers filled out an 11-item school evaluation sheet related to the SES of their students (for example “Most students of the school live in families with financial problems,” “Most students of the school come from underprivileged families,” “Our school has a good reputation”; scale range: 0 [not true at all] to 3 [totally true]; Cronbach's α = 0.85); and student and teacher ratings were correlated (r = 0.57; α = 0.72). This measure is strongly correlated to type of school (r = 0.66), which is a sensitive predictor of the SES of the students in Germany.24 For analysis, we classified SES into population quartiles.
Covariates used to assess social influences on smoking included the smoking status of parents (“Does one of your parents smoke?” [0, no; 1, yes]) and the smoking status of the peers (“How many of your friends smoke?” [0, none; 1, some; 2, most; 3, all]); for the analysis, smoking of peers was dichotomized into none versus any because only 2 pupils indicated that “all” of their friends smoke, and only an additional 40 students answered that question with the answer “most of my friends smoke.” We controlled for personality characteristics that might prompt adolescents to be more attentive to tobacco advertising. Rebelliousness and sensation-seeking were assessed by using 4 items combined into a single index; higher scores indicate greater propensity for rebelliousness and sensation-seeking (Cronbach's α = 0.76)25: “I get in trouble in school,” “I do things my parents wouldn't want me to do,” “I like scary things,” or “I like to do dangerous things.” Response categories were “not at all like me,” “a little like me,” ”pretty much like me,” or “exactly like me.” For the analysis, scores were split at their median. Also included as a covariate was self-reported school performance with the item “How would you describe your grades last year?” Response categories were “excellent,” “good,” “average,” or ”below average.” Furthermore, average television screen time was included as a covariate and measured with the item “How many hours do you usually watch TV in your leisure time?” Response categories were “none,” “about half an hour,” “about an hour,” “approximately 2 hours,” “approximately 3 hours,” “approximately 4 hours,” or “more than 4 hours a day.”
The multivariate associations between the amount of advertising exposure and smoking behavior were analyzed with multilevel mixed-effects Poisson regressions using Stata's “xtmepoisson” command. Poisson regression allows for the presentation of adjusted relative risks and 95% confidence intervals for the relationship between exposure to advertising and smoking onset. Relative risks have the advantage of not being influenced by the prevalence of the exposure. The dichotomized outcome variable, smoking initiation, was regressed on advertising exposure (which was parsed into tertiles) with random effects for school and class. In a first step, unadjusted models were specified with either cigarette advertising, other advertising exposure, or covariates entered as a fixed effect. In the more complex models, all covariates were entered with the 2 advertising exposure types (cigarette versus other commercial products) as fixed effects. In each model, the first (lowest) tertile of exposure to advertising served as the reference category. All analyses were performed by using Stata 11.1.
Participants lost to follow-up were older, reported lower school performance, were higher on sensation-seeking/rebelliousness, had more friends who smoke, had higher exposure to cigarette advertisements, and were more often recruited from schools in the German state of Hamburg.
Descriptive statistics for the advertising exposure measures and covariates are given in Table 1. The sample ranged in age from 10 to 17 years (the majority ranged between 10 and 13 years), and there were slightly more girls in the sample. Approximately 47% of students had 1 or more parents who smoked (according to the German Enviornmental Survey, 46% of Germany's 3- to 14-year-old children lived in households with at least one smoker),26 and 27% of students reported that some of their peers smoked.
Exposure to Advertisements at Baseline
Table 2 illustrates contact frequency (how often the student had seen the advertisement) and cued recall (how often the student correctly identified the brand) rates for all advertised products at baseline. In general and for all products, at least 1-time contact rates were higher than cued recall rates. The 2 most highly recalled cigarette brands were Marlboro and Lucky Strike, for which cued recall rates were 21% and 15%, respectively. The lowest advertisement contact frequency rate was found for F6, a regional German cigarette brand sold mainly in the former German Democratic Republic. Advertisement contact frequency for other products was generally much higher than for cigarette brands. For example, 98% reported ever seeing an advertisement for Kinder Pingui, and 88% correctly identified the brand.
Table 3 shows a zero-order correlation matrix for all variables used in the study. Smoking initiation was significantly associated with having friends who smoke, sensation-seeking, age, SES, school performance, exposure to cigarette advertisements, parental smoking, television screen time, German state, and gender. Exposure to cigarette advertisements was associated with exposure to other advertisements, age, television screen time, sensation-seeking, having friends who smoke, and SES. To account for these differences, we adjusted for these variables in the multivariate analysis.
Association Between Exposure to Advertisements and Teen Smoking Initiation
Thirteen percent (n = 277) of students initiated smoking during the observation period (137 students smoked just a few puffs, 112 students smoked between 1 and 19 cigarettes, 12 students smoked between 20 and 100 cigarettes, and 16 students smoked >100 cigarettes). The incidence of trying smoking was associated with increased exposure to cigarette advertisements: 10% in tertile 1 (low exposure); 12% in tertile 2 (medium exposure); and 19% in tertile 3 (high exposure). Exposure to advertisements for other products than cigarettes predicted smoking initiation in the observation period in the unadjusted models only. In contrast, compared with low exposure to cigarette advertisements, high exposure to cigarette advertisements remained a significant predictor of adolescent smoking after controlling for all covariates in Table 1 (adjusted relative risk: 1.46 [95% confidence interval: 1.08–1.97]; P < .05) (see Table 4). Other predictors of smoking onset included older age, low SES, having friends who smoke, sensation-seeking/rebelliousness, and low school performance. After controlling for covariates, there was no effect of television screen time, having a parent who smokes, gender, or state of residence.
This longitudinal study investigated whether the association between cigarette advertising and youth smoking is attributable to specific cigarette advertising content. We have applied an advertisement exposure measure that involves both advertisement contact frequency as well as cued brand recall to parse exposure to cigarette-advertisement content from advertising content for other products. The study shows that exposure to cigarette-advertisement images was associated with smoking initiation in a prospective design, even after controlling for a number of well-established risk factors. In contrast, exposure to other advertisements at baseline (although it had a bivariate association) was not associated with smoking onset. Therefore, study results suggest that the association between tobacco advertisement and youth smoking initiation is specific to tobacco advertising content and not simply a marker of an adolescent that is generally receptive to marketing.
Why is cigarette advertising aimed at adolescents so powerful? Tobacco companies aim their message at adolescents because this is when most people start smoking.27 Cigarette marketers have created brands with multiple aspirational images,28 each designed to fit the needs common among adolescents. Adolescents are in the process of identity formation, when they face emotional instability and social self-consciousness. Aspirational imagery used in cigarette advertising is especially appealing, because it associates the behavior, smoking, with characteristics adolescents are trying to assimilate, such as masculinity (for boys), thinness (for girls), independence, extroversion, and sex appeal.29,30
This study is subject to limitations inherent in any observational study. Loss to follow-up could affect generalizability; adolescents at lower risk were more likely to be retained. These results may not apply to adolescents with multiple risk factors for smoking. As with any observational study, the results may be biased by unmeasured confounding; that is, an unmeasured risk factor could alter the estimates reported for the association between cigarette advertising and smoking onset. Furthermore, we have not considered general attitudes toward smoking in selecting the baseline sample of adolescents. Finally, we recognize that the measure of exposure to cigarette advertising is a combination of contact frequency and brand recall, both of which could be affected by characteristics of the adolescent and therefore not pure measures of advertisement exposure.
The finding that cigarette advertising predicts smoking in youth could have important public health implications. Five of 6 cigarette advertisements used in the study were advertisements of internationally sold cigarette brands. This fact, and also the fact that only a limited number of countries worldwide have implemented a total ban on tobacco marketing, increases the likelihood that findings from our study may have policy implications beyond the borders of Germany. A total ban of tobacco advertising and promotion around the world is a key policy measure of the World Health Organization Framework Convention on Tobacco Control.31 Under Article 13.1 of the Framework Convention on Tobacco Control, “Parties recognize that a comprehensive ban on advertising, promotion and sponsorship would reduce the consumption of tobacco products.” Data from this study support this measure because study results support the notion of a content-related effect of cigarette advertisements and underline the specificity of the relationship between tobacco marketing and teen smoking initiation.
This work was supported by the Deutsche Angestellten-Krankenkasse, a German health insurance company. Dr Hanewinkel obtained funding for the study.
We thank Dorothea Denker, Mandy Gauditz, Lars Grabbe, Sven Heid, Frank Kirschneck, Carmen and Sarah Koynowski, Corinna Liefeld, Danuta Meinhardt, Marc Räder, Gesa Sander, and Martina Staacken for assessing the data.
This work is dedicated to the memory of our colleague and friend Dr Wolf-Rüdiger Horn, a pediatrician who inspired our work in the smoking-prevention area.
- Accepted November 11, 2010.
- Address correspondence to Reiner Hanewinkel, PhD, Institute for Therapy and Health Research (IFT-Nord), Harmsstrasse 2, 24114 Kiel, Germany. E-mail:
All authors participated in the concept and design of the study, analysis and interpretation of data, and drafting of the manuscript, and all authors approved the manuscript as submitted.
FINANCIAL DISCLOSURE: The authors have indicated that they have no personal financial relationships relevant to this article to disclose.
- SES =
- socioeconomic status
- 2.↵American Academy of Pediatrics, Committee on Environmental Health, Committee on Substance Abuse, Committee on Adolescence, Committee on Native American Child. Tobacco use: a pediatric disease. Pediatrics. 2009;124(5):1474–1487
- Lovato C,
- Linn G,
- Stead LF,
- Best A
- DiFranza JR,
- Wellman RJ,
- Sargent JD,
- Weitzman M,
- Hipple BJ,
- Winickoff JP
- Emri S,
- Bagci T,
- Karakoca Y,
- Baris E
- Peters J,
- Betson CL,
- Hedley AJ,
- et al
- 13.↵National Cancer Institute. The Role of Media in Promoting and Reducing Tobacco Use: Tobacco Control Monograph No. 19. Bethesda, MD: US Department of Health and Human Services, National Institutes of Health, National Cancer Institute; 2008. NIH publication 07-6242
- Neisser U
- 19.↵Deutsches Krebsforschungszentrum. The Tobacco Atlas Germany 2009[in German]. Heidelberg, Germany: Deutsches Krebsforschungszentrum; 2009
- Bondy SJ,
- Victor JC,
- Diemert LM
- Kunter M,
- Schümer G,
- Artelt C,
- et al
- Lange M,
- Kamtsiuris P,
- Lange C,
- Schaffrath RA,
- Stolzenberg H,
- Lampert T
- Pollay RW
- Shibuya K,
- Ciecierski C,
- Guindon E,
- Bettcher DW,
- Evans DB,
- Murray CJ
- Copyright © 2011 by the American Academy of Pediatrics