OUP user menu

Impact of Alcohol Advertising and Media Exposure on Adolescent Alcohol Use: A Systematic Review of Longitudinal Studies

Peter Anderson, Avalon de Bruijn, Kathryn Angus, Ross Gordon, Gerard Hastings
DOI: http://dx.doi.org/10.1093/alcalc/agn115 229-243 First published online: 14 January 2009

Abstract

Aims: To assess the impact of alcohol advertising and media exposure on future adolescent alcohol use. Methods: We searched MEDLINE, the Cochrane Library, Sociological Abstracts, and PsycLIT, from 1990 to September 2008, supplemented with searches of Google scholar, hand searches of key journals and reference lists of identified papers and key publications for more recent publications. We selected longitudinal studies that assessed individuals' exposure to commercial communications and media and alcohol drinking behaviour at baseline, and assessed alcohol drinking behaviour at follow-up. Participants were adolescents aged 18 years or younger or below the legal drinking age of the country of origin of the study, whichever was the higher. Results: Thirteen longitudinal studies that followed up a total of over 38,000 young people met inclusion criteria. The studies measured exposure to advertising and promotion in a variety of ways, including estimates of the volume of media and advertising exposure, ownership of branded merchandise, recall and receptivity, and one study on expenditure on advertisements. Follow-up ranged from 8 to 96 months. One study reported outcomes at multiple time-points, 3, 5, and 8 years. Seven studies provided data on initiation of alcohol use amongst non-drinkers, three studies on maintenance and frequency of drinking amongst baseline drinkers, and seven studies on alcohol use of the total sample of non-drinkers and drinkers at baseline. Twelve of the thirteen studies concluded an impact of exposure on subsequent alcohol use, including initiation of drinking and heavier drinking amongst existing drinkers, with a dose response relationship in all studies that reported such exposure and analysis. There was variation in the strength of association, and the degree to which potential confounders were controlled for. The thirteenth study, which tested the impact of outdoor advertising placed near schools failed to detect an impact on alcohol use, but found an impact on intentions to use. Conclusions: Longitudinal studies consistently suggest that exposure to media and commercial communications on alcohol is associated with the likelihood that adolescents will start to drink alcohol, and with increased drinking amongst baseline drinkers. Based on the strength of this association, the consistency of findings across numerous observational studies, temporality of exposure and drinking behaviours observed, dose-response relationships, as well as the theoretical plausibility regarding the impact of media exposure and commercial communications, we conclude that alcohol advertising and promotion increases the likelihood that adolescents will start to use alcohol, and to drink more if they are already using alcohol.

Introduction

Adolescents are frequent users of alcohol and increasingly consume it in a risky fashion. For example, in Europe, nearly all (over 9 in 10) 15- to 16-year-old students have drunk alcohol at some point in their life (Currie et al., 2004), starting on average just after 12½ years of age. Data from the 2003 European School Survey Project on Alcohol and Other Drugs (ESPAD) found that the average amount of alcohol drunk by 15- to 16-year olds on the last drinking occasion was 60 g (six drinks) (Hibell et al., 2004). Over one in eight (13%) of 15- to 16-year-old students reported being drunk more than 20 times in their life, and over one in six (18%) reported binge drinking (5+ drinks on a single occasion) three times or more in the last month. Binge drinking in young people has increased across much of Europe in the last 10 years, although more so in the early part of this period (Anderson and Baumberg, 2006).

Children and adolescents have greater vulnerability to alcohol than adults. As well as usually being physically smaller, they lack experience of drinking and its effects. They have no context or reference point for assessing or regulating their drinking, and, furthermore, they have built up no tolerance to alcohol. From mid-adolescence to early adulthood, there are major increases in the amount and frequency of alcohol consumption and alcohol-related problems (Bonomo et al., 2004; Wells et al., 2004). Those with heavier consumption in their mid-teens tend to be the ones with heavier consumption, alcohol dependence and alcohol-related harm, including poorer mental health, poorer education outcome and increased risk of crime in early adulthood (Jefferis et al., 2005, Englund et al., 2008; Pitkänen et al., 2008). During adolescence, alcohol can lead to structural changes in the hippocampus (a part of the brain involved in the learning process) (De Bellis et al., 2000) and at high levels can permanently impair brain development (Spear, 2002). Drinking by adolescents and young adults is associated with automobile crash injury and death, suicide and depression, missed classes and decreased academic performance, loss of memory, blackouts, fighting, property damage, peer criticism and broken friendships, date rape, and unprotected sexual intercourse that places people at risk for sexually transmitted diseases, HIV infection and unplanned pregnancy (Bonomo et al., 2001). Adolescents aged 14–17 years with alcohol use disorders show substantially greater brain activation to alcoholic beverage pictures than control youths, predominantly in brain areas linked to reward, desire and positive affect (Tapert et al., 2003). The degree of brain response to the alcohol pictures is highest in youths who consume more drinks per month and report greater desires to drink.

Alcohol advertising is one of the many factors that have the potential to encourage youth drinking. For young people who have not started to drink, expectancies are influenced by normative assumptions about teenage drinking as well as through the observation of drinking by parents, peers and models in the mass media. Research has linked exposure to portrayals of alcohol use in the mass media with the development of positive drinking expectancies by children and adolescents (Austin and Knaus, 2000; Austin et al., 2000). Young people with more positive affective responses to alcohol advertising hold more favourable drinking expectancies, perceive greater social approval for drinking, believe drinking is more common among peers and adults, and intend to drink more as adults (Chen and Grube, 2002). Fourteen-year olds with greater exposure to advertisements in magazines, at sporting and music events and on television are more advertisement-aware than those with less exposure, as are teens who watch more TV, pay attention to beer advertisements and know adults who drink (Collins et al., 2003). Amongst 10- to 17-year olds, the perceived likeability of beer advertisements is a function of the positive affective responses evoked by the specific elements featured in the advertisements. Liking of specific elements featured in beer advertisements, such as humour, animation and popular music, significantly contribute to the overall likeability of these advertisements and subsequently to advertising effectiveness indicated by an intent to purchase the product and brand promoted by the advertisements (Chen et al., 2005).

These studies, however, do not establish whether alcohol advertising actually influences young people's drinking behaviour. Answering this question requires either experimental studies, which are not possible for ethical reasons, or systematic observation of real world effects.

One approach to observation is to look for correlations between the amount of alcohol advertising and the amount of drinking taking place in a particular jurisdiction using econometric methods. It is hypothesized that, if advertising has an effect, drinking rates should shadow temporal variations in the amount of advertising. Establishing such a link, however, is problematic for a number of reasons. First, measures of the amount of advertising, which typically use expenditure on advertising, vary in the accuracy and inclusiveness. For example, in the UK, whilst mass media advertising expenditure has been estimated to be £202.2 million (UK Cabinet Office, 2003), expenditure on promotion more generally (taking in point of sale promotion, electronic communications and other ‘below the line’ activity) is thought to approach £800 million (Drink Pocket Book, 2006). Second, the analysis depends on the construction of a complex model that ascribes values for all the different variables—including price, drinking restrictions and disposable income—as well as advertising (Harrison and Godfrey, 1989; Casswell, 1995; Saffer, 1996) that might be implicated. Third, the duration of advertising effects need to be taken into account: a powerful campaign may continue to have an effect years after it was first deployed. Indeed, advertisers deliberately try to enhance these long-term effects as part of their effort to build brands. Fourth, variations in the amount of advertising tend to be minor (few comprehensive bans have been introduced) so researchers are looking for potentially very small changes in drinking patterns. Finally, and most importantly given our focus on adolescents, measures of the overall amount of advertising do not necessarily give an accurate picture of youth exposure.

To obtain this focus on young people, it is necessary to do research directly with them. Such investigations come in two forms: cross-sectional and longitudinal. Cross-sectional studies take a snapshot of advertising exposure (awareness and/or appreciation) and levels of drinking, and look for correlations between the two. However, because they cannot show whether exposure preceded drinking uptake, they leave open the possibility that any correlation is as likely to reflect drinking encouraging young people to take an interest in advertising, as vice versa.

As Aitken et al. (1988) point out, however, paying attention to advertising presupposes that the viewer is getting some benefit or reward from it—most fundamentally that they are doing the right thing by consuming the advertised product—and advertisers deliberately design their work to provide such rewards (Aitken, 1988). Thus, cross-sectional data can shed a useful light on the role of alcohol advertising in young people's drinking.

Longitudinal studies take the debate a step further by measuring exposure at time A, and how this relates to drinking at time B. Provided potential confounders (such as peer and parental drinking) are controlled for, any correlation indicates a causative relationship. This review therefore focuses on longitudinal studies with young people. It builds on and extends reviews conducted by Jernigan (2006), Smith and Foxcroft (2007) and Booth et al. (2008).

Methods of the Review

Types of studies

We considered studies that examined the association between alcohol advertising and promotion, the portrayal of alcohol in mass media, and adolescent drinking. We included only longitudinal studies in which individuals’ drinking behaviour and exposure to advertising, receptivity or attitudes to alcohol advertising, or brand awareness were measured at baseline and individuals’ drinking behaviours were then measured in one or more follow-ups. Experimental, cross-sectional and time-series or econometric studies were excluded from this review.

Types of participants

Studies that included adolescents 18 years of age or younger were reviewed with the exception of US-based studies, where the legal drinking age of 21 years was taken as the cut-off.

Types of intervention

The ‘intervention’ is alcohol mass media advertising by the industry, including portrayal of alcohol in the mass media, alcohol promotion and media exposure that contained alcohol advertisements. Mass media channels of communication include advertising delivered through television, radio, newspapers, outdoor advertising, posters, etc. Alcohol promotion includes give-aways and items bearing alcohol industry logos. In practice, the measure of exposure to the intervention may not discriminate between specific types of advertising, since adolescents are exposed to many sources.

Types of outcome measures

Self-reported drinking status.

Search strategy

We searched MEDLINE, the Cochrane Library, Sociological Abstracts, and PsycLIT, from 1990 to September 2008, supplemented with searches of Google scholar, hand searches of key journals and reference lists of identified papers and key publications for more recent publications. The search strategy combined the following four sets of terms. Child Search Strategy: Child(MeSH) OR Child* OR Schoolchild* OR School age* OR Kid OR Kids OR Adolescent(MeSH) OR Adoles* OR Teen* OR Boy* OR Girl* OR Minors(MeSH) OR Minors OR Schools(MeSH) OR Primary school* OR Secondary school* OR Elementary school* OR High school* OR Highschool* Or College* OR Universit* OR Young OR Youth*. Alcohol Search Strategy: Alcohol drinking(MeSH) OR Alcohol* drink* OR Alcoholic beverages(MeSH) OR Alcohol* beverage* OR Beer(MeSH) OR Beer* OR Wine(MeSH) OR Wine* OR Liquor* OR Spirits OR Alcohol*. Marketing Search Strategy: Marketing(MeSH:NoExp) OR Marketing OR Advertising as Topic(MeSH) OR Advert* OR Promot* OR Sponsor* OR Television(MeSH) OR Televis* OR TV* OR Radio(MeSH) OR Radio OR Radios OR Motion picture* OR Movie* OR Film* OR Display* OR Billboard* OR Poster OR Posters OR Newspapers(MeSH) OR Newspaper* OR Magazine* OR Mass media(MeSH) OR Internet(MeSH) OR Internet. Longitudinal Studies Search Strategy: Longitudinal Studies(MeSH) OR Longitud* OR Cohort* OR Follow-up* OR Prospectiv* OR Subsequ* OR Wave*.

There were four stages in the review process:

  1. Studies identified in the electronic search were pre-screened for relevance by a reviewer. Articles were rejected if the title and abstract did not focus on the impact of alcohol advertising or promotion on adolescent drinking behaviour. If the article could not be rejected with certainty, the full text was obtained and screened by two reviewers.

  2. Two reviewers independently assessed relevant studies for inclusion.

  3. One reviewer extracted data from included studies using a form and a second reviewer checked these data.

  4. Studies were combined using qualitative narrative synthesis because there was heterogeneity among study designs, type of ‘intervention’ and outcomes measured.

Results

The search strategy resulted in 810 titles, reduced to 729 following deletion of duplicates. Initial assessment of the titles and abstracts reduced the number of papers to 131, further reduced to 29 on closer assessment of the abstract and full text. Sixteen publications reporting on 13 studies met the inclusion criteria. One longitudinal study was excluded, because the use of alcohol at baseline was not accounted for (Wingood et al., 2003). No additional methodological quality criteria were used in selecting papers for inclusion. Table 1 summarizes the studies, describing the alcohol marketing and media exposures, the drinking behaviour outcome measures, the sample and study designs, the survey methods, the baseline sample sizes and follow-up rates, the methods of analyses, the confounders analysed and the outcome at follow-up. The individual studies were not ranked for methodological quality.

View this table:
Table 1

Summary of included studies

The 13 studies included a variety of different age groupings that ranged between 10 and 21 years of age at baseline. Ten studies were conducted in the United States, one in Belgium, one in Germany and one in New Zealand. The years during which data were collected ranged between 1985 and 2005. Baseline sample sizes ranged from 630 to 6522, with a total of over 38,000 at follow-up across the 13 studies.

Two studies investigated the impact of media exposure (television and music videos) on the use of alcohol; three studies, alcohol use in motion pictures; two studies, a range of marketing exposure (including TV, magazines, concession stands at sports or music events, and in store advertisements); two studies, ownership of alcohol branded merchandise; one study, TV alcohol commercials alone; one study, recall and liking of advertisements; one study, outdoor advertising; one study, brand recognition, recall and receptivity to alcohol marketing; and one study, volume of and expenditure on advertisements.

In 10 studies, participants were followed up once after baseline. The duration of the follow-up was 12 months, 18 months, 24 months, 30 months and 12–26 months. One study followed up participants at 8, 16 and 24 months. The New Zealand study reported outcomes at multiple time points, 3 years, 5 years and 8 years. One study evaluated participants at four time points and presents results for follow-up after 21 months taking the multiple time points into account in the analysis (Snyder et al., 2006). Attrition rates varied from 31% to 100% (the sample with 100% follow-up included and analysed all students with alcohol consumption measurements at baseline and follow-up (Casswell and Zhang, 1998). Three studies used imputation for missing data (Ellickson et al., 2005; Collins et al., 2007; Wills et al., 2008); all other studies excluded participants with missing data from the analyses.

All studies measured alcohol use at follow-up. Eight studies provided data on initiation of alcohol use amongst non-drinkers (Robinson et al., 1998; Ellickson et al., 2005; Sargent et al., 2006; Fisher et al., 2007; Henriksen et al., 2008; Hanewinkel and Sargent, 2008; McClure et al., 2008; Wills et al., 2008), three studies on maintenance and frequency of drinking amongst baseline drinkers (Robinson et al., 1998; Casswell and Zhang, 1998; Casswell et al., 2002; Ellickson et al., 2005) and six studies on alcohol use of the total sample of non-drinkers and drinkers at baseline (Connolly et al., 1994; Stacy et al., 2004; Van den Bulck and Beullens, 2005; Snyder et al., 2006; Collins et al., 2007; Pasch et al., 2007).

Study samples included random samples of youth (Snyder et al., 2006; McClure et al., 2008; Wills et al., 2008), randomly selected schools with all participants invited to participate (Stacy et al., 2004; Van den Bulck and Beullens, 2005; Sargent et al., 2006), all elementary schools in a State (Collins et al., 2007; Hanewinkel and Sargent, 2008), all middle and high schools in a city (Henriksen et al., 2008), all participants at six schools, with no information given on how the schools were selected (Robinson et al., 1998), the original sample of participants selected for participation in an RCT (Ellickson et al., 2005; Pasch et al., 2007), all participants of a longitudinal cohort study (Fisher et al., 2007) and a sub-sample of a longitudinal cohort study who had exposure and outcome data available at all follow-up periods (Connolly et al., 1994; Casswell and Zhang, 1998; Casswell et al., 2002).

Measurement of exposure and alcohol use were by self-reported questionnaires in seven studies (Robinson et al., 1998; Stacy et al., 2004; Ellickson et al., 2005; Collins et al., 2007; Henriksen et al., 2008; Hanewinkel and Sargent, 2008), by both face-to-face interview and computer interview in one (Connolly et al., 1994; Casswell and Zhang, 1998; Casswell et al., 2002), by postal survey in one (Fisher et al., 2007) and computer-aided telephone interview in three (Snyder et al., 2006; Sargent et al., 2006; McClure et al., 2008; Wills et al., 2008). One study used digital photography and GPS positioning to ascertain exposure and self-reported questionnaire for consumption data (Pasch et al., 2007).

Connolly et al. (1994) investigated the impact of the number of commercial advertisements recalled at ages 13 and 15 years on average and maximum amounts of alcohol consumed on an occasion and on the frequency of drinking and age 18 years amongst 667 participants in a multi-disciplinary longitudinal study of growth and development in New Zealand. There was no significant relationship with wine and spirit consumption. For males, the number of commercial advertisements recalled at age 15, but not 13, predicted average (P = 0.047) and maximum amounts of beer (P = 0.008) consumed on an occasion. For females, the number of commercial advertisements recalled at age 13, but not 15, predicted the frequency of beer consumption (P = 0.029). Although significant relationships were detected, they could have been due to chance, since results for more than 35 statistical tests were reported. Based on the same cohort, Casswell and Zhang (1998) followed 630 aged 18 beer drinkers until age 21 years, and found that liking of alcohol advertisements at age 18 predicted beer consumption at age 21 [standardized coefficient 0.36 (SE = 0.06, T = 6.6)]. The measure of liking of alcohol advertising was based on responses to three items: ‘alcohol advertisements have plenty of action’; ‘alcohol advertisements show the type of people I admire’; ‘Comparing alcohol adverts generally with other ads, which of the following you most agree with?’ Based on the same cohort, Casswell et al. (2002) studied 714 participants who were alcohol drinkers at ages 18, 21 and 26 years, and found that liking of alcohol advertisements at age 18 did not predict trajectories of quantities of alcohol consumed per occasion for both men and women over the age 18–26 years. Liking of alcohol advertisements at age 18 marginally predicted being in a higher trajectory for the frequency of drinking for men (OR = 1.6, P = 0.0706), but not for women over the age 18– 26 years.

Robinson et al. (1998) studied the impact of media exposure (TV, music video and videotape viewing, and computer and video game use) on initiation of alcohol use and maintenance of drinking among existing drinkers amongst 1533 14- to 15-year olds from six public high schools in California. During 18-months follow-up, 325 (36%) baseline non-drinkers initiated drinking and 322 (51%) drinkers continued drinking. Controlling for the effects of age, gender, ethnicity, and the exposure to other media, each 1-h increase per day in television viewing was associated with a 9% increased risk for initiating drinking [OR = 1.09 (1.01–1.18)]. Each 1-h increase per day in watching music videos was associated with a 31% increased risk for initiating drinking [OR = 1.31 (1.17–1.47)]. During 18-month follow-up, 322 (51%) drinkers continued drinking. There were no significant associations between media exposure and the maintenance of drinking.

Stacy et al. (2004) studied the impact of TV alcohol commercials on alcohol use amongst 2250 12- to 13-year-old school children in California. At baseline, 16% reported drinking beer, 15% wine and 8% three-drink episodes in the past month. At 12-month follow-up, the prevalence was 18% reported drinking beer, 20% wine and 12% three-drink episodes. At 1-year follow-up, each one standard deviation increase in alcohol advertising exposure as measured by the watched TV shows index was associated with a 44% increase in odds of beer drinking (95% CI: 27–61%), a 34% increase in odds of wine/liquor drinking (95% CI: 17–52%) and a 26% increase in odds of consuming three or more drinks on one occasion (95% CI: 8–48%) during the previous 30 days, controlling for covariates related to drinking behaviour. Self-reported frequency of exposure was also positively associated with beer drinking, OR = 1.21 (95% CI: 1.04–1.41), but not to wine/liquor drinking or three or more drinks on one occasion. The cued-recall memory test and draw-an-event memory test did not show significant relationships with any of the outcomes, and, although the relationships were in the direction of positive associations, there was one exception, the draw-an-event memory test being associated with a reduced risk of beer use (OR = 0.86, 95% CI: 0.75–0.99).

Van den Bulck and Beullens (2005) studied the impact of TV and music video exposure on the use of alcohol whilst going out amongst 2546 first- and fourth-year secondary school students in Flanders, Belgium. Two-thirds of students (64%) watched music videos at least several times a week, and about one-third watched daily. The quantity of alcohol consumed while going out at follow-up period related to overall TV viewing (β = 0.068, t = 3.46, P = 0.001) and music video exposure (β = 0.073, t = 3.05, P = 0.004).

Ellickson et al. (2005) studied the impact of exposure to different forms of alcohol advertising on the initiation of alcohol use and the frequency of drinking amongst existing drinkers, and whether exposure to a prevention programme mitigates any such relationship amongst US adolescents aged 12–13 years. Forty-eight percent of 1206 grade 7 non-drinkers consumed alcohol during the previous year at grade 9. Bivariate relationships found a significant impact of all types of alcohol advertisement exposure on initiation of drinking. Controlled for main confounders, including exposure to all different types of advertising and the impact of the prevention programme, exposure to beer concession stands at sports or music events predicted the drinking onset for non-drinkers in the previous 12 months (OR = 1.42, P < 0.05), whereas exposure to TV beer adverts (OR = 1.05, P > 0.05), magazines with alcohol advertisements (OR = 1.12, P > 0.05) and exposure to in-store advertisements (OR = 1.06, P > 0.05) did not. Weekly television viewing, controlled for alcohol advertisement exposure, was inversely related to the onset of drinking, explained as a ‘babysitter’ effect, whereby youth who watch more TV have fewer opportunities to drink. Seventy-seven percent of 1905 grade 7 drinkers consumed alcohol in the previous year at grade 9. Exposure to beer concession stands at sports or music events predicted the frequency of drinking amongst existing drinkers in the previous 12 months (coefficient = 0.09, P < 0.05), as did exposure to magazines with alcohol advertisements (coefficient = 0.10, P < 0.05), whereas exposure to TV beer adverts (coefficient = −0.01, P > 0.05) and exposure to in-store advertisements (coefficient = 0.02, P > 0.05) did not.

Snyder et al. (2006) studied the impact of alcohol advertising expenditures and the degree of exposure to alcohol advertisements (TV, radio, outdoor advertising and magazines) on alcohol use amongst 15- to 26-year olds in 24 Nielsen local geographical media markets (a company that tracks media exposure) in USA. Individuals were randomly sampled within households and households within media markets. Local geographical markets were systematically selected from the top 75 media markets in the US representing 79% of the population. For those aged <21 years, each additional alcohol advertisement seen increased the number of drinks consumed in the previous month by 1% (event rate ratio = 1.01, 95% CI: 1.001–1.021). Each additional dollar per capita spent on alcohol advertisements increased the number of drinks consumed in the previous month by 2.8% (event rate ratio = 1.028, 95% CI: 1.002–1.056). Seeing more or fewer advertisements in a particular month than he or she typically saw was a predictor of drinking (event rate ratio = 1.002, 95% CI: 1.001–1.003). The study has been criticized for the attrition in the study sample (from 1872 at wave one to 588 at wave four), and for confusing correlation with causality (Schultz, 2006; Smart, 2006). However, attrition was greatest among the heaviest drinking segment of the sample, suggesting under-estimation in the findings, and although the study provided associational, prospective evidence on alcohol advertising effects on youth drinking, it addressed limitations of other research, particularly the unreliability of exposure measures based on self-reporting (Snyder and Slater, 2006).

Sargent et al. (2006) conducted a randomized school-based cross-sectional survey, with longitudinal follow-up amongst 2406 non-drinkers at baseline 12–26 months later, to evaluate the impact of exposure to alcohol use in popular contemporary movies and incident alcohol drinking. Baseline median exposure to alcohol use in 601 movies was 8.6 h, [inter-quartile range (IQR) = 4.6–13.5]. Out of 2406 students, 357 (15%) initiated drinking alcohol. Exposure predicted the use of alcohol during the follow-up period (OR = 1.15, 95% CI: 1.06–1.25). The analysis with quadratic exposure effect (OR = 0.996, 95% CI: 0.992–0.999) showed that the relationship between exposure of alcohol use in motion pictures and initiation of alcohol use was stronger among adolescents in lower exposure categories.

Collins et al. (2007) carried out a school-based longitudinal survey that evaluated the impact of exposure of alcohol marketing on beer use amongst 1786 grade 6 students (11- to 12-year olds) 1 year later. Seventeen percent reported past year beer drinking at grade 7. The odds ratios (95% CI) for beer drinking were ESPN cable network (an American cable television network dedicated to broadcasting and producing sports-related programming 24 h a day) 1.08 (0.83–1.42); other sports beer ads 1.19 (1.01–1.40); other TV beer ads 1.13 (0.95–1.34); magazine reading 0.96 (0.87–1.06); radio listening 1.17 (1.00–1.37); beer concessions 1.01 (0.91–1.13); in-store beer displays 1.03 (0.92–1.14); beer promotional items 1.76 (1.23–2.52). The joint effect of exposure to advertisements from all sources: F(8, 28) = 8.36, P < 0.0001, and from three TV sources: F(3, 33) = 3.35, P < 0.05. Twenty percent of youth in the 75th percentile of alcohol marketing exposure at grade 6 reported past year beer drinking at grade 7, compared with 13% in the 25th percentile.

Fisher et al. (2007) conducted a non-random, prospective cohort study to investigate the impact of ownership of or willingness to use an alcohol promotional item on the initiation of alcohol use and binge drinking (five or more alcohol drinks over a few hours at least once over the past year). Out of 3283 girls, 611 (19%) and of 2228, 384 boys (17%) initiated alcohol use. The odds ratio of alcohol initiation during the 12-month period was 1.74 (1.37–2.19) for girls and 1.78 (1.36–2.33) for boys for those who owned or were willing to use an alcohol promotion item compared with those who did or would not. Out of 611 drinking girls, 149 (24%) and out of 384 drinking boys, 112 (29%) engaged in binge drinking. The odds ratio of binge drinking amongst drinkers was 1.79 (1.16–2.77) for girls and 0.87 (0.51–1.48) for boys for those who owned or were willing to use an alcohol promotion item compared with those who did or would not.

Pasch et al. (2007) investigated the impact of exposure of outdoor alcohol advertisements within 1500 feet (457 m) of 63 Chicago school sites of 61 schools that were part of Project Northland Chicago, a randomized controlled trial of an alcohol use prevention programme. On average, each school site had 14.8 alcohol advertisements within 1500 feet (457 m). Out of 2586, 2027 (78%) students followed up were non-users of alcohol at baseline, but initiation of alcohol use was not reported. The exposure to alcohol advertisements at sixth grade did not predict alcohol behaviour amongst sixth grade alcohol users and non-users at eighth grade, but, amongst sixth grade non-users, did predict at eighth grade intentions to use (e.g. ‘do you think you will be drinking alcohol in the next month’), f = 6.29, P = 0.01; and outcome expectancies, f = 4.62, P = 0.03.

Henriksen et al. (2008) used a non-random longitudinal survey to investigate the influence of alcohol advertising and promotions on the initiation of alcohol use amongst 1080 non-drinking students. Twenty-nine percent of never drinkers at baseline had initiated alcohol use at follow-up. Brand recognition, OR = 1.15 (1.02–1.29); brand recall, OR = 1.16 (1.05–1.29); and high receptivity to alcohol marketing, OR = 1.77 (1.27–1.48) predicted initiation. When receptivity to alcohol marketing was controlled, recall and recognition no longer statistically significantly predicted alcohol initiation.

Hanewinkel and Sargent (2008) studied the impact of exposure to alcohol use in movies on initiation of alcohol use amongst 3432 never drinking German adolescents. Estimated mean movie alcohol exposure was 3.2 h, subsequently divided into four quartiles. One-third (33%) of students initiated drinking without parental knowledge and 14% initiated binge drinking (five or more drinks within 2 h) over 12- to 13-month follow-up. Compared with quartile 1, the adjusted RRs (95% CI) for drinking without parental knowledge were 1.42 (1.16–1.74) for Q2, 1.94 (1.65–2.28) for Q3, and 2.0 (1.69–2.37) for Q4; and for binge drinking 1.44 (0.96–2.17) for Q2, 1.95 (1.27–3.0) for Q3, and 2.23 (1.48–3.37) for Q4. The un-adjusted dose–response curve showed that the response was greatest for relatively low exposure adolescents. Adjusting for covariates accentuated this effect, because the attenuation was larger for the highly exposed adolescents; this was probably due to risk factors for alcohol use tending to cluster among the high exposure adolescents who are at risk for alcohol use for reasons other than their excessive media exposure. In another study, Hanewinkel et al. (2008) found a positive dose–response relationship between lack of parental movie restriction and risk of initiation of binge drinking amongst the same sample.

Wills et al. (2008) studied the impact of exposure to alcohol use in movies on ever use of alcohol, binge drinking and alcohol-related problems amongst a random sample of 6522 US 10- to 14-year olds. A previous survey had shown that 83% of movies viewed by the sample, including 57% of movies rated as acceptable for child viewing, depicted alcohol use, with over half (52%), including one in five (19%) of child acceptable movies, containing at least one alcohol brand appearance, exposing the adolescents on average to 5.6 h of movie use and 244 alcohol brand appearances (Cin et al., 2008). In the impact study, viewed alcohol use in movies averaged 31 min at baseline, 34 min at 8-month follow-up, and 30 minutes at 16-month follow-up. Movie alcohol exposure at baseline predicted alcohol use at 8 months (coefficient = 0.1). Movie alcohol exposure between baseline and 8 months did not predict alcohol use at 8 months (coefficient = −0.03) but did predict alcohol problems at 16 months (coefficient = 0.13). Movie alcohol exposure between 8 and 16 months predicted alcohol use at 16 months (coefficient = 0.08). At all times, alcohol use predicted alcohol problems and there were significant indirect and independent effects of movie exposure at baseline, 8 and 16 months on alcohol problems at 24 months. Using the same cohort, McClure et al. (2008) studied the impact of ownership of alcohol branded merchandise (ABM) on initiation of alcohol use and binge drinking. ABM ownership increased from 11% at baseline [the 8-month measurement period reported by Wills et al. (2008)] to 20% 16 months later. Ten percent of adolescents tried drinking for the first time and 5% tried binge drinking during each of the two 8-month periods. There was a reciprocal relationship between susceptibility to alcohol use (three survey items that assessed response to peer offers, intentions and positive expectancies) and ABM ownership. The ownership of ABM at baseline did not have a significant direct impact on alcohol initiation at 8 months (HR = 1.41, 95% CI: 0.98–2.01), nor on alcohol initiation between 8 and 16 months (HR = 1.57, 95% CI: 0.99–2.5), but did on initiation of binge drinking at 8 months (HR = 1.80, 95% CI: 1.28–2.54), but not on initiation of binge drinking between 8 and 16 months (HR = 1.44, 95% CI: 0.90–2.31). New ownership of ABM at 8 months had a significant direct impact on alcohol initiation at 16 months (HR = 2.31, 95% CI: 1.6–3.35) and initiation of binge drinking at 16 months (HR = 2.22, 95% CI: 1.49–3.32).

Discussion

This review identified 13 longitudinal studies that have investigated the relationship between adolescent exposure to alcohol advertising and promotion and drinking. Twelve of the thirteen studies found evidence that such exposure predicts both the onset of drinking amongst non-drinkers and increased levels of consumption among existing drinkers. In each case, researchers controlled for key confounding variables, including family and peer drinking, and relevant demographic variables. The study that did not find an effect on behaviour examined the impact of exposure to outdoor advertising placed within 453 metres of schools (Pasch et al., 2007). This study found an impact of exposure on intentions to drink in the next month.

Seven (Robinson et al., 1998; Ellickson et al., 2005; Sargent et al., 2006; Hanewinkel and Sargent, 2008; Henriksen et al., 2008; Wills et al., 2008) of the eight studies that measured the impact of exposure on initiation of drinking included an interval or continuous level exposure measure, and all seven studies found a dose–response relationship. For example, in the study by Hanewinkel and Sargent (2008), there was a dose–response relationship between hours of movie alcohol exposure and initiation of drinking without parental knowledge and binge drinking, steeper for low hours of exposure than higher; the study by Sargent et al. (2006) found a linear association between movie exposure portraying alcohol use and onset of alcohol use from zero incidence at zero exposure to an incidence of 20% when exposure reached 11 h. Two (Robinson et al., 1998; Ellickson et al., 2005) of the three studies that measured the impact of exposure on maintenance of drinking amongst baseline drinkers included an interval level exposure measure, one of which (Ellickson et al., 2005) found a dose–response relationship with the frequency of drinking. Six of the seven studies (Connolly et al., 1994; Stacy et al., 2004; Van den Bulck and Beullens 2005; Sargent et al., 2006; Snyder et al., 2006; Pasch et al., 2007) on alcohol use of the total sample of non-drinkers and drinkers at baseline included an interval level exposure measure, and all studies found a dose–response relationship. For example, in the study by Stacy et al. (2004), each one standard deviation increase in alcohol advertising exposure was associated with a 44% increase in odds of beer drinking, a 34% increase in odds of wine/liquor drinking and a 26% increase in odds of consuming three or more drinks on one occasion during the previous 30 days; in the study by Snyder et al. (2006) of US individuals aged 15–26 years, for each additional advertisement seen, the number of drinks consumed increased by 1%, and for each additional dollar spent per capita on alcohol advertisements, the number of drinks consumed increased by 3%; in the study by Collins et al. (2007), youth in the 75th percentile of alcohol marketing exposure at grade 6 were 50% more likely to be drinking at grade 7 than youth in the 25th percentile; finally, in the study by Pasch et al. (2007), the greater the exposure to outdoor advertising near schools, the greater the intention to drink (although, this study found no impact on drinking behaviour, possibly due to a lack of statistical power). It is clear, therefore, that longitudinal studies demonstrate that alcohol advertising, amongst other factors, encourages youth drinking.

As explained in the introduction, this review focused on longitudinal studies because the dimension of time makes them a particularly powerful way of untangling cause and effect. Nonetheless, cross-sectional studies, although only providing a snapshot of advertising exposure and levels of drinking, have consistently reported correlations between increased exposure and greater likelihood of current drinking (see Kuo et al., 2003; McClure et al., 2006; Hanewinkel et al., 2007; Hurtz et al., 2007). For example, the cross-sectional study of Hanewinkel et al. (2007) found a dose–response relationship between exposure to alcohol use from popular contemporary movies and alcohol use without parental knowledge and binge drinking in Germany. Therefore, despite their inherently weaker design, cross-sectional studies do corroborate the effects found by longitudinal studies. Advertising influences youth drinking.

Furthermore, both cross-sectional and longitudinal studies are likely to underestimate any effects, because they focus principally on advertising, which is only a part of the promotional effort that is put behind alcohol products. As noted in the introduction, for instance, the most recent estimate for expenditure on alcohol advertising in the UK is actually only a quarter of that for alcohol promotion as a whole. While some of the selected studies looked at promotion (e.g. merchandising) as well as advertising, none looked at the cumulative impact that a coherent and fully fledged ‘marketing communications mix’ (Kotler et al., 2005) may have. This communication effort is, in turn, only part of a company's marketing strategy that also includes price promotions, packaging, distribution and product design.

Three limitations should be considered in interpreting the results of this review. First, it included 13 studies that are comparatively heterogeneous. We controlled for quality by including only longitudinal studies that followed a cohort of individuals. We did not attempt to quantify the quality of other study characteristics. One of the limitations of observational studies is the relationship between the variables of interest and other confounding factors. Whilst all the studies to some extent, measured and controlled for other variables likely to be associated with drinking uptake, it is impossible to know if all relevant variables were measured and adjusted for, and thus not possible to know if residual confounder influenced the analysis.

Second, there is a possibility that publication bias may have affected the studies identified for inclusion. Other cohort studies that examined the relationship between advertising exposure and youth drinking but found no association may not have been published or may have been published with no reference to advertising and so would not be retrieved by our search strategy.

Third, the way in which exposure to advertising was operationalized varied across studies (e.g. receptivity, influence and awareness). An important methodological challenge in evaluating evidence on the effect of advertising on drinking behaviour of adolescents is to achieve standardization and consistency in measuring of exposure to alcohol advertising (for an example of standardization, see Jernigan and Ross (2007)).

Conclusion

This review found consistent evidence to link alcohol advertising with the uptake of drinking among non-drinking young people, and increased consumption among their drinking peers. This evidence comes from high quality longitudinal studies and is corroborated by weaker cross-sectional ones. Because it focuses on mass media advertising, it almost certainly underestimates the impact of wider alcohol promotion and marketing. These findings are not surprising: exactly the same conclusions have emerged from reviews of the impact of tobacco (Lovato et al., 2003) and food (Hastings et al., 2003) marketing on young people.

References

View Abstract