Article Text

Linking mass media campaigns to pictorial warning labels on cigarette packages: a cross-sectional study to evaluate effects among Mexican smokers
  1. James F Thrasher1,2,
  2. Nandita Murukutla3,
  3. Rosaura Pérez-Hernández2,
  4. Jorge Alday3,
  5. Edna Arillo-Santillán2,
  6. Claudia Cedillo3,
  7. Juan Pablo Gutierrez4
  1. 1Department of Health Promotion, Education and Behavior, Arnold School of Public Health, University of South Carolina, Columbia, South Carolina, USA
  2. 2Department of Tobacco Research, Center for Population Health Research, National Institute of Public Health, Cuernavaca, Mexico
  3. 3World Lung Foundation, New York, New York, USA
  4. 4Center for Survey Research, National Institute of Public Health, Cuernavaca, Mexico
  1. Correspondence to Dr James F Thrasher, Department of Health Promotion, Education and Behavior, University of South Carolina, Columbia, SC 29208, USA; thrasher{at}


Objective This study assessed the effects of pictorial health warning labels (HWLs) and a linked media campaign in Mexico.

Methods Cross-sectional data were collected from a population-based sample of 1756 adult smokers, aged 18–55 years, during the initial implementation of pictorial HWLs, which some smokers had seen on cigarette packages while others had seen only the text-based HWLs. Exposure to the campaign and pictorial HWLs was assessed with aided recall methods, and other questions addressed attention and cognitive impact of HWLs, knowledge related to HWL and campaign content, and quit-related thoughts and behaviours. Logistic and linear regression models were estimated to determine associations between key outcomes and intervention exposure.

Results In bivariate and multivariate adjusted models, recall of pictorial HWLs and of the campaign were positively associated with greater attention to and cognitive impact of HWLs, whereas only pictorial HWL exposure was associated with having refrained from smoking due to HWLs. Both recall of pictorial HWLs and of the campaign were independently associated with greater knowledge of secondhand smoke harms and toxic tobacco constituents. Smokers who recalled only the pictorial HWLs were more likely to try to quit than smokers who recalled neither the pictorial HWLs nor the campaign (17% vs 6%, p<0.001).

Conclusions Consistent with other studies, adult smokers' exposure to new pictorial HWLs in Mexico was associated with psychosocial and behavioural responses related to quit behaviour. Exposure to the complementary media campaign was associated with independent additive effects on campaign-related knowledge, and it enhanced psychosocial responses to pictorial HWLs.

  • Tobacco
  • labelling
  • communication
  • media campaign
  • public policy
  • health communication
  • media campaigns
  • qualitative study
  • public opinion polls
  • advertising and promotion
  • low/middle-income country
  • media
  • social marketing
  • global health
  • advocacy

Statistics from

Request Permissions

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


The WHO Framework Convention on Tobacco Control obliges ratifying parties to inform their populations of tobacco-related health risks and to implement policies that discourage tobacco consumption.1 To help achieve these goals, countries have increasingly adopted prominent pictorial health warning labels (HWLs) on cigarette packaging, with 41 countries implementing the policy by the end of 2011.2 HWLs on cigarette packaging that combine pictures and text are more effective than text-only HWLs in engaging smokers, increasing knowledge about smoking risks and promoting cessation-related behaviour.3 Mass media campaigns can also produce these outcomes4–6 and enhance tobacco control policy implementation.7 ,8 Basic principles around optimising ‘marketing mix’ indicate that messaging and imagery on product packaging that is linked to marketing content delivered through other channels can potentiate desired effects.9 In this regard, electronic media campaigns may bring greater narrative depth to static pictorial HWLs, making the outcomes they depict more meaningful and emotionally engaging.10 Research is needed to determine the effects of these interventions and the interactions between them to help countries develop strategies that maximise impacts.

Two studies in Australia have examined media campaigns that complement pictorial HWLs. Repeated cross-sectional surveys of adolescents11 and adult smokers10 before and after the introduction of pictorial HWLs indicated that TV spots linked to pictorial HWLs content increased knowledge addressed in the HWLs and TV spots. Furthermore, adult smokers with prior exposure to pictorial HWLs reported stronger responses to TV spots and greater quit intentions than smokers previously unexposed to the pictorial HWLs to which advertisement were linked.10 Research is needed to determine the effects of complementary media campaigns in low- and middle-income countries, as such campaigns may be particularly cost-effective in these countries due to their novelty and to their ability to capitalise on linkage with HWL messaging and imagery. Because media campaigns can be costly, data on the effects of different campaign strategies should support decision-making around the most effective allocation of resources to reduce tobacco consumption.

Mexican context

In 2008, Mexico passed a law that included adoption of pictorial HWLs that are currently the third largest in the world (ie, 65% of package surface area, compared with 80% in Uruguay and 75% in Canada). Previously, HWLs occupied 50% of the back of the package and included four rotating messages with only textual content. Under the new regulations, the Ministry of Health selects eight new warnings each year and specifies which pair of HWLs the industry must print every 3 months. This rotation frequency is the fastest rotation of novel HWL content in the world. The first two pictorial HWLs focused on the dangers of secondhand smoke (SHS) exposure to children and on premature birth and spontaneous abortion (see figure 1). Content in both pictorial HWLs also called attention to chemicals in tobacco smoke. On 25 September 2010, the implementation period began, and these two pictorial HWLs gradually saturated the market over subsequent months, as has occurred in other countries.12 Because the industry does not have to withdraw HWLs that they print in each quarter, more than two warnings can be found in the market at any particular time after initial implementation.

Figure 1

First two pictorial health warnings in Mexico, which entered the market during the study period. Images presented with permission of the Ministry of Health, Mexico. HWL, health warning label.

A complementary media campaign was aired from 30 September 2010 through 31 October 2010 in the state of Jalisco, Mexico, with most of the media buy in the state capital of Guadalajara. One TV and one radio spot were adapted from a TV spot on SHS effects on children, for which a different execution from the same campaign had tested well in Mexico and other low- and middle-income countries.13 The TV spot showed images of children suffering a variety of health effects and described the toxic chemicals in tobacco smoke. Pictures from the first two pictorial HWLs were interspersed between other still photos, but no explicit reference was made to the new warnings.14 The radio spot similarly described toxic constituents of SHS, harms to children, and encouraged smokers to heed warnings on cigarette packs. Both spots concluded with a call to action (Quit smoking today), but only the TV spot showed the quitline number (see online appendix 1 for transcripts). The media buy was limited and achieved 287 gross rating points (GRPs) for the one TV spot and 547 GRPs for the one radio spot.

The present study aimed to assess the main and interactive effects of pictorial HWLs and a linked media campaign by analysing a cross-sectional survey of adult smokers during the initial implementation period of pictorial HWLs. Because pictorial HWLs slowly saturated the market, smokers who had and had not purchased cigarettes with the new pictorial HWLs were compared, as were those who did and did not recall seeing the campaign.



Cross-sectional data were collected from a population-based representative sample of 1765 adult smokers who lived within the metropolitan area of Guadalajara, Mexico, where the media campaign was aired. The protocol was approved by the ethics review board of the Mexican National Institute of Public Health. Using 2005 census data, a stratified, multistage sampling strategy was used, wherein 86 census tracts with at least 80 inhabitants were randomly selected, followed by the random selection of three block groups within each tract. Each stage involved selection probabilities proportional to the number of households within the unit of selection. Households within each selected block group were then listed, and seven households were systematically sampled and visited up to four times on different days and times of the week in order to establish contact and enumerate household members to determine eligibility. Eligibility criteria included current smoker status (ie, smoked at least once in the previous week and 100 cigarettes in lifetime) and aged 18–55 years. Media ownership was not included as a selection criterion due to high TV ownership (ie, 98% of urban Mexicans have a TV in their household and 85% have a radio).15 Up to one male and one female participant were allowed per household. If seven smokers were not found in selected households, another seven households from the same block group were systematically sampled until seven smokers from each block group were interviewed. Data were collected from 17 October 2010 through 30 November 2012, a period which began approximately 2 weeks after campaign launch (September 30) and 3 weeks after the pictorial HWLs began to appear (September 25).


Dependent variables


Questions about the dangers of SHS included both general health effects not addressed in the campaign (ie, lung cancer, respiratory diseases other than cancer, heart disease, general danger for non-smokers) and health effects among children that were addressed in the campaign (ie, lung diseases, other respiratory illness, ear infections, asthma). For each of these domains, an index was created, summing the number of correct endorsements (range 0–4). An index was also created to reflect knowledge of specific chemical constituents in tobacco smoke mentioned in the campaign or on pictorial HWLs (ie, talium, cyanide, carbon monoxide). Finally, participants were asked if they had heard of a toll-free quitline, as the quitline was mentioned both in the campaign and in the pictorial HWLs.

HWL reactions

Participants' responses to HWLs involved questions validated in Mexico16 ,17 and other countries.18–20 Attention to HWLs was assessed by asking about the frequency of noticing and the frequency of reading or looking closely at HWLs in the previous month (never, sometime, often, very often). Due to the high correlation between these indicators of attention (r=0.53), they were summed to create a single variable, as in prior research.20 Cognitive responses to HWLs were assessed with two questions that assessed the extent that HWLs made the smoker think about health risks and about quitting (not at all, a little, somewhat, a lot). These questions were highly correlated (r=0.64) and summed to create a single variable (range =0–6). Finally, we asked if pictorial HWLs had caused the smoker to refrain from smoking in the previous month, with responses coded as yes or no.

Quit behaviour

A variable was derived to indicate having tried to quit during the period after the introduction of pictorial HWLs. Participants were asked if they had tried to quit in the previous year, and if so, when that quit attempt began. We created a dummy variable to indicate whether the quit attempt start date was after pictorial HWL implementation or not.

Independent variables

Campaign and pictorial HWL recall

Validated, aided recall techniques, used previously in Mexico8 and the USA,21 were used to estimate exposure. Participants were shown each of the two pictorial HWLs (figure 1), and for each picture, they were asked whether they had seen it on cigarette packs they had bought. To assess campaign recall, participants were shown a screen shot from the TV spot and asked if they had seen it. A short description of the radio spot was provided and participants were asked if they had heard it. Participants were then classified into four exposure groups: (1) no recall of exposure; (2) recall of either pictorial HWL, with no recall of campaign materials; (3) recall of the campaign through either channel, but no recall of pictorial HWLs; (4) recall of either pictorial HWL and of either campaign material. Dummy variables were created, with no recall of exposure as the reference group.

Smoking intentions and behaviour

Participants indicated whether they were daily smokers or not, and the amount they regularly smoked each day or, in the case of non-daily smokers, the amount they smoked each week. Smokers were classified as non-daily or at one of three levels of intensity of daily smoking (ie, 0–4, 5–9 and 10 or more cigarettes a day), which generally reflected a uniform distribution. Smokers were also asked their intention to quit (ie, intend to quit in the next month, in the next 6 months, sometime after the next 6 months, no intention to quit), which was recoded to indicate whether they intended to quit in the next 6 months or not.


Participants reported their age, sex, educational attainment (primary or less; secondary; technical school; high school; college or more), monthly household income and marital status. Age and monthly income were re-categorised to have uniform sample distributions across categories. Dummy variables were created for each socio-demographic variable. People with missing income data were classified using a dummy variable so that they were not excluded from analyses that included income.


Data were analysed using STATA, V.11. Except for the sample description, all analyses adjusted for the sampling design and sampling weights, which accounted for the likelihood of participant selection, as well as adjustment to reflect sex differences in smoking prevalence in the Guadalajara metropolitan area, according to the 2008 National Survey on Addictions.22 χ2 Tests and unadjusted and adjusted logistic regression models were estimated when examining dichotomous outcomes. For continuous dependent variables, t tests and unadjusted and adjusted linear regression models were estimated. Adjusted models accounted for socio-demographics, smoking intensity and intention to quit, as well as recall of both the campaign and the pictorial HWLs.


Sample description

The household contact rate was 81% (4212 contacted/5215 approached). Of contacted households, 20% (n=847) refused to participate, 12% (n=494) indicated that no smokers lived in the household and 68% (n=2871) permitted household enumeration. Of the 3459 eligible smokers identified through enumeration, 52% (n=1765) were successfully interviewed. Table 1 shows the sample characteristics.

Table 1

Characteristics of sample of adult smokers, Guadalajara, 2010

Pictorial HWL and campaign recall

Fifty-eight per cent of participants reported seeing at least one pictorial HWL on packages they purchased. Smokers who were younger, more educated and smoked more frequently were more likely to recall pictorial HWLs in both crude and adjusted models (see table 2). The campaign TV and radio spots were recalled by 24% and 12% of the participants, respectively, with 4% recalling both spots and 32% recalling at least one spot. Lower education, moderate income compared with lowest income and higher smoking intensity were all associated with a greater likelihood of recalling the campaign in both crude and adjusted models (see table 3). With regard to the amount of intervention overlap, 28% recalled neither the pictorial HWLs nor the campaign, 40% recalled just the pictorial HWLs, 13% recalled just to the campaign and 19% recalled both.

Table 2

Factors associated with recall of exposure to pictorial health warning labels

Table 3

Factors associated with recall of exposure to campaign

Psychological responses to HWLs in general

All participants were asked about their responses to HWLs in general, without specifying HWL type as pictorial or text only. Bivariate and adjusted linear regression models indicated that recall of the pictorial HWLs only and recall of pictorial HWLs and the campaign were positively associated with attention towards HWLs and with quit-related cognitive responses to HWLs, with a stronger association found for recall of both interventions (see table 4). Although campaign recall did not have a statistically significant independent association with either of these outcomes, examination of the point estimates for these associations, alongside those found for pictorial HWLs only and the combined intervention, is suggestive of an additive effect.

Table 4

Bivariate and adjusted associations between continuous outcomes and recall of pictorial health warning labels (HWLs) and the media campaign

Knowledge related to pictorial HWL and campaign content

Knowledge of SHS was examined with two indexes, one that assessed knowledge of general SHS dangers and the other of SHS dangers specific to children. Recall of pictorial HWLs only and in combination with the campaign were associated with both kinds of SHS knowledge, in both bivariate and adjusted models (see table 4). Recall of the campaign only was associated with child-specific SHS knowledge but not with more general SHS knowledge. Recall of pictorial HWLs only, of the campaign only, and of both interventions were all independently associated with higher levels of knowledge of the toxic constituents of tobacco smoke, showing evidence of an additive effect in the group that recalled both interventions (table 4). Finally, 32% of smokers were aware of the toll-free quitline, with greater awareness in each of the intervention exposure groups compared with the unexposed group (no recall=20%; pictorial HWLs only =33%; campaign only=36%; pictorial HWLs + campaign=45%). This greater likelihood of quitline awareness was statistically significant in bivariate and adjusted logistic regression models, with results indicating that each intervention had an additive effect (see table 5).

Table 5

Bivariate and adjusted ORs for dichotomous outcomes and recall of pictorial health warning labels (HWLs) and the media campaign

Quit-related behaviours

Twenty per cent of participants indicated that HWLs, independent of type, had caused them to refrain from smoking at the moment of lighting a cigarette during the previous month. Compared with those who had no recall of either intervention, report of refraining from smoking because of HWLs was higher across all exposure groups (no recall=11%; pictorial HWLs only=23%; campaign only =22%; pictorial HWLs + campaign=24%), and these differences were statistically significant in bivariate and adjusted models (table 5). Eleven per cent of participants initiated a quit attempt during the campaign period, with the highest rates found among those who recalled only the pictorial HWLs (17%), and the difference between this group and the unexposed group (6%) was statistically significant in bivariate and adjusted models. Nine per cent of those who recalled of the campaign and of those who recalled both the pictorial HWLs and the campaign tried to quit, but this percentage was not significantly greater than those who did not recall either intervention (table 5).


Study results provide evidence in favour of the efficacy of both pictorial HWLs and of complementary campaigns, which is consistent with other studies on this topic.10 ,11 A higher percentage of participants recalled having purchased packages with pictorial HWLs (58%) than having seen or heard the campaign (32%). The novelty of pictorial HWLs and the potentially high number of exposures that accompany cigarette package HWLs may partly explain this finding.23 The lower level of recall found for the campaign is in line with expectations, given the campaign's relatively short duration and its limited media buy, which did not meet recommended levels (390 GRPs)24 for TV. The radio media buy was higher (547 GRPs), although radio is generally considered less effective than TV. Recall of either intervention was associated with more frequent smoking and with middle compared with low household income. However, educational attainment had opposite effects across these exposures. People with lower compared with higher educational attainment were more likely to recall the campaign. This is potentially important since dramatic and evocative campaigns such as this one have previously been shown to produce greater impacts among populations with lower than with higher educational attainment.25 On the other hand, more well-educated adult smokers were more likely to recall pictorial HWLs. This may reflect where smokers from different SES groups purchase their cigarettes, as smokers with higher educational attainment may be more likely to buy cigarettes from formal points of sale (eg, convenience store chains, supermarkets, gas stations). Formal points of sale likely have higher sales volume and turnover of supply, which enables more rapid saturation with newly printed pictorial HWLs than would be found in informal neighbourhood stores run out of private homes, where many Mexican smokers purchase their cigarettes.26 As pictorial HWLs come to saturate the market, all smokers will be exposed to these messages, but future campaigns that are timed to coincide with the implementation of new pictorial HWL content may need to develop media plans that consider how newly printed HWLs appear to reach some populations before others. In our study, the campaign and pictorial HWLs reached slightly different groups of smokers, thus complementing one another, but also reducing the possibility of synergies in some subpopulations.

Study results suggest that both pictorial HWLs and the media campaign enhanced attention and cognitive reactions to HWLs. The former finding is consistent with other research on pictorial HWLs compared with text-only HWLs.16–19 Recall of the media campaign only had an independent statistically significant effect on HWL reactions among smokers who also recalled the pictorial HWLs. This may be due to the limited media buy of the campaign and its link to pictorial HWL content but not the textual HWL content. It is also suggestive of the potential for augmenting pictorial HWL impacts with larger media buys.

Knowledge specific to pictorial HWLs and to the campaign was higher among those who recalled exposure to these interventions. This is important, as informed consumption is a key rationale for pictorial HWLs and mass media public education efforts.27 Furthermore, for the campaign alone, recall was only associated with significantly better SHS knowledge about topics that the campaign addressed. This domain-specific effect suggests that exposure to the campaign was indeed responsible for the improved knowledge. Finally, recall of both the campaign and pictorial HWLs were independently associated with awareness of the toll-free quitline. Media campaigns28–30 and pictorial HWLs31–33 can increase quitline call volume, and countries that implement these interventions with ‘calls to action’ (ie, call the quitline) will need to invest in the quitline infrastructure to respond to an increased volume of calls to the quitline. Since HWLs and mass media campaigns that include such calls to action are clearly effective at generating knowledge of and calls to the quitline, countries do not need to run specific campaigns to generate awareness of the quitline but can instead rely on taglines and call to actions in more general cessation campaigns and HWLs.

Recall of pictorial HWLs was associated with a higher prevalence of quit-related behaviours, which is consistent with other studies,16–19 although ours is the first study of which we are aware that shows an association that is independent of quit intentions. The lack of statistically significant associations between quit attempts and campaign recall may be due to the small media buy, its primary content (ie, child SHS exposure, not cessation) and lack of statistical power due to the small sample of smokers exposed to the campaign. The finding that smokers who recalled the pictorial HWLs and the campaign were no more likely to try to quit than those who recalled neither intervention is more perplexing. Smokers who recalled both interventions may have had lower levels of exposure to pictorial HWLs than smokers who recalled only pictorial HWLs and not the campaign. However, we did not assess exposure intensity. Because new HWLs slowly saturate the market over a period of 6–8 months, our 6-week data collection period was likely too short to allow for meaningful examination of the moment of data collection as a proxy for exposure intensity. Moreover, pictorial HWLs may have their greatest impact on quit behaviour in the initial implementation period when HWL content is still novel. HWL effects wear out,34 but the point at which the effects wear out is unclear, and more research is needed to determine which HWL content is most likely to impede wear out. Although we were unable to address directly these issues, our results are consistent with a recent longitudinal study in Mexico, which found that recent ex-smokers interviewed after pictorial HWL implementation were twice as likely to indicate that HWLs motivated them to quit than recent ex-smokers interviewed before policy implementation (13% vs 26%, respectively).35

Recall of pictorial HWLs and the campaign were associated with additive effects on key psychosocial responses. Coefficients associated with recall of both interventions did not show evidence of multiplicative or synergistic effects that go beyond additive effects. A larger media buy may be necessary to produce such effects, but so might a stronger link between the pictorial HWL content and campaign content. General topical content on toxic tobacco constituents and childhood diseases caused by SHS was shared, but the pictorial HWLs and the campaign did not use the same language and did not visually highlight the connection. The TV spot used both of the images from the first pair of pictorial HWLs, but these were interspersed among other images and did not call attention to the HWLs. The radio spot called for smokers to pay attention to their cigarette packs, but no visual linkage was possible. Similar conclusions regarding the additive but not synergistic effects of these intervention strategies were found in Australia, even though campaign materials were more directly linked to pictorial HWL content.10 While there appear to be benefits to having simultaneous interventions, more research is needed to determine how to maximise their population impact.

Study conclusions are limited by a number of factors, including the cross-sectional nature of the survey. We divided the study population into exposure groups using aided recall methods that have been validated previously8 ,36 including among urban populations similar to the study sample.8 Recall may nonetheless be a biased measure of exposure. We found some socio-demographic differences between exposure groups, although statistical adjustment for these differences did not significantly affect the point estimates or statistical significance of associations between intervention exposure and key outcomes. Hence, exposure assessment appears unlikely to be biased by these factors, although other factors may have biased the results, including unmeasured variables that might account for the associations found here. Since we were unable to observe temporal changes in the outcomes due to exposure, our results are not conclusive regarding the causal impact of the campaign or the pictorial HWLs. Self-report of exposure may have been biased by pre-existing differences in the outcomes of interest, as people who had higher levels of knowledge or quit behaviour may have been more likely to remember exposures, even if their actual exposures were no different from people who did not recall exposures. Nevertheless, our results are consistent with some recent longitudinal research among Mexican smokers that shows similar increases in attention and responses to HWLs, including HWL-specific knowledge and quit behaviour.35

The lower than desired response rates may compromise the generalisability of our results. Nevertheless, educational attainment, household income and civil status in our sample were similar to other population-based studies of urban smokers in Mexico that have better response rates.22 ,37 Intention to quit (33%) was also similar to other studies.38–40 However, participants in our sample indicated higher smoking intensity than found in previous surveys, such as the 2009 Global Adult Tobacco Survey, which found that 50% of Mexican smokers do not smoke daily.38 Only 22% of our sample smoked less than daily, and among those who smoked daily, the amount smoked each day was higher than in other studies. The higher level of addiction in our sample compared with the general population could have resulted in underestimates for the associations found, given that more addicted smokers may be less responsive to tobacco control interventions.

This study provides the first evaluation of pictorial HWLs linked to a media campaign outside of Australia. Results were consistent with previous studies, suggesting the greater impact of pictorial over text-only HWLs. The media campaign was associated with additive effects related to higher campaign-related knowledge and stronger psychosocial responses to pictorial HWLs. It may be necessary to have tighter linkage between the messages and imagery of these two interventions in order to produce synergies that result in greater effects on knowledge and behaviour.

What this paper adds

  • Previous research has shown that pictorial HWLs on cigarette packages and media campaigns promote quit-related psychosocial and behavioural responses among adult smokers.

  • However, few studies have examined the potentially synergistic effects of media campaigns that are linked to pictorial HWLs.

  • Results were consistent with other studies that have found greater knowledge and quit-related psychosocial responses and behaviours among smokers exposed to pictorial HWLs compared with those exposed to text-only HWLs.

  • The media campaign was associated with additive effects related to higher campaign-related knowledge and stronger psychosocial responses to pictorial HWLs.

  • Greater linkage between the messaging and imagery of media campaigns may be necessary to promote more synergistic effects on knowledge and behaviour.


Supplementary materials

  • Supplementary Data

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

    Files in this Data Supplement:


  • Funding This work was supported by the Bloomberg Global Initiative to Reduce Tobacco Use through a grant from the Union against Tuberculosis and Lung Disease, grant number Mexico 2-02, and by the World Lung Foundation. JT's time was partially funded by the US National Cancer Institute, grant number P01 CA138389.

  • Competing interests None.

  • Ethics approval Ethics Committee, National Institute of Public Health, Mexico.

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

  • Data sharing statement No data from this study are currently available for public sharing; however, researchers interested in analysing the data should contact JT or NM.