Theoretical background of stigmatization

Stigmatization depends on access to social, economic, and/or political power [7]. It is worth noting that the early development of non-smokers' advocacy groups coincided with an increase in the concentration of smoking within populations of lower socioeconomic status (SES). As a result, it may have become easier to stigmatize smoking as an undesirable and deviant behaviour [3, 14, 15]. The rise in the stigmatization of smokers and the increasing concentration of non-smokers in groups of higher socioeconomic status may not, therefore, be independent of one another.

As smoking bans have become more common, the social distance and physical separation between smokers and non-smokers has increased. This reduces their opportunities to interact socially and their likelihood of doing so. In some ways, this separation has been entrenched by TCPs, which have pushed smoking, and therefore smokers, out of public spaces, increasing their social exclusion and their sense of being outsiders [16–18]. That separation may also occur in adolescence. By promoting early experience of the stigmatization of smoking at school, STPs may also lead to a ‘conscious and deliberate act’ of separating smokers and non-smokers [19]. STPs denormalize smoking by reducing the visibility of the behaviour, thereby marginalizing smokers [20].

The stereotyping of smokers may also be a consequence of the denormalization of smoking. Stuber et al. showed that 38% of smokers who were interviewed reported feeling that, overall, people ‘think less’ of someone who smokes [21]. This negative image is associated with undesirable characteristics that reduce a person's status in the eyes of the people who stigmatize them [6, 7, 21–23]. Such stereotypes may also apply to adolescent smokers [24–26]. Given that TCPs lead to the denormalization of smoking, stereotyping and discrimination are inevitable corollaries of prevention [27]. To date, however, very little literature has examined the association between STPs and stigmatization in adolescents.

Discrimination, another consequence of the denormalization of smoking, implies a loss of social status [7, 27]. Research shows that some adult smokers report discriminatory treatment by healthcare services, such as being placed lower on waiting lists [3], or in the labour market [5]. Because of the enforcement of STPs, some adolescents may experience discrimination at school due to their smoking status. There is, however, little literature on the stigmatization of adolescent smokers.

Smoking bans and smoke-free culture shift the balance of power towards (young) non-smokers. Unlike most adults, however, adolescents are engaged in building their own identities; they may pay more attention to how others see their behaviour and they may be more receptive to a negative smoking climate [28]. Adolescents who smoke may therefore experience a higher degree of stigmatization than adults [29, 30].

In summary, as smoking becomes increasingly denormalized, smokers may be vulnerable to stigmatization. There is, however, very little quantitative evidence of such a link, and very few studies that compare adolescents in different contexts of smoking regulation. Given that the overall denormalization strategy requires the participation of schools, it is important to know whether stigmatization applies to young smokers at school.

In seven cities with tobacco control policies of varying degrees of strictness, we aimed to measure [1] whether European adolescents perceive a stigmatization of smokers and whether that stigmatization varies according to smoking status and socioeconomic position (SES), and [2] whether stigmatization is greater in school environments in which smoking is denormalized (i.e. those with low smoking rates and strong school tobacco policies).

Measurements

The main outcome variable, overall perceived stigmatization of smokers—in other words, the broad perception of a socially imposed stigma upon smokers—was captured using a standardized survey (see S2 File) and was measured using a short version of Stuber’s scale of stigmatization (devaluation index). It is made up of four items. Two concern stereotyping: (a) most people think less of people who smoke; and (b) most people believe that smoking is for losers. The other two concern discrimination: (c) most non-smokers would be reluctant to date someone who smokes; and (d) most non-smokers would not hire a smoker to babysit their children [32]. The replies (from strongly agree to strongly disagree) were rated from 0 to 3. An average stigma score was calculated by adding together all scores for the four Likert items and dividing by four (mean = 1.62, std = 0.55). The score was then log-transformed to reduce skewness and facilitate parametric tests such as linear regression.

To control for the association between behaviour and stigmatization, we categorized the smoking variables of each stage of smoking status, from initiation to daily use, into six groups: non-smokers, people who had tried smoking at least once, occasional smokers (experimenters), weekly smokers, daily smokers, and ex-smokers [31]. This categorization helped to show how different experiences of smoking are associated with reaction to stigma.

We then identified three smoking environment variables in which power may be shifted between non-smokers and smokers. The first was school tobacco policies, which operated across three dimensions: comprehensiveness (does the STP apply everywhere and to everyone?), enforcement (was the STP actually enforced, and were sanctions associated with breaking the rules?), and communication (was the STP formally communicated by schools and, if so, by what means?). To quantify this variable, the perspectives of adolescents and members of school staff were used to create a school tobacco policy score for each school. In this study, between 24 and 44 school staff members participated in each city. We administered the questionnaire to at least three staff members per school, including, where possible, someone from the management, someone from the teaching staff, and someone in charge of health and prevention matters. Ranging from 0 = ‘low’ to 10 = ‘high’, this score indicated the extent to which the policies were perceived as being comprehensive and enforced, and properly communicated. We have provided, in S2 Table, a detailed list of all items of this STP score. Overall, staff and student perceptions of STPs were significantly correlated (r = 0.46, p = 0.0034). Their scores were also significantly correlated for policy comprehensiveness (r = 0.55, p = 0.0003) and for policy enforcement (r = 0.51, p = 0.0010) [33]. The second variable was the proportion of weekly smokers in each school. The third was weekly smoking among peers, which is a potential indicator of the support a smoker can receive in his/her network to continue the behaviour.

As the SILNE-R study was a social network study, all students were asked to nominate up to five schoolmates they considered to be their best and closest friends. For each adolescent, we computed the proportion of smokers in the first out-degree separation set of friends [34].

As explained above, stigmatization depends on the distribution of economic and political power. We therefore identified two individual socioeconomic position variables that might shift power among adolescents at school, and that might also predict future socioeconomic status (SES). On the basis of each country’s education system, parents’ level of education was classified as low, medium, high, or unknown. Adolescents’ academic performance (low vs high) was based on the students’ self-reported marks from the previous year: ‘Which of the following best describes your marks last year?’. This variable included five initial values based on each country’s academic performance assessment system. Those values were dichotomized into two categories: low and high.

Confounders

As TCPs differ between countries in the European Union, we used the country as a confounder. Countries were ranked according to the Tobacco Control Scale (TCS) developed by Joossens and Raw, ranging from high scores (Ireland-70 and Finland-60) to medium (the Netherlands-53, Italy-51, and Portugal-50) and low scores (Belgium-49 and Germany-37). The TCS includes factors such as taxes, restrictions on smoking in public places, consumer information (e.g. public information campaigns, media coverage, publicizing research findings), comprehensive bans on the advertising of all tobacco products, labels with health warnings on tobacco products, and treatment to help dependent smokers to quit. The scale allocates points to each policy, with a maximum score of 100: prices 30, smoke-free public places 22, spending on public information campaigns 15, comprehensive advertising bans 13, large health warnings 10, cessation support (treatment) 10 [35, 36]. However, the TCS score does not take school smoking bans into account. That is why we created a school tobacco policy score for each school for our analyses (see S2 Table) [33]. We also controlled for age and gender.

Data analysis

After excluding 279 observations due to missing data, we first tabulated the stigmatization items, which were dichotomized into two categories: agree and disagree, according to the adolescents’ smoking status and other covariates. A chi-2 test was included in this analysis. To quantify the association between smoking status, individual socioeconomic position variables, smoking-environment variables, and stigmatization, we then performed multilevel linear regressions, using the school as a random intercept. Due to the hierarchical structure of the data, observations were not independent. We therefore used multilevel models. The school was included in our models as a random intercept to take into account the correlation between different observations for a single school [37].

For linear regression, we used the log-transformed stigmatization score. We also categorized smoking status into three different groups: non-smokers, including ex-smokers and those who had never smoked, occasional smokers, including those who had tried smoking at least once and experimenters, and weekly smokers, including daily and weekly smokers. In Model 1, using the school as a random variable, we regressed the score of the stigma for each variable, and controlled for gender, country, and age. In Model 2, we added two individual socioeconomic position variables (father’s and mother’s education and academic performance) and three smoking environment variables (the percentage of adolescents who smoked weekly per school, the proportion of weekly smokers among peers, and the school tobacco policy (STPs) scores. We stratified the previous analysis by smoking status.

Finally, stigmatization is likely to follow a different pattern in a school with very few smokers than in a school where smoking is rife. Indeed, as shown elsewhere [38], the gradient of smoking across academic performance is much more pronounced in schools with a higher SES than in those with a lower SES. We therefore tested the interaction of Models 1–2 with the school type, using the national classifications of schools’ SES [31]. Statistical analyses were carried out using SAS 9.3.

Main findings

In a unique international survey involving seven European cities, we examined how perceived stigmatization of smokers among adolescents was affected by their smoking environment and the power imbalance between smokers and non-smokers. We found conclusive evidence that smokers in the adolescent population at large were aware of the negative effects of stigmatization, stereotyping, and discrimination. Students who smoked reported less overall perceived stigmatization of smokers than non-smokers did. Perceived stigmatization of smokers was lower among those who were more exposed to smoking among their friends and among those with poorer academic performance. We found no indication that stronger STPs were associated with more stigmatization.

Consistency and interpretation

Our study found that weekly smokers expressed the lowest perceived stigmatization of smoking and that non-smokers expressed the highest. One possible explanation is that adolescents for whom smoking is a key component of their identity may actively resist stigmatization by denying their identity as a smoker. A growing number of experts who warn public health authorities against the use of stigmatization as a preventative tool have highlighted two negative consequences of stigmatization: social isolation from non-smokers and concealment of smoking status [39, 40]. In a qualitative study of adults who engaged in ‘social smoking,’ Whitesel and Shuman showed that smoking in groups allowed individuals to identify as non-smokers even while they continued to smoke [41]. To hide their smoking status, individuals who were unwilling to be categorized as smokers used neutralization techniques that had previously been identified by Becker [41]. For example, adolescents may use neutralization techniques, such as asking others to buy cigarettes for them or obtaining cigarettes from social sources and sharing the packet, to reduce the effects of denormalization and the fear of stigma.

Occasional smokers tend not to smoke alone and tend to use social motivations as a rationale for smoking (‘to be with friends’), largely to feel more in control and to minimize the risk both of becoming dependent [42] and of social isolation, a consequence of stigmatization. Our results showed that, whereas weekly smokers were not protected against stigmatization by having peers who smoked, occasional smokers whose peers were weekly smokers did indeed perceive significantly less stigmatization. We also found that discrimination was perceived less by smokers than by non-smokers. Perceived discrimination against smokers by others (‘not hiring a smoker as babysitter’) was more common than perceived stereotyping of smokers (‘most people think less of people who smoke’).

Our findings indicate that less stigmatization was felt by adolescents who smoked, adolescents whose academic performance was poor, and those who had friends who smoked. This might be consistent with the concept of ‘smoking islands’, i.e. cohesive socioeconomically disadvantaged groups in which smoking is more accepted, either due to a sense of active resistance or because people in these groups feel powerless to quit smoking [21]. Our findings suggest that adolescents who smoked felt the effects of stigmatization less in low-SES schools than in high-SES schools, regardless of the impact of STPs. Again, this suggests that social context contributes to stigmatization. It is consistent with the conclusion of Stuber's study, which found that smokers who live in segregated neighbourhoods perceive less stigmatization [32]. As indicated by Farrimond, smoking remained more accepted in low-SES areas, where rates of smoking were higher. Thus, ‘felt stigmatization’ was weaker among low-SES smokers than among high-SES smokers [43].

Adolescents with the highest individual academic performance or parental SES expressed greater agreement with stigmatization of smoking than those with a lower SES. The greater sensitivity to perceived stigmatization on the part of smokers from high and medium SES backgrounds enables us to understand the differential impact of the ‘social unacceptability’ of smoking within this population. These findings indicate that stigmatization is a mechanism whereby underlying ‘social unacceptability’ may contribute to disparities through ‘denormalization’ strategies [21]. If smokers who are socioeconomically disadvantaged feel less stigmatization, the intended mechanism of social control will not operate as effectively for that population. This may explain why smoking rates in disadvantaged communities are not currently decreasing any further, despite growing stigma [44]. Aside from the school context and the peer context, it could also be explained by the role of the tobacco industry, which has been targeting young smokers of lower SES [45–48]. Current smokers could thus be subjected to a double stigma: smoking and low SES [49]. If so, ‘denormalization’ strategies may eventually exacerbate health inequalities and further marginalize smokers.

Finally, the hypothesis that STPs would be associated with greater perceived stigmatization of smokers by others was not confirmed. This might contradict the literature on the role of political power (STPs) in the stigmatization process. Stigmatization results from the enforcement by social institutions (such as schools) of the use of stigmatization as a preventative tool and from the resulting increase in the social distance between smokers and non-smokers [50]. One possible explanation is that STPs limit the experience of stigmatization to certain aspects of daily life and to certain situations and may ultimately cause only minor inconvenience. Indeed, the stigmatization associated with smoking is reversible: for stigmatization to cease, it is enough for an individual to leave the school premises. STPs were not found to decrease smoking just outside school premises, meaning that smoking remained visible [33]. Moreover, exposure to friends smoking outside school premises limits the effect of denormalization on smoking.

Strengths and limitations

There are several ways in which this paper adds to the previous body of knowledge. First, it draws on a large international survey carried out in 55 schools in seven countries, each of whose tobacco control policies differs with regard to the strictness of its implementation. The design of the study allowed us to capture the role of power in the association between stigma and smoking. This is one of the first studies to apply a validated scale of stigmatization to adolescents (smokers and non-smokers).

Our scale of stigma measures what people believe to be the social norm, not their personal opinions. As Stuber’s scale was designed for adults, its wording was revised in order to make it suitable for surveying adolescents in different countries. This methodological issue could help to explain the large differences between countries. The adaptation of Stuber’s scale and its translation into the languages of the seven European countries may have led adolescents to understand terms such as ‘losers’ in different ways. It could also be argued that Stuber’s scale does not measure real experience of discrimination so much as rational responses to the harmful effects of passive smoking on innocent third parties, such as children (Stuber, Galea, et al., 2008). Given these limitations, the score serves mainly as a general proxy of public stigmatization.