Understanding worldwide youth attitudes towards smoke-free policies: an analysis of the Global Youth Tobacco Survey
- Howard K Koh1,
- Hillel R Alpert2,
- Christine M Judge1,
- Robert W Caughey1,
- Loris J Elqura1,
- Gregory N Connolly2,
- Charles W Warren3
- 1Harvard School of Public Health, Boston, Massachusetts, USA
- 2Center for Global Tobacco Control, Harvard School of Public Health, Boston, Massachusetts, USA
- 3Office on Smoking and Health, Centers for Disease Control and Prevention, Atlanta, Georgia, USA
- Correspondence to Hillel Alpert, Harvard School of Public Health, Center for Global Tobacco Control, 401 Park Drive, Landmark Center, Third Floor East, Boston, Massachusetts 02115, USA;
Contributors All authors contributed to one or more of conception and design, acquisition of data or analysis and interpretation of data, drafting the article or revising it critically for important intellectual content.
- Received 11 July 2010
- Accepted 18 November 2010
- Published Online First 26 January 2011
Competing interests H K Koh, former director of the Division of Public Health Practice, Harvard School of Public Health, is currently the Assistant Secretary for Health for the US Department of Health and Human Services. This article was written prior to his appointment as the Assistant Secretary for Health and does not necessarily represent the views of the Department of Health and Human Services or the United States.
Background Smoke-free policies (SFPs) in public places are increasing globally, but developing countries are lagging behind. Understanding youth attitudes towards SFPs can inform SFP initiatives.
Methods A multilevel logistic regression analysis of data collected from youth aged 13–15 years (2000–2006) who completed the Global Youth Tobacco Survey (GYTS) in 115 countries, primarily in the developing world, was conducted. The analysis examined relationships between support for SFPs and individual-level measures related to smoking status, and exposure to secondhand smoke (SHS), controlling for demographic and environmental factors of interest and country-level policy factors.
Results In all, 77.3% of 356 395 youth in 115 countries favoured SFPs, including majorities of non-smokers (78.7%) and smokers (63.6%). In the multivariable analysis knowledge of smoke harm was the strongest predictor of favouring SFPs (OR 2.42, 95% CI 2.27 to 2.67). Exposure to countermarketing (OR 1.40, 95% CI 1.25 to 1.57) and school anti-smoking education (OR 1.22, 95% CI 1.13 to 1.31) were also positively associated. Current smoking (OR 0.48, 95% CI 0.41 to 0.53), susceptibility to smoking (OR 0.46, 95% CI 0.40 to 0.52) and exposure to tobacco promotion were negatively associated. Significant country-level variation was observed. The presence of any national smoke-free legislation in a country was positively associated with youth favouring such policies.
Conclusions The majority of youth worldwide support, yet lack, smoke-free policies in public places, while being regularly exposed to SHS. Youth support of SFPs is most positively associated with knowledge of the harmful effects of tobacco smoke. Redoubling education efforts represents an opportunity to establish smoke-free environments and improve health of children in developing countries.
- Tobacco smoke pollution/prevention & control
- public policy
- world health*
- cross-sectional studies
- environmental tobacco smoke
Reducing morbidity and mortality from tobacco-related illness remains one of the crucial global health challenges of our time. Exposure to secondhand smoke (SHS) from burning cigarettes, which causes premature death and disease in children and adults who do not smoke, is a major dimension of this challenge. Health conditions caused by SHS exposure include lung cancer and coronary heart disease in adults, as well as sudden infant death syndrome, acute respiratory diseases, middle ear disease and aggravation of asthma in infants and children.1
Comprehensive smoke-free policies (SFPs) in public places can reduce SHS exposure. Currently, countries with such national policies are primarily developed nations.2 The WHO Framework Convention on Tobacco Control (FCTC), an international public health treaty ratified by 168 countries to date, recommends SFPs globally.3 SFPs can improve environmental air quality, reduce SHS-attributable disease, inhibit smoking initiation, encourage cessation and contribute to denormalisation of smoking.1 4 5 Furthermore, they are strongly supported by adult smokers and non-smokers alike in various countries.6–11
Despite the fact that young people are highly susceptible to the dangers of SHS,1 report high rates of exposure,1 12 and represent the next generation whose perspectives will influence future smoke-free policies, information regarding youth attitudes towards SFPs have been virtually absent. Such information is particularly relevant for developing countries, which are in most need of tobacco control progress.
The Global Youth Tobacco Survey (GYTS), a standardised school-based survey jointly developed by the WHO, US Centers for Disease Control and Prevention (CDC) and the Canadian Public Health Association, has been implemented in 132 of the 192 WHO member states. It provides a window into the knowledge, attitudes and behaviours of middle and high school students worldwide regarding tobacco use and policy. A recent GYTS study, which noted that overall, most youth ages 13 to 15 years report recent exposure to SHS and favour SFPs, did not further explore the factors associated with such youth attitudes.13 To deepen our understanding, we conducted a multilevel analysis using GYTS data and country-level policy data. The analyses, with a focus on developing nations, examined specific individual and country-level factors associated with youth attitudes towards SFPs in the context of the concurrent pattern of youth exposure to SHS and coverage by SFPs.
The GYTS, a standardised, structured, school-based questionnaire that includes a ‘core’ set of 56 questions, is self-administered and confidential. Additional questions are available to be used at the discretion of each participating country. The survey was originally developed in English. In each non-English speaking country, the questionnaire was translated and then back translated to ensure accuracy.14
We analysed GYTS data for the years 2000–2006; for countries surveyed more than once, we selected the most recent survey year. Surveys of 115 countries (see Supplementary appendix) asked the question ‘Are you in favour of banning smoking in public places (such as in restaurants, in buses, streetcars and trains, in schools, on playgrounds, in gyms and sports arenas, in discos)?’. The GYTS uses a two-stage cluster design to select a representative sample of students from each country, with sampling methods as described previously.12 13 15 All students present in those selected classes on the day of the survey were eligible to participate.
Youth attitudes towards SFPs were assessed by students' responses (yes or no) to the question regarding favouring the banning of smoking in public places.
Student-level independent variables
Of the 56 core GYTS questions, we identified a priori 18 that could represent potential influences on youth attitudes towards SFPs. We developed 10 measures from these 18 survey questions, based on: (1) literature-derived factors hypothesised to influence youth attitudes towards SFPs5 16–19 and (2) exploratory factor analysis of a priori groupings. We dichotomised these measures using methods previously employed by Warren et al for the measures of current cigarette smoking status, exposure to SHS and susceptibility to smoking (table 1).14 Of note, we identified non-smoking students as not susceptible to smoking when they answered ‘definitely not’ to questions pertaining to their intention ever to smoke.14 20 We included three variables as covariates: student age (ages 13–15), gender and country of residence.
We derived country-level health, education, economic measures from the World Bank, and tobacco control policy-related measures from the 2008 WHO MPOWER Global Tobacco Control Report. Tobacco control policy data were available for 111 of 115 countries examined across the 6 WHO global regions.21 Country-level measures consisted of 2006 life expectancy at birth in years; gross national income per capita; percentage school enrollment; presence or absence of a national agency for tobacco control; and presence or absence of any national smoke-free legislation (defined in the WHO MPOWER report as legislation prohibiting smoking in healthcare facilities, educational facilities, and one or more other places or institutions among government offices, indoor private offices, restaurants, pubs and bars, other indoor workplaces).
‘Developing’ status of countries
We included all GYTS-surveyed countries in the analysis, the vast majority of which are ‘developing.’ Of the 115 GYTS countries surveyed in 2000–2006, 93 are classified as ‘low’ or ‘middle’ income by the World Bank, while an additional 16 countries did not meet the standard of ‘highly industrialised’ by the Organization for Economic Cooperation and Development (OECD).22
Assessment of comprehensive SFPs
Since complete prohibition of smoking in all indoor environments is the only intervention that effectively protects against the harm of secondhand smoke,21 we assessed the presence of comprehensive SFPs in individual countries as defined in the MPOWER report, policies that ban smoking in all healthcare, educational and governmental facilities, and in indoor offices, restaurants and pubs in individual countries. For each WHO region, we calculated the percentage of the population protected by national comprehensive SFPs in the associated countries (table 2). Determinations of the presence of comprehensive SFPs were made for this purpose in distinction from determinations of the presence of any smoke-free legislation (as defined earlier for the purpose of modelling youth attitudes towards SFPs).
We computed global, WHO region-specific and country-specific frequencies of the dependent variable for 115 countries.
We used bivariate logistic regression analysis, adjusting for complex survey design variables (primary sampling unit and stratum) and final weight, to test associations between each of the independent variables (potential influences on youth attitudes towards SFPs) and the dependent variable (youth attitudes towards SFPs).
Generalised linear latent and mixed modelling (GLLAMM) was used for multilevel mixed effects logistic regression of youth attitudes towards SFPs on individual and country-level predictor variables.23 Mixed effects modelling was performed with the random effect country of residence specified as ‘level 2’ in a series of backward stepwise regression analyses. Variables were retained in models when p<0.1, and variables with p<0.05 were considered statistically significant. We retained those variables resulting in 10% or greater effect on regression coefficients or p-values. All final multivariable models included age, gender and country of residence. We used a goodness-of-fit test with properties suitable for logistic regression modelling of data collected with a complex sampling design.24 ORs and 95% CIs are presented for each predictor variable in the final model.
All statistical analyses accounted for the specific sampling frame and weighting. Weights reflect the likelihood of sampling a student based on the student's grade and gender within the population of school children in the same jurisdiction, accounting for non-response patterns.13 The survey design weights generated by CDC were scaled in order to be properly included in the GLLAMM likelihood function as recommended to account for unequal selection probabilities.25 A method of scaling that has been demonstrated to be effective by simulation studies was used so that the new weights sum to the cluster size.25 Standard errors were found to be comparable with and without country-level random effect and scaling.
Stratification and primary sampling units
Schools in each country were stratified into sets of one, two, or three schools, grouped on the basis of size and geographic location. An individual class was the primary sampling unit for strata consisting of a single large school, and an individual school was the primary sampling unit for strata consisting of two or three smaller schools. We combined 18 of the total 6528 strata, which consisted of single primary sampling units, since statistical variance estimation requires more than 1 primary sampling unit per stratum. We limited the combining of strata to within geographic jurisdictions, matching by size and presampled characteristics as closely as possible.26
Full data were available for all variables remaining in the final regression model in 86 of the countries asking specifically about SFPs. We did not impute values for missing data since these resulted essentially from variability between countries in survey design rather than non-response.
The GYTS included full data on the variables ‘family anti-smoking education’ and ‘exposure to SHS’, from only 74 and 77 countries, respectively. These questions were considered potential predictors but were excluded since they would have substantially reduced sample size, as well as the precision of estimates.
The goodness-of-fit test was statistically significant (p<0.001) suggesting that other correlates in addition to those captured in the present analysis may be important.
All statistical analyses were performed using Stata/MP V.11.0 (StataCorp, College Station, Texas, USA).
The median GYTS school response rate per country was 100%, and the median student response rate per country was 87%. From a total of 115 countries asking the GYTS survey question, ‘Are you in favour of banning smoking in public places?’, we obtained data from 356 395 questionnaires representing 61 832 715 youth (after weighting) worldwide. The final multilevel analysis model excluded countries for which data on variables of interest were missing and included data from 190 706 questionnaires representing 16 600 910 students (after weighting) in 86 countries, representing 75% of the countries surveyed.
Table 2 shows the results of univariate analysis indicating that globally, 77.3% of youth favoured SFPs. Most students in each of the six WHO regions favoured them, ranging from 53.8% in Africa to 86.0% in the Eastern Mediterranean. Differences between WHO global regions were statistically significant (Pearson χ2=45.93, p<0.001). In contrast, only 2.9% of the population (including youth) in these countries are covered by comprehensive SFPs (table 2). Reflecting this lack of policy, 54.2% of youth globally report SHS exposure within the past week, with some variation across WHO regions and rates as high as 82.1% in the European Region. Further, while three-quarters (n=115, 75.7%) of countries examined had a national agency for tobacco control, less than half (n=108, 43.5%) of countries had even the lowest level of SFPs as defined by WHO.
Of the 12 measures considered, all but 3 were statistically significantly associated (p<0.001) with favourable youth attitudes towards SFPs: age, gender and parental or peer smoking (table 3). Of note, the majority of youth reported favouring SFPs regardless of response category for each of the nine individual-level measures, including majority support from smokers (63.6%) as well as non-smokers (78.7%) (table 3).
While gender (p=0.477) was not significantly associated with attitudes towards SFPs, this variable as well as age was retained in multivariable models for possible confounding. Youth knowledge of smoke harm was most strongly associated with favouring SFPs (OR 2.46, 95% CI 2.27 to 2.67), followed by self-reported exposure to tobacco countermarketing during the past month (OR 1.40, 95% CI 1.25 to 1.57), self-reported receipt of anti-smoking education in school (OR 1.22, 95% CI 1.13 to 1.31) (table 4).
Current youth smokers were 52% less likely (OR 0.48, 95% CI 0.42 to 0.53) than non-smoking peers to favour SFPs. Non-smoking youth susceptible to smoking initiation were 54% less likely (OR 0.46, 95% CI 0.40 to 0.52) to favour SFPs than those not susceptible. Youth who had been offered a free cigarette by an industry representative were 17% less likely to favour SFPs than their peers (OR 0.83, 95% CI 0.71 to 0.96). Youth who owned a cigarette brand promotional item were 16% less likely (OR 0.84, 95% CI 0.78 to 0.90) to favour SFPs than those who did not. Although no statistically significant difference was observed in bivariate analysis, youth who reported having parents or close friends that smoke were 9% less likely (OR 0.91, 95% CI 0.85 to 0.97) to favour SFPs than their peers, in multivariable modelling (table 4).
Country of residence as a random effect in the multilevel model was statistically significant, with variance 0.72, SE 0.03 (z score=24). To illustrate the degree of variation across countries, youth residing in certain countries of Eastern Europe, such as Albania (OR 4.48, 95% CI 3.74 to 5.37) and the Republic of Moldova (OR 3.37, 95% CI 2.89 to 3.92) were the most likely to favour SFPs, whereas youth in certain countries of Africa, such as Kenya (OR 0.13, 95% CI 0.12 to 0.14) and Eritrea (OR 0.21, 95% CI 0.20 to 0.23) were the least likely to favour SFPs.
Health, education and economic measures (2006 life expectancy at birth in years; gross national income per capita; percentage school enrolment) tested at the country level were not statistically significantly associated with youth attitudes towards SFPs. Two tobacco control policy variables at the country level were however statistically significantly associated with youth attitudes towards SFPs. Youth residing in countries with any smoke-free legislation as defined by WHO (legislation prohibiting smoking in healthcare facilities, educational facilities and one or more other places or institutions among government offices, indoor private offices, restaurants, pubs and bars, other indoor workplaces) were 15% more likely (OR 1.15, 95% CI 1.09 to 1.20), and youth residing in countries with a national agency for tobacco control were 9% less likely (OR 0.91, 95% CI 0.86 to 0.95), to favour SFPs than youth residing in the remaining countries.
Our global analysis finds that sizeable majorities (77% overall) of youth aged 13–15 years in developing countries worldwide, with some variation across WHO regions, favour SFPs regardless of age, gender and smoking status. Using multilevel modelling to consider individual-level and country-level variables, the factors most strongly associated with youth favouring SFPs, controlling for demographics and smoking status, are related to knowledge and education. Also, of note, presence of any smoke-free legislation was associated with an increased likelihood of youth supporting SFPs. Knowledge of the harm of tobacco smoke is the strongest predictor of favourable youth attitudes towards SFPs. Factors most strongly associated with youth not favouring SFPs are current smoking, susceptibility to smoking, being offered a free cigarette sample by an industry representative and owning a cigarette brand promotional item.
The present findings support a growing literature of SHS-related surveys that focus primarily,6–11 27 although not exclusively,4 5 28 29 on adult attitudes. These surveys have sample sizes of up to 8000. A 2006 survey by the European Commission found that an overwhelming majority of European Union citizens supports a smoking ban in offices and indoor workplaces (86%) and indoor public spaces (84%).30 Research based on the 2001 Social Climate Survey of Tobacco Control found that in the US, 89.5% of smokers and 96.7% of non-smokers believed that SHS was harmful to infants and children, and 61% favoured complete restrictions on smoking in restaurants.7 Over 80% of smokers surveyed in Ireland supported the national ban on smoking in public places just after it was implemented in 2004,10 a trend subsequently seen in other countries.6 9 11 One multiple-country survey linked adult support of SFPs to beliefs that SHS is harmful.9 Another multiple-country survey found a positive association between increasing SFPs over time and increasing adult support for these policies.31
Few previous studies have examined youth attitudes towards SHS, however. In particular, none have adopted a global multilevel analysis examining individual-level as well as country-level correlations. Our study notes that despite overwhelming youth support, only about 1 in 30 youth worldwide live in countries now protected by comprehensive SFPs. Not surprisingly, most youth reported recent SHS exposure, including 82% in European countries surveyed. These statistics underscore the public health burden for this vulnerable population, given that that there is no risk-free level of exposure,1 22 and the opportunity for direct engagement by youth themselves to increase SFPs.32
Our findings also suggest that youth residing in countries with any smoke-free policies are 15% more likely to support these policies, a finding consistent with the notion that smoke-free policies appear to be a marker for growing denormalisation and social unacceptability of smoking.4 5 33–35 In the US, smokers living in towns with strong SHS regulations were three times more likely to attempt quitting.33 Additionally, youth living in households with voluntary no smoking rules perceived smoking as less socially acceptable.32 One multilevel analysis showed that smoke-free restaurant laws significantly discouraged youth smoking initiation in the US.36 The present findings might presage future policy support for SFPs in developing nations, especially since an accelerated tobacco-related death disparity is forecast between developed and developing countries.37 Notably, smoke-free nations, numbering 16 currently,21 are almost exclusively developed ones.
Finally, many of the factors found in this study to influence youth attitudes about SFPs also are risk factors for smoking initiation, including knowledge of health effects of smoking, parental or peer smoking, high susceptibility to offers to smoke,16 anti-tobacco countermarketing17 and owning a cigarette brand promotional item.18 Other studies have found that counteradvertising messages about the dangers of SHS and the manipulative practices of the tobacco industry are among those with the highest perceived effectiveness in preventing smoking.5 19
This study is subject to limitations. Our analysis examines responses of students aged 13–15 years who completed the survey in a classroom and does not include youth outside of school. Potential individual level factors associated with attitudes were limited to the GYTS survey questions. Measures of language ability or literacy were not used in the surveys. Additional factors related to tobacco industry influence, such as presence of transnational tobacco companies, are yet to be explored globally. As the survey assessed youth attitudes at a single point in time, future research should examine possible changes over time. Additionally, all responses are self-reported and thus may over-report or under-report behaviour and attitudes, or be subject to social desirability bias.
Nevertheless, our study represents the largest comprehensive analysis to date of attitudes (youth or adult) towards SFPs. Its strengths include a standardised questionnaire and methodology including sampling and data collection procedures, a large sample size, response rates approaching 90% and administration by most developing countries across the six WHO global regions.13 Multilevel modelling allowed for consideration of individual-level and country-level variables that may influence youth support. Furthermore, we focused on studying youth at a formative age when they are vulnerable to initiating tobacco use and addiction and are forming opinions about issues of public concern. Our findings reflect global youth support for adopting FCTC Article 8, which calls for protection against SHS exposure in indoor workplaces, public transport, indoor public places and other public places. Other FCTC provisions, including education, communication, training and public awareness of the dangers of tobacco smoke, and prohibition of the distribution of free tobacco products, especially to minors,3 are each documented in this study to influence, or be associated with, youth attitudes towards SFPs.
In short, most youth in developing countries worldwide desire, but lack, smoke-free policies while being exposed regularly to secondhand smoke. A society's youth attitudes, a window into present or changing social norms, can reflect future policy directions. Redoubling education efforts and furthering supportive youth attitudes towards SFPs represent opportunities to establish smoke-free environments and improve the health of children around the globe.
What this paper adds
The WHO Framework Convention on Tobacco Control recommends smoke-free policies in public places globally. These policies are strongly supported by adult smokers and non-smokers alike in various countries.
Despite the fact that young people are highly susceptible to the dangers of secondhand smoke, report high rates of exposure and represent the next generation whose perspectives will influence future smoke-free policies, information regarding youth attitudes towards these policies have been nearly absent. Such information is particularly relevant for developing countries, which presently are in most in need of tobacco control progress.
The present research uses a multilevel analytic approach to examine individual and country-level factors associated with youth attitudes towards smoke-free policies in 115 nations, primarily in the developing world, in the context of the concurrent pattern of youth exposure to secondhand smoke and coverage by smoke-free polices.
The authors wish to thank Dr James Ware, Dr Claudia Arrigg and Sarah Oppenheimer for reviewing the manuscript and sharing their valuable insights. We also thank Dr Elaine Puleo for biostatistical consulting on the study and Dr E Andres Houseman of the Harvard Catalyst/The Harvard Clinical and Translational Science Center.
Funding This paper was supported by a Flight Attendants Medical Research Institute, Clinical Innovator Award 072085.
Provenance and peer review Not commissioned; externally peer reviewed.