Background India's Cigarettes and Other Tobacco Products Act bans tobacco sales and advertisements within 100 yards of educational institutions. In school-adjacent neighbourhoods in Mumbai, we assessed adherence to these policies and whether tobacco vendor and advertisement densities were associated with students’ tobacco use.
Methods High school students’ tobacco use was measured using a multistage cluster sampling survey (n=1533). Field geographic information systems data were obtained for all tobacco vendors and advertisements within 500 m of schools (n=26). Random-effects multilevel logistic regression was used to estimate associations of tobacco vendor and advertisement densities with ever tobacco use, current smokeless tobacco use and current tobacco use.
Results There were 1741 tobacco vendors and 424 advertisements within 500 m of schools, with 221 vendors (13%) and 42 advertisements (10%) located within 100 m. School-adjacent tobacco vendor density within 100 m was not associated with the tobacco use outcomes, but tobacco advertisement density within 100 m was associated with all outcomes when comparing highest to lowest density tertiles: ever use (OR: 2.01; 95% CI 1.00 to 4.07), current use (2.23; 1.16, 4.28) and current smokeless tobacco use (2.01; 1.02, 3.98). Tobacco vendor density within 200, 300, 400 and 500 m of schools was associated with current tobacco use and current smokeless tobacco use, but not ever use.
Conclusions The tobacco sales ban near educational institutions could be expanded beyond 100 m. Greater enforcement is needed regarding the current bans, particularly because advertisement density within 100 m of schools was associated with all students’ tobacco use outcomes.
- Advertising and Promotion
- Low/Middle income country
- Public policy
- Global health
Statistics from Altmetric.com
Tobacco use prevention and control are critical public health priorities both globally and in India. The prevalence of adult (48% men; 20% women) and adolescent (19% boys; 8% girls) current tobacco use in India is unacceptably high.1 ,2 Tobacco prevention and control are important for all ages, but it is especially important in adolescents because smoking among youth often leads to habitual use during adulthood.3 Nearly half of adult smokers worldwide, with India home to more than 10% of them, are estimated to have begun using tobacco during adolescence.4
India's Cigarette and Other Tobacco Products Act (COTPA) bans the sale of tobacco within 100 yards of educational institutions and bans tobacco advertisements everywhere except at the point-of-sale (POS), where up to two advertisements measuring no more than 60 cm×90 cm are allowed.5 The law effectively bans tobacco advertisements within 100 yards of educational institutions. The assumption behind these policies is that limiting school-neighbourhood access to tobacco products and exposure to tobacco-promoting messages will reduce tobacco use in students. If the assumptions hold true, these provisions will have important implications for population level youth tobacco use prevention in India, which may be relevant for other middle-income and low-income countries in the region.
A handful of studies have researched whether neighbourhood access to tobacco increases the risk of tobacco use in adults6 ,7 and youth,8–14 but none to our knowledge have examined the relationship in middle-income or low-income countries where diverse marketing strategies are used and a variety of tobacco products are available. Studies testing the hypothesis that neighbourhood access to tobacco increases the risk of tobacco use in youth have been equivocal.8–14 The studies were school-based, the effect sizes were small and the findings were inconsistent with respect to tobacco use outcome. Two studies showed increases in experimental smoking8 and smoking initiation,10 ,11 but not current smoking. In contrast, another study reported that current smoking was associated with neighbourhood density of tobacco vendors.9 Others found that vendor density was associated with intention to smoke11 and that tobacco-using youth were more likely to purchase cigarettes themselves rather than by using social contacts.12 These studies used different methods of measuring neighbourhood access to tobacco outlets, for example, through direct observation,9 ,12 government lists of licensed tobacco retailers8 and other methods.10 ,11 They suggest that youth tobacco use might be impacted by neighbourhood access due to illegal tobacco sales, change in smoking intention via promotional influences13 ,14 and/or via perceived and actual ease of access.
The literature shows that exposure to tobacco advertisements in general15 and in stores13 ,14 ,16 ,17 is associated with youth tobacco use, including in India.18–20 These studies suggest that exposure to tobacco advertisements promote beliefs favouring tobacco including positive attitudes about using and beliefs that using is normative. All but one of the studies14 examined self-reported exposure to tobacco promoting messages in community environments and the media. In contrast, the findings reported below are from direct observations of tobacco advertisements at the POS and at all other locations in school-adjacent neighbourhoods.
We assessed whether the presence of tobacco vendors and advertisements at the POS and other locations near high schools in Mumbai, India was associated with students’ smoking and smokeless tobacco use outcomes. In addition, we examined the level of adherence to COTPA's ban on the sale and advertising of tobacco within 100 yards of educational institutions and advertising at locations other than the POS. These bans, if adhered to, might be effective in curbing students’ tobacco use risk by reducing access, intentions to use and positive normative beliefs about tobacco. We used field GIS (Geographic Information Systems) data collection to directly measure the location of tobacco vendors and advertisements in school-adjacent neighbourhoods and linked these data with population-based high school student survey data. We hypothesised that (a) there would be a significant number of tobacco vendors and advertisements within 100 m of schools; (b) there would be tobacco advertisements at locations other than at the POS; and (c) student tobacco use would be positively associated with the density of tobacco vendors and advertisements within 100–500 m of schools. If confirmed, our findings will have direct relevance to COPTA enforcement efforts as well as further policy development about where tobacco can be sold and advertised in India.
In 2010, following the methods outlined by the Global Youth Tobacco Survey,21 we used a self-weighted multistage cluster sampling design where each school and each class within schools had equal probabilities of selection. A list of all public and private schools in Mumbai was obtained from the Mumbai District Education Department. This list served as a sampling frame for the first stage. We sampled 26 public and private high schools using sampling probabilities based on the number of students in each school. Once we obtained permission from each school, administrators provided a list of all 8th, 9th and 10th standard classes at the school, which served as the sampling frame for the second stage. From the list of classes, one 8th, one 9th and one 10th standard class was randomly sampled at each school. All students in each sampled class were eligible to complete an in-class self-administered questionnaire in either English or Marathi. The final sample included 1533 students. The school-level and class-level response rates were 100% and the student-level response rate was 99%.
The questionnaire was adapted from the Global Youth Tobacco Survey-India, which has been thoroughly pilot tested and fielded. Nevertheless, our questionnaire was pilot tested with one class each of 8th, 9th and 10th standard students, at which time the questionnaire took an average of 33 min to complete. We revised the questionnaire based on students’ feedback and results of the pilot test.
Institutional review board approvals were obtained from all collaborating institutions. School-level consent was obtained by providing the head of each school with a verbal description of the study, an information sheet and a letter of support from the Mumbai District Education Department. Once school-level consent was obtained, a passive parental consent procedure was used to obtain student consent. Students were provided with a description of the study and consent information describing that participation in the study was completely voluntary and confidential. Students were asked to fill out a consent form if they wanted to volunteer for the study.
GIS data collection
During the same time period, field GIS data were collected to obtain the latitudes and longitudes of each school's perimeter and of all tobacco vendors and advertisements within a 500 m radius of each school's perimeter. A 500 m radius was selected so that we could examine associations of tobacco vendor and advertisement densities with students’ tobacco use at 100 m increment buffer zones up to 500 m and because this was the maximum feasible area to map within project time and resource constraints. Research assistants trained in field GIS data collection walked along all roads near each school and mapped all tobacco vendors and advertisements using ESRI's ArcPad 8.0 and Trimble Juno GIS enabled handheld computer. All advertisements visible from streets or sidewalks were recorded including those at the POS. Advertisements included boards and posters displayed on the outside of stores and on walls, doors or other surfaces (figure 1). Typically locations displayed one or two boards or posters, which were counted as a single advertisement. We did not record violations to COTPA provisions regarding advertisement content, number and size.
Ever tobacco use was defined as a positive response to the question ‘Tobacco can be smoked as cigarette, bidi, cigar, chutta or dhumti; or it can be smoked in a hukka, chillum, pipe, etc. It can be chewed as gutka, pan masala, betel-quid, khaini, mawa or zarda; or applied as mishri, gul, bajjar, snuff, tobacco toothpaste, tobacco tooth powder etc. Have you ever tried or experimented with any such form of tobacco, even once or twice?’ Current smoking tobacco use was defined as a positive response to the question ‘During the past 30 days, did you smoke tobacco in any form?’ Current smokeless tobacco use was defined as a positive response to either of the following two questions: ‘During the past 30 days, did you chew tobacco in any form?’ or ‘During the past 30 days, did you apply tobacco in any form?’ Current tobacco use was defined as using smoking and/or smokeless forms of tobacco in the last 30 days.
The number of tobacco vendors and advertisements near each school (ie, density) was measured using procedures outlined elsewhere.8 Projections of 100, 200, 300, 400 and 500 m buffer zones around the perimeter of each school were spatially joined to the projection of geocoded tobacco vendors and advertisements, yielding the number of vendors and advertisements within each school's five different buffer zones. The shape of the buffer zones followed the shape of school perimeters. The data for each buffer zone were then grouped into tertiles.
The following covariates were included: age, gender, religion (Hindu, Muslim, Buddhist or other), monthly receipt of pocket money, hopelessness (measured using a single item adapted from the U.S. Behavioral Risk Factor Survey, which asked whether respondents experienced prolonged feelings of hopelessness or sadness for 2 or more weeks in the last 12 months (http://apps.nccd.cdc.gov/BRFSSQuest)), ease of access to tobacco, parental tobacco use, peer tobacco use and each school's annual fee structure as a proxy for student-level socioeconomic status.
First, we calculated frequencies of student-level study variables, including missing values. Second, we calculated measures of central tendency and spread for variables measuring density of tobacco vendors and advertisements. Third, we used analysis of variance to assess whether vendor and advertisement density were associated with school-level annual fee structure. Fourth, we measured the correlation between the school-level prevalence of each measure of tobacco use and the density of tobacco vendors and advertisements. Finally, since the data were hierarchical, with students nested within schools and neighbourhoods, we used a multilevel modelling framework. Specifically, random-effects logistic regression22 ,23 was used to estimate the association between the measures of vendor and advertisement density and measures of tobacco use (ever tobacco use, current tobacco use, current smoking tobacco use and current smokeless tobacco use), while controlling for the covariates listed above. Only students who responded to all student-level variables were included in the regression analysis (n=1320). All statistical analyses were conducted using Stata 12.0 with an α level of 0.05. We used the command xtlogit for multilevel regression and the option cluster to account for the sampling design.
Table 1 shows the distributions of student-level variables. About 63% of students were 14 or more years of age, there were more men than women (59% vs 41%), the majority were Hindu (59%) or Muslim (18%) and received some pocket money (52%). About 28% reported that either one or both parents use tobacco and more than 40% reported some level of peer tobacco use.
A total of 1741 tobacco vendors and 424 tobacco advertisements were identified within 500 m of schools, with 221 vendors (13%) and 42 advertisements (10%) located within 100 m of schools. Eighty-five per cent of schools (22/26) had at least one tobacco vendor and 54% (14/26) had at least one tobacco advertisement within 100 m of their perimeters. About 11% of advertisements (47/424) were located in places other than the POS, about 46% of schools (12/26) had at least one such advertisement within 500 m and about 4% (1/26) had at least one within 100 m. Table 2 shows that an average of eight vendors (Median: 5; Range: 0–28) and 1.5 advertisements (Median: 1; Range: 0–9) were present within 100 m of schools. There were many vendors (Mean: 60; Median: 49; Range: 2–199) and advertisements (Mean: 15.4; Median: 12.5, Range: 1–64) within 500 m of schools. Table 3 shows that the school-level fee structure was negatively associated with tobacco vendor density and advertisement density within 100 and 500 m of schools (p<0.05).
Figure 2 shows the location of each sampled school in Mumbai, and illustrates the school-level prevalence of current tobacco use and the school-level density of the tobacco vendors within 100 m. There was a moderate correlation between these two variables (r=0.34, p=0.08), but not with vendor density within 500 m (r=0.30, p=0.14). There was a significant moderate correlation between school-level prevalence of current smokeless tobacco use and vendor density within 100 m (r=0.42, p=0.03) and a borderline significant correlation with vendor density within 500 m (r=0.37, p=0.07). Though three of these coefficients were not significant at the 0.05 level, they are notable, given a sample size of 26.
Table 4 shows that students in schools with higher tertiles of vendor density within 200–500 m were at increased risk of either current tobacco use or current smokeless tobacco use as compared with students in schools with the lowest tertiles of vendor density. For example, student in schools that had 200 m tobacco vendor densities in the second (OR: 2.25; 95%: 1.19, 4.25) and third tertiles (2.05; 1.02, 4.13) were at higher risk of current tobacco use, but not ever tobacco use or current smokeless tobacco use. Students in schools with higher densities of tobacco vendors within wider buffer zones were at higher risk of current tobacco use (for 300 and 400 m buffer zones) and current smokeless tobacco use (for 300, 400 and 500 m buffer zones), but not ever use.
Table 5 shows that tobacco advertisement density within 100 m was associated with all tobacco use outcomes when comparing students in school with the highest tertile of advertisement density with the lowest tertile: ever tobacco use (2.01; 1.00, 4.07), current tobacco use (2.23; 1.16, 4.28) and current smokeless tobacco use (2.01; 1.02, 3.98). However, advertisement density was not associated with any tobacco use outcomes when it was measured using wider buffer zones.
We did not report regression results for current smoking tobacco use because the relatively low prevalence resulted in unstable estimation of regression coefficients
Though there was great variability, Mumbai school-adjacent neighbourhoods on average had a large number of tobacco vendors and advertisements. The hypothesis that student tobacco use would be positively associated with the density of tobacco vendors and advertisements within 100 to 500 m of schools was largely supported by the results. The density of tobacco advertisements was a more important risk factor for students’ tobacco use within the 100 m buffer zone than was the density of tobacco vendors, although, importantly, the latter was positively associated with current tobacco use and current smokeless tobacco use when vendor density was measured at wider school-adjacent areas, up to 500 m. In addition, vendor density was ecologically correlated with school-level prevalence of tobacco use outcomes.
In addition to providing access to purchasing tobacco, as suggested by other researchers,16 the POS environment may serve as an important venue for tobacco promotion via the display of tobacco products and advertisements, especially within close proximity of schools. The density of tobacco advertisements within 100 m of schools, the area where COTPA bans the sale, and consequently, the advertisement of tobacco, was a significant risk factor for all student tobacco use outcomes measured in this study. The results also indicate that there was a substantial lack of adherence to COTPA policies on the sale and advertising of tobacco within 100 yards of educational institutions.
It is unclear why student-level tobacco use was associated with advertisement density but not vendor density within close proximity of schools (ie, 100 m). As students go to and from school, they may readily and frequently be exposed to tobacco advertisements within 100 m of schools. They, therefore, might be strongly influenced by those tobacco-promoting messages. In contrast, students may not risk purchasing tobacco in close proximity of their school because other students, teachers and administrators may observe them. Students may prefer to purchase tobacco from further away. Unfortunately, we did not ask students who used tobacco where they purchased tobacco.
These results have important policy implications. First, they suggest that COTPA bans on tobacco sale and advertising near educational institutions may be effective in reducing high school students’ tobacco use risk. Tobacco vendors were present within 100 m of nearly all sampled schools, and within this close proximity, tobacco advertisements were present around a majority of schools. About 1 in 10 advertisements were displayed at locations such as telephone booths, tailor storefronts and general stores that did not sell tobacco, suggesting that greater education and enforcement efforts on placement of such advertisements would likely decrease youth exposure to tobacco promoting message. Increasing efforts to implement, enforce and promote adherence to COTPA bans on tobacco sale and advertising near schools would likely help reduce the risk of tobacco use in high school students. Low-income students may be particularly impacted because schools with lower annual fees had on average more advertisements and vendors near them. Moreover, given that there were significant increased risks of all student tobacco use outcomes with increasing density of tobacco advertisements within 100 m of schools, two options may be prudent. The first is to completely ban all advertising even at the point of sale and strongly enforce the ban. The second, less stringent option, is to explicitly ban tobacco advertisements within 100 m of schools with stricter fines and penalties, and at the same time promote and diligently enforce current bans within close proximity of educational institutions. Furthermore, the area around schools where tobacco sales are banned could be expanded because tobacco vendor density within 200–500 m of schools consistently increased student tobacco use risk. The tobacco industry has argued that banning tobacco sales within 500 m of educational institutions would, in densely populated areas such as Mumbai, effectively leave no areas where tobacco could be sold. In light of the results of this study, independent studies of this claim have to be conducted and policy makers should strongly consider increasing the banned area to more than 100 m.
In contrast to studies reporting that greater density of tobacco vendors near schools increases student experimental smoking risk, but not current smoking8 ,10 ,11 we found that it was associated with an increased risk of current tobacco use, particularly the smokeless form. Our study found no significant association with ever tobacco use. The majority of tobacco products consumed in India are smokeless, which are widely available and cheaper than cigarettes and even bidis. Greater tobacco vendor density near schools may increase neighbourhood access to cheap smokeless products relative to smoking products and would subsequently have a greater influence on student smokeless tobacco use in comparison with tobacco smoking.
Strengths and limitations
There are several notable strengths to this study. First, we directly measured the locations of each tobacco vendor and advertisement using field GIS data collection, while other studies have largely relied on existing data on the location of tobacco vendors, and have not measured location of advertisements. Second, we measured locations of vendors and advertisements during the same time period as the student surveys. In past studies, often the neighbourhood exposure variables were measured after the outcomes. Third, we tested the hypotheses within a policy and cultural context where they had not been examined before, and in a region where a high disproportionate number of tobacco users live. This study, therefore, provides new data on important tobacco use prevention and control policy questions that are relevant to the Indian context and can serve as a foundation for future research in the region.
There are several factors that limit the validity of the study. Most importantly, although the data were population-based, they were collected cross-sectionally thereby limiting conclusions regarding causality. For example, schools with high student tobacco use prevalence may attract tobacco vendors and encourage display of tobacco advertisements in adjacent neighbourhoods. Second, we did not measure student reports of exposure to tobacco advertisements and vendors in their school-adjacent neighbourhood. Students may instead be more influenced by the tobacco vendor and advertisement environment in their home neighbourhoods. Third, tobacco use was measured using self-reports. However, such measures have been widely used and have shown good validity and reliability when self-administered and measured anonymously within classroom settings.24
This study suggests that India's COTPA bans on the sale and advertising of tobacco within 100 yards of educational institutions will be effective in reducing tobacco use risk in high school students. Exposure to tobacco advertisements within close proximity of educational institutions seems to be a particularly important risk factor. Greater promotion, adherence and enforcement of India's current COTPA sale and advertisement bans near educational institutions would likely result in lower levels of tobacco use in high school students. There may be value in banning tobacco advertisements altogether and increasing the size of the area around educational institutions where tobacco sales are banned.
What this paper adds
First, this paper reports that there is widespread lack of adherence to India's bans on the sale and advertising of tobacco near educational institutions.
Second, it shows that tobacco vendor and advertisement density near educational institutions may be important risk factors for current tobacco use, particularly smokeless tobacco use, indicating that the bans are likely effective strategies to reduce youth tobacco use risk.
This study was conducted with support from the Fulbright-Nehru Scholar Program and the Jonsson Cancer Center Foundation. The project would not have been possible without the hard work of the field staff as well as the cooperation of the Municipal Corporation of Greater Mumbai and Office of the Director of Education—Mumbai.
Contributors RM originated the study concept and design, conducted the data analysis and wrote the paper. MP assisted with developing the survey protocol, provided critical feedback on data analysis and critically edited the paper. SP assisted with developing the mapping protocol and critically edited the paper. PG, WM and SS were responsible for the scientific integrity of the study, provided guidance on the study concept and design and critically edited the paper. LR managed the data collection and training, provided critical input on the study design and critically edited the paper. MP created the GIS measures and critically edited the paper.
Funding Fulbright-Nehru Scholar Program and the Jonsson Cancer Center Foundation.
Competing interests None.
Ethics approval University of California, Los Angeles (UCLA), Healis, Tata Memorial Hospital.
Provenance and peer review Not commissioned; externally peer reviewed.
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.