Introduction

Youth smoking remains a serious public health problem despite the recent decline in some countries, including the US, where substantial resources have been devoted to curb the pandemic [1–3]. The Global Youth Tobacco Survey (GYTS), a school-based survey of students aged 13–15 conducted in 131 countries between 1999 and 2005, showed that 8.9% of students currently smoked; and one in five nonsmoking students reported that they were likely to smoke by the following year [4]. Several national surveys suggested that adolescent smoking increased rapidly during the past few decades in China [5–7]. For example, smoking prevalence among men aged 15–19 increased from about 18% in 1996 to 21% in 2002 despite a decrease in smoking prevalence in the general Chinese population by 1.8% during the same period [6]. A study conducted in 2007 on passive smoking and its risk factors among 25,600 adolescents in 32 regions of China showed that 6.3% currently smoked, 23.1% had experimented with smoking, and 57.5% had been exposed to secondhand smoke [8]. Of particular concern is that the early initiation of smoking increases the likelihood of smoking dependence in adults [9–11] and causes more negative health outcomes. For example, starting to smoke before the age of 15 doubles one’s risk of developing lung cancer compared to starting to smoke at the age of 20 [12,13].

Multiple theoretical models have been developed to explain the initiation of adolescent smoking [14,15]. Among the most influential were Leventhal and Cleary’s four-stage smoking onset construct—preparation, initiation, becoming, and maintenance [16]—and Flay’s five-stage model—preparatory stage, initial trying, experimentation, regular use, and nicotine dependence [17]. During the preparation stage, children or adolescents are exposed to influences and opportunities to smoke, gain knowledge, and form beliefs and expectations about smoking and its effects. During initiation or initial trying, children become involved in smoking for the first time, smoking only a few cigarettes. Gradually, such children or adolescents indulge in smoking as a regular but still infrequent habit. Once they reach the maintenance or nicotine dependence stage, they smoke daily or almost daily, and their attempts to quit trigger withdrawal symptoms [18]. Similar to the development of other deviant behaviors through “social learning,” adolescent smoking behavior is acquired, in part through observing the behavior of role models in the social environment [19]. Experimentation may be directly influenced by the attitudes and smoking behaviors of their parents, siblings, teachers, and peers [20–24]. The school provides a crucial setting where social learning may take place [24]. In a school where smoking restrictions are absent and teachers smoke, students may perceive smoking as socially acceptable and therefore may be more likely to smoke [24].

Although evidence suggests that the majority of smokers start using tobacco between the ages of 11 and 13 [25], very little research has focused on elementary school students in China. Exceptions include an earlier investigation conducted in 1988 in Beijing [26]. In past decades the tobacco industry has increasingly advertised their products to young people through multiple channels. Many primary schools have been funded by tobacco companies as part of China’s Hope Project, a charity that was initiated to support schools in poor rural areas [27]. Such sponsorship by the tobacco industry may support a positive image of tobacco, promote the sale of tobacco products, and jeopardize the implementation of school-based interventions [28]. The Chinese government, however, signed and ratified the WHO Framework Convention on Tobacco Control (WHO FCTC) in 2005 and has subsequently initiated some programs to curb the smoking pandemic and reduce exposure to secondhand smoke. In particular, in 2010 the Ministry of Health (MOH) and the Ministry of Education (MOE) jointly released guidelines for schools to establish smoke-free campuses [29]; however, the extent to which the MOH and MOE have implemented or enforced these policies remains unclear.

In this study, we used data from a recent survey of elementary school students in Ningbo, China, and investigated smoking experimentation among youths aged 9–12, a younger group than those investigated in most previous studies, including the GYTS. We examined how smoking experimentation among elementary school students was associated with tobacco-related knowledge and attitudes, smoking behaviors of their parents, peers, and teachers as well as smoking policies and programs of the schools they attended.

Methods

Results

The descriptive analysis (Table 1) suggests that the risk of experimentation with smoking was significantly different in boys and girls (p<0.001, not shown in the table). With respect to knowledge about the harmfulness of tobacco, 96.0% of boys and 96.8% of girls reported smoking was harmful, but only 80.4% of boys and 81.5% of girls thought secondhand smoke was harmful. Approximately 29.5% of boys and 25.7% girls thought that distributing cigarettes to guests was polite. Only 10.9% of boys and 7.6% of girls thought that their parents would not mind whether they smoked when they were older. The majority of students (68.8% of boys and 68.5% of girls) reported that some family member had smoked in front of them. Regarding the tobacco control practices at school, 30.3% of boys and 25.9% of girls had seen teachers smoke in the previous week, and more than 80% of students often saw antismoking information on campus. About 55% of students reported received tobacco control educational materials during the previous semester. Table 1presents additional descriptive information related to these variables by gender and smoking status; the characteristics differ significantly by smoking status, especially among boys.

Table 2 presents the results of the logistic regression models. As shown in Model 1, elementary school students in suburban and rural areas had a 50% higher risk of experimentation with smoking than their counterparts in urban areas (OR=1.50; 95% CI: 1.03-2.18). The risk of experimentation with smoking among girls was significantly lower than that among boys (OR=0.34; 95% CI: 0.27-0.43); moreover, lack of awareness that smoking was harmful (OR=1.96; 95% CI: 1.28-3.00) and indifference to teenage smoking (OR=1.40; 95% CI: 1.15-1.71) predicted a higher risk of experimentation with smoking.

As results from Model 2 suggest, the perception that it was polite to give cigarettes to guests (OR=1.78; 95% CI: 1.35-2.36) and having family members who had smoked in front of them (OR=1.86; 95% CI: 1.35-2.55) predicted a higher risk of smoking experimentation. In addition, the students who thought their parents would not mind whether they smoked when they were older (OR=1.47; 95% CI: 1.05-2.05) and the students who were not sure about their parents’ attitudes regarding their future smoking (OR=1.67; 95% CI: 1.16-2.39) were more likely to have experimented with smoking than those who thought that their parents would mind.

In Model 3 peer-related influences on experimentation with smoking were examined. All three risk factors, including the belief that boys who smoke have more friends (OR: 1.93; 95% CI: 1.37-2.72), the belief that girls who smoke have more friends (OR: 2.74; 95% CI: 1.63-4.59), and having close friends who smoked (OR=4.47; 95% CI: 3.37-5.94) predicted higher risks of experimentation with smoking.

In Model 4 the effect of school-based tobacco policies and programs were examined. Students who saw a teacher smoking on campus were more likely to have experimented with smoking (OR=1.87; 95% CI: 1.49-2.35). Students who reported that school personnel had not distributed antitobacco materials during the previous semester were more likely to have experimented with smoking (OR=1.55; 95% CI: 1.16-2.09) when compared with students who reported receiving the materials; however, neither frequent exposure to antismoking information on campus nor receiving education on refusal skills at school predicted the likelihood of experimentation with smoking.

Model 5 shows the associations of experimentation with smoking and all factors simultaneously. According to this model, the region (suburban/rural vs. urban) and the students’ attitude toward smoking no longer predicted the risk of smoking experimentation. Controlling for other factors attenuated the associations between the risk of experimenting with smoking and family-related effects, peer-related effects, and school effects; but most of these associations remained significant at the 0.05 level.

Discussion

The rate of smoking experimentation for the sampled elementary school students in Ningbo was substantial, with an average rate of 19.5% among boys and 7.5% among girls. The significantly higher rate of experimentation with smoking among boys reflects gender differences in smokers in the adult Chinese population. According to the 2010 Global Adult Tobacco Survey (GATS), 52.9% of Chinese men in contrast to only 2.4% of women were current smokers [35]. Similar gender differences in smoking behaviors were found among other Asian students [36]. For example, a study reported that prevalence of smoking was higher among male students (22.3%) than among females (5.5%) in Malaysia [36]. Such a low smoking prevalence among females was likely related to cultural norms or stigma associated with female smoking [37]. Girls have also been found more likely to be “hidden smokers” in Asian culture [38]. This rate of experimenting with smoking is higher than that from a previous study conducted in 2006 in six Chinese cities, including Beijing, Shanghai, Tianjin, Shenzhen, Puyang, and Changsha, which showed that the rate of experimental smoking among similar-grade students was 10.0% among boys and 4.7% among girls [39]. The differing results in the two studies may simply reflect regional differences; however, evidence from many other countries including Asian region suggests that the rate of experimenting with smoking among elementary school students is rising [40–43]. A recent study in Taiwan showed that 34.9% of third- and fourth-grade students from mountain schools reported that they had tried cigarette smoking [44].

Prior research suggests that the initiation of smoking among adolescents may be related to parental and peer relations [11,45–48]. For example, adolescents who identified their parents as upset and thus responding negatively to their child smoking are less likely to smoke [49]. This is consistent with our findings that students who think their parents will not mind whether they smoked when they were older are more likely to have experimented with smoking. Our study also confirmed a positive association between experimenting with smoking and smoking behaviors of peers among very young students in Taiwan and South Korea [44,50]. Specifically, youth behaviors are heavily shaped by social interaction with peers; in fact, the utility of a given activity depends on the actions of their peers [51,52]. A friend who smokes may provide a young person with access to cigarettes or exert pressure on that individual to initiate smoking [44,50,53]. Although a youth may select friends who smoke because smoking peers are more accepting of others who smoke (selection effect), studies have provided evidence in support of the peer effect (i.e., having friends who smoke causally increases the risk of smoking initiation) instead of the selection effect [54,55]. Previous studies conducted in the US have shown that peer education programs that trained older students to become positive role models for elementary students were more effective compared to didactic programs led by researchers and teachers [56,57], and such studies may warrant further evaluation in the context of China.

Ample evidence suggests that the school environment plays an important role in smoking behaviors among students [57,58]. According to a national survey on knowledge, attitudes, and behaviors of teachers, 50.2% of male teachers in China smoke. The majority of these teachers (81.4%) smoke on campus, and 48.1% of them smoke in front of students [59]. Our results suggested that the smoking behaviors of teachers as role models significantly increase the risk of smoking experimentation among students. In 2010, the Chinese MOH and MOE jointly issued a policy in which smoking would be completely prohibited on elementary, junior high, and high school campuses [29]; however, the present study, based on data from the latest survey in Ningbo, suggests that this policy has not been effectively translated into practice.

In summary, findings from this study suggest that regardless of the recent regulations released by the Chinese MOE and MOH aimed at creating smoke-free campuses in China, early exposure to smoking is likely to remain a serious public health problem among elementary school students; in addition, the “smoke-free” goal is unlikely to be achieved in the absence of strict implementation of such regulations. Specific guidelines are expected from the Ministries or other Chinese authorities regarding tobacco control measures, the training of personnel, and the monitoring and evaluation of implementation and cessation services offered to faculty, staff, and students who smoke [58,60]. In addition, our findings, along with lessons from numerous studies in other countries, suggest that the health-knowledge deficit and other factors, including smoking behaviors of parents and peers as well as positive attitudes towards smoking, are crucial to the initiation of smoking [1,61]. Interventions aimed at curbing the initiation of youth smoking may therefore require a more comprehensive approach that includes tobacco-free schools, smoke-free homes, and smoke-free public places as well as training young people to resist negative influences from their peers, families, and communities [1,61]. In particular, programs that identify youth with a propensity to use tobacco and prevent them from initiating smoking may be promising and warrant further evaluation [57].

This study has several limitations. First, the smoking behaviors were self-reported, and students may have been reluctant to disclose their experiences with smoking as suggested by previous studies comparing self-reports and biochemical measures [62]. Second, peers, family, and society as factors in this study are loosely defined groups of variables, which should not be interpreted equivalently as some “latent” factors generated from factor analysis or similar techniques. Third, despite having carefully consulted the GYTS design, the Ningbo study was designed primarily to gather benchmark information for the subsequent implementation of tobacco prevention programs at the sampled schools; and the questionnaire was therefore heavily modified to serve this purpose, considerably limiting the feasibility of comparing our findings with those from GYTS-based studies.