Smoking habits formed in adolescence not only affect one’s smoking behaviors in adulthood but are also related to other risky behaviors and negative health/social outcomes including alcohol and/or illicit drug use, academic problems, mental health problems, violence, and risky sexual behaviors [1,2]. Moreover, 80% of adults who had ever tried a cigarette reported first smoking before the age of 18 [3]; therefore, adolescent smoking prevention should be a top priority to manage the public health problems related to cigarette smoking.

Previous studies suggested that smoking initiation in adolescence is closely associated with subsequent smoking behaviors including the amount and frequency of cigarette smoking, difficulties in smoking cessation, and nicotine dependency [4-15]. Many of these studies were conducted in adults and investigated the relationship between age at smoking initiation and their smoking behaviors in adulthood [4-9]. In addition, interventions to stop smoking were found to be more effective in adolescence than in adulthood because the long-term behavioral habituation in adults is difficult to change or improve.

Several studies with smoking interventions in adolescents have been performed, but some methodological limitations exist. First, important confounders closely related with smoking behaviors including interpersonal factors (friends, parents, and/or siblings who smoke), socioeconomic status (education and income level), deviant behaviors, and psychological well-being were not considered in most previous studies. For example, several studies considered only sex, age, and race/ethnicity as confounding factors [11-13,15]. Other studies included academic performance or the number of friends who smoked in addition to age, sex, and race/ethnicity [10,14]. Moreover, a convenient sampling method with no representative study population was another limitation of previous studies [10,11].

In addition, previous studies have revealed conflicting results. For example, the duration since smoking initiation was found to be negatively associated with daily smoking in adolescence, but without significance [11]. A recent prospective study suggested that smoking initiation later in adolescence may lead to subsequent heavier cigarette consumption and also criticized other studies for their methodological limitations including a cross-sectional design, recall bias for age at smoking initiation, and dichotomous or limited categorical variables that measured age at smoking initiation [10]. In addition, a vulnerable period may exist that increases an adolescent’s risk of subsequent smoking behaviors [10].

Most previous studies were conducted in Western countries; only one study of this nature was conducted in Japan, but with similar limitations [15]. Therefore, this study was conducted to report the current smoking levels by age at smoking initiation among adolescents from a representative sample of urban-dwelling Koreans.

Table 1 describes the frequency of the general characteristics for the total study population. Most participants reported trying their first cigarette in grade 9 (28.3%) followed by grade 8 (24.5%), grades 10 to 11 (17.6%), grades 1 to 6 or earlier (17.4%), and grade 7 (12.2%). Among the 764 current smokers, 524 (70.5%), 449 (60.4%), and 176 (23.7%) participants were frequent smokers, daily smokers, and heavy smokers, respectively.

Table 2 shows the results of the univariate logistic regression model that identified factors associated with each smoking level. The statistically significant factors associated with all the smoking levels were male gender, attending a vocational high school, frequently drinking alcohol, level of perceived stress, having many close friends who smoke, and initiating smoking at a younger age. The factors related with frequent smoking were having a father who had ever smoked, and higher grade. Factors related with daily smoking were higher grade and having ever purchased cigarettes for their parents and someone else. Last, factors related with heavy smoking were a low level of perceived academic performance, having siblings who smoke cigarettes, and having ever purchased cigarettes for their parents and someone else.

After adjustment for significant confounders in the univariate analysis, adjusted ORs for each level of smoking tended to increase with younger ages of smoking initiation (p for trend <0.001) when compared to students who started smoking in grades 10 to 11. Adjusted ORs for frequent smokers increased from 2.24 (95% CI, 1.30 to 3.87) to 3.90 (95% CI, 2.02 to 7.56) for those who started smoking from grade 9 to grades 1 to 6 or earlier. For daily smokers and heavy smokers, the adjusted ORs increased from 1.56 (95% CI, 0.92 to 2.62) to 3.17 (95% CI, 1.70 to 5.92) and from 2.56 (95% CI, 1.21 to 5.42) to 5.67 (95% CI, 2.61 to 12.30), respectively, for those who started smoking from grade 9 to grades 1 to 6 or earlier (Figure 1).

This study found earlier ages at smoking initiation in adolescence to be strongly correlated with subsequent adolescent smoking levels including both smoking amount and frequency. These results are consistent with generally accepted concepts from nationally representative surveys from the United States [12]. Moreover, our study included important confounders such as interpersonal factors and other covariates, which were not considered in previous studies [17].

However, the results of this study conflicted with those of recent studies that found smoking initiation at a later age in adolescence to possibly lead to subsequent heavier cigarette consumption [10,11]. However, those studies used a convenience sampling method, thus their results cannot be generalized to all adolescents. In addition, very limited covariates were adjusted for in their statistical models, including sex, age, and the number of friends who smoke [10,11]. In our study, the most robust variable was the number of closest friends who currently smoked. Moreover, all confounding variables except for perceived family income level were significantly associated with smoking amount and frequency in our analysis. Therefore, at least seven covariates were adjusted for in our model that was used to confirm the association between age at smoking initiation and subsequent smoking behaviors.

Adolescents who start smoking at an early age smoked more frequently and were heavier smokers than those who started at later age. As age at smoking initiation is inversely associated with nicotine dependence, people who start smoking earlier in life suffer more difficultly in quitting smoking than those who start later in life [3-5,7,9,18]. In addition, the success rate of smoking cessation in adolescents is low. In a nationally representative longitudinal study from the US, only 4% of current adolescent smokers are estimated to successfully quit smoking per year [19].

The evidence from this study and that of previous studies stress the importance of early interventions for smoking prevention, especially in adolescents. Thus, it is also meaningful to estimate the appropriate age to intervene for effective smoking prevention. In a randomized controlled trial, early school based interventions for smoking prevention starting from grade 1 (aged 5.3 to 7.7 years) significantly lowered the smoking rate in early adolescence by 1.5 to 1.8 times compared to students who did not receive the intervention [20]. Considering this positive effect after a smoking prevention program and our results that earlier smoking initiation is related with increased amount and frequency of cigarette smoking, smoking prevention programs should be implemented as early as possible. If early prevention programs are not possible, programs should be implemented at least before reaching the critical period when programs become less effective, and the critical period for implementation could be determined by considering the following.

First, smoking prevention programs are likely to be most effective if implemented at least before the majority of adolescents start smoking. According to a report by the US Surgeon General, more than 60% of people who have ever tried a cigarette or are currently regular smokers first smoked in grade 9 or earlier, and most current smokers first tried cigarettes in grades 8 to 9 [3]. Likewise, more than half of our study participants started smoking in grades 8 to 9. Therefore, implementing a smoking prevention program before grade 8 is recommended.

Second, it is necessary to consider the variation in smoking levels by age at smoking initiation. In our study, frequency of smoking among all three smoking levels increased gradually with age at smoking initiation. However, the odds of being a frequent, daily, or heavy smoker increased sharply in one point in time for each smoking level. For example, compared to those who started smoking in grades 10 to 11, the ORs of being a frequent, daily, or heavy smoker among those who started smoking at grade 9 were 2.24 (95% CI, 1.30 to 3.87), 1.56 (95% CI, 0.92 to 2.62), and 2.56 (95% CI, 1.21 to 5.42), respectively. In addition, compared with those who started cigarette smoking at grade 8, the ORs of being a daily or heavy smoker among those who started cigarette smoking at grade 7 were 1.43 or 1.65, respectively, which is in higher than that among frequent smokers. However, the odds of smoking among those who started in grade 6 or earlier compared with those who started in grade 7 ranged from 1.11 to 1.14 across all three smoking levels. Accordingly, these results suggest that if smoking initiation were prevented through grade 8 with the help of smoking prevention programs, the proportion of those who progress to become daily or heavy smokers would be effectively reduced. To this end, intervention programs should be implemented at least before grade 8.

Third, students in the transition period from elementary (grades 1 to 6) to middle school (grades 7 to 9) experience psychological distress, decreased academic achievement, and adjustment problems due to being exposed to new environments [21]. In addition, this period overlaps with the onset of puberty that can also cause emotional stress. As a result, students in this age range are at a higher risk of deviant behaviors including substance abuse than students in elementary school are. A previous study also suggested that if smoking prevention programs in all grades is impossible, programs should at least be introduced in the transition year from elementary to middle school [22]. For those reasons, smoking prevention programs should begin at least by late elementary school.

There are some limitations in our study. First, this cross-sectional study cannot determine whether a causal relationship between age at smoking initiation and subsequent smoking levels exists. Second, information on age at smoking initiation was obtained by retrospective recall, so recall bias may have influenced our results. Finally, the number of participants who started smoking before grade 6 was too small to allow for stratification into smaller subgroups by grade. Therefore, we could not identify a relationship between age at smoking initiation and subsequent smoking levels in these students. Despite these limitations, a major strength of this study is that it considered important covariates including interpersonal factors, socioeconomic status, deviant behaviors, and psychological well-being using a regionally representative sample of urban-dwelling Korean adolescents.

In conclusion, this study found that earlier smoking initiation among adolescents increases the likelihood of becoming frequent, daily, and heavy smokers. Therefore, preventing or delaying cigarette smoking is important using appropriate smoking prevention programs. Based on the results of this study, we suggest that smoking prevention programs should begin at least by late elementary school.