INTRODUCTION
The World Health Organization (WHO) has emphasized the importance of controlling tobacco consumption and supporting smoking cessation efforts. As a result, they have designated the ‘World No Tobacco Day’ to be observed annually on 31 May, starting from the year 19881,2. The significance given to the issue of tobacco consumption by various relevant organizations both domestically and internationally indicates that smoking is a global problem concerning every country. Smoking is considered a major cause of serious health problems and life-threatening conditions such as chronic obstructive pulmonary disease (COPD), lung cancer, cardiovascular diseases, and others. If individuals with chronic diseases continue to smoke, it can lead to more severe complications3. Therefore, the issue of tobacco consumption can be considered a significant problem that has a profound impact on the health of the population, causing premature illness and death. Statistics reveal that over 50000 Thai people die annually from smoking-related causes. Additionally, the marketing strategies employed by the tobacco industry contribute to an increasing number of new smokers among the youth, who are more likely to continue smoking throughout their lives. Apart from the health issues faced by the population, the burden on the country’s healthcare system in treating tobacco-related illnesses results in significant economic losses. The estimated economic loss from smoking-related healthcare costs amounts to 74.9 billion THB4,5.
Strategies to combat tobacco usage are based on three approaches: creation, reinforcement, and collaboration. Under the approach of creation, advocates and volunteers were trained to raise awareness and create a sense of consciousness regarding the hazards and consequences of smoking. The reinforcement approach focuses on enhancing skills and knowledge among current smokers to reduce and quit smoking and prevent new smokers from starting6,7. It also involved strengthening the social support system responsible for controlling tobacco consumption, such as exemplary leaders and community influencers. This collaborative approach emphasized collective action and collaboration to empower and drive the initiatives. This included a network of smoke-free community leaders, smoke-free network alliances, youth groups, and social organizations working together to establish control measures within the community such as designated no-smoking areas, employment policies, and interventions for social groups or members who smoke 8-10.
However, despite these efforts, the number of new, young smokers has continued to increase over the past decade, while the age of smoking initiation has shown a declining trend. Additionally, the implemented measures have not been successful in reducing the number of new smokers among youth. According to a report by the Ministry of Public Health in June 2019, there were 10.7 million Thai young smokers out of a total youth population of 55.9 million, accounting for 19.1% of that population. Among these, 9.4 million were regular smokers (16.8%) and 1.3 million were occasional smokers (2.3%)4,11.
From the problems mentioned above, this study aims to ascertain the critical factors that influence smoking initiation among Thai youth, aiming to address the rising issue of new adolescent smokers. By leveraging an advanced feature selection approach, this research seeks to identify key social and behavioral drivers behind smoking behaviors12-14. Through the analysis of a comprehensive statistical database of tobacco consumption among the Thai population from 2004 to 2021, the study aims to reveal meaningful correlations between smoking habits and risk factors, offering valuable insights into the patterns and predictors of youth smoking initiation.
METHODS
Study design and study participants
This study utilized a cross-sectional design using pooled secondary data collected by the Tobacco Control Research and Knowledge Management Center (TRC) and the National Statistical Office (NSO) of Thailand. The data were gathered from nationally representative surveys conducted between 2004 and 2021. The study population consisted of adolescents aged 15–18 years, totaling 30600 participants, representing a diverse demographic and geographical distribution across Thailand. The data collection process ensured anonymity by not including personal identifiers, encouraging honest responses from participants about their smoking and alcohol consumption behaviors, as in other studies15,16.
Outcomes and variables
The primary outcome of this study was smoking status, categorized as either ‘smokers’ or ‘non-smokers’. Smokers were defined as adolescents who reported smoking any form of tobacco, while non-smokers were those who reported no tobacco use. Independent variables included demographic factors (age, sex, education level, and marital status), geographical location, position in the family hierarchy, peer influence, purchasing behavior, and environmental exposure to smoking-related cues (e.g. cigarette butts in public spaces). These variables were selected to assess their association with smoking initiation. Definitions for each variable were standardized according to the survey guidelines, ensuring consistency in data interpretation.
Data source
Survey data on the smoking and drinking behavior survey among the Thai population were from Thailand TRC and NSO, 2004–2021. Data collection began in 2004 as a cross-sectional survey representative of the Thai population regarding smoking and alcohol consumption behaviors. The survey was conducted anonymously, with variables coded numerically, ensuring no redundancy among values. For the present study, we selected data on smoking behavior, specifically for the age range of 15–18 years only, comprising 36067 individuals. Additionally, we categorized the dependent variables into two groups: 1) smokers and 2) non-smokers, and chose independent variables to characterize the smoking behavior of adolescents.
Data analysis
Descriptive statistics were used to summarize the characteristics of the study population. Bivariate analyses, including chi-squared tests and t-tests, were performed to identify potential associations between smoking status and independent variables. Variables showing significant associations in bivariate analyses were included in multivariable logistic regression models to adjust for potential confounders. Adjusted odds ratios (AORs) with 95% confidence intervals were calculated to determine the strength of association between the predictors and smoking behavior. All analyses were conducted using statistical software, ensuring robust handling of the large dataset and controlling multicollinearity and other potential biases.
Moreover, during data analysis, the research team developed a novel feature selection algorithm called Adaptive Chaotic Binary Grey Wolf Optimization (ACBGWO), the methodological approach of which is presented in detail in the Supplementary file.
RESULTS
From the data presented in Tables 1–4, the results include six primary data groups related to Thai adolescents aged between 15 and 18 years. These data encompass general smoking prevalence among the Thai population, environmental observations of smoking behaviors, exposure to anti-smoking media messages, observations of cigarette promotions, awareness of smoking-related diseases, and purchasing behaviors.
Table 1
General data on smoking prevalence among the Thai population, bivariate analysis of cross-sectional data examining the data on smoking consumption behavior among teenagers in Thailand, 2004–2021 (N=36067)
Table 2
Noticing anyone smoking nearby or any cigarette butts in any public places, bivariate analysis of cross-sectional data examining the data on smoking consumption behavior among teenagers in Thailand 2004–2021 (N=36067)
Table 3
Noticing information about the danger of smoking cigarettes or that encourages quitting smoking in any media and any types of cigarettes promotions, bivariate analysis of cross-sectional data examining the data on smoking consumption behavior among teenagers in Thailand 2004–2021 (N=3606)
Table 4
Noticing any advertisements or signs that encourage smoking in any place and awareness of diseases caused by smoking tobacco, bivariate analysis of cross-sectional data examining the data on smoking consumption behavior among teenagers in Thailand 2004–2021(N=36067)
The dataset comprised 36067 records, with 2848 individuals (7.90%) identified as smokers and 33219 individuals (92.10%) as non-smokers. The gender breakdown revealed disparities; among males, 2781 individuals (15.35%) reported smoking, compared to only 67 females (0.37%), indicating a significant gender difference in smoking prevalence.
Education level showed a strong association with smoking prevalence. Individuals with primary education had the highest prevalence of smoking at 11.42% (1036 individuals), followed by those with lower secondary education at 6.41% (1472 individuals). These findings highlight that lower education level is correlated with higher rates of smoking initiation among adolescents.
Marital status also influenced smoking behavior. Married individuals exhibited a smoking prevalence of 13.54%, compared to 7.46% for single individuals. This suggests that marriage may be associated with higher smoking rates, potentially due to life transitions or stressors.
The acquisition of cigarettes among adolescents was divided into two primary methods: obtaining cigarettes without purchase or receiving them for free, and purchasing cigarettes themselves. A significant 93.43% (33696 individuals) of participants reported acquiring cigarettes without purchasing them, while 6.57% (2371 individuals) purchased cigarettes directly. Notably, those who purchased cigarettes themselves were exclusively smokers, indicating the importance of regulating sales to minors.
Observations of cigarette butts in public areas were most frequently reported in fresh food markets or community spaces, with 5621 individuals noting their presence. This suggests that environmental exposure to smoking-related cues in everyday settings may normalize smoking behaviors among adolescents.
Anti-smoking media campaigns showed varied effectiveness. A total of 5719 individuals (15.86%) reported noticing visual and textual warnings on cigarette packs. However, the majority of smokers (28977 individuals or 80.34%) indicated that they did not notice or pay attention to these warnings, likely due to accessing cigarettes through informal channels or rolled cigarettes that lack packaging.
Awareness of smoking-related diseases was evaluated among the participants. The most commonly recognized diseases included chronic obstructive pulmonary disease (COPD) at 22.27%, laryngeal cancer at 21.58%, and oral cancer at 21.48%. Despite this awareness, the data suggest that knowledge of health risks alone may not sufficiently deter smoking behavior among adolescents.
Additionally, the data on retailer practices revealed that 98.14% (35397 individuals) of cigarette vendors did not request identification when selling cigarettes, indicating weak enforcement of age restrictions. Only 1.74% of vendors requested identification, so most vendors allowed minors easy access to tobacco products.
The findings presented in Tables 1–4 emphasize key demographic, social, and environmental factors associated with smoking behavior among Thai adolescents. These results provide critical insights into patterns of smoking initiation, serving as a basis for evidence-based intervention strategies.
Factor analysis and key findings
This study employed the Adaptive Chaotic Binary Grey Wolf Optimization (ACBGWO) algorithm to analyze factors contributing to smoking initiation among Thai adolescents. The primary goal was to identify critical variables predicting the likelihood of smoking initiation, focusing on social and environmental influences. Variables such as geographical location, purchasing habits, and exposure to smoking-related cues (e.g. cigarette butts in public areas) were evaluated. The feature selection process effectively narrowed the analysis to seven key factors most strongly associated with smoking initiation. These included age at first smoking exposure, cigarette accessibility, and the presence of warning labels. The ACBGWO algorithm facilitated the extraction of meaningful insights from a large dataset of 36067 participants, providing a comprehensive understanding of the social and environmental factors influencing youth smoking behaviors in Thailand.
Experimental results
As shown in Table 1 of the Supplementary file, the ACBGWO algorithm achieved an accuracy of 99.63% with a low standard deviation of 0.0479, demonstrating its robustness and reliability in identifying key factors. Among the selected features, cigarette accessibility was highlighted as a critical predictor, with adolescents who purchased cigarettes themselves exhibiting a much higher likelihood of becoming regular smokers. The analysis also revealed that early exposure to smoking is a key factor in predicting long-term smoking habits. Adolescents who began smoking before the age of 15 years demonstrated stronger tendencies to continue smoking into adulthood. Additionally, environmental cues, such as the presence of cigarette butts in public spaces, were associated with increased smoking initiation rates.
DISCUSSION
The results of this study shed light on key social factors contributing to smoking initiation among Thai adolescents aged 15–18 years. This is particularly significant given the alarming trend of increasing youth smokers over the last decade, as highlighted by the Ministry of Public Health of Thailand. Despite various anti-smoking campaigns, such as the ‘Quit Smoking, Discover Happiness’ initiative, smoking remains prevalent among the youth, which suggests that the factors influencing smoking initiation are deeply entrenched in the socio-cultural environment17.
One of the key findings in this study was the role of environmental factors. Geographical location, for example, played a significant role in predicting smoking behavior, with adolescents in regions such as the Northeastern part of Thailand showing a higher prevalence of smoking compared to other regions. This aligns with global findings that rural and semi-urban areas tend to have higher smoking rates due to less stringent enforcement of tobacco laws and greater access to cigarettes through informal channels18,19. Additionally, the presence of cigarette butts in public spaces like markets was found to be a trigger for smoking initiation. This can be explained by the theory of environmental cues, which suggests that visual reminders in one’s environment can increase the likelihood of engagement in risk behaviors like smoking. The role of social relationships also emerged as a critical factor in smoking initiation. Adolescents who had family members or friends who smoked were significantly more likely to start smoking themselves. This is consistent with social learning theory, which posits that behaviors are learned through observing others, particularly in close social circles. The normalization of smoking within families or peer groups reduces the perceived risks associated with smoking, making it easier for adolescents to experiment with cigarettes. This finding suggests that tobacco control policies should expand their focus to include family-based interventions, such as educating parents about the risks of smoking around their children and the importance of quitting as a role model for their children.
The study also found that age at first exposure to cigarettes was a strong predictor of future smoking behavior. Adolescents who began smoking at an early age were more likely to become regular smokers. This is in line with previous research that shows that early initiation of smoking increases the likelihood of addiction, as younger individuals are more vulnerable to the neurochemical effects of nicotine. This underscores the need for preventive measures aimed at delaying the onset of smoking, such as school-based anti-smoking programs that target children before they are exposed to cigarettes. It also highlights the importance of strict enforcement of age restrictions on the sale of tobacco products.
Another important finding of this study was the role of cigarette purchasing behavior. Adolescents who purchased cigarettes themselves, rather than obtaining them from others, were more likely to continue smoking regularly. This suggests that the act of purchasing cigarettes may serve as a reinforcement mechanism, making the behavior more habitual. Moreover, the study found that most retailers in Thailand do not ask for identification when selling cigarettes to minors, which highlights a major gap in the enforcement of tobacco control laws. This calls for stronger regulation and monitoring of cigarette sales, especially in convenience stores and markets where adolescents are likely to buy cigarettes.
The study’s use of the ACBGWO algorithm to identify significant features related to smoking behavior also proved to be highly effective, achieving an accuracy rate of 99.63%. This suggests that advanced machine learning techniques can be instrumental in identifying patterns in large datasets, such as those used in this study. By focusing on the most significant factors, future interventions can be more targeted and effective in reducing smoking rates among adolescents.
Limitations
However, it is important to consider the limitations of this study. First, the reliance on self-reported data may have introduced biases, particularly in the underreporting of smoking behaviors due to social desirability. Adolescents may have been reluctant to disclose their smoking habits, particularly in light of the negative social attitudes toward smoking. Future research could address this limitation by incorporating biochemical verification methods, such as cotinine testing, to confirm self-reported smoking status. Additionally, the retrospective nature of some of the data, particularly regarding smoking history and dependence levels, may have led to recall bias, where participants may have inaccurately remembered or reported their past behaviors. Finally, the study period spans data collected between 2004 and 2021, which may limit the current applicability of findings. Changes in societal attitudes, tobacco policies, and youth behaviors over the years could affect the relevance of the conclusions drawn from historical data in today’s context. However, despite the passage of time since the data collection, it is important to note that the fundamental landscape of smoking behaviors and cessation efforts among Thai adolescents has not significantly changed. While there may have been minor developments, no substantial shifts in smoking initiation patterns or tobacco control measures have been widely observed or reported. Therefore, the findings of this study remain highly relevant, and the core dynamics that influence smoking initiation among adolescents are expected to persist in a comparable manner, making this research valuable for current and future tobacco control strategies.
Implications
Despite the limitations, the findings of this study have important implications for tobacco control policies in Thailand. The identification of key factors, such as geographical location, social relationships, and purchasing behavior, provides valuable insights into the dynamics of smoking initiation among Thai adolescents. By targeting these factors, public health interventions can be more effective in curbing the rise of new smokers. For example, strengthening the enforcement of tobacco sales regulations, creating smoke-free public spaces, and implementing family-centered interventions could significantly reduce the number of new smokers in Thailand.
CONCLUSIONS
This study contributes to the growing body of literature on adolescent smoking behavior by using advanced feature selection techniques to identify the most significant predictors of smoking initiation. The findings underscore the importance of addressing environmental and social factors in tobacco control efforts and highlight the need for more targeted and comprehensive interventions to prevent smoking among Thai youth. As smoking continues to pose a major public health challenge in Thailand, it is crucial that policymakers and public health officials take immediate action to address the root causes of smoking initiation and promote a smoke-free future for the next generation.
