INTRODUCTION

Smoking has been recognized as an urgent threat to human health with millions of people dying annually from smoking-related diseases, and increased risk of increased susceptibility to cardiovascular diseases, adverse effects on reproductive and developmental health, and a potential link to diabetes and other conditions¹. In recent years, the Chinese government has implemented a series of nationwide tobacco-control measures, establishing a multi-level tobacco-control framework². However, the smoking rate among Chinese males is still about 50.5%, with a significant gender disparity with males the primary smokers³. Smoking is deeply ingrained in male social practices, regarded as a symbol of masculinity and assisting interpersonal interactions⁴. This is partly attributable to the cultural importance of smoking in Chinese social settings, where cigarette gifting significantly contributes to the high smoking prevalence and perceived social benefits often outweigh health risks, thereby fostering a socially accepted smoking behavior⁵.

Socioeconomic status (SES) is widely acknowledged as a structural determinant of smoking behavior. In high-income countries, smoking prevalence and intensity are inversely associated with SES. Research indicates that individuals with lower SES exhibit higher rates of smoking and face an elevated risk of mortality, even when smoking at low intensity. Individuals with low SES are more inclined to adopt unhealthy habits such as smoking, drinking, insufficient physical activity, and unhealthy eating patterns⁶. Conversely, high-SES groups typically exhibit healthier behaviors⁷.

However, studies in developing countries reveal a positive association between SES and tobacco use^8,9. Studies have shown that individuals from lower income households exhibit a higher likelihood of smoking, which can subsequently lead to increased poverty and health costs⁸. Research in China indicates that individuals with lower socioeconomic status (SES) are more likely to engage in unhealthy lifestyles, which in turn is associated with a higher risk of death and cardiovascular diseases. Individuals with lower SES exhibit a lower probability of engaging in health-risk behaviors, whereas higher-SES groups are more prone to smoking⁹.

Why does cigarette smoking spread so differently through these two kinds of societies over time? It can be attributed to the fact that China and other high-income countries are at different stages of tobacco diffusion¹⁰. China is still at the early stages, where smoking is still regarded as a symbol of status and masculinity⁴. In high-income countries, higher socioeconomic groups begin to quit smoking following health awareness and anti-smoking campaigns, leading to a decline in male smoking prevalence. But smoking prevalence remains persistently high among lower socioeconomic groups, creating widening health disparities based on socioeconomic status¹¹.

Over the past decade, Internet penetration in China has significantly increased, with the number of internet users growing from 564 million in 2012 to over 1.079 billion by June 2023, and the internet penetration rate soaring from 42.1% to 76.4%¹². Internet use typically includes online chatting, shopping, entertainment, and gaming. Meta-analysis evidence demonstrates that increased exposure to tobacco advertising and promotional content on the Internet elevates the likelihood of smoking initiation, since social media, virtual communities, online games, and streaming platforms all offer opportunities for social learning of smoking^13,14. Digital communities may facilitate the spread of cigarette use¹⁵. And spending more time on social media is significantly associated with more smoking behaviors among young people¹⁶.

A robust positive SES gradient in Internet use has been well documented¹⁷. Advanced digital skills, such as using the Internet for news, job searching, product information acquisition, and work-related activities, are strongly related to education level and household income¹⁸. High-SES individuals typically possess superior digital literacy, thus enabling them to actively search for, evaluate, and subscribe to smoking-cessation or health-promotion content, thereby reducing their exposure to pro-tobacco messages¹⁹. In contrast, low-SES individuals tend to engage more in online games and entertainment, which may contribute to worse Internet addiction and more cigarette consumption²⁰.

In addition, high-income groups are also more likely to obtain online health information, conferring additional advantages in health communication and tobacco control^21,22. High-SES groups are in a better position to offer social support, exert normative pressure against addictive behaviors, and moderate the effect of Internet use on healthy behaviors²³. Adolescents from low-SES families are more likely to exhibit problematic internet use, as evidenced by a significant positive link between low SES and excessive social media use. However, this link does not exist among adolescents from high-SES backgrounds, suggesting that higher SES may act as a protective factor against addiction risk²⁴. Thus, SES appears to exert a systematic weakening effect on Internet-induced addictive behaviors.

Thus, this study aims to systematically investigate whether SES and Internet use are, respectively, associated with daily cigarette consumption (DCC) among adult men in China. Specifically, we aim to assess whether higher individual SES – typically measured by a combination of education level, occupational category, and household income – is associated with an increase in daily cigarette consumption. Concurrently, it will examine whether the frequency of individual Internet use is positively correlated with DCC. Furthermore, this study will analyze the potential moderating role of SES in the relationship between Internet use and smoking behavior. That is, it will assess whether SES can enhance, attenuate, or alter the strength of the impact of Internet use on Chinese men’s DCC.

METHODS

RESULTS

Table 1 illustrates the temporal trend in male smoking behavior. DCC decreased markedly, from 8.8 cigarettes in 2014 to approximately 8.2 in 2016 and 2018, and subsequently dropped sharply to 7.3 in 2020 and 6.9 in 2022. Over the nine-year period, average consumption decreased by approximately two cigarettes, indicating a 21.7% reduction.

Log of household income exhibited a steady increase from 8.7 in 2014 to 10 in 2022. Years of schooling rose from 8.2 years in 2014 to 8.7 years in 2018, experienced a slight decline to 8.5 years in 2020 and 2022. Internet use, a sum of five indicators, rose sharply from 1.4 in 2014 to 3.6 in 2022, reflecting the sustained expansion of broadband infrastructure and the widespread adoption of digital devices (more details can be found in the Supplementary file).

Table 2 reports the results from the fixed-effects negative-binomial regression models. First, household income displays a similarly stable, positive association with DCC in Model 1 (Baseline model: DCC, all SES indicators, internet use, and full control variables). One-unit rise in log of household income raises DCC by about 3.1% (incidence rate ratio, IRR=1.031; 95% CI: 1.023–1.040, p<0.001). Increases in the year of schooling result in higher DCC (IRR=1.013; 9 5 % CI: 1.009–1.017, p<0.001). Occupational type also matters: relative to workers, farmers (IRR=0.873; 95% CI: 0.844–0.902, p<0.001) smoke significantly less. The non-employed males have the lowest levels of DCC (IRR=0.587; 95% CI: 0.562–0.615, p<0.001), underscoring the workplace as a key smoking-promoting environment. Internet use is also significantly associated with DCC (IRR=1.034; 95% CI: 1.027–1.040, p<0.001).

To assess how social structure conditions the effect of Internet use on smoking, Model 2 (Main effects + household income × Internet use interaction) introduces an interaction term between household income and Internet use. It shows that household income amplifies the Internet–smoking correlation (IRR=1.005; 95% CI: 1.001–1.009, p<0.01); wealthier men experience a larger increase in daily cigarette consumption for a given increment in Internet use. Model 3 (Main effects + years of schooling × Internet use interaction) further includes an interaction term between years of schooling and Internet use. Each additional year of education attenuates the positive correlation between Internet use and DCC by 0.2% (IRR=0.998; 95% CI: 0.996–0.999, p<0.001); the smoking-inducing effect of Internet use is thus weaker among highly educated men in China. Model 4 (Main effects + non-employment status × Internet use interaction) tests the interaction effect of employment status on the Internet-smoking relationship. It shows that non-employed men smoke fewer cigarettes than the employed for a given increment in Internet use (IRR=0.962; 95% CI: 0.947–0.977, p<0.001). Model 5 (Full model: all main effects + three SES–Internet interaction terms) includes these three interaction effects on DCC. The results are similar to those in the previous three models, indicating that the interaction effects of SES with Internet use are relatively robust.

DISCUSSION

In light of China’s ongoing lifestyle transformation and the entrenched alcohol- and tobacco-friendly culture, this study aims to explore how SES and Internet use interact to shape the daily cigarette consumption of Chinese men, using a secondary individual-level panel dataset from the China Family Panel Study (2014–2022).

This study reveals that household income exhibits an ‘anti-gradient’ pattern, also observed in developing countries^8,9. Rising household income and higher level of education are both associated with increased cigarette consumption. While schools emphasize the harms of tobacco and maintain a relatively low smoking prevalence, the transition from student to employee exposes Chinese men to new social settings where cigarettes often function as social currency, thereby increasing their likelihood of smoking.

We noted that individuals out of the labor market smoke significantly fewer cigarettes than their employed counterparts, highlighting the workplace as a crucial context that promotes smoking²⁶. The occupational gradient is primarily driven by the disparity between farmers and non-agricultural workers; within the non-agricultural sector, differences across occupational strata are negligible. Moreover, education weakens the association between Internet use and smoking. Each additional year of schooling reduces the marginal effect of Internet use on cigarette consumption. Highly educated men tend to apply the Internet for self-improvement and selectively screen out pro-tobacco content. Further analysis indicates that individuals with a lower level of education are more vulnerable to detrimental online social interactions²⁷. Therefore, digital health initiatives and clinician-patient online interaction programs should place greater emphasis on the subgroup with lower level of education.

Our study also noted that household income amplifies the growth in cigarette consumption driven by Internet access. As the income of male consumers rises, their online activities, including online social interaction and digital entertainment, lead to greater cigarette consumption. Consequently, individuals with low income who do not utilize the Internet, demonstrate the lowest level of cigarette consumption. These findings are inconsistent with prior research that reported the protective function of SES in addiction risk²⁴. Therefore, Internet-based smoking cessation interventions with precise targeting and advanced digital smoking cessation tools should prioritize male internet users from higher socioeconomic backgrounds²⁸.

We noted that the effect of internet use on smoking behavior is significantly weaker among the non-employed population. The majority of this group consists of students, unemployed individuals, and retirees, who typically have a more limited scope of online activities. As a result, they are exposed to fewer tobacco advertisements and pro-smoking stimuli, which reduces the strength of the association between Internet use and cigarette consumption. Policymakers should prioritize safeguarding the newly employed individuals from tobacco-related content, to maintain their relatively low risk of smoking.

CONCLUSIONS

This study elucidates the complex interplay between SES, Internet use, and cigarette consumption among Chinese men. It reveals an ‘anti-gradient’ pattern where higher household income and education level correlate with increased smoking, largely driven by workplace socialization and the role of cigarettes as social currency. Crucially, it demonstrates that SES significantly moderates the impact of digital engagement on smoking habits. Education serves as a protective buffer, enabling individuals to filter out pro-tobacco content, whereas higher income exacerbates the smoking-inducing effects of online entertainment and social interaction. Conversely, non-employed individuals exhibit lower consumption due to restricted online exposure.