Adolescents who abuse tobacco are more vulnerable to long-term nicotine dependence, and more than 75% of current smokers started smoking during puberty^[1]. This early initiation is particularly concerning because the prefrontal cortex, which governs executive functions and complex behavior, is not fully mature until age 25, making it especially susceptible to the neurotoxic effects of smoking^[2]. The smoking rate among Chinese secondary school students decreased from 5.9% in 2019 to 4.2% in 2023^[3]. However, easy access to cigarettes remains a critical obstacle to further reducing the smoking rate^[4]. China’s tobacco control strategy goes beyond general population measures to include specific policies protecting minors, such as the Tobacco Monopoly Law (1991) and the Law on the Protection of Minors (2006), which explicitly prohibit tobacco sales to minors. A 2020 revision of the Law on the Protection of Minors, which came into effect in June 2021, introduced specific penalties for non-compliance and significantly enhanced the enforceability of the legislation^[5].

Recent studies have primarily focused on the availability of tobacco products near schools in highly developed cities. For example, an average of 3 tobacco stores were found within 100 m of each school in Changsha^[6]. In addition, the China National Youth Tobacco Survey (NYTS) revealed that the percentage of secondary school smokers who succeeded in purchasing tobacco decreased from 83.3% in 2019 to 71.7% in 2023^[5,7]. Despite these indications, there is still limited evidence regarding whether policies banning tobacco sales to minors are effectively enforced. To address this gap, we used data from the 2019, 2021, and 2023 China NYTS and a logistic difference-in-differences (DID) regression analysis^[8] to examine the impact of the revised tobacco laws implemented in June 2021. Our findings provide key evidence regarding the effectiveness of bans on tobacco sales to minors and may inform strategies to prevent underage smoking in China.

The China NYTS was a nationally representative survey that covered 31 provinces of mainland China and used a multi-stage stratified cluster random sampling design. Additional details of this survey have been published elsewhere^[4]. The 2019, 2021, and 2023 waves were conducted in the same schools between September and December of each respective year. All three surveys were approved by the Institutional Review Board of the Chinese Center for Disease Control and Prevention (2019: No. 202008, 2021: No. 202110, 2023: No.202301).

The main point of interest for our DID analysis was a revised tobacco policy implemented in June 2021. A two-category variable was created for the time period: the pre-policy revision group (2019–2021) and the post-policy revision group (2023), with the policy change period beginning in the calendar year of implementation. We selected 17- and 18-year-olds based on their similar physical characteristics and differing legal eligibility for tobacco purchase. The 17-year-olds were assigned to the intervention group and the 18-year-olds to the control group.

The primary outcome, denial of cigarette purchase due to age, was assessed by asking students who had bought cigarettes in the past 30 days, “During the past 30 days, did anyone refuse to sell you cigarettes because of your age?” Binary (yes/no) responses were recorded. Covariates included weekly pocket money, categorized into ≤ 10, 11–30, 31–50, or > 51 RMB; smoking behavior, where respondents who had not smoked in the past 30 days were defined as nonsmokers, with further categories for frequency (1–9, 10–19, and 20–30 days) and intensity (≤ 5, 6–10, 11–20, > 20 cigarettes per day) for smokers; and the unit of cigarette purchase (pack, individual sticks, or carton) among current smokers.

All analyses accounted for the complex sampling design by using a three-step weighting scheme. The point estimate and 95% confidence interval (CI) for each parameter was calculated using the final weights. Differences in denial rates across demographic and smoking-related subgroups (including gender, residence, geographic region, school type, pocket money, cigarette use, smoking frequency, smoking intensity, and unit of purchased cigarettes) were assessed using the Rao-Scott test. These same variables were included as covariates in the adjusted regression models. Linear regression models were applied to identify the change in the denial rate from 2019 to 2023. The Jackknife replication method was used to calculate the relative percentage change.

The effect of the policy revision was estimated using a DID approach that compared pre-post (2019–2021 vs. 2023) changes in denial rates between 17-year-olds (the intervention group, protected by the law) and 18-year-olds (the control group, not protected). The DID estimate was derived from the interaction term between group and time variables in the logistic regression models. This design relied on the critical identifying assumption that, in the absence of the policy, outcome trends for both groups would have followed parallel paths^[9]. We developed two types of models for each relationship, running both unadjusted and adjusted versions. Model 1 included all students who reported purchasing cigarettes in the past 30 days. Model 2 restricted the sample to those who had both purchased and smoked cigarettes in the same period. Analogous models in the sensitivity analysis compared individuals aged 16–17 to those aged 18–19. All analyses were conducted using SAS 9.4. The threshold of significance was P < 0.05 and all tests were two-sided.

The China NYTS surveyed 288,192 secondary school students in 2019 (response rate: 94.8%), 269,250 in 2021 (response rate: 95.9%), and 261,454 in 2023 (response rate: 95.7%). Across all three waves, the mean age of participants was 15 years (range: 13–16). In 2023, among all secondary school students who purchased cigarettes during the previous 30 days, 29.6% reported being denied purchase due to age, which is lower than previously reported in Thailand and Brazil. The denial rates were lowest in Zhejiang and Shanghai for secondary and middle school students and lowest in Hubei, Jilin, and Henan for high school students (Figure 1).

Figure 1.  Denial rates in different provinces of China 10.10.

Between 2019 and 2023, the denial rate increased by 13.9% (95% CI: 11.4%, 16.3%), with a 7.4% (95% CI: 4.6%, 10.1%) increase occurring between 2021 and 2023. The absolute increase in the denial rate was greater for middle school students (8.4%, 95% CI: 4.7%, 12.0%) than for high school students (5.6%, 95% CI: 2.1%, 9.0%) (Table 1, Supplementary Table S1), despite the latter having a significantly higher baseline smoking prevalence^[10]. This discrepancy may indicate that enforcement has been either stricter for middle school students or that retailers are more likely to comply when facing younger adolescents. Given that rigorous age verification is a cornerstone of effective enforcement—as demonstrated by Japan’s successful proof-of-age system for tobacco vending machines—the development and adoption of an electronic verification system in China could be a pivotal next step to sustain and amplify these gains. The development of such a system in China would likely enhance the accuracy of age verification processes.

While denial rates for 17- and 18-year-olds did not differ significantly in 2019 and 2021, a significant gap emerged by 2023 (23.1% vs. 19.1%). The increase in the denial rate for 17-year-olds from 2021 to 2023 was significant (5.3%, 95% CI: 1.2%, 9.3%), whereas the change for 18-year-olds was not statistically significant (3.4%, 95% CI: –2.5%, 9.2%) (Table 2). Although several policies banning tobacco sales to minors have been implemented since 1999 in China, more than 75% of 17-year-olds still successfully purchased cigarettes in 2023. Internationally, raising the legal purchasing age is an increasingly adopted strategy to protect youth, as seen in New Zealand’s landmark generational ban. We suggest that China promptly raise the legal age for purchasing tobacco.

The event studies showed that pre-policy trends in denial rates between 17- and 18-year-olds were relatively parallel (Supplementary Figure S1), validating the DID design. Following the implementation of the revised law, the DID analysis found no statistically significant effect on cigarette purchase denials among the targeted minors. After adjusting for covariates including residence, geographic region, school type, sex, pocket money, smoking frequency, smoking intensity, and tobacco purchasing pattern, the change in the denial rate for 17-year-olds from 2019 to 2023 did not significantly surpass that for 18-year-olds (Model 1: Odds ratio [OR] = 0.84, 95% CI: 0.52, 1.34, P = 0.46; Table 3). This null association was consistent among smokers who had purchased cigarettes in the past 30 days (Model 2: OR = 0.78, 95% CI: 0.48, 1.26, P = 0.31; Table 3). Sensitivity analyses using alternative age groupings yielded similar results (Supplementary Table S2). These results suggest that the enhanced penalties introduced in 2021 have not yet translated into a statistically discernible increase in the denial rate for the protected cohort (17-year-olds) relative to the control group (18-year-olds). To our knowledge, no other concurrent nationwide policy changes would specifically explain a differential shift in denial rates between these ages. The persistently high rate of successful cigarette purchases among minors indicates that stricter penalties alone are insufficient. Effective reduction in youth access to tobacco likely requires more systematic enforcement, such as integrated electronic age-verification systems, to ensure compliance at the point of sale.

This study benefited from a large, nationally representative sample of Chinese secondary school students and was the first to employ a DID design to evaluate this specific policy, enhancing the robustness of our trend analysis. However, several limitations should be considered. The nationwide ban on tobacco sales to minors precluded a traditional control group, so our DID approach instead compared 17- and 18-year-olds. Age was self-reported without verification against birth dates, which may have caused some exposure misclassification and potentially led to an underestimation of the policy effect. Furthermore, our data captured only purchases from conventional retail outlets, excluding tobacco obtained online or through informal sources. Consequently, our findings may not fully reflect overall tobacco access among minors.