Introduction

China has the highest number of tobacco users in the world, and accounts for over 40% of the world’s tobacco consumption.1 The prevalence of current smoking in China in 2010 was 54.0% among men and 2.6% among women, accounting for 318 million adults aged 18 years and older (304 million men and 14 million women).2 Tobacco use is estimated to cause one million premature deaths in China each year, which is projected to rise to two million in 2030 and three million in 2050, unless widespread tobacco control policies and programmes are immediately put into effect.3 Of these premature deaths, about 0.4 million were cancer deaths in 2013.4

Cancer is the leading cause of death in China,5 and smoking is the leading modifiable risk factor for cancer.4 Strong national and local tobacco control policies and programmes are critical to reduce this growing cancer burden in China. However, while China has made advances in tobacco control in recent years, progress has remained limited in scope and effect. In 2005, China ratified the WHO Framework Convention on Tobacco Control.6 Since then, at least 18 cities in China have adopted comprehensive smoke-free laws prohibiting smoking in public places and workplaces.7 In particular, the Beijing Smoking Control Ordinance, which came into effect in June 2015, has provided a model for successful tobacco control efforts.8 9 However, most of the other cities still lack comprehensive smoke-free laws, policies and programmes and continue to face challenges with both implementation and enforcement.7 Tobacco control policies also remain particularly weak in provinces with high smoking prevalence, such as the Yunnan province.10 Moreover, progress on substantive national and local tobacco control has been hampered by strong opposition from the state-controlled tobacco industry.6

Despite geographical variations in smoking prevalence,11 differences in the burden of cancer attributable to tobacco use across provinces in China remain unknown. Province-specific smoking-attributable mortality is particularly valuable for public health advocates and local policy makers to take effective action against tobacco use. Herein, we estimate the number and proportion of cancer deaths attributable to active smoking and second-hand smoking in adults older than 30 years in each of the 31 provinces/autonomous regions/municipalities (hereafter, provinces) in Mainland China.

Methods

Results

In 2014, an estimated 383 167 cancer deaths in adults 30 years or older in China were attributable to smoking (both active and second-hand smoking), with 342 854 of these deaths occurring in men (23.8% of all cancer deaths, 95% CI 23.3% to 24.3%) and 40 313 in women (4.8%, 95% CI 4.5% to 5.2%) (table 1). Of these deaths, about 6180 lung cancer deaths among men (5.9% of lung cancer deaths among non-smokers; 1.4% of all lung cancer deaths) and 20 170 among women (11.5% of lung cancer deaths among non-smokers; 10.2% of all lung cancer deaths) were attributable to second-hand smoking (online supplementary etable 4).

Among men, the absolute (non-standardised) proportion of cancer deaths attributable to smoking ranged from 14.6% in Xinjiang to 26.8% in Tianjin and Shandong; and was over 26.0% in five provinces (Tianjin, Shandong, Chongqing, Heilongjiang and Hunan) and less than 20.0% in five other provinces (Xinjiang, Tibet, Guangxi, Ningxia and Shanghai). Among women, the absolute proportion ranged from 1.8% in Jiangxi to 14.9% in Heilongjiang. In five provinces (Heilongjiang, Tianjin, Jilin, Inner Mongolia and Liaoning) the PAF was >10.3%, but is 6.1% or less in all other provinces (table 1).

The age-standardised and cancer site-standardised PAF among men is 19.0% or higher in all provinces except Xinjiang (15.7%), being highest in Guizhou (26.0%) (figure 1 and online supplementary etable 4). Among women, the standardised PAF ranged from 1.6% in Jiangxi to 9.6% in Heilongjiang. Overall, provinces with the highest PAFs differed by sex. Among men, they are mostly located in the South-West, whereas among women, they are in the North-East (figure 1). Tibet, Ningxia and Xinjiang are among 6 provinces with the lowest standardised PAFs for men but among the 10 highest for women.

• Download figure
• Open in new tab
• Download powerpoint

Figure 1

Rank of age-standardised and cancer site-standardised proportion of cancer deaths attributable to smoking in each province of Mainland China, 2014. The proportion was standardised for each cancer site and age group, with active and second-hand smoking both included.

The pattern of absolute PAFs (figure 2) and age-standardised PAFs (online supplementary etable 4) by cancer site for active smoking is generally comparable across provinces. Guizhou is the province with highest absolute and standardised PAFs for all evaluated cancer sites among men. Tianjin, Heilongjiang, Jilin, Inner Mongolia and Tibet rank as the top five among women.

• Download figure
• Open in new tab
• Download powerpoint

Figure 2

Provincial-level absolute proportion of deaths attributable to active smoking for site-specific cancer in Mainland China, 2014. Five minor sites include: mouth, pharynx, bladder, larynx and pancreas.

The proportion of lung cancer attributable to second-hand smoking is higher in women than in men across all 31 provinces (figure 3). Among men, the PAF is the highest in Heilongjiang, Guizhou, Hainan, Shanxi, Tianjin and Jilin and lowest in Jiangxi. Among women, Guizhou, Shanxi, Gansu, Yunnan and Jilin have a higher PAF, and Guangxi has the smallest. The rankings of provinces in terms of absolute and standardised PAFs for lung cancer attributable to second-hand smoking are very similar (online supplementary etable 4).

• Download figure
• Open in new tab
• Download powerpoint

Figure 3

Provincial-level absolute proportion of lung cancer deaths attributable to second-hand smoking in Mainland China, 2014. The proportion were reported using the non-smokers lung cancer deaths and all lung cancer deaths as references, respectively.

By age group, the PAFs are higher in men aged 55–59 years and women aged 80–84 years for all active smoking-related cancers, while the peak age of PAFs is about 25 years younger for second-hand smoking, that is, 30–34 years for men and 55–59 years for women (online supplementary efigure 3). Using alternative definitions for smoking did not substantially change the results (online supplementary etable 5).

Discussion

In 2014, about 0.4 million cancer deaths (16.8% of all cancer deaths) in adults 30 years and older in China were attributable to active and second-hand smoking. More than 98% of these deaths in men were attributable to active smoking, compared with 50% in women. PAFs were higher in South-West China for men and North-East China for women, consistent with patterns in smoking prevalence. Interestingly, while the PAF for smoking was lowest for men in Xinjiang, Ningxia and Tibet, these three minority autonomous regions were among the provinces with highest PAF for women.

Disparities in PAFs by sex and province appear to be primarily driven by differences in smoking prevalence in corresponding provinces, age groups and sex. Smoking prevalence among Chinese men started to substantially increase in the second half of the 20th century and peaked in the 1990s (61.4%).18 Despite some declines, however, male smoking prevalence is still high (52.9% in 2010).6 However, smoking is much less accepted for women in China and smoking prevalence among women remains relatively low, especially in more traditional, rural areas in South China.19 Smoking among women is higher in North-East China and in urban areas like Beijing and Tianjin, where female smoking is more socially acceptable.20 21 By birth cohort, in the China Kadoorie Biobank Study, smoking prevalence peaked in men born in the 1950s–1960s, while among women, smoking prevalence gradually decreased from the oldest age group (born in the early 1930s) to the youngest age group in that study (born in the late 1960s).3 13 There was a similar pattern in the NNHS data used in this study. Correspondingly, PAFs for active smoking in this study were higher in men aged 55–59 years (born during 1955–1959) and women aged 80–84 years (born during 1930–1934). Cigarettes are still regarded as a common gift and a form of social currency in China, especially in rural areas where awareness of the health effects of tobacco use is lower than in urban areas.22 As in most other countries, smoking prevalence in China is associated with lower socioeconomic and educational levels.23 24 In some metropolitan areas, such as Beijing and Shanghai, smoking ban in public areas was implemented and public support for tobacco control has grown substantially.25 However, in other provinces, notably those where tobacco is grown, such as Yunnan, Hunan and Guizhou, tobacco use is considerably higher.26 27 Thus, attitudes and cultural acceptability of smoking, socioeconomic status and education level, and public sector support for tobacco control policies and programmes, appear to be key determinants of patterns of smoking prevalence, and the resulting cancer burden, among the 31  provinces in Mainland China.

Despite high smoking prevalence, the current proportion of cancer deaths attributable to smoking is lower in China compared with some other countries. For example, while the PAF for lung cancer mortality associated with active smoking in men was around 45% in this study, it is around 90% in USA, the UK, France, Japan and Korea.28 Methodological differences, such as using the smoking impact ratio to calculate the burden of cancer associated with smoking, may have contributed to this variation. However, differences in stages of the cigarette epidemic and competing risks in China versus other countries may be greater contributors to this variation in PAFs.29 In comparison with Western populations where the epidemic is more advanced, the observed RR for lung cancer and several other cancers are quite modest in China (eg, 2.5 in China and about 25 in USA for lung cancer).13 30 However, probably due to an earlier age of initiation, the RRs associated with smoking along with the associated PAFs have increased over the past two decades.3 It is likely that the magnitude of this association will substantially increase in China over time as the full impact of tobacco smoking emerges, a pattern also observed in several Western countries.30 Lung cancer incidence is relatively high among non-smokers in China as compared with other countries, likely as a result of exposure of competing risks factors such as second-hand smoking, ambient air pollution and other sources of indoor air pollution.31 32 Generally, estimates from our study were consistent with previous estimates from China for smoking-related PAFs for all cancers4 13 and site-specific cancers.33 34 However, slightly higher overall PAF estimates for active smoking in 2005 reported by an earlier study (32.7% in men and 5.0% in women)35 could, in part, be explained by higher RRs used in that study. For example, the RR for ever-smoking and lung cancer in that study was 5.7 for men and 5.0 for women, based on a case-control study in Shanghai,35 whereas it was about 2.5 for current smoking in our study, based on a large-scale prospective study.13

Our study is the first analysis that systematically assesses sex-specific, age-specific and site-specific cancer burden attributable to smoking in 31 provinces across Mainland China. However, our study has some limitations. First, we used RRs for specific cancer sites based on reported associations among men due to high uncertainty in reported RRs among women in the China Kadoorie Biobank Study.13 However, the reported RRs for the association between smoking and all cancers were similar in men (1.44) and women (1.42). Further, we took into account smoking intensity to reduce potential bias due to lower smoking intensity among women in China. Second, the RRs for lung cancer death reported by the China Kadoorie Biobank Study were slightly higher than that for incident lung cancer,3 13 but we used RRs for cancer incidence in our study as they were based on higher numbers of cancers. Third, we aggregated all other cancers into one group due to small numbers, and this group included some cancer sites with less than sufficient evidence for a causal association with smoking, which may overestimate the total number of attributable deaths in this study. However, the latter aggregated group constituted only 5.1% of all attributable cancers among men and 4.6% among women, thereby limiting possible overall impact on our findings.

We found that a considerable proportion of cancer deaths in all provinces in China, notably among men, are attributable to tobacco use. However, there are also substantial differences in this proportion across provinces. As the tobacco epidemic remains at an earlier stage in China than that compared with North America and Europe, the full impact of tobacco smoking patterns in recent decades on cancer mortality may have not have been realised yet. Thus, if no tobacco control action is taken to immediately reduce tobacco use with quality implementation of targeted policies and programmes, the burden for cancer and other smoking-related diseases will continue to grow dramatically. As illustrated by the experience of Beijing’s strictest smoking ban, policies and programmes in Beijing appear to be having an impact on reducing smoking behaviour,9 36 this experience could serve as a model for other regions in China, including rural areas and tobacco-growing provinces. Further, the MPOWER tobacco control initiatives recommended by the WHO Framework Convention on Tobacco Control include monitoring tobacco use, protecting people from tobacco smoke, offering help to quit tobacco use, warning about the dangers of tobacco, enforcing bans on tobacco advertising and promotion, and raising taxes on tobacco products. These efforts need high-level support and should be required and promoted at the national and provincial levels, including joint efforts from the local governments, multiple levels of the Chinese Centre for Disease Control and Health Inspection Bureaus, as well as greater support for non-government organisations, private enterprises and civil society organisations to promote strong tobacco control policies and programmes, and help China stem the quickly rising tide of preventable cancer deaths attributable to tobacco use.