Tuberculosis (TB) is one of the leading causes of death in the world and remains a major public health burden in many developing countries [1]. From worldwide epidemic of TB in 2006, around 9.2 million new cases and 1.7 million deaths were expected every year, of which 0.7 million cases and 0.2 million deaths were in HIV-positive people [2].

As a risk factor of TB, tobacco smoking has increased substantially over the past three decades, especially in developing countries [3, 4]. Globally, TB and smoking are simultaneously increasing, both of which could damage the lungs, and interact at an immunologic and cellular level [5]. Studies investigating the association between smoking and TB have been published since 1918 [6]. Both passive and active exposures to tobacco smoke have been shown to be associated with TB infection and the transition from being infected to developing TB disease. Moreover, cigarette smoking is also associated with the prognosis of TB. A cohort study conducted in Hong Kong found that significantly more current smokers developed TB and subsequently died within the follow-up period than ex-smokers and never-smokers [7]. Thomas investigated predictors of recurrence among TB patients in South India and showed that a higher relapse rate was independently associated with smoking (OR: 3.1, 95% CI: 1.6–6.0) [8].

However, the association between smoking and TB, which has been shown to exist in different studies with different ethnic background, has not yet received sufficient attention in terms of TB care standards and research. Though both smoking and TB are targeted by major international prevention and control efforts, there has been little research on the measures and effects of smoking cessation among TB patients [5].

In the present study, we aim to investigate the cigarette smoking behavior among TB patients and smoking cessation after diagnosis, and to identify factors associated with smoking relapse among quitters. It might be helpful to provide policy directions on tobacco control for current TB program.

Apart from HIV/AIDS, tobacco smoking is the only major cause of death that is increasing rapidly [10]. It is estimated that smoking will cause about 10 million adult deaths from all causes in 2030 and most of the increased tobacco-related deaths will take place in Asia, Africa and South America [10]. China has the largest production and consumption of tobacco worldwide [11]. Although the government has announced its decision to ratify the WHO's Framework Convention on Tobacco Control and series of measures have already been taken, it is still at the beginning of its 'long way' towards improving the population's health status by reducing tobacco consumption in China [12].

Tobacco contains more than 4,500 compounds in the particulate and vapor phases which comprise five known human carcinogens and many toxic agents [13]. Long-term inhalation of tobacco smoke alters a wide range of immunological functions, resulting in significantly increased risk of heart disease, lung cancer, microbial infections and delayed recovery from these diseases [13]. Though the underlying biological mechanism is unclear, strong associations between tobacco smoking and TB have been proved in several areas [14]. Results from our study also corroborated these former reports. Furthermore, a dose-response relationship between cigarette smoking and TB was also demonstrated in the present study: with the increase of daily cigarettes consumption and duration of smoking, the risk for TB also increased accordingly.

Given that smoking is indeed causally associated with TB disease, the clinical manifestations could also be influenced by the underlying mechanism of its action [15]. A cohort of newly diagnosed TB cases were followed up from their discharge after completion of treatment and it was found that smoking was associated with the relapse of TB with OR of 2.53 (95% CI: 1.23–5.21), even after adjustment for the socioeconomic variables [16]. Although there is lack of evidence on the direct effects of smoking cessation on TB treatment outcomes, available data suggest that smokers are less adherent to TB treatment, and thus at higher risk for default and persistent infectivity [5, 17]. In the present study, we also found that patients without smoking cessation had higher risk for non-adherence to anti-tuberculosis treatment. Non-adherence has been proved in several studies to contribute to prolonged infectiousness, drug resistance, relapse and death [18]. Thus, intervention for tobacco smoking is not only a public health issue among general population, but also an essential activity in TB control program.

It should be noticed that behavior of tobacco use is very difficult to change, even with medicinal aids for cessation. Only a small proportion of smokers stop smoking successfully on their own [19]. Thus smoking cessation support should be incorporated with TB control programs. As attitudes and knowledge can change, change of smoking behaviors needs to be reinforced regularly. Repeated brief cessation advice has been shown to be a feasible and inexpensive addition to routine TB case management [20]. Thus, offering advice to TB patients when they seek healthcare can influence unhealthy behaviors. The central strategy recommended by WHO for controlling TB is DOTS. However, smoking cessation has not been involved in the DOTS framework in China. TB patients lack access to smoking intervention services and social supports, resulting in lower cessation and higher relapse rate. It is essential to revise current TB treatment guidelines and provide regular medical advice on smoking behaviors involved in standard practice of DOTS and other TB control programs. However, additional research is needed in the form of pilot study to determine the feasibility of this intervention, random controlled study to test methods in various settings, and evaluation study to determine effects after widespread application of such practices [21].

One intriguing finding from the present study was that the higher risk of smoking relapse occurred in the period of 6–9 months and 12–15 months after diagnosis. It might be related with the traditional anti-tuberculosis treatment period and patient's recovery from disease. Thus interventions on smoking cessation should be focused on such special risk periods.

There are several potential limitations of this study. Firstly, our findings are based on the results of observational investigations. Though we have performed multivariate regression model and stratified analyses to control the potential confounders, other factors including economic status, intensive contact with TB patients, nutrition intake, as well as exposure to other's smoke may also confound the association between smoking and TB. Secondly, we enrolled TB patients with phone number and interviewed those who were successfully contacted through telephone. Selection bias should not be neglected. Thirdly, recall bias would also influence the results. Compared with healthy controls, TB patients could attribute disease to smoking and aggrandized it, resulting in the overestimate of smoking effects. Fourthly, a major limitation of this study was that smoking status was based on the patient's self-report rather than the detection results of nicotine levels. On one hand, some patients may decline to admit to smoking, especially if a connection between the disease and smoking has been accepted by people in the community, which will underestimate the relation between cigarette smoking and TB. On the other hand, TB patients would be more likely to falsely admit to quitting smoking for fear of disappointing the interviewers, resulting in the overestimate of smoking cessation proportion among TB patients.

Cigarette smoking is associated with TB in the Chinese. Interventions of smoking cessation among patients were insufficient in rural areas of China. Physicians and DOTS providers should be actively involved in smoking cessation activities. Regular and repeated medical advices on smoking behaviors are recommended to be included in DOTS practice.