Health Policy and Planning Advance Access published online on November 12, 2007
Health Policy and Planning, doi:10.1093/heapol/czm043
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Hierarchical linear modelling of smoking prevalence and frequency in China between 1991 and 2004
1Associate Research Scientist, Pacific Institute for Research and Evaluation, Louisville, KY 40208, USA.
2Professor of Epidemiology, Dean of College of Public Health, Zhengzhou University, 40 Daxue Road, Zhengzhou, Henan 450052, P.R. China.
* Corresponding author. Pacific Institute for Research and Evaluation, 1300 South 4th Street, Suite 300, Louisville, KY 40208, USA. E-mail: tpan{at}pire.org
 |
Abstract |
|---|
 Top Abstract IntroductionMethodsFindingsDiscussionConclusionReferences |
|---|
This study uses the hierarchical linear modelling (HLM) growth curve technique to explore predictors of the change in the prevalence and frequency of cigarette smoking in China between 1991 and 2004. Using nationally representative data, the study introduces a number of previously unanalysed variables at both the individual and the community level. The findings show that a number of factors are associated with the change in both the prevalence and frequency of smoking in China. In addition, there is a trend of decreasing prevalence of smoking in China after the effects of other covariates are adjusted. Finally, the free market cigarette price has an inconsistent relationship with the change in the prevalence and frequency of smoking, which further reveals the daunting task of tobacco control for public health scholars and policymakers in China.
Key Words: Smoking, China, prevalence, predictors of smoking
KEY MESSAGES
|
|
Introduction |
|---|
|
TopAbstractIntroductionMethodsFindingsDiscussionConclusionReferences |
|---|
According to the Chinese Association on Tobacco Control (CATC), China comprises 20% of the world's population but currently produces 42% of the world's tobacco products, consumes 31% of the world's cigarettes (or 1.8 trillion cigarettes per year), and is home to 320 million smokers according to a consensus of various estimates. Secondhand smoke affects 600 million people in China, including 200 million children (Best Foundation 2006). Facing this increasing public health problem and pressure from the World Health Organization (WHO), the Chinese parliament approved the International Treaty on Tobacco Control with the WHO on 28 August 2005, formally declaring the country's intention to enforce tobacco control nationwide.
Unlike smoking in many other countries, cigarette smoking in China is unique in having several interesting characteristics:
- There is a very high prevalence among males (66.9%) and very low prevalence among females (4.9%) (Levy 2006
).
- Smoking prevalence among medical doctors is also very high (at 56%) according to a 1984 survey (CATC).
- There is a rapidly increasing but still low prevalence of smoking among teenagers who, ironically, seem to have healthy (desirable) attitudes toward cigarette smoking. Past studies have shown that less than 10% of youth believe that teen smokers have a glamorous and tough image according to some local case studies (Li et al. 1999; Hesketh et al. 2001).
- The average age of smoking initiation is declining, with an earlier study indicating that it has dropped from 28 to 25 (Yang 1999).
- There is a special group of so-called ‘social smokers’, who smoke only when they are with friends in social settings where peer pressure is very high and cigarette smoking is important to reinforce relationships, but the social smokers abstain from smoking while being alone because of the harmful health effects (Pan 2004).
- There is a strong awareness of smoking as a bad habit (88% of adult smokers believe that it is harmful to their health), yet there is also a persistently very low desire to quit (only 14% would like to quit) among adult smokers (Gong et al. 1995).
- Smoking and cigarette offering is a social necessity and serves a very important social function in building and/or reinforcing friendships in Chinese culture (Pan 2004).
- Expensive cigarettes, along with alcohol products, are very popular gifts exchanged among Chinese families, relatives and those they intend to bribe in hope of returned favours; cigarette offering is frequently used to reinforce or initiate connections in China, where connections (also referred to as Guanxi) are frequently more important than other qualifications to obtain benefits such as housing, promotion and employment (Bian 1994).
While smoking and its connection with various diseases has been well-documented by previous research, epidemiological studies of smoking problems in China have been sporadic (Hu et al. 2005). A number of studies have provided the prevalence of smoking among both the general population and various sub-segments of the population in China. A review of the literature suggests that previous studies have focused on the overall population (Gong et al. 1995), rural-to-urban migrants (Chen et al. 2004), college students (Xiang et al. 1999; Chen et al. 2004), farmers (Cheng et al. 2003), industry population (Wang and Dobson 1992), urban families (Zhu, 1999) and urban workers (Pan 2004). A number of other studies also have explored various aspects of smoking by teenagers (Zhu et al. 1996; Cheng 1999; Li et al. 1999; Zhang et al. 2000; Hesketh et al. 2001; Liu 2003; Cheng 2004; Fielding et al. 2004; Shakib et al. 2005).
Further, extensive research has explored the correlates of smoking prevalence and frequency in China. Gender, education, marital status, types of employer and Communist Party membership (Gong et al. 1995; Pan 2004) have been linked to smoking prevalence in the Chinese population. Maternal smoking, paternal smoking, low paternal education level and a poor self-reported academic record (Li et al. 1999; Zhang et al. 2000; Hesketh et al. 2001) also have been linked to smoking status based on other regional or case study samples. Furthermore, researchers have identified the correlates of smoking frequency among variables such as marital status, household size (Pan 2004), stress and attitudes (Sun and Shun 1995; Xiang et al. 1999), curiosity and loneliness (Xiang et al. 1999). Most of these studies, however, use local cross-sectional data that do not address longitudinal trend and environmental factors beyond the household or peers.
A few small-scale smoking intervention studies have been documented to show varying results (Loke and Lam 2005). Zheng et al. (2004) reported what they believed to be the first evaluation of a smoking cessation programme called ‘Project EX’. The authors observed a 10.5% 30-day quit rate, a 16% reduction in daily cigarette consumption at post-test and a 33% reduction at 4-month follow up. Another study by Yang et al. (2004) showed that the ‘Quit and Win’ intervention achieved a 27.7% 1-year quit rate in select areas based on a random sample of 1298 participants. Abdullah (2005) observed a 52% prevalence rate of intent to quit in Hong Kong. In an intervention targeting parents of sick children in Hong Kong, Chan et al. (2005) reported a 1-month quit rate of 7.5% in the intervention group compared with 2.5% in the control group.
In summary, although the serious nature of the smoking epidemic and the factors associated with smoking have been explored by various studies, the methodology and sample representation vary greatly across studies. Additionally, past studies rarely went beyond the peers of smokers to examine the household and community factors that may be associated with both the prevalence and the frequency of cigarette smoking by individuals. Due to a lack of longitudinal data, few studies have identified the factors associated with change in smoking prevalence or smoking frequency in China. This study was designed to contribute to the literature by examining the predictors of change in prevalence and frequency of cigarette smoking using nationally representative panel data.
|
Methods |
|---|
|
TopAbstractIntroductionMethodsFindingsDiscussionConclusionReferences |
|---|
Data
The study data were collected by the Carolina Population Research Center in collaboration with the Chinese Center for Disease Control and Prevention. In the original study (China Health and Nutrition Survey), a multi-stage, random cluster process was used to draw the original sample surveyed in the provinces of Guangxi, Guizhou, Heilongjiang, Henan, Hubei, Hunan, Jiangsu, Liaoning and Shandong across a 13-year span between 1991 and 2004. Approximately 4000 households were surveyed in 1991, 1993, 1997, 2000 and 2004, with some households dropping out and some households added each year (see Table 1). The replacement of households in subsequent years followed the principle that the sample for each year remained diverse, with variation found in a wide-ranging set of socio-economic factors (income, employment, education and modernization) and other related health, nutritional and demographic characteristics so that it remained nationally representative (see Beydoun and Popkin 2004 for detailed information regarding the data). To be consistent with past literature on smoking research, only those who are 15 years of age or older are included in the analysis.
|
Table 1 shows that only 3222 individuals reported repeated measures data for all five data collection points while all other respondents have data missing in at least one year. In order to make use of all the data available, data from the 5 years were stacked vertically with a time variable created to identify the year in which the data were collected for each individual. This vertical stacking of data eliminates missing value rows for those years in which data are not available; it also allows hierarchical linear modelling (HLM) software to still ‘borrow’ information from the available data in estimating variable coefficients (Bryk and Raudenbush 1992
). The final stacked sample includes 50 106 cases. Measures
Table 2 presents the descriptive statistics of all the variables included in this study. The coding information for these variables is also presented in the table. Note that the time variable in Table 3 is coded so that the year 1991 is equal to –2, 1993 equal to –1, 1997 equal to 0, 2000 equal to 1 and 2004 equal to 2. This effect-coding scheme will create a roughly normal distribution for this time variable to satisfy the normality assumption for multivariate linear models. This variable will show the impact of time on the prevalence and frequency of smoking across the 13-year span after the effects of other variables are adjusted.
|
|
The selection of those variables are partially based on previous literature and partially based on theoretical speculations. Previous literature (Pan 2004
) has indicated that gender and types of occupation are all associated with prevalence of smoking and these variables are included in the model for statistical control. Urban residence is included in the model to test whether difference in residential location is a predictor of the prevalence of cigarette smoking after other variables are statistically controlled. Previous literature has established that education is associated with smoking (Xiang et al. 1999), but other variables such as physical activity at work are rarely analysed, although the literature on Western populations has documented its association with higher prevalence of smoking. The number of smokers in a household was introduced to see whether exposure to smokers in a household is associated with smoking prevalence of others in the family. Community-level variables
Community variables included are: proxy community economic development stage, local free market cigarette price, community size in thousands of households, and average daily wage of workers in a community. A proxy community economic development stage was included in the model to adjust for the variation in economic development stage because China's economic reform during the past decades has created drastic stratification in economic development among different regions. This variable is measured by the road condition of the community: mostly dirt road = 0; mostly paved road = 2. This measure is an effective proxy of economic development stage because in the past two decades, the road conditions of the communities have been closely associated with a symbol of economic development: the mode of transportation. If a community has mostly dirt roads, the implication is that there are mostly animal-driven carts or motorized tractors available for transportation in an agricultural economy. On the other hand, a community reported to have mostly paved roads implies the wide availability of sedan-size cars which requires smoothly-paved roads. Therefore the road condition in a community is a fairly good indicator of a community's economic development stage. The local free market price of cigarettes is a key variable included in the model as the literature has long established the importance of cigarette price and its connection with smoking (Wang et al. 2006). Finally, the average daily wage of workers is the average amount of Yuan a regular worker is expected to earn for each day of work in a community. This variable is included in the model to adjust the variation of wages across communities and also to see the impact of the prevailing wage in a community on cigarette smoking measures.
Analysis
Since the data are nested in nature with individuals nested in communities and since that data were collected from the same people (with some missing data) over time, HLM growth curve modelling techniques were used to analyse the stacked data. The HLM modelling technique was developed for data with a nested structure: students nested in schools, individuals nested in communities, etc. Because the individuals nested in a school or community share the same characteristics of the same level 2 unit (school or community), there may be potential non-independence that may negatively affect the estimated impact of individual-level predictors. HLM can correct this problem by allowing the level 1 coefficient to vary randomly at level 2. In addition, the HLM model was designed to allow the researchers to test how level 2 predictors affect the change of level 1 predictor's coefficient across level 2 units. The HLM model is explained in much detail in Bryk and Raudenbush (1992) and Leyland and Goldstein (2001), including all assumptions, technique of estimation and other statistical information.
Note also that all individual-level covariates were measured each time a respondent was surveyed, therefore they are all time-variant and thus do not need to interact with the time variable. The estimated coefficients for each variable represented the impact of that variable over time rather than observed association in cross-sectional data (Bryk and Raudenbusch 1992). This is one of the major contributions of this study: to illustrate whether the variables in the study are associated with the change in the prevalence and frequency of cigarette smoking. It should be noted that community-level data were also time-variant, meaning that they were collected for different years and could change over time. However, the interaction between community-level variables and the time variable also were created to determine whether the impact on the prevalence and the frequency of cigarette smoking impacted the slope (or the coefficient of time) variable.
To ensure statistical validity, collinearity diagnostics were conducted by running linear regression with all the independent variables in the model. The validity check suggests that the tolerance levels were all above 0.50, indicating that there were no collinearity problems among all the independent variables. Since there is no available theory regarding which of the level 1 variables should be set to vary randomly across communities, we first allowed all coefficients of individual-level variables to vary randomly at the community level to estimate a random component for each variable. Then only the statistically significant random components were left to vary randomly at level 2, while the others were fixed or assumed to be zero (i.e. not varying randomly across communities).
Only odds ratios are presented for the smoking prevalence model because the logit coefficients are only intuitively meaningful while odds ratios can show not only the direction of the impact or association by predictors, but also the extent of the impact or association as they can also be interpreted as effect sizes in logit models that analyse dichotomous dependent variables. If an odds ratio is larger than 1, it would indicate a positive impact of the predictor on the probability of being a smoker, while a smaller than 1 odds ratio would indicate a negative impact. Since smoking prevalence is a dichotomous variable, the Bernoulli distribution setting was selected for this variable in HLM. Since the number of cigarettes smoked is a count variable that may not follow a normal distribution, a Poisson distribution setting was chosen in the HLM software when analysing this dependent variable. Effect sizes in terms of Cohen's correlation r are presented to provide better comparison across studies and to better assess the extent of impact by independent variables.
|
Findings |
|---|
|
TopAbstractIntroductionMethodsFindingsDiscussionConclusionReferences |
|---|
Table 3 presents the odds ratio and the level of statistical significance for the HLM growth curve model, with current smoking prevalence and smoking frequency as the dependent variables. Note that, overall, the prevalence rate of smoking was 31%, 30%, 28%, 29% and 27% for 1991, 1993, 1997, 2000 and 2004, respectively, showing a slight declining trend.
The model for the prevalence of smoking shows that, over time, older age is associated with an increase in the probability of smoking (odds ratio = 1.23, P 
0.01). Over time, males are 67 times more likely to smoke than females (P 0.01). As expected, urban residence (odds ratio = 1.14), currently working (odds ratio = 2.53), and having a second job (odds ratio = 1.50) have contributed to the increase in the probability of smoking among the respondents in the 13-year span between 1991 and 2004 (P 0.01). Although higher number of smokers in a household is a positive predictor of an increase in smoking prevalence (odds ratio = 2.02, P 0.01), larger household size is a negative predictor of an increase in smoking prevalence (odds ratio = 0.92, P 0.01). Conversely, community size seems to contribute to the increase in the probability of smoking (odds ratio = 1.02, P 0.01). The free market local cigarette price seems to increase the probability of smoking (odds ratio = 0.97, P 0.05) as its main effect, but it interacts with time to cause a decrease in the probability of smoking (odds ratio = 0.97, P 0.05) over time.
The second model in Table 3 analyses the impact of the same variables on the frequency of cigarette smoking: number of cigarettes smoked per day. Age (r = 0.11), male gender (r = 0.30), currently working (r = 0.12), having secondary occupation (r = 0.09), and higher number of smokers in a household (r = 0.13) are all positively associated with a higher frequency of smoking (P 0.05) over time. On the other hand, urban residence (r = –0.02), household size (r = –0.12), community size (r = –0.08), and average community daily wage (r = –0.02) all have a negative impact on the frequency of smoking in China (P 0.05).
|
Discussion |
|---|
|
TopAbstractIntroductionMethodsFindingsDiscussionConclusionReferences |
|---|
Although previous literature established the drastic difference in smoking prevalence across the gender line (Levy 2006
), it is still surprising to see that after adjusting for the effects of other variables, the net impact of male gender on the change in smoking prevalence is much greater than expected, suggesting that smoking among males is a deep-rooted problem that future prevention or intervention programmes must address (Zheng 2004). It is interesting to see that the level of physical activity at work is not a statistically significant predictor of change in the prevalence of smoking, while urban residence is a significant predictor, implying that smoking is equally prevalent regardless of the level of physical demands at work over time, and urban residents have a higher probability of smoking compared with their rural counterparts. It is encouraging to see that the time variable has a statistically significant negative effect on the prevalence of smoking, suggesting that over time the model predicts a decrease in smoking prevalence after the effects of other variables are statistically controlled. This declining trend is consistent with the most recent observation (Levy 2006) and is important for China where there are currently more than 300 million smokers who consume one-third of the cigarettes in the world.
The community-level variables such as economic development stage differences and wage differences do not seem to influence the prevalence of smoking over time, and the interaction terms are not statistically significant either. This result suggests that smoking is a deep-rooted problem in Chinese culture that is prevalent in all different regions with different stages of economic development (Sun and Shun 1995; Yang 1999; Zhu 1999). It is also interesting to observe that as time goes by, community size has a negative impact on the probability of smoking, while free market cigarette price interacts with time to have a negative impact on the probability of smoking. These results suggest that the probability of smoking tends to decrease in larger communities over time. This finding may suggest that, in larger communities, people's health awareness has increased, and thus over time a slight decline in smoking prevalence has occurred, which is also evident nationally in China after decades of a drastic overall increase in the prevalence rate of cigarette smoking.
Interestingly, higher cigarette price is found to be associated with an increase in the probability of smoking (main effect). Over time, however, the interaction between free market local cigarette price and time seems to cause a decline in smoking prevalence. This inconsistent finding may be explained by referencing the culture of smoking in China and the convention economic theory. The positive association between cigarette price and smoking prevalence (main effect) can be explained by the fact that smoking serves a social function (Pan 2004). In Chinese culture, smoking not only serves as a connection-building tool, it is also a show of status, with different brands of cigarettes symbolizing varying social strata. In the initial years when cigarettes were widely available due to mass production, it was common for smokers to try to smoke only the most expensive cigarettes they could afford to demonstrate their social and economic status. Offering more expensive cigarettes to friends and business contacts remains a gesture of heightened hospitality and sincerity in the relationship-building process. One example is that when the Marlboro brand was first introduced in China, higher import tariffs made the brand prohibitively more expensive than local cigarettes. Instead of discouraging sales, the purchase of Marlboro cigarettes was considered a rare treat and became the choice of expensive gifts as bribes or shows of status and respect when offered to friends or business contacts. The positive association between cigarette price and the prevalence of smoking in Table 3 may imply that this culture of smoking/offering expensive cigarettes is still prevalent in China.
Still, China has witnessed a decline in smoking prevalence among males in recent years. The interaction effect between free market cigarette price and time may be due to the fact that as the cigarette market matures and as the price of cigarettes continues to rise, the supply-and-demand economic theory has finally overtaken the culture of smoking in that, over time, the free market price of cigarettes is starting to interact with time to cause a downward direction in the prevalence of smoking.
There are also some inconsistent results when comparing the two HLM analysis models in Table 3. Urban residence, which is positively associated with the increase in smoking prevalence, is positively associated with the decrease in smoking frequency, suggesting that while urban residents tend to be more likely to smoke over time, they tend to smoke less compared with their rural counterparts. Also inconsistent is the time variable. Although time is associated with a decrease in the probability of smoking prevalence over the 13-year span, it is positively associated with the frequency of smoking despite the effect size being very small. This result suggests that, over time, although smoking prevalence has yielded a much-needed decline, the number of cigarettes smoked per day has increased slightly. Furthermore, although the free market price of cigarettes is positively associated with the prevalence of smoking, it does not have any statistically significant association with the frequency of smoking. Finally, although average daily wage in a community does not impact the prevalence of smoking, it does have a small negative impact on the frequency of smoking over time.
Note that most of the effect sizes are small in the model with the frequency of smoking as the dependent variable, suggesting that those variables’ impact on the frequency of smoking is far less than that of the change in prevalence of smoking. This result is consistent with previous literature findings (Pan 2004) which reveal that while some socio-demographic variables are good predictors of smoking prevalence, they do a poor job of explaining the variation in smoking frequency for which the predictors are still not properly identified in the Chinese cultural settings. It is worth noting that the variables having the largest effect on the change in smoking frequency are male gender, currently working and household size.
This study has a few weaknesses. As a secondary data analysis study, the construction of some measures was constrained by the original coding of the survey. Some key variables of interest had to be approximated using the best knowledge of the existing literature. In addition, the lack of household-level factors makes it impossible to explore other family factors that have been documented to affect individual smoking behaviour. Although the calculation of some independent variables may not be very precise and the lack of previously proven predictors of smoking in the model may be another source of criticism, this study is one of the few rigorous quantitative analyses available and the findings should provide a better understanding of the smoking problems in China.
|
Conclusion |
|---|
|
TopAbstractIntroductionMethodsFindingsDiscussionConclusionReferences |
|---|
Several tentative conclusions can be reached based on the study findings. Overall, the results show that the smoking problem in China is unique to Chinese culture in many ways. Contrary to study results in other countries, level of education, level of physical activity and community economic development stage do not have any impact on the change in either smoking prevalence or smoking frequency. This result confirms that smoking in China is a very widespread problem across different social classes and regions at different stages of economic development. The interesting impact of free market cigarette price on smoking prevalence over time further corroborates that smoking is a culture-driven issue in Chinese society. Any future prevention or intervention programme must be very sensitive to the social and cultural function of cigarette smoking. The lack of association between level of education and the change in cigarette smoking measures over time may not mean that educational intervention approaches will prove ineffective. Proper education on the harmful health effects of smoking, stress management skills, refusal skills and alternative relationship-building techniques (other than smoking or cigarette offering) may still be an effective strategy to reduce both the prevalence and the frequency of smoking in China.
Although it is discouraging to observe the positive association between local cigarette price and the prevalence of smoking, it does not necessarily mean that increasing the price of cigarettes is not an effective policy intervention. It is rather encouraging to see that free market cigarette price does interact with time to contribute to the decline in the prevalence of smoking. Currently, cigarette prices in China are still very low due to lack of government restriction and the presence of hundreds of different brands and cigarette production companies. Furthermore, there is no minimum-age law regarding cigarette purchase or smoking, although some policy interventions such as banning of cigarette advertisements have been in place. It is possible that, combined with the establishment of a minimum-age law for cigarette purchase, stricter enforcement of no-smoking laws and an increase in cigarette taxation may result in reduced access to cigarettes, which in turn may lead to lower prevalence and frequency of smoking.
Finally, the overall results present a grave challenge for practitioners and health officials to reduce smoking in China as smoking is rapidly becoming the most widespread public health problem in the country. The inconsistent and conflicting impact of some variables presented in Table 3 make it a particular challenge to combat the smoking problem in China where smoking has a strong root in cultural practices. Future research should explore various prevention/intervention strategies such as family-based, policy-based or educational approaches to examine the feasibility of different prevention and intervention programmes. Past literature has offered little evidence on what factors to target. The longitudinal analysis in this study has identified a number of factors that are associated with the prevalence and frequency of smoking over time. Some of these factors may be targeted by future prevention/intervention programmes. Despite its limitations, such as a lack of other household variables and geographical coverage, this study can be useful for policymakers or practitioners in formulating effective strategies to reduce smoking in China.
|
References |
|---|
|
TopAbstractIntroductionMethodsFindingsDiscussionConclusionReferences |
|---|
Abdullah AS, Mak YW, Loke AY, Lam TH. Smoking cessation intervention in parents of young children: a randomised controlled trial. Addiction (2005) 100:1731–40.[CrossRef][Web of Science][Medline]
Best Foundation. Secondhand smoke affects 600 million in China (2005) accessed 15 April 2006. Online at: <http://www.jointogether.org/news/headlines/inthenews/2006/secondhand-smoke-affects-600.html>.
Beydoun MA, Popkin BM. The impact of socio-economic factors on functional status decline among community-dwelling older adults in China. Social Science and Medicine (2004) 60:2045–57.[Web of Science]
Bian Y. Guanxi and the allocation of urban jobs in China. The China Quarterly (1994) 140:971–99.
Bryk AS, Raudenbush SW. Hierarchical linear models: applications and data analysis methods (1992) Newbury Park: Sage.
Chan SS, Lam TH, Salili F, et al. A randomized controlled trial of an individualized motivational intervention on smoking cessation for parents of sick children: a pilot study. Applied Nursing Research (2005) 18:178–81.[CrossRef][Web of Science][Medline]
Chen X, Li X, Stanton B, et al. Cigarette smoking among rural-to-urban migrants in Beijing, China. Preventive Medicine (2004) 39:666–73.[CrossRef][Web of Science][Medline]
Chen X, Li X, Stanton B, et al. Patterns of cigarette smoking among students from 19 colleges and universities in Jiangsu Province, China: a latent class analysis. Drug and Alcohol Dependence (2004) 76:153–63.[CrossRef][Web of Science][Medline]
Cheng TO. Teenage smoking in China. Journal of Adolescence (1999) 22:607–20.[CrossRef][Web of Science][Medline]
Cheng TO. Peer, parental, and commercial influences on cigarette smoking among Chinese youths. Journal of the National Medical Association (2004) 96:691–2.[Medline]
Cheng Y, Jin Y. Prevalence survey of smoking pattern among peasants in China. Wei Sheng Yan Jiu (2003) 32:366–68. 406.[Medline]
Chinese Association on Tobacco Control. Introduction to tobacco control in China (2005) accessed 22 April 2006. Online at: <http://www.cash-prc.com>.
Fielding R, Chee YY, Choi KM, et al. Declines in tobacco brand recognition and ever-smoking rates among young children following restrictions on tobacco advertisements in Hong Kong. Journal of Public Health (2004) 26:24–30.[CrossRef][Web of Science][Medline]
Gong YL, Koplan JP, Feng W, et al. Cigarette smoking in China: prevalence, characteristics, and attitudes in Minhang District. Journal of the American Medical Association (1995) 274:1232–5.
Hesketh T, Ding QJ, Tomkins A. Smoking among youths in China. American Journal of Public Health (2001) 91:1653–5.
Hu J, Galeone C, Lui R, et al. Smoking and lung cancer in Harbin, northeast China. Annals of Oncology (2005) 16:1605–8.
Leyland AH, Goldstein H. Multilevel modeling of health statistics (2001) New York: Wiley.
Levy D. The role of public policies in reducing smoking and deaths caused by smoking in China: results from the China Tobacco Policy Simulation Model (2006) New York: World Health Organization.
Li X, Fang X, Stanton B. Cigarette smoking among schoolboys in Beijing, China. Journal of Adolescence (1999) 22:621–5.[CrossRef][Web of Science][Medline]
Liu X. Cigarette smoking, life stress, and behavioral problems in Chinese adolescent. The Journal of Adolescent Health (2003) 33:189–92.[CrossRef][Medline]
Loke AY, Lam TH. A randomized controlled trial of the simple advice given by obstetricians in Guangzhou, China, to non-smoking pregnant women to help their husbands quit smoking. Patient Education and Counseling (2005) 59:31–7.[CrossRef][Web of Science][Medline]
MacKay J. China's tobacco wars (1992) accessed 22 April 2006. Online at: <http://multinationalmonitor.org/hyper/issues/1992/01/mm0192_06.html>.
Pan Z. Socioeconomic predictors of smoking and smoking frequency in urban China: evidence of smoking as a social function. Health Promotion International (2004) 19:309–15.
Shakib S, Zheng H, Johnson CA, et al. Family characteristics and smoking among urban and rural adolescents living in China. Preventive Medicine (2005) 40:83–91.[CrossRef][Web of Science][Medline]
Sun W, Shun J. Smoking behavior among different socioeconomic groups in the work place in the People's Republic of China. Health Promotion International (1995) 10:261–6.
Wang HK. The impact of tobacco expenditure on household consumption patterns in rural China. Social Science and Medicine (2006) 62:1414–26.[CrossRef][Medline]
Wang WQ, Dobson AJ. Cigarette smoking and sick leave in an industry population in Shanghai, China. International Journal of Epidemiology (1992) 21:293–7.
Xiang H, Wang Z, Stallones L, et al. Cigarette smoking among medical college students in Wuhan, People's Republic of China. Preventive Medicine (1999) 29:210–5.[CrossRef][Web of Science][Medline]
Yang G, Fan L, Tan J, et al. Smoking in China: findings of the 1996 National Prevalence Survey. Journal of the American Medical Association (1999) 282:1247–53.
Yang Y, Jiang Y, Yang X, et al. Analysis on main factors for successful quitting: study on the one-year follow-up for Chinese "Quit and Win" in 2002. Wei Sheng Yan Jiu (2004) 33:478–80.[Medline]
Zhang L, Wang W, Zhao Q, Vartiainen E. Psychosocial predictors of smoking among secondary school students in Henan, China. Health Education Research (2000) 15:415–23.
Zheng H, Sussman S, Chen X, et al. Project EX–a teen smoking cessation initial study in Wuhan, China. Addictive Behaviors (2004) 29:1725–33.[CrossRef][Web of Science][Medline]
Zhu BP, Liu M, Shelton D, et al. Cigarette smoking and its risk factors among elementary school students in Beijing. American Journal of Public Health (1996) 86:368–75.
Zhu Z. Smoking problems in the inhabitants of Beijing: an epidemiological study of 3,000 families. Chinese Mental Health Journal (1999) 13:105–6.
Accepted for publication 13 August 2007.
CiteULike 
Connotea 
Del.icio.us What's this?
This article has been cited by other articles:
|
|
 |
|
  D. Ding, M. F. Hovell, M. Ji, C. R. Hofstetter, P. Zheng, H. Fu, and S. C. Hughes Employment and social "determinants" of smoking in urbanizing China: A representative survey Nicotine Tob Res, July 1, 2009; 11(7): 779 - 784. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 A. Millman, D. Tang, and F. P. Perera Air Pollution Threatens the Health of Children in China Pediatrics, September 1, 2008; 122(3): 620 - 628. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||