The Database of Chinese Important Newspapers http://www.cnki.net has indexed major Chinese newspapers since 2000. The database covered more than 500 newspapers, 7.95 million papers, cover 168 topics and 3600 subgroup until December 31, 2010. But not all newspapers were indexed in 2000; the number of newspapers indexed in the database has increased continually. So the increase of newspapers indexed in the database may confound the analysis of the growth of newspaper coverage. We therefore only searched news articles from 185 newspapers indexed continuously from January 1, 2000 to December 31, 2010(See additional file 1). The search keywords were fixed as 'xiyan'(smoking) OR 'xiangyan'(cigarettes) OR 'yancao'(tobacco) OR 'kong yan' (tobacco control), OR 'wu yan' (tobacco free) OR 'jin yan' (tobacco prohibition) with Chinese and English expansion. Each article obtained from the database was scrutinized to ensure that it related to tobacco control, and all without tobacco control related content were excluded. A total of 1149 articles published in 102 continually published newspapers were thus located and analyzed. This compared with Liu et al's 152 national and 363 local newspapers indexed at any point in their 2000-June 2008 study [11].
Coding
Each article was coded for the newspaper in which it was published and date of publication (year and month), and front page (whether - yes/no -- the article was published on the front page in the newspaper). Two reviewers (JLG and PPZ) read each article and coded for the following variables: tobacco control topic, type of article, and point or view (or slant) regarding tobacco control.
To ensure reliability of the coding process, we selected a 10% random subsample (115 articles) to assess the inter-coder reliability using the Kappa statistic. The median value of κ across the coded variables was 0.96 (tobacco control topic: 0.97, type of article: 0.98, point of view: 0.93 respectively), indicating high inter-coder reliability [12]. In the case of disagreement between the two coders, a third coder was used to help make the final decision.
Tobacco control topics
Tobacco control topics were coded to reflect the six WHO MPOWER categories [13], as well as three additional categories used in previous studies [14, 15]. The topics consisted of the following:
(1) Monitoring: articles about tobacco use including (i) prevalence of tobacco use; (ii) impact of policy interventions on use; and (iii) tobacco industry marketing, promotion and lobbying;
(2) Protection: articles about protecting people from secondhand tobacco smoke, such as the establishment of regulations prohibiting smoking in public places, on public transport, in schools and hospitals;
(3) Offering help: articles on (i) smoking cessation including primary health-care services; quit lines; pharmacological therapy;
(4) Warning: articles focused on the dangers of tobacco use;
(5) Enforcement, enforce bans on tobacco advertising, promotion and sponsorship;
(6) Raising taxes: articles on tobacco
(7) Youth access, purchase and possession;
(8) Large scale smoke free activities, such as the 2008 smoke-free Olympics, the 2010 smoke-free World Expo, and the 20010 smoke-free Asian Games.
(9) Miscellaneous items which do not match any of the aforementioned topics.
Type of article
Articles were classified into three types [4, 16]: (1) news stories (a factual account of an event or issue); (2) editorials (a nonfactual account = opinion of an event or issue written by newspaper staff); (3) letters to the editor (usually written to the newspaper by a member of the community).
Point of view
Point of view-or slant -- which denotes the overall tone of the article [5, 17], was categorized into: (1) Pro-tobacco Articles that would be understood by such a person as conveying a view of smoking as in some way positive, relatively unimportant compared to other issues, or which were critical of laws, policies and individuals advocating restricting or opposing smoking; (2) Anti-tobacco Articles judged as being likely to be read by a person favorably disposed toward smoking control, as enhancing the general view that smoking and those institutions, interests, laws and policies that support it are undesirable; (3) Neutral/balanced. A neutral or balanced article was one either made no value judgments about tobacco use or control, or else provided a clear balancing of such judgments.
Origin of article
We coded articles as being either national or local in origin. Specifically, we distinguished the origin of an article by coding whether the story was covered by a national newspaper or wire service (e.g., People's Daily, Xinhua News Agency) or by a local newspaper (e.g., Beijing daily, Guangdong Daily).
An example of coding:
September 02, 2010 Source: People's Daily Page:8
"No smoking" law comes into effect in south China's Guangzhou
A local smoking-control law came into effect Wednesday in Guangzhou, the capital of south China's Guangdong Province. The law covers all of Guangzhou City. There, smoking in most public places, like offices, conference rooms, halls and elevators, is strictly prohibited. Places of business larger than 150 square meters or having more than 75 seats may designate an area for smokers. Those who break the law will be fined 50 yuan (about 7.35
U.S. dollars) and businesses not meeting their obligations will be fined up to 30,000 yuan.
However, some doubt the law is enforceable.
......
"A quick smoke can take less than one minute. How can an enforcement agency come in such little time?" a different pedestrian asked a Xinhua reporter.
Hu Angang, an economics professor at Tsinghua University, said....... "Putting goals into the national plan shows great political will from state leaders. So the whole of society, especially local governments, will implement it without hesitation. The plan is a binding document for the government," said Hu.
Shanghai began implementing a regulation to ban smoking in 12 types of public places March 1 in an effort to have a smoke-free World Expo.
......
One of the world's largest tobacco producing and consuming nations, China manufactures about 100 billion packets of cigarettes each year.
The article was published in People's Daily on September 02, 2010, and only reported the local smoking-control law in Guangzhou without making value judgments about tobacco use or control, so it was coded as:
Tobacco control topic: protection
Type of article: news stories
Point of view: neutral/balanced
Origin of article: national article
Year: 2010
Month: September
Front page: No
Statistical analysis
The analyses reported here are based on the relative frequency of articles within each of the coded variables. We used chi-square analysis to determine associations between categorical variables as appropriate. Where there were significant associations we examined a contingency table including standardized residuals to determine what was driving the significant result. Standardized residuals convert the difference between the observed value and expected value into a z-score. Standardized residuals larger than two, therefore, indicate large differences between observed values and expected values that drive a significant chi-square result [2]. In addition, Poisson heterogeneity tests were used to test whether the number of articles in each year was the same [18].