Settings
The Asia Best Workplace Mainland China (ABWMC) programme was a cross-sectional survey to support companies in building healthy workplaces through policy, infrastructure, and culture. The ABWMC programme was designed by Peking University and organized by the American International Assurance Company. We invited companies to join the programme using a purposive selection method. The inclusion criteria for participating companies were as follows: (1) legal companies registered in China; (2) at least 100 full-time employees; and (3) agreement to participate in the programme.
Population
We used the 2018 and 2019 ABWMC programme data. The total sample size was 29,148, of which there were 14,195 from the 2018 survey and 14,953 from the 2019 survey.
Sampling
The human resource departments of each company delivered the questionnaires to all employees. All employees who were (1) aged 18 years old or above and (2) full-time employees were invited to participate in this programme.
Data collection procedure
A four-stage method was applied in this survey. In the first stage, experts at Peking University designed standardized questionnaires, including sociodemographic information, smoking-related behaviour and quitting intentions (Additional file 1: Research Questionnaire). In the second stage, an online survey system was developed by Ipsos Inc., and a specific internet link was generated. The self-check function of the online survey system automatically identified missing data, logical errors and illegal characters. In the third stage, the Human Resources Departments of each participating company delivered the internet link to all employees. After signing the informed consent online, the employees completed and submitted the questionnaire. All participants knew that the statistical analyses would be anonymously conducted, and their information would be used for research purposes. At the final stage, the submitted questionnaires were reviewed by the staff at Peking University, and the respondents were contacted for clarifications if any problems were detected. This study was approved by the Peking University Institutional Review Board (number: IRB00001052-18,055).
Measurements
SHS exposure and workplace SF policies
The presence of SHS exposure was measured using the question “How many days per week do you usually suffer from SHS exposure at the workplace for more than 15 min a day? A: almost every day; B: 4–6 days; C: 1–3 days; D: never”. Respondents who answered D were considered to be working without SHS exposure. The presence of a workplace SF policy was measured using the question “Does your company have SF policies? A: no SF policies; B: SF policies that permit smoking in parts of the indoor area; C: SF policies that completely ban smoking inside the building; D: I have no idea”. Respondents who answered C were considered to be working under an SF policy.
Smoking and quitting intentions
Smoking was measured by the following question: “Do you smoke now? A: yes, every day, B: yes, only occasionally, C: I have quit smoking, D: never”. Respondents who answered A or B were considered current smokers. Quitting intentions were measured using the question “Are you going to quit smoking? A: yes, within a month, B: yes, within 6 months, C: yes, but not within 6 months, D: no plan for quitting”. Respondents who answered A, B or C were considered intending to quit smoking.
Smoking harm awareness
Smoking harm awareness was measured by the following question: “To the best of your knowledge, which diseases can be caused by smoking? A: stroke, B: heart disease, C: lung cancer, D: cardiovascular disease, E: chronic obstructive pulmonary disease, F: asthma, G: I don’t know.” Respondents who chose all answers from A to F were considered to have smoking harm awareness.
Other control variables
In addition to these key variables, other variables were collected as control variables, such as sex, age, marital status, ethnicity, education, chronic diseases, job position and night-shift duty.
Data analysis
Our data have a hierarchical structure; therefore, we first tried to use hierarchical linear modelling by setting individual- and company-level factors. This type of analysis will consider that workers' responses are correlated within companies. We ran four standardized models (null model, random coefficients regression model, intercepts as a model, slopes as an outcomes model). However, when we finished the null model, we found that the intraclass correlation coefficient (ICC) was too low (0.051, lower than 0.059), which indicates that only approximately 5.1% of the total variation was attributable to differences among companies/clusters [18]. In other words, we can use the usual method to perform analyses. Therefore, we used logistic regression for our statistics.
Logistic regression was used to estimate the association of workplace SF policies with individual smoking and quitting behaviours. The specification of our empirical model was as follows:
$$\mathrm{SHS exposure }\left(\mathrm{or smoking}\right)= {\upbeta }_{0+}{\upbeta }_{1} \mathrm{SF policy}+\upgamma {\mathrm{X}}_{\mathrm{i}}+{\upgamma }_{t}+{\varepsilon }_{it}.$$
The dependent variable was either SHS exposure or smoking. The explanatory variable SF policy was a dummy that indicates whether an employee was working for a company with an SF workplace policy (if yes = 1; otherwise = 0). \(\gamma Xi\) is a variety of other control variables; \({\gamma }_{t}\) is the year-fixed effects to control for year-specific factors, and \({\varepsilon }_{it}\) is the error term. We also separately performed regression with the sample of each year.
To determine whether such a policy had an association with smoking harm awareness and quitting intention, we changed the dependent variable and ran the following regression:
$$\mathrm{SHA }\left(\mathrm{or QI}\right)={\upbeta }_{0+}{\upbeta }_{1} \mathrm{SF policy}+\upgamma {\mathrm{X}}_{\mathrm{i}}+{\gamma }_{t}+{\varepsilon }_{it.}$$
Our analyses used the responses to the questionnaire for whom the variables of interest were available, with no imputation for missing data. All statistical analyses were performed using SPSS 19.0.
Role of the funding source
The funders of the study had no role in the design of the study, collection, analysis, or interpretation of the data, or the writing of the paper.