EMA for Tobacco Control Policy Research

• Study Design The proposed study is a time-based system-triggered EMA which will collect real-time data about exposures to smoking-related cues and behaviors. With a smartphone app installed in the mobile phones of the participants, they will each day complete at least 3 pairs of EMA prompted randomly for only 2 days (a working and a non-working day). The prompting time will be sampled from the time when the participants are outdoor, so that more exposures to the tobacco HW, POS tobacco displays, and smoking hotspots can be captured. The EMA includes questions of the self-reported exposure to and/or responses to tobacco HWs, POS tobacco displays, smoking hotspots, craving intensity, and other smoking-related outcomes.
• Procedures The clinic counselors will briefly introduce all the clients in the clinics about the study during their visits. The clients who are interested will be referred to the on-site recruitment staff to screen for the eligibility. In the setting of retail shops and smoking hotspots, the recruitment staff will proactively approach the smokers. Eligible participants will be given more details of the study and invited to give a written consent for participation. Then, they will complete a baseline questionnaire administered by the recruitment staff. Afterwards, the staff will assist the participants to install the EMA app in their mobile phones, and ask the participants to select one working day and one non-working day, and then select at least 3 time windows for the EMA on each study day. These time windows can be their outdoor time when they are more likely to be exposed to the smoking-related locations, such as traveling to work, lunch time, or shopping. Then, the staff will set up the EMA schedule in the app for the participants, and train the participants to answer the questionnaire. A pocket-sized instruction card of using the app will be provided. The EMA will not interfere the smoking cessation intervention they receive in the clinic.

To protect the privacy of the participants, several strategies will be adopted. First, participants will be reassured that all data are anonymous in the data collection and the analysis. All data will not be accessed by the clinic counselors. Second, no personal data (e.g. name and telephone numbers) will be collected in the app.

The EMA validity largely relies on the participants' compliance, therefore an attractive incentive is needed. If they complete all the EMAs in the 2-day period, they will be given HK$200 dollar.

After the data collection, the participants will be reminded to uninstall the app. They will be contacted to receive the incentive and complete the last questionnaire face-to-face.

3 . EMA app The EMA app will be developed based on the app used in our aforementioned pilot randomized trial, which has a user-friendly interface and is easily operated. Users need to input at least 3 time windows (each for at least an hour) in each study day. In each time window, the app will firstly prompt the user at a random time to answer a set of questionnaire in the app, including questions on (1) exposure and responses to tobacco HW, (2) exposure to tobacco POS displays, (3) exposure to smoking hotspots, and (4) smoking and quitting outcomes. After 30 minutes of completing the first questionnaire, the app will prompt again the user to answer another set of questionnaire, which only assess smoking and quitting outcomes. The design of EMA pair within 30 minutes facilitates the analysis of causality, because we can capture some smoking or cravings preceded by the environmental exposures. Each questionnaire will take no more than 1 minute to complete. If an EMA prompt is not responded by the user, 2 additional prompts in every 5 minutes will be generated to remind the user to complete. The EMA will regard non-response as missing data if these additional prompts are not responded. All the above data will be sent to our server in each day.

• Statistical analysis Data will be analyzed with SPSS or Stata for Windows. The first research question can be addressed by comparing the frequency/proportion of the 4 HW-related outcomes pre- and post-implementation of the new warnings. To take into account the clustering between the outcomes within each participant, generalized estimating equation model will be used to assess the effect of new HWs on the noticing and responses to the HWs. To answer the second and third research questions, each EMA pair is the analysis unit, which is nested within each participant. The 3 outcome variables include the craving intensity (continuous variable), purchase of tobacco (binary) and smoking (binary). In the main analysis, linear mixed model (continuous) and mixed logistic regression model (binary) which allow for multiple observations in each subjects and account for clustering of data within subjects will be used to explore the association between each exposure variable and the outcome variables, adjusted for the variables collected at baseline. Significance level will be defined as p<0.05.

• Consent Participation in the study is voluntary. The recruitment staff will explain to the potential subjects who agree to join the study, before the written consent is requested. The subjects will be assured that they can withdraw from the study anytime without any prejudice, and all the information will be kept confidential and results will be reported in an aggregate format. Subjects are required to sign the written consent form.

Inclusion criteria

• daily smoker before enrolment in the clinics,
• aged 18 years or older,
• had a mobile phone,
• his/her mobile phone can access the Internet,
• he/she has no plan to travel overseas in the coming week.