Data source and study population
Data were obtained from the 2015 Global Adult Tobacco Survey (GATS) in Vietnam, which is a cross-sectional nationally representative survey of 8,996 Vietnamese participants who were ≥ 15 years old. Standardized approaches for sampling method, questionnaire design, data collection, data management, and ethical considerations were used for conducting GATS. The questionnaire addressed 10 sections related to the World Health Organization’s MPOWER measures to assist countries with tobacco control: (1) demographic characteristics, (2) tobacco smoking, (3) electronic cigarette use, (4) smokeless tobacco use, (5) cessation efforts, (6) secondhand smoke exposure, (7) economics, (8) media exposure, (9) knowledge, attitudes and perceptions, and (10) pictorial graphic health warnings and tax stamps on cigarette packs.
Tobacco users were defined as who reported that they currently smoke any kinds of tobacco (e.g., cigarette, bamboo waterpipe, smokeless tobacco, etc.) on a daily basis or less than daily. After excluding non-tobacco users and occasional tobacco users, women, and those < 18 years old, 1,600 adult male tobacco users were available for inclusion in this study. We excluded the non-daily tobacco users because of lacking several information on smoking behaviors such as first use of a cigarette/waterpipe after waking among non-daily smokers, which was a well-known factor (i.e., nicotine dependence) associated with the intention to quit. After excluding 18 tobacco users who used these other minor types of tobacco products (shisha, smokeless tobacco, e-cigarettes and cigars) and 231 users who had incomplete information on concerned variables, 1,351 tobacco users were included in the final analysis, including 966 users of cigarettes only, 256 users of bamboo waterpipes only, and 129 dual users (Fig. 1).
Description of variables
The main outcome of this study is intention to quit. All tobacco users who planned to quit “within the next month”, “within the next 12 months”, or “someday but not in the next 12 months” were classified as having the intention to quit. Those who answered “not interested in quitting” were classified as not having the intention to quit, which was also defined in previous study .
The perceived harm from tobacco use were independent variable, including knowledge of whether cigarette and waterpipe use causes serious illness (both do not cause severe illness, only waterpipe causes severe illness, only cigarette causes severe illness, or both products cause severe illness), and perceived harm from waterpipe use versus cigarette smoking (less harmful, equally harmful, or more harmful).
For covariates, the factors groups associated with intention to quit were illustrated in conceptual diagram in Fig. 2. For individual level, demographic characteristics included age (18–24, 25–44, 45–65, or ≥ 65 years), ethnic group (Kinh-major ethnicity, or others—minor ethnicity such as Thai, Tay, Nung, Dao, etc.), residential area (rural or urban), education level (primary school or less, secondary school, high school or higher), marital status (unmarried; married; or separated, divorced or widowed). Occupation was classified as professionals or managers (e.g., legislators, senior officials, or managers; high qualified professionals; or technicians or associate professionals), skilled laborers (e.g., members of armed forces; service workers; shop and market sales workers; skilled agricultural and fishery workers; craft and related trade workers; or plant and machine operators and assemblers), semi-skilled laborers or clerks (e.g., elementary occupation, clerks, drivers, or guardians), and others (e.g., student, homemaker, retired, or unemployed).
Information on tobacco use behaviors, including type of tobacco products used (cigarettes, waterpipe, or dual user) and age at tobacco use initiation were obtained. Among daily tobacco users, the number of cigarettes smoked and waterpipe sessions per day were asked for cigarette smokers and waterpipe users, respectively. Time to the first use of cigarette or waterpipe tobacco after waking (≤ 5, 6–30, 31–60, or > 60 min) was determined.
For household level, information on the number of family members, having children at home (yes or no), and tobacco use regulation at home (no ban, partial ban, or comprehensive ban) was obtained. No ban on smoking in household was defined as smoking is allowed in every room inside of home or there are no rules on smoking ban. Partial ban was defined as smoking is allowed in some rooms inside of home or smoking is generally not allowed inside of your home but there are exceptions. Smoking is never allowed inside of the home was comprehensive ban.
For community factors, tobacco control policy also was measured. Data were also recorded regarding whether exposure to pictorial health warnings on cigarette packs, anti-smoking campaigns or encouragement to quit information, and advertisements or signage to promote tobacco products within the last 30 days in locations such as newspapers, television, radio, or internet.
The frequency distribution for each variable by intention to quit was performed and the collinearity of variables was evaluated. Multiple logistic regression analysis was used to evaluate the association between intention to quit and perceived harm on cigarette and bamboo waterpipe tobacco. The final model was selected after consideration of collinearity of variables of individual-level factors, adjustment for potential confounders including individual-level factors (age group, educational level, marital status), age at smoking initiation, number of cigarette smoked/waterpipe sessions used per day, time to the first use of cigarette or waterpipe tobacco after waking), household-level factors (smoking ban at home, having children at home), and community-level factors (Exposed to anti-smoking campaigns or encouragement to quit information, and exposed to advertisements/signage to promote tobacco products in the last 30 days), and assessment of model fit. Because of lacking the standardized measurement of intensity for both cigarette and waterpipe smoking, the number of cigarettes smoked and a number of waterpipe sessions used daily for cigarette users and waterpipe users was measured, respectively. Therefore, we combined two such variables into a single one to adjust in multiple logistic regression model among the whole study population. Subgroup analysis stratified by tobacco users was performed, and the reference group of a variable on perceived harm from waterpipe use versus cigarette smoking was changed. For cigarette-only users, perceived harm from cigarette smoking compared with waterpipe tobacco use was asked; in contrast, for waterpipe-only users, perceived harm from waterpipe use compared with cigarette smoking was measured. To examine the selection bias due to excluding 231 observations having missing information on concerning variables, we did the sensitivity analysis of factors associated with intention to quit by tobacco user groups, shown in Supplemental Table 1. Both descriptive and analytical statistical approaches were applied using weights. All statistical analyses were performed with STATA (version 14.0) software, and values of p < 0.05 were considered statistically significant.