Study design

We conducted a secondary data analysis using data collected in the parent trial M2Q2, a two-arm double-blinded randomized control trial with smokers recruited at four rural community centers in Hung Yen province in Northern Vietnam (NCT03567993) by the Vietnam Institute of Population, Health and Development (PHAD)^14,15. Participants and research staff were blinded to allocation to intervention or comparison group via our text messaging system, with research staff also being blinded to allocation while assessing outcomes.

Ethics

This study involving humans was approved by the UMass Chan and Population, Health, and Development Institution Review Boards and monitored by a Data and Safety Monitoring Board. The studies were conducted in line with local legislation and institutional requirements. Participants provided written informed consent to participate in this study.

Parent trial

The objective of the parent trial was to examine the effectiveness of a 6-month texting intervention to promote the use of a Vietnam Ministry of Health-based Quitline (Bach Mai Quitline, located in Northern Vietnam) and improve smoking cessation. The intervention group received peer-written motivational text messages promoting Quitline use and smoking cessation for six months, along with biweekly messages assessing smoking status and interest in Quitline referral, while the comparison group received only biweekly smoking status assessments via text messages. The parent trial was conducted among 750 participants, with 1 participant withdrawing at baseline, with 369 participants remaining after six-month follow-up for the intervention group, and 372 participants remaining for the control group. Retention rate was 98.8%. The intervention group had significantly higher verified Quitline use (18% vs 1%) but no significant difference in carbon monoxide–verified smoking cessation rates at six months (both ~28%). The methods and full results of the parent trial have been published^14,15.

Setting and population

Participants were recruited at their community health centers by study staff and local collaborators. After providing written consent forms and meeting the eligibility criteria of the study, all participants were administered a baseline survey by trained staff. All survey instruments were translated into Vietnamese by a certified translator. Translators are native Vietnamese speakers and members of the American Translators Association. Translated documents underwent editing by an accredited language expert to confirm accuracy. All translated materials received certification of accuracy¹⁵. Participant inclusion criteria were: 1) a resident of the selected commune; 2) current smoker; 3) able to receive texts and read text (literate); 4) not cognitively impaired; 5) not a participant who helped develop the motivational texts used in the intervention; and 6) not a family member of another participant in the study. Since rates of smoking in Vietnam are very low for women¹⁶, sex was not an eligibility criterion for the parent study and resulted in a sample that was 100% male.

Data collection

Baseline data included demographics and smoking-related behaviors. WTP use was assessed via the question: ‘What other tobacco products do you smoke: Waterpipes? Yes/No’. Baseline data also assessed chronic comorbidities (heart disease, lung disease, hypertension, diabetes, presence of ulcers, kidney disease, liver disease, anemia, cancer, depression, arthritis, back pain, other rheumatic disease), quality of life (QOL), and Smoking Self-Efficacy SEQ-12¹⁷. The primary outcomes of the parent trial were 6-month CO-verified smoking cessation and Quitline usage, with self-efficacy on the SEQ-12 as a secondary outcome. Smoking cessation was defined using the 7-day point prevalence question at six months, verified by carbon monoxide breath testing (≤6 ppm), and Quitline use was confirmed through call records collected by the Quitline.

Data analysis

Our analyses corresponded to our research questions. To explore which sociodemographic factors were associated with WTP use, participants’ demographics and smoking characteristics were compared between WTP users and non-WTP users, using chi-squared or Fisher’s exact test for categorical variables and t-test for continuous variables. Baseline variables included age, education level, marital status, number of adults and children in a household, financial anxiety status in the last 12 months, number of chronic diseases, self-perceived health, and inpatient hospitalization in the last six months. Univariate logistic regression was conducted to examine associations between each independent variable and WTP use. Variables were selected for inclusion in the multivariable adjusted logistic regression model determined by both a priori knowledge and statistical criteria (p<0.25 in univariate analysis). Model fit was examined using the goodness-of-fit test and an area under the receiver operator characteristic (ROC) curve. We then examined how smoking characteristics differed between WTP users and non-WTP users. Next, to evaluate the impact of WTP use on trial outcomes, we estimated mixed-effects logistic regression models for: 1) CO-verified smoking cessation at six months, and 2) verified Quitline use. For each outcome, we used both an unadjusted model and an adjusted model. In the adjusted models, WTP status was the primary predictor, and we adjusted for the same covariates identified in the univariate screening (p<0.25), which included age, education level, employment, and comorbidities. All mixed-effects models accounted for clustering at the commune level. Finally, we examined the association between WTP and the Quitline usage group using a similar modeling approach. This analysis was restricted to the intervention group, as only intervention participants received active referral to the Quitline. Baseline analyses were completed with our total sample of 749 participants, while analyses based on follow-up had fewer participants due to loss to follow-up or missing data values.

An alpha level of 0.05 was used as a criterion to determine statistical significance. All analyses were performed using SAS software (version 9.4; SAS Institute, Cary, NC). Complete case analysis was used to manage missing data.

Additional covariates

We included the following covariates in our model for adjustment: age, education level, employment, and number of chronic diseases. Age was grouped into four categories: 30–39, 40–49, 50–59, and ≥60 years. Education level was also grouped into four categories: Primary school, completed secondary school, completed high school, and vocational college/university or higher. Employment was defined as self-employed, farmer, other, and paid work. Finally, the number of chronic diseases was grouped as: 0, 1, or ≥2 comorbidities. These covariates were selected through both statistical approach and a priori knowledge. Variables such as age, education level, employment, and number of chronic diseases can represent aspects of life outside of the healthcare system that may influence health behaviors and outcomes^18,19.

Sociodemographic factors are associated with WTP use

Out of 749 individuals, a total of 398 individuals who smoke (53%) reported WTP use at baseline. Education level was significantly associated with WTP use. A larger proportion of WTP users completed secondary school compared to non-WTP users (52% vs 38%) (Table 1). Non-WTP users were more likely to have attained higher education levels, such as vocational college or university (20% vs 14%). There was no significant difference in other demographic characteristics between the two groups in terms of age, marital status, employment status, household composition, or health status.

In the models, the older age groups had lower odds of WTP use compared to those <30 years (Table 2). Completing high school (AOR=0.61; 95% CI: 0.43–0.86), or vocational college/university or higher (AOR=0.47; 95% CI: 0.30–0.73) was associated with statistically significant odds of WTP use compared to completing secondary school, while completing primary school was associated with decreased odds, but lacked significance. Individuals reporting at least one comorbidity had higher odds of WTP use compared to those with none, but this was not significant.

Smoking behaviors by WTP use

Several tobacco use behavior differences emerged between waterpipe users and non-users (Table 3). WTP users smoked fewer cigarettes per day on average than non-users (11 vs 16) and reported smoking sooner after waking. Specifically, 53% of WTP users smoked within 5 minutes of waking compared to only 22% of non-WTP users (p<0.001). Additionally, WTP users spent significantly more money on tobacco products weekly compared to non-WTP users. WTP users first smoked tobacco at the age of 18 years, while non-WTP users first smoked tobacco at 19 years (p=0.02). Thirty percent of non-WTP users were more likely to work in environments with smoking restrictions compared to 20% of WTP users (p=0.002). Furthermore, baseline smoking self-efficacy scores (SEQ-12) were significantly higher among non-WTP users compared to WTP users (p=0.004).

WTP use and smoking cessation

WTP users had lower odds of CO-verified smoking cessation than non-users (Table 4).

WTP and Quitline use among the intervention group only

Among intervention participants, the adjusted models indicated that WTP users had lower odds of verified Quitline use compared to non-WTP users, but this association was not statistically significant (Table 5).

Strengths and limitations

To our knowledge, this is the first study examining sociodemographic associations of WTP use among smokers in rural Northern Vietnam. We report recent RCT data for public health surveillance and policy in Vietnam, where nationally representative data on waterpipe use are limited, especially in rural areas. The study has several limitations. We used data from a parent trial focused on smoking cessation with specific inclusion/exclusion criteria of the study participants and relied on questionnaires, which may limit the generalizability of the findings and introduce potential recall bias. The usage of complete case analysis may have limited our handling of missing cases in our models. Our sample was exclusively male, which, while consistent with the demographics of our target population¹⁶, may limit overall generalizability. There could have been potential unmeasured factors that we did not collect in the parent study (e.g. strong social norms and ritualistic habits associated with WTP use)³¹ that may impact study findings. Additionally, the potential for reverse causation exists, given that WTP use and smoking-related behaviors were both measured at baseline. Despite these limitations, our findings provide a strong foundation for further understanding behaviors and factors of use among WTP users in Northern Vietnam.

Public health implications and future directions

Overall, WTP was significantly associated with specific smoking behaviors, such as product usage and smoking frequency, as well as environmental factors like workplace smoking rules, but did not affect smoking cessation outcomes in the context of a mobile health intervention. Given the high prevalence of WTP use among rural smokers in Vietnam, our findings underscore the urgent need for public health campaigns that specifically target dual users, addressing both cigarette and WTP behavior. Future studies may need to focus on qualitative exploration into the perceptions of WTP among smokers, particularly among older persons with established WTP habits, those with lower levels of education, and younger individuals with lower socioeconomic status, in order to develop and test targeted mHealth interventions, such as incorporating specific peer messages designed for waterpipe users. Additionally, studies should focus on establishing a longitudinal relationship to better understand patterns of WTP habits over time. As mHealth smoking cessation interventions expand in LMICs, a nuanced understanding of subgroups like WTP users will be critical to designing effective public health strategies.