Intervention: description of the program

The TWTQ program is a six-week hybrid self-help smoking cessation intervention designed to enhance user engagement through a combination of structured planning, cognitive-behavioral strategies, and personalization. Its design was informed by core principles of cognitive psychology, particularly regarding motivation, self-regulation, and habit change. These principles guided the integration of preparatory steps, repeated actions, and self-monitoring components, with the aim of supporting gradual behavioral transformation. Delivered as a comprehensive kit (Figure 1), it includes a printed instruction booklet, step-by-step guides, customizable worksheets, a timeline calendar, advice sheets, and ‘Great Ideas’ cards for recording essential strategies and reflections. These cards are gradually filled in and inserted into a compact box that mimics the dimensions of a cigarette pack, serving as a symbolic and practical replacement for cigarettes throughout the program.

Figure 1

The ‘Two Weeks To Quit’ program kit, prospective observational study, Paris, France, February 2023 to April 2024 (N=97)

Digital resources complement the physical materials, with QR code-accessible coaching videos created by experts in psychology, dietetics and cognitive-behavioral therapy. These videos are designed to provide users with professional guidance and motivational support at each stage of their journey. Participants also have access to a private Facebook support group, moderated by the lead researcher, fostering peer encouragement and shared experience throughout the process. The program follows a structured timeline of 28 steps, always starting on a Monday to align with the natural weekly rhythms known to support the establishment of new healthy habits²⁸ and is divided into two distinct phases (Figure 2). By integrating symbolic tools, structured guidance, therapeutic content, and social support, TWTQ offers a multidimensional, user-driven approach to quitting smoking and maintaining long-term behavioral change.

Figure 2

Structure of the ‘Two Weeks to Quit’ program: Two-phase progression over six weeks, prospective observational study, Paris, France, February 2023 to April 2024 (N=97)

Study population

Adults aged 18–60 years, French-speaking, and residing in Paris, France, who self-identified as regular smokers wishing to quit tobacco use, and who had access to a smartphone, an internet connection, and a Facebook account, were eligible for inclusion. Pregnant women and individuals with polyaddictions were not included. Participants who withdrew before the end of the study or did not complete the program evaluations were excluded from the analysis. Written informed consent was obtained from all participants prior to inclusion.

Recruitment was carried out primarily through pharmacies and on Instagram, via calls for participation over a 14-month period, from February 2023 to April 2024. Pharmacies were specifically targeted based on the assumption that smokers motivated to quit would go there to purchase nicotine replacement products, as they do during the ‘Moi(s) sans tabac’ campaign²⁹ – a national public health initiative in France encouraging smokers to quit collectively each November – when they pick up quit kits.

Measurement tools

At baseline, participants completed a sociodemographic questionnaire including seven items. They were asked to indicate their gender (female, male, or other), their age in years, their current marital status (single, cohabiting, married, in a civil union, divorced or separated, or widowed), and their highest level of education (secondary education, high school diploma or equivalent, Bachelor’s degree, Master’s degree, or other). They were also asked to report their employment status (employed, unemployed, on medical leave, retired, student, or other), their current housing situation (tenant, owner, hosted by someone else, living in a shelter, or homeless), and their native language. All responses were treated as categorical variables, except for age, which was treated as a continuous variable. No grouping or recoding was applied in the analysis.

Participants also completed the Fagerström test for nicotine dependence (FTND)³⁰, a 6-item instrument used to assess the intensity of physical nicotine dependence. The total score ranges from 0 to 10 and was treated as a continuous variable in the main analyses. For descriptive purposes, the following categories were used: very low dependence (0–2), low (3–4), moderate (5), high (6–7), and very high (8–10).

The Hospital Anxiety and Depression Scale (HADS)³¹ was used to assess psychological symptomatology. It consists of 14 items divided into two subscales – anxiety and depression – each ranging from 0 to 21. Following standard interpretation thresholds, scores were categorized as follows: 0–7 indicating no symptomatology, 8–10 suggesting borderline or doubtful symptomatology, and ≥11 reflecting clear symptomatology. These subscale scores were treated as continuous variables in the analyses.

Motivation to quit smoking was assessed with a single Likert-scale item ranging from 1 (‘not at all motivated’) to 5 (‘extremely motivated’), which was treated as a continuous variable. Participants were also asked to report the number of previous quit attempts, recorded as a continuous variable.

A weekly questionnaire was sent to participants during the six-week program. This questionnaire was created specifically for this study and has not undergone prior psychometric validation. It included self-reported smoking status with the exact quit date when applicable, as well as binary questions on whether they had followed the order and pacing of the program steps during the past week. It also included binary items on the use of nicotine replacement therapies and electronic cigarettes, with open-ended responses specifying the type of product, the nicotine dosage, and the frequency of use. Engagement with key program components – customizable tools, exercises, QR-code coaching videos, and participation in the Facebook group – was measured using 5-point Likert scales (from ‘never’ to ‘very often’). Participants were also asked to rate their level of anxiety on a 0–10 scale, and were invited to provide personal remarks about their experience each week.

Procedure

Participant data were collected at baseline, weekly during the six-week program, at the end of the program, then again two months later (i.e. three months after the theoretical quit date) and five months later (i.e. six months after the theoretical quit date). Weekly assessments were conducted every Sunday during the program. The questionnaire was sent by email each Sunday, with a reminder the day before (Saturday). In the event of non-response, up to two additional email reminders were sent. These weekly data allowed us to analyze participant engagement and adherence to the program timeline (step order and pacing). At the end of the program, all participants completed the Fagerström test for nicotine dependence (FTND) and answered the question ‘Have you quit smoking?’ to report their smoking status. Abstinence was defined as a ‘yes’ response to this question combined with an FTND score of 0. Since some individuals may report a score of 0 on the FTND despite continuing to smoke occasionally, we took care to distinguish abstinent participants from non-abstinent participants with a zero score. For this reason, only FTND scores of 0 reported by participants who answered ‘no’ to the abstinence question were included in longitudinal analyses of tobacco consumption. This approach ensured that changes in nicotine dependence could also be evaluated among participants who did not achieve full abstinence, while clearly excluding abstinent profiles from the analysis of consumption levels.

For the purpose of analysis, participants were classified into two groups based on their level of participation. The ‘Respondents’ group included individuals who completed the full six-week program, responded to the weekly assessments, and completed the post-program FTND. The ‘Non-respondents’ group included all other participants, including those who discontinued the program before completion (dropouts) as well as those who failed to respond to the weekly assessments or post-program evaluation. This classification was used to compare baseline characteristics and program outcomes.

Statistical analysis

Data were analyzed using SPSS software. Descriptive statistics were used to characterize the sample and adherence patterns. Differences between participants who completed the program and those who did not were assessed using independent samples t-tests for continuous variables. A multiple logistic regression analysis was conducted to identify predictors of smoking cessation, including adherence to the program timeline and engagement with the program’s tools. In addition, paired samples t-tests were used to assess changes in FTND scores over time (baseline, end of program, two months, and five months post-intervention). Correlational analyses (Pearson’s r) were conducted to explore associations between engagement with individual tools and changes in FTND scores. Two multiple linear regression analyses were also conducted: the first to identify predictors of short-term reduction in FTND score at the end of the program, and the second to assess whether engagement with program tools predicted the maintenance of FTND score reduction at five months. A significance threshold of p<0.05 was applied to all tests.

Participants

The initial sample included 97 participants, the majority of whom were women (53.6%), held a postgraduate degree (62%), and were employed (68%). The mean age was 38 years (SD=12.5, range: 19–61). Overall, motivation to quit smoking was high, with 58% of participants reporting being highly (4/5) or very highly motivated (5/5). Nicotine dependence, measured using the Fagerström score, indicated a moderate average level of dependence (mean score=4, SD=2.4). More specifically, 32% of participants were moderately dependent (score 5–6), 28% were mildly dependent (score 3–4), 28% were non-dependent (score 0–2), and 12% were highly dependent (score ≥7). Participants reported an average of three previous quit attempts, with considerable interindividual variability (range: 0–20). Psychologically, 58% showed signs of anxiety or suspected anxiety disorders, and 17% reported depressive symptoms (Table 1).

Participant typology based on study engagement

A comparison of baseline data revealed several differences between the two groups. Respondents (N=47) showed a slightly lower nicotine dependence at baseline (Fagerström score=3.5) than Non-respondents (N=50; mean score=4.3), but this difference was not statistically significant (p=0.293). They included a higher proportion of women (66.0% vs 53.6%), though this difference was also non-significant [χ²(1)=1.43, p=0.232]. However, education level was significantly higher among Respondents (80.9% with postgraduate education vs 62.0%; χ²(1)=4.16, p=0.041). Respondents had made more previous quit attempts (mean=3.32 vs 2.90), but this difference was not significant (p=0.425). Anxiety and depression scores were slightly lower in Respondents (HADS anxiety: 8.89 vs 8.91, p=0.975; HADS depression: 4.32 vs 4.86, p=0.359). Motivation to quit was similar in both groups (4.00 vs 3.97, p=0.895).

Engagement with program tools and adherence to the timeline

Respondents engaged with the program’s tools throughout the six-week period, although their engagement declined over time. Completion of the exercises was reported by nearly 70% of participants in week 1, but gradually decreased to 30% by week 6. Video usage, which was high at the beginning (75% of participants watched them), followed a similar trend, with 30% still watching them by week 4 and 70% having stopped by week 6. The customizable tools were actively used at the start of the program (53% usage in week 1, average score=2.2) and remained in use by a quarter of participants in week 6 (25%, average score=0.9). The Facebook group was rarely utilized from the outset (87% had never accessed it in week 1), and its use remained marginal at the end of the program (95% had never accessed it in week 6). As for adherence to the program steps, 66% of participants were still following the order of the instructions in week 6, compared to 91% in week 1, reflecting a relative maintenance of program structure. The pacing of the steps showed consistent stability throughout the six weeks (56% in week 1; 52% in week 6), suggesting that participants followed the program at their own rhythm without significant fluctuation

Effects of the program on respondents’ smoking status

The ‘abstinent’ smoking status at the end of the program was defined by a score of zero on the FTND and a ‘yes’ response to the question ‘Have you quit smoking?’. Among the 47 Respondents, 27 (57.4%) reported being abstinent at the end of the program. All abstinent participants indicated a quit date that matched the one planned by the program, namely two weeks after the start of the intervention. At follow-up, 23 participants (48.9%) remained smoke-free two months after completing the program, and 19 (40.4%) maintained abstinence five months after the program ended. When analyzed using an intent-to-treat approach – considering all non-respondents as smokers – the five-month abstinence rate was 19.6% (19 out of 97 participants). Although this is a more conservative estimate, it reflects a realistic measure of the program’s effectiveness in routine conditions and supports its relevance as a scalable self-help intervention.

Predictive factors of smoking cessation and maintenance

A logistic regression analysis was conducted to identify the predictors of immediate smoking cessation at the end of the program. Three factors significantly predicted abstinence: adherence to the prescribed step order (p=0.032), adherence to the recommended pace (p=0.043) and participants’ age (p=0.047), with older participants being more likely to succeed. In contrast, the use of customizable tools, exercises, videos, the Facebook group, or anxiety levels were not significantly associated with short-term cessation. To explore whether the effects of the program were maintained over time, a multiple regression was used to assess the predictive value of engagement with the program’s tools on abstinence at five months. None of the predictors – use of customizable tools, exercises, videos, or the Facebook group – were significantly associated with smoking status at five months. The regression model was not significant (R²=0.183; p=0.246), suggesting that long-term outcomes may be influenced by factors outside the program’s scope, such as intrinsic motivation or social environment. Interestingly, a counterintuitive finding emerged: higher engagement with customizable tools in the early weeks was associated with a greater likelihood of relapse at five months (p=0.024). This may reflect a compensatory effort among participants experiencing more difficulty with quitting.

Changes in tobacco consumption among non-abstinent respondents

In parallel with the assessment of abstinence, we examined whether the program also had a measurable effect on tobacco consumption among participants who did not quit smoking entirely. The objective was to determine whether a significant reduction in tobacco consumption could be observed, even in the absence of full abstinence. To assess this, we used the FTND, administered at four time points: before the program (baseline), immediately after completion, and at two and at five months post-program. At baseline (Session 0), Respondents (N=47) had a mean FTND score of 3.5 (SD=2.36), which was slightly lower than the total sample (mean=4.28, SD=2.37). After six weeks in the program (Session 1), the mean score among Respondents had decreased to 0.87 (SD=1.65), representing a statistically significant reduction [t(46)=7.71, p<0.001], with a large effect size (Cohen’s d=1.12). This suggests a strong impact of the program on reducing tobacco consumption, regardless of abstinence status. Beyond these immediate effects, we conducted a longitudinal analysis to determine whether the reduction in tobacco consumption was maintained over time (Table 2). The mean FTND score remained very low at the follow-up at two moths (mean=0.39, SD=0.87), and although a slight increase was observed at five months (mean=1.03, SD=2.04), it remained substantially below baseline levels. Paired-sample t-tests confirmed that the reduction was statistically significant both at two months [t(35)=7.94, p<0.001] and at five months [t(30)=4.52, p<0.001], with no significant difference between the post-program score and the follow-up scores. These findings suggest that the program’s effects on tobacco consumption were not only immediate but also sustained. Participants appeared to have internalized the benefits of the intervention in a durable way, with no significant rebound in consumption scores during the five-month study period.

Predictive factors of the FTND score reduction

A multiple linear regression analysis was conducted to identify the predictors of reduction in tobacco consumption, measured by the difference in Fagerström scores between baseline (Session 0) and the end of the program (Session 1). The dependent variable was a continuous score (ΔFagerström=S1-S0), where higher negative values indicated a greater reduction in consumption. The overall model explained 24.1% of the variance in consumption reduction (R²=0.241). Adherence to the prescribed order of program steps was significantly associated with a greater a greater consumption reduction (β=0.593, p=0.003; 95% CI: 0.211–0.976). In contrast, adherence to the pacing of the steps was not a significant predictor (β=0–0.160, p=0.402; 95% CI: -0.542–0.222). The use of customizable tools was also a significant predictor of consumption reduction (β=0.681, p=0.002), whereas the use of exercises (β= -0.072, p=0.766) and video content (β= -0.148, p=0.326) had no significant effect. Finally, the use of the Facebook group remained marginal and was not associated with any measurable impact on consumption scores.

Predictive factors for long-term effectiveness

To explore which factors might predict the maintenance of consumption reduction at five months post-program, a multiple linear regression was conducted. The aim was to assess whether engagement with program tools could predict Fagerström scores at the follow-up at five months. The model included four predictors: Facebook group usage, customizable tools usage, exercise completion, and video viewing. The dependent variable was the Fagerström score at five months, and the analysis was conducted on the 31 participants with available follow-up data (Table 3). The overall model was not statistically significant [F(4, 26)=1.45, p=0.246, R²=0.183]. Customizable tool usage, which had previously predicted short-term reduction, was not associated with Fagerström scores at five months. Similarly, no significant associations were found for the use of exercises, videos, or the Facebook group.

Additionally, a logistic regression was conducted to assess predictors of abstinence status at five months (abstinent vs non-abstinent). The model included engagement variables, anxiety level, and age. None of the predictors reached significance, though one effect was notable: higher use of customizable tools was significantly associated with a greater likelihood of relapse (p=0.024). Age also approached significance (p=0.071), suggesting a possible trend toward higher abstinence among older participants.

Use of nicotine replacement therapies and electronic cigarettes

No participant was using nicotine replacement therapy (NRT) or electronic cigarettes (EC) at the time of inclusion, and none had initiated a quit attempt prior to starting the program. General guidance on the use of cessation aids was provided as part of the program, including encouragement to seek advice from a pharmacist. Participants were also invited to try NRT products during the preparatory phase (the first two weeks of the program) to become familiar with them prior to their potential quit date.

Use of NRT and EC was controlled for in the analysis. Two simple linear regressions were conducted to assess their association with abstinence and reduction in Fagerström scores. Neither NRT use (β=0.29, p=0.379) nor EC use (β=0.13, p=0.494) was significantly associated with smoking reduction or abstinence at the end of the program. These results suggest that the program’s effectiveness was primarily driven by structural and motivational factors rather than pharmacological ones. Consequently, these variables were not included in the multiple regression model. Moreover, the use of NRT and EC was not significantly associated with either sustained abstinence or changes in tobacco consumption assessments at 2 and 5 months.

Limitations

This study has several limitations that should be acknowledged. First, although the six-week program showed positive outcomes, the follow-up period was limited to five months post-intervention (six months post theoretical cessation date). While this timeframe allows for a medium-term assessment, prior research has demonstrated that cessation outcomes may continue to fluctuate up to one year after the quit date¹. Therefore, to better evaluate the sustainability of results, future studies should include extended follow-up periods up to 12 months post-intervention.

Second, smoking status and tool usage were self-reported, which introduces the risk of response and recall bias. Biochemical verification (e.g. CO monitoring) and digital tracking of tool engagement would strengthen the reliability of future evaluations by reducing the risk of misclassification associated with self-reported smoking status and tool usage.

Third, the study did not include a control group. As such, it is not possible to establish a causal relationship between participation in the program and the outcomes observed. We can only formulate hypotheses regarding the program’s effectiveness. Furthermore, the results cannot be directly compared to other interventions or to natural quit rates. While the positive outcomes suggest that the TWTQ program may be effective, controlled trials are needed to confirm these findings and isolate the specific contribution of each component of the intervention.

Fourth, the regression model assessing long-term predictors of tobacco consumption did not identify significant relationships with tool usage, possibly due to limited statistical power. This highlights the need for larger sample sizes and more granular measurement tools to understand what drives long-term abstinence.

Additional limitations should be acknowledged. First, the absence of certain covariates in the analyses may have led to residual confounding. Second, participants were primarily recruited through community pharmacies, which may have introduced selection bias. This recruitment method may have attracted individuals who were already more motivated to quit smoking or who had specific expectations about the program, limiting the generalizability of the findings. Third, a progressive decline in participant engagement was observed during the program, raising questions about the factors influencing long-term adherence. Moreover, the use of the FTND to assess changes in tobacco use may itself represent a methodological limitation. Although widely used to evaluate nicotine dependence, the FTND was not originally designed to measure changes in smoking behavior over time, especially among participants who reduce consumption without achieving full abstinence. Future studies could benefit from incorporating complementary measures to better capture nuanced patterns of tobacco use. Finally, the relatively small sample size and geographical specificity of the study (urban area in France) limit the external validity of the results. In addition, if loss to follow-up was not random but associated with both the intervention and smoking outcomes, this could have introduced attrition bias, further challenging the interpretation of long-term effects.