Cohabitation with a smoker and efficacy of cessation programmes: the mediating role of the theory of planned behaviour

Abstract

Objective

The present research sought to examine whether cohabitation with a smoker undermines smoking cessation among people engaged in a cessation programme and whether the components of the Theory of Planned Behaviour (TPB) act as mediating mechanisms.

Design

A prospective longitudinal study with online questionnaires was conducted among smokers living in Switzerland who enrolled in a 6-months smoking cessation programme.

Main outcome measures

Cohabitation with a smoker and the TPB constructs were assessed 10 days after the start of the programme (T1; N = 820). Smoking abstinence was measured at T1, and at 3-months (T2; N = 624) and 6-months follow-ups (T3; N = 354).

Results

Results showed that living with a smoker decreased the odds that smokers remained abstinent throughout the cessation programme. Furthermore, we found that cohabitation was negatively associated with subjective norm. Afterwards, subjective norm predicted intention to maintain smoking cessation, which, in turn, predicted smoking abstinence. Such mediation effects persisted at each time point.

Conclusion

The present research provided evidence that living with other smokers at home can lead to greater risks of relapsing among people engaged in a cessation programme. We discussed the role of smoking-related norms in the efficacy of cessation interventions.

Supplemental data for this article is available online at https://doi.org/10.1080/08870446.2022.2041638 .

Keywords:

• Cohabitation with a smoker
• smoking
• cessation interventions
• theory of planned behaviour
• abstinence

Declaration of interest statement

No potential competing interest was reported by the authors.

Data availability statement

All data concerning this research is openly available at https://osf.io/vc8xw

Notes

1 Given that research has extensively shown that most relapses occur within the first weeks after a quit attempt (Herd et al., 2009; Hughes, 2007), we had planned to administrate the first questionnaire to participants 10 days after the quit date to evaluate their smoking status at that point. Unfortunately, for practical reasons, we were not able to provide participants with two questionnaires a few days apart and we had then to incorporate the other measures only 10 days later. We acknowledged this limitation in the discussion section.

2 As part of a larger study investigating the psychological predictors underlying the efficacy of online smoking cessation interventions, it is important to note that other variables were assessed in every follow-ups in addition to those described here, the results of which are not reported in the present manuscript (see Desrichard et al., 2021; Falomir-Pichastor et al., 2020). Also, note that this research has been approved by the ethical committee of the authors' faculty (approval number: PSE.20190304.01).

3 Attrition rate was 23.9% at T2 (n = 196; of them, 51.5% did not cohabitate with a smoker) and 57.9% at T3 (n = 475; of them, 56.6% did not cohabitate with a smoker). At T1, while there was no attrition with respect to smoking abstinence (all the participants responded to the corresponding item), average rates of missing data were 10.7% (n = 88) for attitude, 13% (n = 107) for subjective norm, 11% (n = 90) for PBC, 10.9% (n = 89) for behavioural intention, and 5.4% (n = 44) for dependence. There were 44 missing data (5.4%) on the cohabitation variable and 284 (34.6%) regarding the use of replacements.

4 In the supplementary materials, we provided the results of multi-group CFA and SEM examining whether or not the measurement and structural models are invariant between those who cohabitated with a smoker and those who did not.

5 Based on modification indices, we freely estimated the parameter with the largest and most substantial modification value (MI = 486.27, EPC = .42). Note that this modification remains theoretically sound with the assumptions of the TPB (see Ajzen, 2020) and that the difference between the uncorrected and corrected models was significant (Δχ² (1) = 99.19, p < .001).

6 To ensure reliability of this model, we compared it to more restrictive alternative models. In the first alternative model, we constrained the direct paths from cohabitation to attitude and PBC. This model led to a good fit (χ2 (111) = 274.07, p < .001; RMSEA = .053; sRMR = .058; CFI = .96), but significantly differed from our model (Δχ² (6) = 39.048, p < .001). In the second alternative model, the direct paths from cohabitation to attitude, norm, and PBC were constrained. We found a good fit (χ2 (117) = 301.481, p < .001; RMSEA = .055; sRMR = .067; CFI = .96), but a significant difference still appeared with our model (Δχ² (12) = 63.216, p < .001). In a third alternative model, we freely estimated the direct paths from cohabitation to intention and constrained the paths going to abstinence. This model fitted poorly to the data (χ2 (120) = 1309.569, p < .001; RMSEA = .138; sRMR = .164; CFI = .72) and statistically differed from our model (Δχ² (15) = 626.276, p < .001). In sum, these analyses suggested that our model exhibited a better fit in comparison with alternative models and was retained as the most optimal model to explain the data.

Additional information

Funding

This research has been funded by The Swiss Tobacco Control Fund.