Household Composition and Smoking Behavior in a Prospective Longitudinal Australian Cohort

Abstract Introduction This study estimates the extent to which individuals’ smoking cessation and relapse patterns are associated with the smoking behavior of their household members. Aims and Methods Longitudinal data on household members’ smoking behavior was sourced from a representative sample of 12 723 Australians who ever reported smoking between 2001 and 2019. Controlling for a rich set of confounders, multivariate regression analyses were used to predict the likelihood of smoking cessation and relapse given other household members’ smoking status and their relationship type. The models were then used to forecast smoking prevalence over 10 years across different household types. Results Individuals living with a smoking spouse were less likely to quit (OR 0.77 [95% CI 0.72;0.83]) and more likely to relapse (OR 1.47 [95% CI 1.28;1.69]) compared to those living with nonsmoking spouses. Subsequently, the proportion of smokers living with other smoking household members increased by 15% between 2011 and 2019. A 10-year forecast using the smoking cessation and relapse models predicts that, on average, smokers living with nonsmokers will reduce by 43%, while those living alone or with a smoking partner will only reduce by 26% and 28% respectively. Conclusions Over time, those who are still smoking are more likely to live with other smokers. Therefore, the current cohort of smokers is increasingly less likely to quit and more likely to relapse. Smoking projection models that fail to account for this dynamic risk may overstate the downstream health benefits and health cost savings. Interventions that encourage smoking cessation at the household level, particularly for spouses, may assist individuals to quit and abstain from smoking. Implications The current and future paradigm shift in the smoking environment suggests that smoking cessation and relapse prevention policies should consider household structure. Policies designed to affect smoking at the household level are likely to be particularly effective. When estimating the long-term benefits of current smoking policies intrahousehold smoking behavior needs to be considered.