Skip to main content
Open Access Publications from the University of California

Using Cluster Analysis to Investigate the Association between Smoke-free Homes, Light Smoking, and Cessation Behaviors

  • Author(s): Kempster, Jennifer Amy
  • Advisor(s): Pierce, John P
  • et al.

Background: Smoking cessation rates were stable in the 20 years to 2011. This is surprising given the concurrent desirable trends in many cessation-related behaviors, including smoke-free homes (SFH) and cigarettes per day (CPD). This could be the result of patterns of characteristics and behaviors having different associations with cessation than the individual associations may suggest. Understanding such associations could give new insights into the smoking population and help to improve policy and interventions.

Objectives: To identify homogeneous smoker sub-groups and compare their one-year odds of quitting behaviors and cessation-related behaviors, CPD and SFH. To identify trends in these behaviors in the sub-groups.

Participants: Adult smokers (n=2569) who completed the longitudinal 2010-11, Tobacco Use Supplement of Current Population Survey; 1447 had no SFH and 1799 consumed 10+ CPD at baseline.

Methods: Logistic regression identified variables associated with quitting, which were used in cluster analysis to identify smoker sub-groups. Sub-group predictive validity was tested for four outcomes: quit attempt (QA), 30-day abstinence (30D-A), SFH implementation, and reduced CPD.

Results: Four cluster subgroups were characterized by their combination of smoking intensity, nicotine dependence, and SFH status: high dependence home smokers (n=700, ref.), high dependence smokers with a SFH (n=458), low intensity smokers (n=446), and low dependence moderate-heavy smokers (n=488). One non-homogenous cluster grouped racial/ethnic minority smokers (n=477). QAs were higher than in the reference group than all other sub-groups. Given a QA, only low intensity smokers (ORadj=2.18, p=0.01) and low dependence moderate-heavy smokers (ORadj=2.15, p=0.04), both containing predominantly low dependence smokers, were more likely to report 30D-A.

High dependence smokers with a SFH were no more likely to reduce to <10 CPD than high dependence home smokers (19.7% vs 16.3%; ORadj=0.68, p=0.07). Low dependence moderate-heavy smokers were equally likely to introduce a SFH as high dependence home smokers (ORadj=1.21, p=0.09).

Conclusions: Contrary to findings in the literature, having a smoke-free home is not associated with a reduction to light smoking among heavier smokers. Both smoke-free homes and a reduction in consumption are diffusing fairly rapidly in all sub-groups of the US smoking population.

Main Content
Current View