Smoking Cessation and Quality of Life: Insights From Analysis of Longitudinal Australian Data, an Application for Economic Evaluations
Moayeri F., Hsueh YS., Dunt D., Clarke P.
Objectives: A number of studies have shown an association between smoking habit and quality of life, but these have mainly involved cross-sectional data. This study takes advantage of longitudinal panel data to estimate the effect of the transition from “smoker” to “ex-smoker” status (smoking cessation) on health-related quality of life (HRQoL), measured by SF-36, in an Australian general population sample. Methods: Panel data from 13 waves (2001-2013) of a nationally representative longitudinal survey of Household Income and Labour Dynamics of Australia (HILDA) were used; 1858 respondents (5% of total HILDA sample) who experienced only 1 cessation event in their HILDA life were selected. HRQoL trajectories elicited by SF-36 (0-100 scale, worst to best health) were modeled before and after cessation events using a piecewise (segmented) 2-way fixed-effect linear regression, adopted to capture within-person differences. This enabled measurement of changes of regression slopes and intercept while controlling time-invariant characteristics (eg, country of birth, gender) and time-varying changes in health status. Results: Annual pre-post intervention improvements were estimated for the following dimensions: role physical 0.65 (95% CI 0.62-1.24), bodily pain 0.48 (95% CI 0.10-0.86), general health 0.55 (95% CI 0.2-0.9), and the physical component summary score 0.22 (95% CI 0.01-0.04). Immediate effects (discontinuity at the time of cessation) of smoking cessation existed for bodily pain –1.5 (95% CI –2.52 to –0.40) and general health 1.82 (95% CI 1.01-2.62). The effects for mental health domains were not significant. Conclusions: Adjusting for all unmeasured time-invariant confounders and controlling the effect of time, this study revealed the varied effects of smoking cessation on HRQoL; it has positive effect on physical and general health but nonsignificant effect on mental aspects. Preference-based utility measures based on SF-6D capture changes that can be measured in several of the domains of the SF-36.