To What Extent Do Age, Period, and Cohort Patterns Account for Time Trends and Social Inequalities in Smoking?

Abstract

Aims: This paper evaluates nonconfounded and independent age, period, and cohort effects on time trends and social disparities in smoking behavior. Methods: Data from nine waves of the German Epidemiological Survey of Substance Abuse (ESA) conducted between 1980 and 2009 were used. A total of N = 73,782 individuals aged 18 and older were included in the analyses. Using cross-classified random-effects models, fixed effects of age (level 1) as well as random effects of periods and cohorts (level 2) on 30-day smoking prevalence and average number of cigarettes per day were estimated. Analyses were stratified by socioeconomic status (SES). Results: Independent of SES, positive and curvilinear age effects emerged. Prevalence and amount of smoking declined sharply over time and revealed an inversion of the social gradient in 2003. This was due to a stronger decrease of smoking in the highest SES group compared to the other groups. Cohort effects were nonlinear and inconsistent. Conclusion: The emergence of increasing social disparities in smoking implies the need to specifically address the situation of low-SES groups when developing tobacco control measures.