Abstract
Background: Cigarette smoking is a major preventable cause of morbidity and premature mortality world-wide. It is well known that the risk of becoming a smoker is socially patterned and differentially distributed in the population. In particular, a socioeconomic gradient in smoking has been frequently described such that rates increase with declining socioeconomic status (SES). A consequence of this gradient is that the health burden of smoking falls disproportionately on lower socioeconomic strata, which in turn is reflected in their lower life expectancy. However, little is known about the causal mechanisms that might account for this gradient in smoking.
Smoking initiation typically occurs during adolescence. Thus, antecedents to adolescent smoking have become an important focus in social epidemiology. Such research, while abundant, has reported inconsistent findings on the socioeconomic patterning of adolescent smoking. Of particular interest to our understanding of the generation of socioeconomic inequalities in health is whether the socioeconomic gradient in adult smoking is replicated amongst their offspring, and if so, what might be the underlying mechanisms contributing to this phenomenon. This thesis addresses these questions by drawing on data collected during the course of the Christchurch Health and Development Study (CHDS)- a longitudinal study of a birth cohort of 1265 children born in the urban region of Christchurch, New Zealand, during 1977.
Objectives: The substantive aim of this thesis was to identify, aided by empirical evidence and sociological theory, plausible mechanisms implicated in the potential causal pathway between parental SES and adolescent smoking, developing a hypothetical model to test on data available from the CHDS.
Results: Initial analyses revealed the presence of a statistically significant linear association between parental SES at birth and respondent smoking at age 21. Stratified analyses by gender and ethnicity, and linear logistic regression models to test for interaction effects, revealed that this association was the same for both sexes and ethnic groups.
Six potential pathways were identified from the literature: parental smoking, parental attitudes to smoking, early smoking experimentation, conduct disorder, educational achievement, and affiliations with smoking peers. After fitting a series of linear logistic regression models, only three were found to mediate the association between parental SES and adult offspring smoking. These were: parental smoking, educational achievement and affiliations with smoking peers. This pattern was replicated and confirmed with a further exercise using log-linear modelling. A log-linear model fitted to the data revealed that the association between parental SES and young adult smoking was explained by three pathways: directly by way of parental smoking and educational achievement, and indirectly via the effect of parental smoking on affiliations with smoking peers.
Conclusion: This analysis of the CHDS data highlights how parental smoking and educational achievement, and to a lesser extent, affiliations with smoking peers, are potential points for intervention in reducing the socioeconomic gradient in young adult smoking. However, the association of these causal mechanisms with parental SES at birth suggests that fundamental socioeconomic differences and inequality are important and may hold promise in addressing health behaviours associated with SES.