The effect of closure on the relationship between ADHD symptoms and smoking initiation: A moderation model using Add Health data


Wise, Barbara (2015). The effect of closure on the relationship between ADHD symptoms and smoking initiation: A moderation model using Add Health data.


Background: Tobacco is the leading cause of preventable and premature death in the US. Eighty-eight percent of adult smokers initiated smoking in adolescence. Adolescents with ADHD are at high risk for smoking. Peer influence been long been identified as a major influence in smoking initiation in adolescence; peer relationships among children with ADHD have been demonstrated to be dysfunctional. Does low social integration increase the risk of smoking initiation among adolescents with ADHD, and if so, is that relationship curvilinear? Burt theorizes that moderate levels of closure (social integration) in an individual’s social networks are optimal for a variety of outcomes. This study examined whether moderate levels of social integration were protective against smoking initiation among adolescents with ADHD, compared to those without ADHD. Methods: Design: Observational, longitudinal secondary analysis of a large nationally representative longitudinal database. Sample: Data was drawn from the National Longitudinal Study of Adolescent to Adult Health (Add Health) data. Measures: The Dependent Variable is smoking initiation between waves I and III. The Independent Variables are ADHD related: inattentive and hyperactive subtypes, measured as symptom frequency, where all participants had a score; and dichotomous ADHD variables where 1= met diagnostic criteria and 0= not met. The ADHD symptoms were derived from retrospective self-reported ADHD symptoms in childhood, asked of participants in wave III. Moderators included five measures of social network position that serve as proxies for closure: in-degree, out-degree, Reach, Reach3, and Bonacich’s Centrality. Analysis: Multilevel logistic regression, accounting for school smoking prevalence and size at level 2. Conclusions: ADHD hyperactive symptoms as a continuous variable was associated with a modest 12% increase in the likelihood of initiating smoking (OR=1.12, 95% CI 1.04-1.22), controlling for all other variables. Inattentive symptoms and ADHD categorical subtypes (ADHD inattentive, hyperactive, and combined) were not significantly associated with initiation of smoking when controlling for other variables in the model. A significant interaction was found between Bonacich’s centrality and inattentive categorical ADHD, where those with inattentive ADHD were significantly less likely than those without ADHD to initiate smoking if they had moderate levels of centrality, compared to high or low levels of centrality (OR 3.1, 95% CI 1.1-8.8). No other significant interactions were identified among the other proxy closure measures investigated with any other ADHD measures. This study contributed to the field by examining five proxy closure measures and whether they moderate likelihood of initiating smoking in a curvilinear fashion. Further research needed includes basic descriptive studies of the social networks of those with ADHD, including both closure measures and bridging, as well as testing whether moderate levels of bridging are protective for smoking among those with ADHD. In addition, research investigating the implications of social network position on other known outcomes of ADHD, such as academic success, career success, delinquency and criminality, etc. would be relevant and useful.


Reference Type


Book Title



Wise, Barbara

Year Published


Volume Number



Ohio State University

Reference ID