The Smoke Ring : are your friends a drag? Social network analysis and tobacco use

  • Raglan Maddox

    Student thesis: Doctoral Thesis


    Background Smoking is the single most preventable cause of morbidity and mortality within Australia. While there have been reductions in smoking in Australia, Aboriginal and Torres Strait Islander people are twice as likely as non-Indigenous people to smoke. This study (the Smoke Ring Study) comprises two components: a systematic review to examine the influence of social networks on tobacco use; and a prospective mixed-method study. The prospective study explored and assessed the evidence on Aboriginal and Torres Strait Islander social networks and tobacco use and also Action Area 1 of the Australian Capital Territory (ACT) Aboriginal and Torres Strait Islander Tobacco Control Strategy 2010/11–2013/14 (the ACT Strategy). Development and implementation of components of the ACT Strategy commenced in 2010,with engagement of local Aboriginal and Torres Strait Islander community organisations and development of community communications commencing in 2012. Methods Systematic review The systematic review followed the Preferred Reporting Items for Systematic Reviews and Meta- Analyses (PRISMA) guidelines. The review searched the following databases: CINAHL (Cumulative Index to Nursing and Allied Health Literature); Informit Health Collection; PsycINFO; PubMed/MEDLINE; Scopus / Embase; Web of Science; and Wiley Online Library. A narrative approach was used to summarise the 279 papers that were included in the systematic review. This systematic review helped to inform and focus the primary data collection that formed part of the Smoke Ring Study. Prospective mixed-method study The prospective study used a mixed-method pre- and post-test design, pre- and postimplementation of the ACT Aboriginal and Torres Strait Islander Tobacco Control Strategy. The study used a panel survey (n=204 baseline; n=103 follow-up),individual interviews (n=10 baseline; n=9 follow-up) and focus groups (baseline: 3 focus groups,40 participants; follow-up: 3 focus groups, 30 participants). Logistic regression and social network analyses were used for the survey. Grounded theory was used to analyse the interviews and focus groups. Results Systematic review Synthesis of the findings of the systematic review indicated that social network structures, positions and relationships influence tobacco use (that is, initiating, maintaining and ceasing tobacco use). Social network analysis is relevant to tobacco use given that social relations and social contexts impact on the decision to smoke or not to smoke. Prospective mixed-method study Baseline data from the prospective study identified a prevalence of smoking of 36.4% (95% CI,27.8– 44.9) among the Aboriginal and Torres Strait Islander community in the ACT region—a figure that is significantly higher than that for the general Australian population (which is approximately 15%). While participants were not representative of the ACT Aboriginal and Torres Strait Islander population, a broad cross-section of the Aboriginal and Torres Strait Islander community in the region, covering a wide range of smoking behaviours, participated in the prospective study. The mean age in the study was 35 years (12 to 75 years of age). The sample was 65% female and 35% male and reported a median household income category of $67,600–$83,199 per annum. Household size ranged from one to seven people and 47% of participants had completed education to year 12 or above at baseline. At baseline, logistic regression models were used to determine factors significantly associated with smoking. Two independent variables made a unique, statistically significant contribution to whether respondents smoked:  completing education to at least year 12 or equivalent (p=0.003) (OR=21.5; 95% CI,2.9-158.7); and  the number of housemates who smoke (p=0.046) (OR=11.8; 95% CI,1.1-132.2). Social network analysis at baseline revealed that the total participant-nominated network (that is, the social network that participants in the survey claimed to belong to) included sub-groups that were mostly inaccessible through recognised relationships—i.e. connected via a small number of relationships—and there was significant difference between smokers’ and non-smokers’ networks. When smoking and non-smoking networks were examined separately, it was found that the average distance between connected smokers and non-smokers was 2.8 and 2.7 steps or relationships respectively. This indicated that, when considered independently, smoking and non-smoking networks were more cohesive than the total network. Members of the total network were a mean distance of 11.0 steps away from each other. At follow-up, there was a statistically significant difference (p=0.007) in the number of smokers (42.9% and 44.4% of the network at baseline and follow up) and non-smokers (21.1% and 22.7% of the network at baseline and follow up) who reported that their best friend was a smoker. This also suggested some polarisation, or independence among smoking and non-smoking groups respectively. Themes from the study, but specifically the qualitative analysis at baseline and follow up included:  social normalisation of smoking;  tobacco being convenient and easy to obtain;  role modelling; and  smoking being seen as a way to facilitate social interactions. The results that were obtained from the study indicated that the ACT Strategy may have had an impact on smoking behaviour, noting that other local and national tobacco control measures have also been implemented. Therefore it is not possible to attribute changes specifically to the Strategy. Among Aboriginal and Torres Strait Islander people in the ACT, there was a reduction in smoking, an increase in the number of people who had never smoked and a decrease in the number of participants who reported incorrect perceptions that ‘some cigarette brands were more harmful than others’. Limitations This thesis has a number of limitations. The systematic review may have incurred publication bias, and included studies with different methods, different settings and at various points in time. In relation to the primary data collection, the use of a survey name generator question may not have provided a complete list of participants’ networks. The prospective study also used self-reported measures of smoking and network characteristic behaviours and the study’s attrition at follow-up was also a limitation. Conclusions The Smoke Ring Study was the first mixed-method longitudinal study to utilise social network analysis to examine Aboriginal and Torres Strait Islander social connections and how they impact on smoking. This study demonstrated that achieving at least a year 12 level of education was protective against smoking. It also supported the hypothesis that exposure to smokers in one’s social network strongly influenced smoking behaviours. It would appear that having a best friend who smoked was strongly associated with whether a person was a smoker. These findings imply that social networks can facilitate smoking behaviours, providing insight into the nuanced nature of social networks. They also suggest that good work has been undertaken as part of the ACT Strategy to reduce smoking prevalence. However, more work is required. The findings demonstrate that there is a need to focus policy, program and service delivery on smoking networks in order to reduce smoking rates and on non-smoking networks to minimise smoking uptake.
    Date of Award2015
    Original languageEnglish
    SupervisorRachel Davey (Supervisor), Tom Cochrane (Supervisor), Ray Lovett (Supervisor), Anke Van Der Sterren (Supervisor) & Joan Corbett (Supervisor)

    Cite this