Study design and theoretical framework
A fundamental qualitative descriptive design [
19] was utilised to enable a comprehensive, low-inference description of phenomena for the purpose of informing program development. The study was guided by social ecological theory [
20] which provides a set of principles for understanding the “interrelations among diverse personal and environmental factors in human health and illness” (p. 283) [
21]. A social ecological perspective was selected since it is consistent with Aboriginal notions of health and wellbeing [
22]. The social ecological framework recognises that behaviour is affected by multiple levels of influence including personal factors, interpersonal processes, organisational factors, community factors and public policy [
23]. Given the central role of the family in traditional and contemporary Aboriginal living [
24], the framework was adapted to include family as a unique and proximal level of influence. In the context of AHWs, the ‘organisation’ level was the health service in which AHWs were employed. Hence, the factors associated with smoking in AHWs were framed in six levels of behavioural influence: personal factors, family, interpersonal processes, the health service, the community and public policy.
Data collection and study sample
We pursued a purposive sampling plan that included AHWs with a range of smoking histories (never smoked, current smoker, ex-smoker) in addition to health service managers and tobacco control personnel. A balanced gender representation was considered desirable, as was a varied geographical distribution of respondents (including staff from metropolitan, regional and remote health services). The views of non-AHWs (i.e. health service management, tobacco control personnel and other health staff) were sampled to gain further insight into the broader socio-ecological (e.g. interpersonal, organisational, community and policy level) features that could influence AHWs’ smoking behaviours. In applying this sampling framework, we captured the direct experience of smoking from AHWs and utilised shadowed data’ in two ways to enhance sampling analysis [
25]. During interviews AHWs referred to other AHWs sharing similar experiences (‘generalised self’) and those not like them (‘generalised other’). In addition, perceptions of those external to AHWs (i.e., health service and tobacco control personnel) were included to “make the domain and various components of the phenomenon understandable more quickly” [
25] (pg. 291). Here, external perspectives provided insight into the range of social-ecological factors perpetuating AHW smoking.
Since relationships and trust are central to working respectfully with Aboriginal communities, the researchers sought the mentorship of the State-wide Tackling Tobacco Coordinator who was a respected elder employed at AHCSA. He championed the project in his role and acted as a cultural mentor for the researchers. Health services that employed AHWs were contacted including both community-controlled Aboriginal health services and government-controlled health services. Fifteen site visits to metropolitan, regional and remote health services were undertaken in conjunction with the Tackling Tobacco Coordinator where possible, to provide the research team and the project with credibility. The recruitment strategy was guided by two Aboriginal members of the investigating team who advised when to visit health services based upon their knowledge of local challenges, opportunities and competing organisational demands. During site visits, the team presented information about the research project and engaged in informal discussions about smoking with local health staff. As a sense of trust developed, AHWs and health service management were invited to formally participate in the study. Tobacco control personnel were recruited individually (via telephone) and interviewed in the workplace in one instance and via telephone in another instance. Recruitment continued until the sampling plan had been achieved and data saturation occurred.
Semi-structured interviews were the primary data source. Thirty-four interviews were conducted between August 2009 and August 2010, including 20 interviews undertaken with government-controlled health service employees and 14 interviews undertaken with community-controlled health service employees. Interviews were conducted until no new data emerged, with the final three interviews yielding no new information. Interview duration (mean 51 [range 16–133] minutes) was dependent upon availability of time and readiness to share information. Table
1 describes the 23 AHWs, 9 other health service staff and two tobacco control personnel that participated.
Table 1
Participant characteristics of stakeholders who participated in interviews (n = 34)
Participants | 23 | 9 | 2 | 34 |
Gender (M/F) | 10/13 | 3/6 | 2/0 | 15/19 |
Location | | | | |
Metropolitan | 9 | 2 | 1 | 12 |
Rural | 13 | 6 | 1 | 20 |
Remote | 1 | 1 | 0 | 2 |
Ethnicity | | | | |
Aboriginal | 23 | 3 | 2 | 28 |
Other | 0 | 6 | 0 | 6 |
Smoking History | | | | |
Never Smoked | 1 | 5 | 0 | 6 |
Ex-smoker | | | | |
Action | 3 | 0 | 0 | 3 |
Maintenance | 6 | 0 | 2 | 8 |
Current smoker | | | | |
Precontemplation | 3 | 0 | 0 | 3 |
Contemplation | 7 | 3 | 0 | 10 |
Preparation | 3 | 1 | 0 | 4 |
There were five interview schedules developed to guide semi-structured interviews, including four for AHWs with a smoking history (i.e. ex-smokers, smokers not ready to quit, smokers unsure about quitting, and smokers ready to quit) and a final one for both non-AHWs and AHWs who had never smoked. The interview schedules enquired into AHWs’ personal history of smoking (e.g.
Can you tell me how you started smoking?), their reasons for smoking (e.g.
Can you tell me about why you smoke now?) in addition to the situational circumstances of smoking for typical week and weekend days including when they smoked, with whom, where the smoking occurred, and the factors that triggered smoking (e.g.
Can you walk me through a typical day during the week to describe your smoking, starting from when you get up?). Questions further explored experiences with quit attempts. Quitters were asked to reflect on their previous smoking history and discuss their experiences of quitting. Health service managers and AHWs who had never smoked were asked about smoking in the workplace, using questions such as “
Do many people at the health centre smoke? Where and when do they smoke?”. The interviewers used clarificatory, exploratory, amplificatory and explanatory probes [
26] to understand the smoking experience and the meaning ascribed to the individual, social, cultural and environmental factors related to smoking.
Focus group discussions were used to clarify emergent findings from the analysis of interview data and provide greater insight into understanding the organisational, social and community contexts related to smoking. The dynamic of group discussions provided a collective perspective regarding smoking in AHWs and the acceptability of smoking in the health service that was not illuminated in individual interviews. The discussion schedule for the focus groups was compiled after preliminary analysis of the first 22 interviews and was internally reviewed by an Aboriginal team member. Participation by AHWs as well as a range of other health service staff was welcomed in order to capture a broad range of perspectives. To recruit participants to focus groups, we visited health services, presented information about the study, and then invited all staff present to participate in a group discussion around smoking. In total, 17 health service staff (including 4 AHWs, 3 home and community care workers, 2 trainee enrolled nurses and 8 other health service staff) who were 53% female participated in three focus groups (mean 57 [range 45–71] minutes) conducted in regional locations in April and May 2010.
The interviewers were non-Aboriginal non-smokers who were experienced in conducting individual and focus group interviews and in undertaking research in Aboriginal communities. Interviews and focus groups were digitally recorded and transcribed verbatim by an external provider. Audio recording was not mandatory, and two participants elected not to be recorded. Further, one interview with a regional participant was conducted over the telephone. In these instances notes taken by the interviewer were used in analysis. All volunteers were provided with information about the nature of the research and gave written informed consent. Ethical approval for the study was granted from the Aboriginal Health Research and Ethics Committee, Aboriginal Health Council of South Australia, in addition to the Human Research Ethics Committees of the University of South Australia and SA Health, Government of South Australia.
Data analysis
Analysis was undertaken using Nvivo 8 software (QSR International Pty Ltd, 2008). First, digital recordings were reviewed to ensure that the mood and emotions of the respondents were held within the data during analysis. Next, a brief narrative summary was generated for each interviewee. Transcribed text from interviews and focus groups and notes taken during unrecorded interviews were then analysed using a qualitative content analysis approach, based on the method of Graneheim and Lundman [
27]. Sections of transcribed text that dealt with similar issues or fields of inquiry were first divided into content areas. Next, data were divided into meaning units that included words, sentences or paragraphs related in content and context. Thereafter, categories were created that represented internally homogeneous and externally heterogeneous [
28] groups of meaning units that expressed the manifest content of the data [
27]. Following the advice of Kitzinger [
29], the views of individuals were distinguished from group consensus. Once the manifest content of the data had been analysed, the underlying themes evident in and emerging from the data were also identified. Themes are “a thread of an underlying meaning through condensed meaning units, codes or categories, on an interpretative level” (p.107) [
27].
The first author conducted an analysis of the full dataset. Next, MC undertook a secondary review of the subdivision of data and the content and labelling of each category. A series of discussions were held between AD and MC to refine the content, labelling and meaning of categories. Given that AD and MC are non-Aboriginal researchers, further review by Aboriginal team members and AHWs was paramount. Hence, AD and MC consulted with an Aboriginal investigator (AC) and the Tackling Tobacco Coordinator (HS) to review and refine the categories. AHW representatives who were members of the Aboriginal Primary Health Care Workers Forum were then presented with the findings and invited to make amendments. An example of a revision that occurred during this process was an amendment to a family-level factor from ‘dis-integration of family’ to ‘breakdown of family dynamics’. Finally, collective clarification was sought regarding policy level factors at a meeting of the project’s Working Group (that included AHWs, health service chief executive officers, tobacco control personnel and other key personnel from tobacco-related community organisations). The active engagement of Aboriginal partners in the data interpretation process enriched the findings and acted to further strengthen relationships.