Setting
The community setting for this healthy lifestyle intervention program was a remote multilingual Aboriginal community (over 20 language groups) of an approximate population of 2,500 in Northeast Arnhemland, Australia. Access to the community is restricted: permission to visit is required by law. Air- and water-based travel between the community and mainland locations is costly. The largest source of income is derived from government payments, although many community members acquire paid employment through a Community Development Employment Program (CDEP). The community was established by the Methodist Overseas Mission in the early 1940's and, while experiencing influences of westernisation since then, community members have retained traditional cultural practices. One of the most pressing issues for community organisations has been the retention of culture for future generations in the face of westernisation, but overall community life remains rooted in traditional culture[
18]. The community has diverse services including cultural, educational, recreational, and health services provision.
The Healthy Lifestyles Project
The Healthy Lifestyles Project was initiated as a participatory research project [
19] between the target community and Menzies School of Health Research in response to a community-wide concern about the escalating burden of chronic disease. A planning committee for the Healthy Lifestyles Project, representing various community groups and agencies, formed in July 2001. This led to the voluntary screening of community members aged 15 years and older for type 2 diabetes and cardiovascular risk factors [
20]. Screening was completed in March 2002. Community-wide feedback and discussion of screening results was then used as the basis for developing and implementing intervention strategies.
To prevent CVD and the development of type 2 diabetes, the Healthy Lifestyles Project actively promoted a healthful diet, physical activity and smoking cessation and prevention. Part of the original intent of the project was to activate and enable a coordinated community-directed approach to increase the allocation of community resources to prevention activities. Existing community initiatives that supported these healthy lifestyle messages were identified. The aim was to build and support these initiatives by strengthening inter-organisational linkages. Many community organisations and agencies became involved in planning and implementing activities that advanced project goals.
Data Collection
Prevention activities implemented in the first three intervention years of the Healthy Lifestyles Project (between January 2002 and January 2005) were examined. A two-page activity monitoring form aided collection of data on intervention activities at regular intervals. Forms were completed through a researcher-assisted interview and participant observation process. Open-ended questions were used to obtain information on activity objectives, an activity description, participant recruitment, and to identify organisations that were taking a primary role in decision making related to activity planning and implementation. Given the decentralised approach to CVD and type 2 diabetes prevention programming in the community, interviews were also conducted with members of diverse organisations and professionals working in the local health centre and other agencies. Interviews commenced with health centre staff and representatives of community organisations and government agencies; other organisations and groups involved in implementing healthy lifestyle interventions were identified through a snowball sampling approach which continued until no new organisations implementing healthy lifestyle activities could be identified. Some interviews were done in small sharing circles with 3-4 representatives of the same organisation. This information was translated onto each activity implementation form during the course of the interview. The interviewer's interpretation of the information was verified by those interviewed - either verbally or by reviewing a hard copy of the monitoring form. Data were collected from approximately 30 community stakeholders. Documents such as organisational reports, community newspapers, and program materials were collected to identify activities that may have been missed (resulting in follow-up interviews to properly complete activity implementation forms) and to verify information for those activities already identified. Strong social networks between community members and researchers facilitated the identification of relevant activities.
The protocol was reviewed and clearance provided by the Human Research Ethics Committee of the Northern Territory Department of Health and Community Services and Menzies School of Health Research.
Coding Scheme
Following the ecological coding scheme developed by Richard [
9,
12] and refined by Lévesque [
10], information obtained on the activity description and objectives was used to code for intervention
targets and intervention
strategies, information on participant recruitment was used to code for intervention
setting, and the information on organisations was used to develop an index of organisations taking a lead or primary role in activity planning and implementation.
Intervention targets refer to the sub-group in the community intended to benefit from the intervention, or for whom health behaviour change was designated. Five types of targets are identified: 1) Individuals (IND), 2) Interpersonal environment (INT), 3) Organisations (ORG), 4) Community (COM), and 5) Political players/systems (POL). In this scheme, the health program is annotated as the HP.
All intervention activities implemented by the HP are directed towards an ultimate target, or IND. The IND represents the primary individual beneficiary or those ultimately designated for change. Thus, health education activities like making brochures available for community members in health clinics on the health risks of smoking are annotated as HP→IND. To be consistent with the communities' prevention efforts, the Healthy Lifestyles Project designated two ultimate targets for their intervention activities: children and community members aged 15 years and older. However, if a single activity was directed towards both groups, children were coded as the ultimate target. Activities were analysed respecting the ultimate target of each activity.
In the case of an intervention strategy including more than one intervention target, a distinction is made between the ultimate target and
proximal target(s). A
proximal target represents any intermediate entity or entities (i.e., INT, ORG, COM or POL), designated for change through implementation of a given intervention activity. Where there is a proximal target designated for change (X), the intervention pathway is specified as 'indirect' and intervenes on the ultimate target through another medium (e.g., HP→ X → IND). For example, children's eating habits at home (IND) can be influenced by engaging mothers in workshops to build their knowledge and skills to purchase healthy food at the grocery store (INT). This, then, gives HP→INT→IND. A networking strategy involves the linking of at least two targets by the program team (HP→[X-X]→IND). One such example is bringing together organisations in a coalition to create a bike path to benefit school children (HP→[ORG-ORG]→IND) [
10].
The overall intervention strategy represents the sequencing of one or more targets joined either in a direct transformation relationship (i.e., direct transfer of information or resources to the intended target) or in an indirect relationship (i.e., linking at least two targets). Intervention strategies are aggregated into three categories of activities: 1) traditional health education (HP→IND); 2) networking (HP→[X-X]→IND); and 3) indirect transformation (HP→X→IND).
The intervention setting is defined as the social system(s) in which persons/entities designated for change are reached. Four types of settings were designated in the coding procedure: 1) Organisation, 2) Community, 3) Society (i.e., state/territory or nation) and 4) Supra-national (i.e., link of two or more societies). Since the traditional extended family structure is strong in this community and a potentially important mechanism for reaching community participants, "Family" was added as a fifth setting for coding activities.
The community organisations, institutions and agencies taking a primary or lead role in activity planning and implementation were identified and listed on the activity implementation form. This enabled determination of the number of lead organisations per activity.
Analytic Procedure
The analytic phase commenced with the training of two raters in the ecological coding procedure. Information on the activity monitoring forms was coded independently by both raters. Inter-rater agreement was estimated by coding a random sample of 25 activities. Inter-rater reliability was found to be Kappa = 0.76 (p < .0.001), 95% CI (0.58, 0.94). During the coding process, disagreements were noted and resolved through discussion. Frequencies of intervention strategies, intervention settings and intervention strategy types were assessed by year and for differences across the three intervention years. In addition, each organisation was given a score from 0 to 4, with higher scores indicating higher ecological complexity, based on the algorithm developed by Richard [
9,
12]. A score of 0 was given to an organisation that employed only one intervention strategy, independent of setting and type of strategy. A score of 1 was given to an organisation that employed at least two intervention strategies that did not include HP→IND, regardless of the number of settings in which the strategies were implemented. Scores of 2, 3, and 4 were given to organisations employing an HP→IND intervention strategy and at least one other intervention strategy within 1, 2, and 3+ settings, respectively. The number of lead organisations was assessed by year and across the intervention years as well. PEPI (Version 4) and SPSS (Version 13.0) software were used for descriptive and chi-square analyses. Statistical significance was set at 0.05.