The process used to identify and classify studies was consistent with Cochrane methods for systematic searches [5].
Inclusion criteria
To capture evidence of the transfer and implementation of Indigenous Australian health services and programs, studies were included in this review if they evaluated, described or reviewed Indigenous Australian health services or programs and were published between 1992 and 2011 (inclusive) in the peer review or grey literature. A substantial proportion of Indigenous health research is published in the grey literature, making it an important source [6]. In cases where a relevant study was published in both the peer review and grey literature, we included the grey literature only if it referred to a discrete aspect of a service or program not included in its peer reviewed counterpart. Services, programs and innovations were defined as systematic actions and approaches taken to address an identified Indigenous health need [7]. Health was defined broadly according to the Indigenous Australian definition which includes physical, mental, emotional and spiritual wellbeing [8].
Search strategy
A two-step search strategy, summarised in Figure 1, was utilised. First, electronic databases Informit, Infotrac, Blackwells Publishing, Proquest, Taylor and Francis, JStor, Medline and the Australian Indigenous HealthInfoNet were separately searched (last date: 25 November 2011) for citations that included the following terms in the title, abstract or MeSH heading: Aborigin* OR Indigen* OR Torres AND health AND service OR program* OR intervention AND Australia. We identified 1554 references (after removal of duplicates). Second, the reference lists of 19 Indigenous health-related literature reviews, identified through database search, were examined. This process identified an additional 75 references.
Classification of studies
The 1629 references identified in step 1 were classified in a four step process.
Step 1: Identification of studies for exclusion: We excluded studies that were 1) not Indigenous-Australian-specific; 2) not related to the provision of a service or program; or 3) duplicates. Given that some services and programs changed their names during the 20 year timeframe and were cited in different ways, the elimination process may have underestimated the number of duplications. Step one excluded 318 publications.
Step 2: Identification of transfer studies and type of transfer: The remaining 1311 references, which documented 1098 programs and services and 19 reviews, were entered into an Excel spreadsheet. They comprised 309 peer reviewed papers and books/book chapters and 1002 reports and websites. Abstracts were searched by one author (JM) to classify studies according to whether transfer and implementation was: 1) the focus of the study, 2) considered as one of several key themes, or 3) not addressed. If an abstract suggested (but did not make explicit) transfer, the conclusions were also searched. Step 2 identified a total of 119 “transfer studies” (9.1% of 1311). Transfer studies (n = 119) were further classified by three authors (KT, RB and JM) to identify the extent to which they focused on the transfer and implementation of a health service or program in Indigenous healthcare settings, with an initial inter-rater agreement of 82.4%. The studies for which there was a discrepancy were re-evaluated until consensus was reached by the three authors. The process of transfer described in the studies which focused on transfer was classified according to the theory described previously as a: 1) hierarchical, centrally-driven; 2) decentralised and participatory, or 3) informal grass roots process [2].
Step 3: Classification of studies: The 119 transfer studies (which documented 97 services or programs) were then classified as evaluative or descriptive studies. Impact/outcome evaluation studies were defined as those that informed understanding about the effectiveness or acceptability of Indigenous health services or programs. Process evaluation studies were those which measured reach, satisfaction, quality and implementation (how to produce change). Descriptive studies were “descriptions of methods or processes .... in which no data-based evaluation was reported” [9, 10]. Studies which reported both process and impact/outcome measures were classified as impact/outcome evaluations. Step 3 found 37 studies (31.1% of 119) which reported impact measures.
Step 4: Quality of studies: The likely extent of scientific rigour of the 37 impact/outcome evaluation studies was assessed in terms of whether they: 1) had been peer-reviewed, and 2) used an experimental design. Peer-review was included as a quality indicator since papers published in the scientific literature have been subject to peer review while those published in the grey literature most likely have not. As per Sanson-Fisher et al.. [9], peer-reviewed studies were then classified as either controlled experimental designs (randomised and non-randomised controlled trials) or non-experimental (cohort/longitudinal analytic studies, case–control studies, single group pre-post or post-evaluation measurement, and other).
Statistical Analysis
The 95% confidence intervals (95% CI) for proportions were calculated as exact binomial confidence intervals. Test comparisons between counts and proportions were conducted using SPSS (IBM) version 20 and Fisher’s exact test with an alpha level of 5%.