The Delphi method [19] is a way of determining the consensus of a group of experts on a particular topic. It is particularly helpful in developing guidelines where the use of other research methods are not appropriate, e.g. randomised controlled trials. Development of the current guidelines involved four steps: (1) formation of the expert panel, (2) literature search and survey development, (3) data collection and analysis, and (4) guidelines development.
Step 1: Panel formation
In line with other similar Aboriginal and Torres Strait Islander mental health first aid Delphi studies (e.g. [12]), this study utilised one expert panel consisting of professionals with experience researching or treating gambling problems in Aboriginal and Torres Strait Islander people. The decision to use only one expert panel was made because the field of Aboriginal and Torres Strait Islander gambling is small and it was thought that it would have been difficult to recruit enough Aboriginal and Torres Strait Islander people to a ‘lived experience’ and ‘affected other’ panel to produce meaningful results. Therefore, the selection criteria were:
-
18 years or over, AND
-
A gambling help professional or researcher who is informed about Aboriginal and Torres Strait Islander gambling, AND
-
Have a minimum of 2 years’ experience in the field of Aboriginal and Torres Strait Islander gambling problems.
The aim was to recruit a minimum of 30 people to the panel, which is within the typical Delphi panel size of 15–60 experts [20], allowing for reliable consensus to be reached.
Twenty-two participants with experience in working with Aboriginal and Torres Strait Islander people with gambling problems completed all three rounds. The participants had an average age of 46.4 (SD 10.48) and 12 were female and ten male. They were from the following Australian states and territories: ACT (n = 1), NSW (n = 2), SA (n = 1), VIC (n = 7), NT (n = 2), QLD (n = 6), TAS (n = 1) and WA (n = 2). They worked in an Aboriginal gambling service (n = 7), a general gambling service (n = 5), a local health service (n = 4), a community service (n = 1), or other mental health setting (n = 3), or as a researcher (n = 4) (note: some may have worked in more than one service). Seven of the participants were Aboriginal and none were Torres Strait Islander. Three participants had experienced gambling problems themselves, while 10 had supported a family member and 11 a fellow community member who experienced gambling problems. The retention rate for participants completing all three rounds was 84.6% (26 participants completed Round 1).
Step 2: Literature search and survey development
The Round 1 survey included two types of questions – items from a previous international study to develop guidelines for assisting people with gambling problems in developed English-speaking Western countries [17] and items derived from a targeted literature search described below. The methodology of the study to develop the gambling guidelines for English-speaking Western countries is described in detail elsewhere [17]. Three hundred and forty-seven items from the previous survey that received a consensus rating of at least 50% were used in the Round 1 survey of this current study.
In order to further inform the content of the initial survey, a systematic search of the ‘grey’ and academic literature was conducted in July 2015 to gather statements about how to help an Aboriginal or Torres Strait Islander person with gambling problems. The website search was conducted using Google Australia, the book search was conducted using Google Books and the journal search was conducted using Google Scholar and PubMed. See Table 1 for the search terms.
In line with other similar Delphi studies (e.g. [17, 21, 22]), the first 50 websites, 50 books and 50 journal articles were retrieved. The decision to examine the first 50 for each search term was based on a previous Delphi study that found that the quality of the resources declined rapidly after the first 50 [23]. After duplicates were removed, the remaining sources were reviewed for relevant information, as were any links appearing on the websites. Websites, articles and books were excluded if they did not contain information about how a member of the public can recognise and help an Aboriginal or Torres Strait Islander person who has gambling problems. A total of 24 websites, articles and books were included and used to develop the Round 1 survey. Figure 1 summarises the results of the literature search.
A working group, consisting of staff from Mental Health First Aid Australia, the University of Melbourne and an Aboriginal and Torres Strait Islander mental health first aid expert (who is Aboriginal) translated the results from the literature search into helping statements that were clear, actionable, and contained only one idea. These statements, plus the items from the recent Delphi study to develop guidelines for helping a person with gambling problems from a developed English-speaking Western country [17], were used to form the first of three questionnaires that were administered to the expert panel via SurveyMonkey.
In this study, a distinction was made between the subclinical symptoms of problem gambling and gambling disorder (a diagnosis). We use the term gambling problems, defined as gambling activities where the person struggles to limit the amount of time or money spent on gambling, leading to adverse consequences for the person, their family, or the community. This includes someone whose gambling problems are at a clinically diagnosable level [10]. This definition was used because it is not feasible or preferred that members of the public (e.g. family or friends) diagnose pathological or disordered gambling, and because the study sought to identify the signs of a range of gambling problems (from at risk gambling through to problem gambling). Also, if family, friends and co-workers can recognise, identify and address gambling problems earlier, severe gambling harms may be prevented.
Step 3: Data collection and analysis
Data were collected in a survey administered over three rounds between January and April 2016. In the survey, participants were asked to rate each of the helping statements, using a 5-point scale (‘essential’, ‘important’, ‘don’t know/depends’, ‘unimportant’ or ‘should not be included’), according to whether or not they thought the statement should be included in the guidelines. In Round 1, participants also could provide qualitative data in the form of suggestions for new helping statements. See Additional file 1 for copies of the three rounds of the survey.
The statements were analysed and categorised as follows:
-
1.
Endorsed. The item received an ‘essential’ or ‘important’ rating from 90 to 100% of participants.
-
2.
Re-rate. The item received an ‘essential’ or ‘important’ rating from 80 to 89% of participants.
-
3.
Rejected. The item did not fall into either the endorsed or re-rate categories.
These cut-off criteria were chosen by the working group because there was only one panel and lower cut-off percentages would have yielded too many statements, making the guidelines impractical to use.
The following criteria were used to determine whether the participants’ comments would be translated into new helping statements: (1) the idea was actionable and understandable, (2) it was not a repeat of an item in the first survey, and (3) it was within the scope of the project. This new content was translated into helping statements for the Round 2 survey. The Round 2 survey also included Round 1 items that needed to be re-rated. Participants were given a summary of Round 1 that included a list of the items that were endorsed and rejected, as well as the items that needed to be re-rated in Round 2. The summary included the panel percentages of each rating, as well as the specific panel member’s scores for each re-rated item. This allowed the participants to compare their ratings with the expert panel’s consensus rating and decide if they wanted to maintain or change their answer when re-rating an item.
The procedures for Rounds 2 and 3 were the same as Round 1 with several exceptions. Round 2 consisted of new items from the Round 1 comments. There was no opportunity for comments in Round 2 or Round 3, and if a re-rated item did not receive an ‘essential’ or ‘important’ rating by 90% or more of the panel, it was rejected. Round 3 only contained items introduced in Round 2 that needed to be re-rated, according to the above criteria.
Step 4: Guidelines development
The endorsed items were written into continuous prose to form the guidelines. The first author drafted the guidelines and the working group edited the draft to produce the final guidelines document. This document was presented to the expert participants for comment and final endorsement.