Here results of consultations are summarized and the general principles and key features underpinning the survey items and their delivery in the App are described.
General principles
Consultation suggested that the App would need to:
-
1)
Be suitable for an Indigenous Australians aged from 16 to old-age, including those who are unfamiliar with computers or tablets;
-
2)
Be suitable for individuals in urban through to isolated or traditional areas;
-
3)
Help individuals to be comfortable telling their drinking story (e.g. how often or how much);
-
4)
Reassure participants of confidentiality;
-
5)
Work offline, then data from each iPad can be ‘pushed’ to a secured computer server at the University of Sydney when WIFI is available;
-
6)
Provide a de-identified summary of completed surveys periodically (sex, age, community, drinking status), with remote access to data for principal investigators; and
-
7)
Be comparable with (some items of) existing national and international alcohol surveys or screening tools.
Suitability of existing approaches to measure alcohol consumption
Examples of national and international alcohol surveys were reviewed, and potential items that might be adapted for use on the App were discussed with Indigenous and non-Indigenous experts. International tools included: Alcohol Use Disorders Identification Test (AUDIT) [15]; Composite International Diagnostic Interview Version 7.1 (CIDI) [16]; Alcohol, Smoking and Substance Involvement Screening Test Version 3.0 (ASSIST) [17]; 2007 Gender Alcohol and Culture: An International Study Survey Version 6.1 (GENACIS) [18]; and the International Alcohol Control Policy Evaluation Study (IACS) [19]. Australian tools included: the 2008 National Aboriginal and Torres Strait Islander Social Survey (NATSISS) [20]; the 2013 National Drug Strategy Household Survey [21]; the Indigenous Risk Impact Screen (IRIS) [22]; the Harms From Others’ Drinking Study [23]; and a community survey on alcohol consumption in Indigenous populations in remote Western Australia [24].
Two other internationally validated approaches to assessing alcohol use were considered, as they appeared to have particular relevance to Indigenous Australian contexts. The ‘Timeline follow back’ approach [25] encourages the individual to recall where they were and who they were with, to help elucidate a detailed history of drinking. The interviewer works backwards, day-by-day for the past month or up to a year. This recounting of real life context and linking of drinking to events and people was considered relevant and approachable in an Indigenous context. However, substantial time is needed to use this approach in its entirety. Another approach, the “Finnish” method, only enquires into the last four drinking occasions [26] and so is less time consuming. Again, the Finnish method asks the person to think over the events or context that were associated with drinking, and does not assume regularity of drinking pattern. However, it is possible that the past four drinking occasions may not be typical of the rest of the year: for example, a person might have 4 days of heavy drinking associated with a trip to the city, while the rest of the year was spent in a ‘dry’ (alcohol-restricted) area.
Item selection was then guided by Indigenous and non-Indigenous colleagues’ advice, for example that:
-
1)
It should take no more than 20 min to complete the survey App (due to competing demands on participants);
-
2)
The app should make it easier for individuals to describe their drinking without requiring skills in numeracy or literacy; and
-
3)
Survey items should cater to a range of drinker types (e.g. including those who drink episodically).
For quantifying alcohol consumption: 10 items are enquired into. This consisted of:
-
1)
Any alcohol consumption in the last year;
-
2)
Frequency of consumption in the last year (The Alcohol Use Disorders Identification Test [AUDIT] Q1-modified);
-
3-6)
The frequency and timing of the last four occasions of drinking and what was consumed, using the ‘Finnish method’ [25], combined with elements of ‘time line follow back’ [24] to help participants remember where they were and who they were with in each drinking occasion;
-
7)
Reasons why participants sometimes might drink more;
-
8)
Quantity and types of alcohol consumed in a heavy drinking occasion (24-h period);
-
9)
Length of the longest period of no drinking in the last year; and.
-
10)
Frequency of consuming five or more standard drinks (24 h period) in the last year? (AUDIT Q3-modified) with visual cues to quantify.
We were advised that AUDIT-Q2 was problematic as it asks into ‘usual’ consumption, and in some traditional regions, the concept of ‘usual’ does not exist. The app also collected other data on alcohol use behaviours that are the not the focus of this paper (on: alcohol dependence, harms to self or others and treatment access).
Presenting questions in a conversational way
Indigenous colleagues and other clinicians stressed the importance of asking questions in a conversational manner. This was based on clinical and research experience, and on the work of one author (Assan and colleagues) on training clinicians in the use of the Indigenous Risk Impact Screen (IRIS; a screening tool for alcohol, drug and mental health issues developed by and for Indigenous Australians [22]). Accordingly, sensitive survey items were introduced with a short scenario (where appropriate), to assist the individual to reflect on their own life experience (see Table 2).
To ensure privacy, each participant would be presented with an iPad and headphones and be supported by an Indigenous research assistant to open the survey. The individuals would then work through the questions, with a research assistant sitting a little distance away in case problems or questions arose.
Reference points used to ‘anchor’ time periods
Time is not universally understood as a linear concept in Australia [10, 11]. In traditional communities time of year may be marked more by the seasons or a tree flowering, or times when shops are shut rather than by a calendar. So reference points were used to help individuals to anchor their answers in time. Based on a small group discussion focused on this issue at the workshop, reference points that are widely recognised across Indigenous Australia were agreed upon. As a result, the ‘last 12-months’ is divided into quarters with the help of four key time points: a) Christmas or New Year (December/January); b) Easter (April); c) National Aboriginal and Islander Day of Celebration (NAIDOC) week (July); and d) Australian Football League (AFL) or National Rugby League (NRL) grand finals (September/October; see Table 2). There was consensus that individuals who do not celebrate Christmas or Easter, or who do not follow sport, would know when in a calendar year these events occur.
The survey app calculates which reference point (for ‘in the last 12-months’) to use depending on the date when the App is being completed. This then enables an individual to focus on what they were doing, for example, at ‘Easter last year’, rather than trying to remember what they were doing ‘12-months ago’. A visual timeline was used to allow respondents to select dates moving back in time, of their four recent drinking occasions. The reference points are converted to dates ‘behind’ the App for data analysis.
Response categories for questions on frequency of drinking
Indigenous colleagues and other clinicians advised that response categories typically used in alcohol surveys posed difficulties, as they are reliant on individuals counting days, weeks or months [10]. Instead, modified response categories were used that included colloquial English that would be commonly understood by the target population (e.g. ‘once in a blue moon (less than once a month)’ instead of ‘less than monthly’; see Table 2).
Asking about pattern and quantity of drinking on the iPad App
The last four occasions approach [26] was adapted for a user-friendly and visual approach. This combined elements of ‘Timeline follow back’ [25], and was seen by our advisers as compatible with a conversational or story telling approach.
On the iPad screen, a retrospective “grog diary” appears as a strip. The participant selects when (in the last 12-months) each of their last four drinking occasions occurred. The time periods displayed on the first screen are: “Yesterday, 2 days ago […. up to], 1 week ago etc”. The user moves backwards in time to select the day. The App uses the timing of drinking and the quantity selected (see below) to calculate average quantity consumed.
In addition, to better describe drinking which may stop and start according to geographic location or circumstance, participants are asked about their longest gap without alcohol in the last 12-months (indicating the actual length of time using the same retrospective grog diary and nominating reasons for this dry period).
Each individual is also asked about a ‘heavy’ drinking occasion: “In the last 12 months, when you drank a lot of grog, would you ever drink more than [this--]?”. An image depicting the largest amount of alcohol that the person reported consuming in the last four drinking occasions is then shown. If the person responds that they sometimes drink more than that amount, they can select items of alcohol to describe their level of consumption at that higher level of drinking. The individual then reports how often they drink a lot for them (see ‘Response categories for questions on frequency of drinking’ above).
Identifying the type of alcohol a person drinks
Some drinkers are not familiar with the names for some categories of alcohol type, for example, ‘fortified wines’, but rather they identify type of alcohol by its brand or container. To address this issue, a simplified classification of alcohol types was agreed on: beer, wine, port or sherry, spirits or other. Pictures of common local brands in each alcohol type would be displayed. The ‘other’ category included cocktails, methylated spirits and drinks not listed elsewhere.
A listing of common alcohol brands and drinking containers was created for each alcohol type in each surveyed state. With the help of colleagues from those states, this guide was refined to reflect popular alcohol brands but also sufficient choice in each alcohol type. State-based drinking preferences were reflected. For example, “XXXX” was a beer choice made available for Queensland individuals, but instead, “West End” appeared in SA. It was not possible to present every choice, so research assistants were instructed to encourage participants who cannot find their choice to select an alcohol type of similar strength to their preferred brand, or to choose the “other” category.
Beer posed particular challenges as Indigenous colleagues and other clinicians reported confusion around terms such as ‘regular strength’ versus ‘mid strength’ or ‘low alcohol’ [21]. The term ‘low carb’ was sometimes incorrectly understood to mean ‘low alcohol’. To reduce confusion, pictures of several actual beer brands were used. Brand recognition is typically strong. For example, workshop participants advised that in more isolated settings, some types of alcohol are known by the colour of the packaging. So, if hand drawn images were to be used, extra care would need to be taken to ensure comprehension of brand names in different geographical settings.
Drink containers
Indigenous colleagues stressed the importance of offering a broad range of containers from which alcohol might be consumed. For example, many individuals do not drink wine from a wine glass, especially in remote communities. Instead they may use a container sold for other purposes, ranging from a pannikin (metal mug; 355 mL), slurpee/slushy cup (490 mL), empty water bottle (600 mL), through to a large soft drink bottle (1.25 L; see Fig. 2).
Working out individual consumption based on a share of what the whole group drank
Indigenous and non-Indigenous clinicians reported that when collecting an alcohol history, some clients spontaneously report what the whole group had to drink, rather than on what they alone consumed. The clinician then assists the individual to estimate their share. So, when asked about the last drinking occasion, the App enables the participant to choose to describe what they consumed as an individual or to describe what the group drank (see Fig. 3).
After selecting when in the last 12-months the last drinking occasion took place (on the retrospective grog diary), an individual is asked: “How many people were you drinking with?”. A ‘slider’ (from left to right) enables the individual to select: “Just you” through to a group of “10+ people”. The accompanying image changes as the slider is moved to show the number of people in the drinking group. The individual then chooses 1) To work out how much the whole group drank or 2) To work out how much they themselves drank.
Then, for example, if an individual selects that the last time they drank, the group consumed ten cases or slabs (24-cans) of full strength beer. The App then asks: “Sometimes when people drink together they all have the same amount. Other times some drink more than others. That day, what was your share?” In the screenshot below (Fig. 3), an individual reported a total of five people in their drinking group. So, the next screen defaults to showing an image of two out of ten cases as that individual’s share (i.e. the App assumes an equal share of beer on this occasion). The individual can then slide a meter to adjust their individual portion. Invisible to the user, each container is divided into 10 to calculate standard drinks. For example, of the ten cases of full strength beer consumed by the group, the individual consumed two cases (see Fig. 3). Estimated at 4.8% alcohol/volume at 0.789 g/mL, this equates to 45.5 standard drinks consumed by that individual on that occasion.
Modifying AUDIT Q3
This item was modified to be in keeping with Australian drinking guidelines (i.e. to ask about consuming five drinks or more on an occasion) [7]. Also, instead of relying on participants to convert what they drink into ‘standard drinks’, the App dynamically produces an image of 50 g or more of ethanol based on the type of alcohol that each individual consumed the most of (i.e. the maximum grams of alcohol) from their last four drinking occasions (see Table 2).
Image and audio options
When a survey is started, each participant identifies their gender. The images, audio and lifestyle references are then matched with that gender. For example, female participants predominantly see women and girls in the images on screen and the audio is spoken by a woman. Original artwork featured on the app was commissioned from an artist employed by the Aboriginal Drug and Alcohol Council SA.
Two language options are offered in the Stage 1 version of the App – English or Pitjantjatjara, a language of Indigenous Australians used in a region of South Australia, Northern Territory and Western Australia. To begin with, two experienced Indigenous alcohol and other drug professionals recorded the English female and then the English male audio. During recording, there was further refinement of the wording of survey items, for example, to check that phrasing was comfortable. Suggestions made by these clinicians during the recording process were checked at the time with three researchers (Lee, Conigrave and Perry).
The Pitjantjatjara language program at the University of South Australia facilitated translation of survey items from English to Pitjantjatjara (including ‘back translation’ of key items). Audio was provided by two Pitjantjatjara language speakers/interpreters (a male and a female). This team and a researcher (Perry) met face-to-face to workshop the translation of survey items. Where there was differing opinion, clarification was sought on intended meaning from a researcher (Perry or Conigrave), until consensus was reached. Efforts were made to ensure suitability of survey items for Pitjantjatjara speakers from adolescence to old age. Further workshopping of survey items occurred in the recording studio. To ensure consistency, one researcher (Perry, who speaks some Pitjantjatjara) was present for one of the English and both of the Pitjantjatjara (male and female) recordings. Suggestions made during the recording process were checked at the time by two researchers (Lee and Perry).