Design & setting
A cross-sectional survey of health care consumers attending outpatient clinics (see Additional file 1) within one large, regional Australian tertiary referral hospital was conducted. Data was collected over a two-month period in 2016–2017. This study was approved by the Hunter New England Human Research Ethics Committee, and registered with the University of Newcastle Human Research Ethics Committee.
Eligible participants were aged 18 years or older; attending an outpatient appointment at the participating hospital, either as a patient or person accompanying a patient; had sufficient English to complete the survey; and judged to be mentally and physically well enough to complete an iPad survey. Younger as well as older participants were included, as younger adults are also likely to be involved in decision making about whether loved ones should be assessed for dementia (e.g. their parents). Participants who indicated they had received a diagnosis of dementia or had completed the survey at a previous appointment were excluded.
Trained research assistants approached potential participants in the outpatient clinic waiting rooms to introduce the study and provide a printed information sheet. Research assistants checked eligibility and provided willing participants with an introduction to the iPad touch screen technology, where required. Participants completed a consent form electronically via the iPad. The age and gender of non-consenters was recorded by research assistants to enable examination of consent bias.
Consenting participants were asked to complete an anonymous online survey on a web-connected iPad while waiting for their appointment. Research assistants provided participants with assistance using the iPad where requested. Participants who were called in for their appointment prior to finishing the survey were invited to continue with the survey after their appointment if they wished to. A unique identifier was used to allow participants to save responses and enable completion of the survey after their appointment. QuON software was used to program and administer the survey via a secure server . QuON is a web-based survey software application that allows for sophisticated branching of questions using pre-programmed algorithms, to ensure survey questions are tailored based on the participant’s previous responses .
Development of the survey involved an iterative process of review and refinement, including: 1) Critical appraisal of the relevant literature examining the advantages and disadvantages of early diagnosis of dementia; 2) A small focus group with six members of the community aged over 65 years with no previous diagnosis of dementia to expand on data from Step 1, and explore perceived advantages and/disadvantages of early diagnosis; 3) Development of survey items and peer review by four public health researchers; and 4) Pilot testing of the final draft survey with an initial 20 participants (recruited as per the methods previously described), with further refinements made as necessary.
The survey questions (see Additional file 2) were embedded within a larger survey which also included questions about health-related decision making. The larger survey took approximately 10 min to complete. The following items are reported here:
Participants self-reported their age, gender, education, employment status, and relationship status.
Experience with dementia
Participants were asked if they themselves had previously been diagnosed with dementia, whether they know someone with dementia, and if so, their relationship to the person.
Preferences for timing of diagnosis disclosure
A lay description of dementia was provided: “Dementia (sometimes called ‘Alzheimer’s’) affects thinking, behaviour and the ability to perform everyday tasks. There is currently no cure or treatment to reverse or stop the progression of dementia.” Following this, two scenarios were presented to elicit participants’ preferences for receiving a diagnosis of dementia. Participants were first asked ‘Imagine you see your doctor and test results show that you have dementia. Given there is no cure, when would you want your doctor to tell you that you have dementia?’. Participants were asked to select one of four response options available for selection (1 = As soon as possible, 2 = Not until my symptoms got worse or made me really worried, 3 = Only when my family thought it was necessary to tell me, 4 = I would not want to be told at all). Participants were then asked to indicate the reasons for their preference. The list of reasons presented were tailored depending on the response selected for the scenario. Participants were able to select as many reasons as applied. Participants who selected response option 1 were presented with reasons associated with advantages of early diagnosis (e.g. So I could have more time to be involved in decisions about my future care), while participants who selected response options 2–4 were presented with reasons associated with disadvantages of early diagnosis (e.g. It would cause me to feel depressed). Participants also had the option of entering their own reasons into the survey. Participants who indicated they had a spouse or partner who did not have a diagnosis of dementia were then presented with the second hypothetical scenario which asked ‘Imagine your partner/spouse sees their doctor and test results show that they have dementia. Given there is no cure, when would you want to them to be told that they have dementia?’. The same response options and follow up questions were presented as described above (with appropriate wording changes).
Statistical analyses were conducted using IBM SPSS Statistics v23  and SAS software v9.4 . Consent bias was examined using chi-square tests to compare age and gender of non-consenters versus participants. Data from the iPad survey was analysed using descriptive statistics (percentages and 95% confidence intervals for the true proportion) to examine preferences regarding the timing of diagnosis disclosure for dementia and reasons for these (Aim 1). Given low cell counts within the preference for disclosure variable, response options regarding preferences for timing of diagnosis disclosure were collapsed into two groups: 1) as soon as possible; and 2) later or no disclosure, which incorporated all other response options. An exact logistic regression was conducted with the binary outcome (as soon as possible vs. later or no disclosure) to explore whether socio-demographic characteristics and previous experience with dementia were associated with preferences for diagnosis disclosure (Aim 2). The following variables were included in the regression as possible predictors: age, gender, education, employment status, and experience with dementia. Agreement between preferences for the self versus preferences for their spouse was examined using a test of tetrachoric correlation (Aim 3).