Investigating the preferences of older people for telehealth as a new model of health care service delivery: A discrete choice experiment

Abstract

Introduction

Telehealth approaches to health care delivery can potentially improve quality of care and clinical outcomes, reduce mortality and hospital utilisation, and complement conventional treatments. However, substantial research into the potential for integrating telehealth within health care in Australia, particularly in the provision of services relevant to older people, including palliative care, aged care and rehabilitation, is lacking. Furthermore, to date, no discrete choice experiment (DCE) studies internationally have sought the views and preferences of older people about the basic features that should make up a telehealth approach to these services.

Methods

Using a DCE, we investigated the relative importance of six salient features of telehealth (what aspects of care are to be pursued during telehealth sessions, distance to the nearest hospital or clinic, clinicians’ attitude to telehealth, patients’ experience of using technology, what types of assessments should be conducted face-to-face versus via telehealth sessions and the costs associated with receiving telehealth). Data were obtained from an online panel of older people aged 65 years and above, drawn from the Australian general population.

Results

The mean age for 330 study participants was 69 years. In general, individuals expressed strong preferences for telehealth services that offered all aspects of care, were relatively inexpensive and targeted specifically at individuals living in remote regions without easy access to a hospital or clinic. Participants also preferred telehealth services to be offered to individuals with some prior experience of using technology, provided by clinicians who were positive about telehealth but wanted all or some pre-telehealth health assessments to take place in a hospital or clinic. Preferences only differed by gender. Additionally, respondents did not feel that telehealth led to loss of privacy and confidentiality.

Discussion

Our findings indicate a preference amongst respondents for face-to-face pre-telehealth health assessments and, thereafter, a comprehensive telehealth model (in terms of services offered) targeted at those with some technological know-how as a substitute for attendance at hospitals and clinics, especially where these health facilities were far away from older people’s homes. The findings may be usefully incorporated into the design of future telehealth models of service delivery for older people.

Keywords

Telehealth rehabilitation palliative care aged care discrete choice experiment

Introduction

Telehealth involves the use of ‘information and communications technologies to deliver health care and transmit health information over both long and short distances’.¹ There are many ways telehealth technology can be used.² For example, and as demonstrated by our group in the Flinders Telehealth in the Home (FTH) trial,^3,4 a patient may have a consultation with their health care provider via video conference from their home, instead of travelling to an appointment. Other examples include the use of health information technology for the management of chronic disease and medication (e.g. electronic medical records),⁵ use of individual mobile devices for monitoring and transmitting data on physiological indicators following a fall or injury (e.g. smart phones and tablets)⁶ and the use of environmentally based devices for detecting falls in community-dwelling older people (e.g. sensors).⁷

Across different health systems, telehealth approaches can potentially improve quality of care and augment conventional treatment programs. For instance, a recent systematic review assessing the effectiveness, acceptability and costs of interactive telehealth (telemedicine) in differentiated clinical conditions reported evidence of an association between telehealth and improved quality of life (QoL) for heart failure patients.⁸ The review also showed a positive relationship between telehealth and lower glycated haemoglobin and blood pressure in people with diabetes.⁸ The benefits of telehealth have also been observed in terms of improved clinical outcomes,^9–11 reduced mortality,¹² decreased hospital utilisation,^13,14 enhanced patient preferences,¹⁵ increased service and patient time¹⁶ and reduced use of personal assistance services.¹⁷ Evidence from a large recent randomised controlled trial (the UK Whole System Demonstrator Trial) indicated that, despite the absence of evidence of cost-effectiveness,¹⁸ telehealth is associated with lower mortality rates and emergency visits.⁹

As people age they are more likely to need health care, and the majority of individuals requiring specialist rehabilitation and palliative services are older people aged 65 years and above. The Productivity Commission Inquiry Report in 2011 ‘Caring for Older Australians’¹⁹ states that ‘fundamental reform is required’ to respond to current and future challenges that exist in Australia’s aged care system. These challenges include a significant increase in the number of older people, an increasing incidence of age-related disability, and disease and rising expectations about the type and flexibility of care that is received.²⁰ The introduction of a telehealth model of care potentially allows more people from a wider area to access specialist health care services (e.g. in rehabilitation, aged care and palliative care) by negating or minimising the need for travel to service providers.^9,21 Distance from metropolitan services can present a significant barrier to accessing subacute health care services (rehabilitation, geriatrics and palliative care).²² Regardless of geography, travel to and attendance at health appointments can be also stressful and physically taxing, particularly for older people and people with dementia or significant disabilities,²³ as well as their carers.

Research from our group^4,16,24 and elsewhere^17,24–26 has also shown that telehealth services can replace some in-person visits while enhancing patient outcomes. Additionally, there is a growing evidence base for the usefulness of telehealth in providing health and allied services, particularly for older people, with recent systematic reviews reporting benefits in various contexts including aged care (improved QoL),²⁷ geriatrics and gerontology (improved health outcomes and patient satisfaction),²⁸ rehabilitation (improved functional improvement and reduced rick of hospital admission)²⁹ and palliative care (reduced need for hospital admissions and cost savings).²⁹ Several previous studies have focused upon older people’s perceptions and acceptance of telehealth.^30–34 However, less is known about the relative importance of different attributes related to telehealth service provision such as geographic proximity to health services or costs to individual consumers.

This paper reports on the first study internationally to employ discrete choice experiment (DCE) methodology³⁵ to assess older people’s preferences in relation to the salient features of telehealth care. We also sought to identify the extent to which individual characteristics, such as living arrangements and presence or absence of a long-term disability, may or may not influence individuals’ preferences for attributes of telehealth service delivery.

Methods

DCE methodology

DCE is a stated preference quantitative technique originating in mathematical psychology which is designed to establish the relative importance and impact of individual attributes, or characteristics, upon the overall utility of a good or service.³⁶ In a DCE, respondents are presented with a sequence of hypothetical scenarios (choice sets) made up of two or more competing alternatives that vary according to the attributes that define them.³⁵ For each choice set, respondents choose their preferred scenario that, based on the Lancasterian framework,³⁷ represents the alternative with the highest utility.³⁵ From these responses, the probability of choosing an alternative as a function of the attributes and other factors can then be calculated, allowing this method to go beyond the traditional qualitative assessments and provide quantifiable data depicting the strength of respondents’ preferences for particular attributes.³⁵ Unlike traditional ranking and rating exercises, DCE also goes further to provide quantitative information on marginal rates of substitution across attributes, trade-offs between attributes and predicted probabilities of uptake or demand for particular alternatives.³⁵ An added advantage of DCE-generated preferences for alternative scenarios is the ability to apply them within the framework of economic evaluation to inform health care management decision-making.³⁸

Development of attributes and the DCE survey

In a prior qualitative study, a series of qualitative interviews with 17 participants (older patients aged between 60 and 92 years, spouses and carers of patients receiving rehabilitation treatment in Metropolitan Adelaide³⁹) and focus groups with 44 health care and allied services professionals were conducted between December 2012 and November 2013. Of the 44 professionals, 36 were from the rehabilitation and palliative care sector (29 metropolitan and seven country) while eight were from the residential aged care sector (all country). Using thematic and constant comparative analysis,^40–42 consistent with good practice guidelines for the development of attributes and levels for a DCE study,^43,44 salient service characteristics important to participants were determined and formulated into attributes and attribute levels for presentation within a DCE. Initially, eight broad potentially important telehealth attributes were identified, and these were reduced, through discussions with the research team, to six salient service features or characteristics, each with three levels defining increasing gradations of telehealth-based health services, as shown in Table 1. Some of these final attributes align with the most important constructs (in terms of their strength of predicting behavioural intention) within the extended united theory of acceptance and use of technology (UTAUT2) designed to aid understanding of acceptance and use of technology amongst consumers.⁴⁵ First, the telehealth model was a priori assumed to be effective and lead to benefits and, therefore, had a high degree of ‘performance expectancy’, defined as the extent to which using technology will lead to benefits to consumers.⁴⁵ Specific UTAUT2 constructs⁴⁵ reflected within the DCE attributes (with the related attributes in parentheses) were ‘social influence’ (clinicians’ attitude to telehealth), ‘habit’ (technology-experience levels of patients targeted by telehealth) and ‘price value’ (cost of telehealth to you).

Table 1.

Attributes and attribute levels.

Attribute	Description	Attribute levels	Variable name used in models
Aspects of care in telehealth sessions	What sort of aspects of care are clinicians able to pursue during telehealth sessions?	• All aspects of care • Most aspects of care • A few aspects of care	• All_CareAspects • Most_CareAspects • Few_CareAspects
Distance to nearest hospital or clinic^a	How far from your home is the nearest hospital or clinic which could be used as an alternative to telehealth sessions?	• Less than 15 km away • Between 15 and 100 km • More than 100 km	• LessThan_15km • Bet_15to100km • MoreThan_100km
Clinicians’ attitude to telehealth	What is the clinicians’ attitude to using telehealth?	• Very positive • Moderately positive • Uncertain	• Very_Positive • Moderately_Positive • Uncertain
Technology-experience levels of patients targeted by telehealth	What sort of experience of using technology should individuals targeted by telehealth services have?	• A lot of experience • Some experience • No experience	• Lots_Experience • Some_Experience • No_Experience
Assessments related to telehealth sessions	What sort of assessments need to be done face-to-face in the clinic and not via telehealth?	• All assessments • Some assessments • No assessments	• All_Assessments • Some_Assessements • No_Assessments
Cost of telehealth to you^b	How much you need to pay towards the cost of providing telehealth services.	• AU$0 • AU$40 • AU$80	Cost_telehealth (continuous variable)

Note: Effects coding utilised for all variables except for ‘Cost_telehealth’.

^aWhereas proximity of patient residence to health services has been measured in terms of both travel time and distance in the literature (e.g. Bliss et al.⁴⁶), we chose the latter in this study as the former is more variable– for example, it may depend on the mode of transport used. Further, Australian telehealth policy focuses on distance (geographical eligibility) rather than time and specifies that for a location to qualify for telehealth services generally, ‘…there must be 15 km by road between a patient and a specialist, consultant physician, or consultant psychiatrist’.⁴⁷

^bThe levels for this attribute (i.e. AU$0, AU$40 and AU$80) represent, respectively, the lowest telehealth cost possible (with full Medicare concessions), the average MBS fee for telehealth services related to palliative care, aged care and rehabilitation⁴⁸ and the maximum average cost per telehealth consultation charged within the Flinders Telehealth in the Home (FTH) trial^3,4 which funded this study.

A DCE survey was developed for our study population and, following ethical approval for the study granted by the Flinders University Social and Behavioural Research Ethics Committee, the survey was piloted with 10 older people. It was subsequently revised to improve phraseology and question layout, resulting in a final version of the survey which included a preamble that: (i) defined the terms ‘telehealth services’, ‘rehabilitation’, ‘aged care’ and ‘palliative care’; (ii) provided examples of telehealth services available in Australia; and (iii) presented specific examples of telehealth services specific to rehabilitation, palliative care and aged care. The rest of the survey comprised four sections. Section A contained nine attitudinal statements relating to the administration and delivery of telehealth services within rehabilitation, palliative care and aged care, presented using a Likert-scale format. Section B consisted of the DCE whereby individuals were presented with a series of choice questions in which respondents were asked to indicate which of two hypothetical telehealth services (comprising differing levels of the salient attributes) they would prefer, again within rehabilitation, palliative care and aged care. A forced-choice design was used in order to arrive at a more efficient design.³⁵ Sections A and B were considered as complements of each other, with the former used a ‘warm-up’ exercise for the latter. Section C contained questions about participants’ demographic characteristics and Section D was made up of questions on QoL. QoL measures included the EuroQoL EQ-5D 5 level (EQ-5D 5L), which is a generic-preference-based measure of health-related QoL, and the Older People’s Quality of Life brief questionnaire (OPQoL-brief), which is a non-preference-based measure of health-related and broader QoL. For both measures, a higher score is associated with a higher QoL.

A D-efficient D_z-error measure DCE design (with zero priors) was created in Ngene software.⁴⁹ To minimise participant burden, three blocks of six choice sets (each with two alternatives) were developed for administration to DCE participants. An example DCE choice set presented to participants is shown in Table 2. We used a ‘main effects only’ design with interactions tested within the analysis.

Table 2.

Example telehealth choice set for older people.

Attribute	Package 1	Package 2
1. How much are clinicians able to do in telehealth sessions?	Most aspects of care.	A few aspects of care.
2. Distance from nearest hospital or clinic for patients targeted by telehealth	Less than 15 km away from your home.	Between 15 and 100 km away from your home.
3. Clinicians’ attitude to telehealth	Moderately positive.	Very positive.
4. Technology-experience levels of patients targeted by telehealth	Patients with some experience of using technology.	Patients with no experience of using technology.
5. Assessments related to telehealth sessions	No assessments occur face-to-face.	Some assessments occur face-to-face.
6. Costs of telehealth to you	AU$0 per telehealth consultation.	AU$40 per telehealth consultation.

DCE participants and administration of surveys

Older people aged 65 or over were recruited from the Australian general population by ‘PureProfile’, an Australian online panel company that specialises in conducting online polls and surveys with members of the general community. A wide selection of participants were drawn from PureProfile’s panel of online account holders and chosen to enable maximum variation with regards to a number of important characteristics including age, gender, ethnicity, living arrangements, income levels, education level and employment status from both metropolitan and country areas in Australia. The inclusion of country participants provided valuable information about the preferences of people living in country areas for whom telehealth has the potential to improve access to specialist health services. The questionnaires were administered via an online portal in July 2014.

Statistical and econometric approaches

Established scoring algorithms provided by the respective instrument developers were used to score responses to the EQ-5D 5L and OPQoL-Brief. The EQ-5D 5L was scored using Australian value sets developed by Norman et al.⁵⁰ We generated descriptive statistics and conducted statistical tests of differences for responses to statements relating to the administration and delivery of telehealth services, and to questions on participants’ demographic characteristics and QoL.

Our econometric approach was based on a standard random utility framework⁵¹ which assumes that participants choose the alternative that maximises their utility. The respondents were first treated as different subgroups (female or male, whether one lived alone or not, whether DCE participants had gone beyond high school in their education or not, whether or not participants had used the internet for health-related purposes in the three months prior to the study and whether or not they had a long-term disability). The subgroups were chosen because they have been shown to impact on technology-use in the literature.^52–54 After testing for poolability using the Swait-Louviere log-likelihood ratio (SL) test of equal model parameters,⁵⁵ eligible subgroups were then analysed as a pooled sample. Failure to reject the null hypothesis of equal model parameters implies that the preferences of groups tested do not differ in terms of preferences and scale.⁵⁵ The utility (U) of alternative j for individual n in choice set k was specified as:

U_{njk} = β_{n} X_{jk} + ɛ_{njk}

where X was a vector of observed attributes, β was the corresponding coefficient (parameter) and

ɛ_{njk}

was an error term which was assumed to be an independent and identically distributed extreme value.

In the first instance, a traditional conditional logit (clogit)⁵⁶ was fitted to the data. Three types of regression models were then used to account for heterogeneity. Firstly, the heteroscedastic clogit (clogit-het) was used, which tests for observed scale heterogeneity only.⁵⁶ Secondly, the mixed logit (MXL) was applied, which accounts for unobserved taste or preference heterogeneity only,⁵⁷ and finally, the generalised multinomial logit (G-MNL) was used, which takes account of both unobserved preference and scale heterogeneity simultaneously.^58–60 A number of studies have shown that incorrectly restricting preferences to be homogeneous may lead to biased parameter estimates for individuals^58,61,62 with policy implications in terms of the optimal implementation of results from a DCE.^58,63 The statistical fit of the clogit, clogit-het, MXL and the G-MNL was assessed using the Akaike information criterion (AIC) statistic, with lower values implying a superior fit.⁶⁴ The aim was to focus on results from the model with the most superior fit whilst accounting for any heterogeneity present in the data. The predicted probability of each combination of attribute levels being the preferred package was simulated using estimates from the model coefficients.^60,65 We also estimated the proportion of the participant population for whom particular telehealth attributes had a positive or negative effect on their choice of telehealth package.

In all models, observed heterogeneity was explored via subgroup analysis based on individual characteristics. All models were first specified using main effects only and thereafter using both main and interaction effects. The key interactions considered were between telehealth attributes and the respondent’s location when completing the DCE in order to control for any potential effect of location on participant preferences. The location dummy was equal to 1 if one was in a metropolitan area and equal to 0 if not. We also tested the appropriateness of including the cost attribute as a linear and continuous effect within the regression models by adding and then examining the statistical significance of the quadratic term of this attribute. An insignificant quadratic term would suggest that the linear assumption was appropriate.⁶⁶

Internal consistency was measured using a test of non-satiation⁶⁷ based on a choice of restaurant (A or B) presented as a practice question. All levels of attributes for restaurant A (in terms of distance from respondent’s home, menu, size of meal and cost) were better than those for restaurant B. Individuals were only deemed rational if they chose restaurant A. This test was not based on telehealth scenarios because the attributes and levels we included within the telehealth scenarios did not easily lead to a test of dominance of one scenario over another. There was no missing data to account for in the analysis. A significance level threshold of 5% (0.05) was assumed as the criterion for determining statistical significance in all analyses.⁶⁸ All analyses were conducted in Stata version 13.1.⁶⁹

Results

Demographic, internet-related health care use and QoL details

The results relating to the demographic characteristics, health and QoL, as well as selected information on health care service provision, are presented in Table 3. A total of 330 respondents provided data for this study. Slightly over half of the study sample was female (171, 52%) or lived in a metropolitan area (182, 55%). The majority did not have a long-term disability (198, 60%) and also reported not having used the internet in the three months prior to the study for health-related purposes (192, 58%). The mean and median ages of our sample were both 69 years. The mean EQ-5D 5L score for the entire sample was 0.73 but higher for certain subgroups (i.e. 0.76 for individuals who had not used the internet in the three months prior to the study for health-related purposes compared to 0.69 for those that had, and 0.86 for older people without a long-term disability compared to 0.64 for those with a long-term disability). The mean and median OPQoL-brief scores for the entire sample were 55.85 and 57, respectively. Similarly, the OPQoL-brief scores were higher for the two subgroups that also recorded higher EQ-5D 5L scores.

Table 3.

Demographic, internet-related health care use and quality of life details.

	Gender		Did participant use the internet for health-related issues in previous 3 months?		Have a long-term disability?		Entire sample N = 330
	Female	Males	Yes	No	Yes	No
	N = 171	N = 159	N = 138	N = 192	N = 198	N = 132
Mean (median) age in years∼	69 (69)	70 (69)	70 (69)	69 (69)	70 (69)	70 (69)	69 (69)
Mean (median) QoL (EQ-5D score)	0.72 (0.80)	0.72 (0.81)	0.68 (0.77)	0.75 (0.85)	0.61 (0.70)	0.88 (0.92)	0.72 (0.81)
Mean (median) QoL (OPQoL score)	56.36 (57.00)	55.30 (56.00)	54.67 (55.00)	56.69 (58.00)	53.78 (55.00)	58.95 (60.00)	55.85 (57.00)
Age groups in years (%)
65–69	86	84	85	85	84	86	85
70–74	14	14	14	14	15	14	14
75+	0	1	1	1	1	0	1
Living arrangements (%)
Living on their own	35	18	24	59	32	20	27
Living with your spouse	56	77	69	64	62	73	66
Living with other family	6	2	4	4	2	6	4
Living with others (not family)	3	3	3	3	4	2	3
Highest qualifications (%)
Primary School	4	1	2	2	2	2	2
Some Secondary School	15	15	11	18	15	16	15
Completed Secondary School	23	17	21	19	18	23	20
Some additional training	36	40	36	39	39	36	38
Undergraduate degree	18	17	18	16	15	20	17
Postgraduate degree	5	10	10	6	11	3	8
Ever used telehealth before? (%)
Yes	4	1	3	2	2	3	2
No	96	99	97	98	98	97	98
Location
Metro	56	55	54	56	51	61	55
Country	44	45	46	44	49	39	45
Mean (median) number of times in previous 3 months that respondent has visited a specialist’s room	1.4 (1.0)	1.6 (1.0)	2.0 (1.0)	1.2 (0)	2.0 (1.0)	1.0 (0)	1.5 (1.0)

Note: ^aQoL = quality of life ^bEQ-5D = EuroQoL EQ-5D 5 level and OPQoL = Older People's Quality of Life brief questionnaire

The majority of older people who participated in the DCE survey indicated that they lived with their spouses (66%) while 27% lived alone. This pattern was similar across all three subgroups previously described. A higher proportion of older people were educated beyond high school (63%), with the pattern again replicated across the subgroups and the majority in the entire sample and across all subgroups (96%–99%) reported never having used telehealth before. Finally, respondents were asked to indicate approximately how many times they had visited a specialist for the treatment of a health condition in the three months prior to the study. The mean and median numbers of times were 1.5 and 1, respectively.

Responses to attitudinal questions

Supplementary Material Table 1 presents information on participants’ levels of agreement, with nine statements relating to the administration and delivery of telehealth services. Responses were analysed according to all pairs of subgroups outlined in the methods section, in addition to respondents’ location, and show that the majority of participants (range 71%–94%) either agreed or strongly agreed with the first six statements. Further, there was more agreement than disagreement (41% versus 18%) with the statement that ‘Health examinations need to occur face-to-face in a clinic and not via telehealth’. Lastly, there was more disagreement than agreement with the statements that ‘Telehealth leads to loss of privacy and confidentiality’ and that ‘Telehealth should only be offered to people living in the country or in a rural area’ (51% versus 13% and 41% versus 28%, respectively).

When analysed according to subgroups, the only statistically significant differences in preferences expressed were in terms of agreement with the following statements: ‘a good understanding between patients and telehealth clinicians is important’ (higher amongst females and those without a long-term disability); ‘health examinations need to occur face-to-face in a clinic not via telehealth’ (higher amongst individuals with prior experience of using telehealth or health-related internet services) and ‘an initial face-to-face health clinic consultation needs to occur prior to telehealth’ (higher amongst individuals with prior experience of using telehealth). Other significant differences in agreement were seen for ‘telehealth leads to loss of privacy and confidentiality’ (lower for individuals with prior experience of using telehealth); ‘when services are hard to access, telehealth is a good alternative’ (higher amongst those who reported health-related internet use in the three months prior to the study) and ‘telehealth monitoring by clinicians will improve patients motivation’ (higher amongst those who reported health-related internet use in the three months prior to the study).

DCE results – test of consistency and comparisons between regression models and subgroups

All respondents in this study chose the most dominant option in the test of consistency and, therefore, passed the test. The results of the clogit-het presented in Supplementary Material Table 2 do not show any evidence of scale heterogeneity. The results of the G-NML (available upon request from the authors) confirmed this finding. However, the results from the MXL confirmed the presence of some preference heterogeneity. In addition, the statistical fit (assessed by the AIC) for the MXL models was better than that for all other models, suggesting that the former was an improvement over other models. Therefore, only DCE results from the MXL are presented below.

The main DCE results are shown in Table 4. The χ² statistics from the SL tests for equality of model parameters for four of the five pairs of subgroups considered in the study (defined by living arrangements, educational level, prior usage of the internet for health-related purposes and presence/absence of long-term disability) were all lower than the χ² critical value of 26.296 (based on 5% level of significance and 16 degrees of freedom). This, therefore, meant that analysing the data relating to these four subgroups as a pooled sample was appropriate. On the other hand, χ² statistics for gender was higher than the χ² critical value, suggesting significant differences between preferences for males and those for females. The quadratic terms of the cost attribute in all models were statistically insignificant, suggesting that it was appropriate to specify this attribute as a linear and continuous variable.

Table 4.

Mixed logit regression estimates.^a

Attribute levels	Gender		Living arrangements		Education level		Did participant use the internet for health-related issues in previous 3 months?		Did participant have a long-term disability?		TOTAL sample	TOTAL sample (% positive)^b
Attribute levels	Female	Male	Living alone	Living with others	Up to High School	Post-High School	Yes	No	Yes	No	TOTAL sample	TOTAL sample (% positive)^b
Mean
Bet_15to100km	0.709 (0.304)*	0.778 (0.289)**	2.149 (1.216)	0.617 (0.194)**	1.293 (0.721)	0.634 (0.209)	0.824 (0.000)**	0.513 (0.174)**	0.507 (0.208)*	0.747 (0.272)**	0.614 (0.154)**	99%
MoreThan_100km	−0.121 (0.294)	0.392 (0.296)	−1.660 (1.118)	0.213 (0.190)	−0.423 (0.506)	0.348 (0.221)	0.432 (0.000)	−0.101 (0.185)	0.154 (0.237)	0.129 (0.264)	0.126 (0.160)	53%
Most_CareAspects	0.944 (0.366)	0.589 (0.307)	2.340 (1.116)*	0.503 (0.203)*	1.482 (0.725)*	0.386 (0.213)	0.705 (0.000)*	0.356 (0.187)	0.467 (0.224)*	0.608 (0.298)*	0.491 (0.164)**	81%
All_CareAspects	1.102 (0.350)**	1.292 (0.346)**	2.026 (1.116)	1.000 (0.225)**	1.232 (0.737)	1.194 (0.286)	1.548 (0.000)**	0.635 (0.197)**	0.991 (0.237)**	0.847 (0.284)**	0.898 (0.172)**	82%
Some_Experience	0.499 (0.270)	0.627 (0.256)*	−1.095 (1.503)	0.437 (0.175)*	1.411 (0.754)	0.251 (0.182)	0.698 (0.000)	0.195 (0.165)	0.400 (0.201)*	0.420 (0.233)	0.374 (0.143)**	99%
Lots_Experience	−0.141 (0.319)	0.499 (0.307)	0.232 (1.260)	0.168 (0.188)	0.183 (0.490)	0.139 (0.217)	0.081 (0.000)	0.268 (0.185)	0.270 (0.219)	0.000 (0.292)	0.173 (0.163)	57%
Some_Assessments	1.257 (0.338)**	0.582 (0.232)*	5.551 (2.482)*	0.752 (0.166)**	0.860 (0.471)	0.861 (0.197)	1.028 (0.000)**	0.578 (0.156)**	0.805 (0.185)**	0.494 (0.216)*	0.757 (0.148)**	95%
All_Assessments	0.795 (0.312)*	0.380 (0.267)	0.114 (0.835)	0.539 (0.187)**	0.656 (0.481)	0.502 (0.221)	0.655 (0.000)*	0.349 (0.186)	0.650 (0.203)**	0.099 (0.243)	0.498 (0.158)**	68%
Cost_telehealth	−0.041 (0.009)**	−0.059 (0.012)**	−0.224 (0.090)*	−0.038 (0.006)**	−0.078 (0.036)*	−0.041 (0.006)	−0.050 (0.000)**	−0.034 (0.005)**	−0.041 (0.006)**	−0.040 (0.008)**	−0.040 (0.005)**	−81%
Moderately_Positive	1.713 (0.401)**	1.107 (0.337)**	9.313 (4.067)*	1.043 (0.219)**	2.516 (1.272)*	1.090 (0.244)	1.027 (0.000)**	1.066 (0.207)**	1.113 (0.250)**	1.239 (0.343)**	1.052 (0.168)**	99%
Very_Positive	1.508 (0.452)**	1.213 (0.430)**	8.256 (3.769)*	0.966 (0.251)**	3.776 (1.883)*	0.932 (0.278)	0.928 (0.000)*	1.028 (0.243)**	1.148 (0.340)**	1.315 (0.420)**	1.031 (0.206)**	82%
Standard Deviations
Most_CareAspects	0.399 (0.420)	−0.074 (0.477)	−1.299 (0.966)	0.003 (0.327)	−1.672 (0.994)	0.300 (0.381)	−0.382 (0.000)	0.140 (0.325)	0.197 (0.437)	0.089 (0.926)	0.020 (0.321)
All_CareAspects	−1.082 (0.446)*	−1.045 (0.417)*	7.859 (3.390)*	0.485 (0.367)	−1.350 (1.149)	1.009 (0.328)	−0.990 (0.000)	0.465 (0.363)	0.630 (0.497)	1.044 (0.394)**	0.827 (0.285)**
Some_Experience	0.218 (0.456)	0.296 (0.621)	−5.389 (2.425)*	−0.289 (0.278)	−0.269 (0.711)	0.391 (0.349)	−0.084 (0.000)	−0.248 (0.364)	0.522 (0.367)	−0.232 (0.336)	0.098 (0.282)
Lots_Experience	2.340 (0.620)**	−1.474 (0.530)**	11.810 (5.078)*	−1.402 (0.362)**	3.561 (1.787)*	1.357 (0.426)	2.435 (0.000)**	0.728 (0.475)	1.476 (0.403)**	1.632 (0.493)**	1.383 (0.270)**
Some_Assessments	1.003 (0.458)*	−0.681 (0.471)	−4.708 (2.135)*	−0.780 (0.235)**	1.566 (0.750)*	−0.457 (0.454)	1.085 (0.000)**	0.370 (0.288)	−0.331 (0.390)	0.935 (0.397)*	0.710 (0.229)**
All_Assessments	1.404 (0.513)**	−0.685 (0.408)	1.243 (1.208)	0.660 (0.306)*	−0.894 (0.802)	1.131 (0.396)	0.042 (0.000)	−0.940 (0.321)**	−0.542 (0.422)	−0.338 (0.462)	−0.649 (0.283)*
Cost_telehealth	0.048 (0.012)**	0.050 (0.011)**	0.135 (0.055)*	0.042 (0.007)**	0.090 (0.039)*	0.037 (0.007)	0.044 (0.000)**	0.031 (0.005)**	0.035 (0.007)**	0.042 (0.009)**	0.038 (0.005)**
LL	−436.266	−373.03	−209.098	−612.774	−299.938	−518.083	−327.087	−494.839	−474.537	−348.112	−830.715
AIC	908.532	782.059	454.197	1261.548	635.876	1072.166	690.174	1025.679	985.073	732.225	1697.431
N	89	241	123	207	171	159	138	192	198	132	330
Observations	890	2410	1230	2070	1710	1590	1380	1920	1980	1320	3300
LL for S-L Test^c	42.839		17.686		25.388		17.578		16.133

LL = Log Likelihood; AIC = The Akaike information criterion.

In the simulation-based technique, 500 Halton draws were run (both models based on total sample). Figures are coefficient (standard errors). *Coefficient statistically significant at 5%. **1% level of significance.

The proportion of respondents for whom a telehealth attribute has a positive or negative effect on preference for a telehealth package.

S&L test = Swait-Louviere Test where the sum of LL statistics of complementary subgroups was subtracted from that for the whole sample (–830.715). The χ² statistics from the Swait-Louviere likelihood ratio tests for equality of model parameters for the four of the five pairs of subgroups (living arrangements = 17.686; education level = 25.388; prior internet usage for health-related purposes = 17.578; and long-term disability = 16.133) were lower than the χ² critical value of 26.296 (based on 5% level of significance and 16 degrees of freedom). The LL statistic for gender (42.839) was, however, higher than the critical value. Consequently, the data relating to the first four subgroups were analysed as a pooled sample in all models, while that relating to gender was analysed separately for males and for females. A similar result was obtained when the Swait-Louviere Test was applied on the conditional logit model.

DCE results – attributes important in choice of telehealth packages

These are also shown in Table 4. In the simulation-based technique, 500 Halton draws were run. In choosing a telehealth package (for the pooled sample), participants expressed a strong preference for telehealth services targeted at individuals for whom the nearest hospital or clinic that could serve as an alternative to telehealth services was between 15 and 100 km away from their homes and for services focussed on individuals with some experience of using technology. In addition, there was a strong preference for telehealth services associated with lower costs, as well as telehealth services where a clinician pursued all or most aspects of care during a telehealth session, where all or some of the health assessments took place in a clinic prior to a telehealth session and where clinicians were very or moderately positive about the telehealth service. These results did not differ according to respondents’ location (metropolitan or rural; see Supplementary Material Table 3).

Sub-group analysis revealed some differences in preferences according to gender. Compared to females, males were relatively more concerned that telehealth services should be made available to those living at a greater geographical distance from the nearest hospital or health care facility, that all aspects of care should be covered within telehealth sessions, and that patients had prior experience with technology and the cost of telehealth. Females were more concerned about having some pre-telehealth assessments take place in a clinic and the clinicians’ attitude towards telehealth. As also shown in Table 4, only a few standard deviations for the subgroups and the sample as a whole were statistically significantly different from zero which suggests that, despite the differences in sub-group preferences identified above, there was an absence of substantial preference heterogeneity in the data in general.⁵⁷ Assuming a normal distribution for random parameters in the MXL model results, it was also possible to estimate the proportion of the participant population for whom particular telehealth attributes had a positive or negative effect on their choice of telehealth package. All attributes, with the exception of the cost attribute, were found to have a positive effect on choice.

Table 5 presents the predicted probabilities, together with 95% confidence intervals, associated with choosing a particular telehealth package for the 10 most preferred packages. The 95% confidence intervals of the preference scores and probabilities overlap, suggesting that there were no statistically significant differences between preferences for the top 10 preferred telehealth packages. This implies that these packages were all highly valued by respondents. The most preferred package was: one that targeted individuals for whom the nearest hospital or clinic that could serve as an alternative to telehealth services was between 15 and 100 km away from their homes; one where clinicians pursued all aspects of care during a telehealth session; one that was targeted at individuals with some experience of using technology; one where some of the health assessments took place in a clinic prior to a telehealth session; one that had a low associated cost; and one where clinicians were moderately positive about the telehealth service. In general, respondents were willing to make trade-offs between all attributes, except those related to distance to the nearest clinic or hospital and cost of telehealth.

Table 5.

Predicted probabilities for the top 10 preferred telehealth packages.

Distance to nearest hospital or clinic	Aspects of care in telehealth sessions	Technology-experience levels of patients targeted by telehealth	Assessments related to telehealth sessions	Cost of telehealth to you	Clinicians’ attitude to telehealth	Preference score^a (95% CI)	Probability^b (95% CI)	Rank
Bet_15to100km	All_CareAspects	Some_Experience	Some_Assessments	Low	Moderately_Positive	3.654 (2.108, 5.200)	0.108 (0.062, 0.154)	1
Bet_15to100km	All_CareAspects	Some_Experience	Some_Assessments	Low	Very_Positive	3.633 (2.011, 5.255)	0.108 (0.060, 0.156)	2
Bet_15to100km	All_CareAspects	Lots_Experience	Some_Assessments	Low	Moderately_Positive	3.453 (1.868, 5.039)	0.102 (0.055, 0.149)	3
Bet_15to100km	All_CareAspects	Lots_Experience	Some_Assessments	Low	Very_Positive	3.432 (1.771, 5.093)	0.102 (0.052, 0.151)	4
Bet_15to100km	All_CareAspects	Some_Experience	All_Assessments	Low	Moderately_Positive	3.395 (1.829, 4.962)	0.101 (0.054, 0.147)	5
Bet_15to100km	All_CareAspects	Some_Experience	All_Assessments	Low	Very_Positive	3.374 (1.732, 5.016)	0.100 (0.051, 0.148)	6
Bet_15to100km	Most_CareAspects	Some_Experience	Some_Assessments	Low	Moderately_Positive	3.247 (1.716, 4.778)	0.096 (0.051, 0.141)	7
Bet_15to100km	Most_CareAspects	Some_Experience	Some_Assessments	Low	Very_Positive	3.226 (1.619, 4.832)	0.095 (0.048, 0.143)	8
Bet_15to100km	All_CareAspects	Lots_Experience	All_Assessments	Low	Moderately_Positive	3.194 (1.588, 4.800)	0.095 (0.047, 0.142)	9
Bet_15to100km	All_CareAspects	Lots_Experience	All_Assessments	Low	Very_Positive	3.173 (1.491, 4.855)	0.094 (0.044, 0.144)	10

CI: confidence interval.

Preference scores were calculated by summing up the model coefficients for every combination of attribute levels.

The probability that each combination of attribute levels is the most preferred scenario was calculated as the preference score for that particular attribute divided by the sum of all preference scores.

Discussion

To our knowledge, this is the first study that has sought to utilise DCE methodology to assess the attitudes and preferences of older people for a telehealth approach to a number of health-related services. Our findings indicate that all six attributes identified during the first phase of our project and then tested on Australian older people were significant in determining the choice of a telehealth package. However, having clinicians who were very or moderately positive about telehealth services as well as having a comprehensive list of services provided by these clinicians were the strongest determinants of this choice, while the cost of the service (preference being for a cheaper one) was the weakest. When contrasted against telehealth services currently available in Australia where only select Medicare-funded telehealth services are available to patients outside major cities (e.g. specialist video consultations),⁷⁰ our results from the DCE (Table 4) show that older people want all health services suitable for delivery via telehealth to be provided. However, it was also clear that respondents feel that telehealth should not completely replace necessary in-person contact with clinicians, preferring that all or some of the initial assessments take place in a clinic prior to a telehealth session. These findings confirm other research suggesting that telehealth should be seen as a supplement to, rather than a substitute for, traditional care; providing additional services that otherwise would not or could not be provided,^71–73 These results are highly relevant for policy makers as they present empirical evidence to indicate what basic features should make up a telehealth approach to rehabilitation, aged care and palliative care services from the perspectives of older people. These findings may also be incorporated into the framework of economic evaluation, by combining the DCE results with information relating to the costs associated with the provision of preferred telehealth service configurations, in order to provide an assessment of the cost effectiveness of models of telehealth care.³⁸ Such cost information would be particularly useful given that the appropriate valuation of health and non-health outcomes has been identified as one of the challenges to the economic evaluation of telehealth.⁷⁴

The mean and median ages for participants in this DCE were both 69 years, representative of the eligibility criteria for the study – a survey targeted at those aged 65 years and over. However, the proportion of those aged 65–69 years (85%) was much higher than that reported for the general population in the Australian 2011 census (31%).⁷⁵ This is not surprising as our target sample of people who use the internet is most likely to be that of the younger old. Compared to the people aged 65 years and over in the general population,⁷⁵ our sample had more individuals who had completed additional training (38% versus 14%) or university education (25% versus 18%). Again, this is an artefact of the sample studied – those who use the internet and are, therefore, more likely to have higher levels of education as they are required to have some level of technological know-how. Australian population norms for the EQ5D 5L are as yet available. When compared to the EQ-5D 3-level scores found in the general Australian for older people aged 65–74 years (mean score of 0.82) and for those aged 75 + (mean score of 0.80),⁷⁶ however, older people in our sample had slightly lower QoL as measured by the EQ-5D 5L (mean score of 0.73). This may be because the EQ-5D 5L is more sensitive and has smaller ceiling effects⁷⁷ which therefore allows for more intermediate QoL scores to be registered. Our sample was, however, representative of older Australians in other characteristics reported in the 2011 census⁷² and in other Australian research.^78,79 The majority of people in our sample (52%) were female which is fairly representative of the 54% figure reported in the census. Similarly, the proportions of individuals in our sample living alone (27%) or with their spouses (66%) were similar to those in the general population (25% and 56%, respectively). Further, 60% of our sample reported having a long-term disability which falls within the 10%–68% range reported in the census. Though the proportion of individuals in our sample who lived in a metro area (55%) was lower than that in the general population (65%), census statistics⁷⁵ suggest that the proportion of just the younger old (which made up the bulk of our sample) is likely to be much lower than 65%, which may explain our figures. Nationally representative OPQoL-Brief mean scores are also not available in Australia, but the mean score for our sample (55.85) was similar to those found in a recent study that explored the QoL of older people receiving rehabilitation services (N = 21) in South Australia using the same instrument (mean score 54.6).⁷⁸ The mean score for those who reported having a long-term disability (and may, therefore, have probably needed health and other interventions) was 53.78, which is much more comparable to the score from the South Australian study.⁷⁸ It is notable that 138 older people (48%) reported having used the internet in the three months prior to the study for health-related purposes. While the ‘2012–13 Multipurpose Household Survey’ for persons aged 65 years or over⁷⁹ did not specifically ask if the type of online activities undertaken at home by older persons were health-related and what percentage of users accessed such information, 50% of these activities involved accessing government services which may have included those that were health-related.

The vast majority of older people (98%) had not used telehealth services before and this is reflective of the relative infancy of the telehealth approach as it applies to health care for older people and the knowledge gaps in terms of determining older people’s technology needs.⁸⁰ It was also evident that older people want to engage with telehealth service clinicians who are themselves positive about the service. This is in line with research that shows that champions of telehealth play an important role in its development and acceptance.^81–84 The responses to the attitudinal statements also showed that older people had specific views about how telehealth should be administered, which also filtered through into their responses to the DCE questions. In particular, they felt that telehealth services should be targeted at people for whom hospital services are hard to access but mainly due to distance to hospitals or clinics and not necessarily due to living in the country or rural areas. Defining need on the basis of lack of proximity to health institutions rather than on rurality per se aligns itself well with the core objective of many telehealth services: ‘the delivery of health care services where distance is a critical factor’.²¹ Further, DCE participants underlined the need for a good rapport to exist between clinicians delivering these services and patients. The importance of ‘user friendliness of information and communication technology services’ used to support older people has also been highlighted elsewhere.^85,86 This outlook may also explain why the majority in our sample felt that telehealth would lead to an improvement in ‘patients’ motivation and willingness to comply with health care recommendations’ without the risk of loss of privacy and confidentiality. This is notable as older Australians have cited lack of confidentiality as a concern when using telehealth services in another study.⁸⁶ While there was a good proportion that felt that all health examinations need to occur face-to-face in a clinic and not via telehealth, the majority of participants were of the view that at least the initial health consultation prior to telehealth sessions needs to occur in a clinic, in line with other research.^71–73

The results from the DCE highlighted that, in order of the strength of preference, study participants favoured telehealth services: (i) where clinicians were very positive or moderately positive about the telehealth service; (ii) where a clinician pursued all or most aspects of care during a telehealth session; (iii) where all or some of the health assessments took place in a clinic prior to a telehealth session; (iv) targeted at those for whom the nearest hospital or clinic that could serve as an alternative to telehealth services was between 15 and 100 km away from their home; (v) targeted at individuals with some experience of using technology; and (vi) that had a low associated cost. The results from the MXL regression model also show that statistically significant telehealth attributes had a positive effect on the decision to choose telehealth packages for at least 57% of the respondents (except cost which had a negative effect on 81% of the respondents). The latter result is not surprising, given that most health services in Australia are provided through Medicare with a zero or reduced cost to patients at the point of use, as has been seen in other countries with similar health funding systems.⁸⁷

This study had some limitations. First, some studies have shown that strategies used to choose among alternatives vary with age,^88,89 leading to inconsistent choices being made by older respondents. Even though all respondents in our study passed the test of consistency (suggesting that it was plausible to assume that preferences expressed by these respondents were rational), this test was not based on telehealth scenarios, which would have been more consistent with the rest of the DCE. However, the results of this test were still in line with our other research that shows that it is possible to get consistent responses from older respondents^90,91 and that cognitive decline due to old age does not have a significant effect on the consistency of responses to a DCE survey.⁹² Second, participants were essentially a self-selected group who were able to use the internet and also part of PureProfile’s online database. They may, therefore, not be entirely representative of older people in the Australian general population, especially as the proportion of Australian internet users aged 65 years and over is about 46%.⁷⁹ However, we did achieve wide representation across Australia and the study participants were reflective of a broad range of socio-demographic characteristics. Third, in developing the DCE attributes, the older people interviewed were all rehabilitation patients and their views may not be completely generalisable to those of older people in the Australian general population. However, we also conducted qualitative interviews with carers and focus groups with health care and allied professionals which, when combined with the views of experts involved with our study, helped to develop more representative DCE attributes.

Conclusions

The findings from this study are significant for policy makers as they represent the first empirical evidence about what basic features, from older people’s perspectives, should make up a telehealth approach to a range of services, including palliative, aged and rehabilitation care services, as revealed in the responses to the DCE forced-choice questions. Telehealth programmes in Australia, and internationally, have been shown to be associated with improvements including lower costs and reduced inconvenience while accessing specialist health services, better clinical outcomes, reduced mortality and hospital utilisation, improved access to services and improved quality of clinical services.^9–14,93 In addition to views including that telehealth does not lead to loss of privacy and confidentiality or that telehealth should not only be offered to people living in the country or in a rural area revealed through responses to the attitudinal statements (Likert scale), our findings from the DCE revealed respondents’ choice of telehealth attributes in order of strength of preference as well as the predicted probabilities associated with choosing particular telehealth packages. These findings indicate a preference amongst respondents for a comprehensive telehealth model (in terms of services offered) targeted at those with some technological know-how as a substitute for attendance at hospitals and clinics, especially where these health facilities were far away from older people’s homes. This is in line with the federal government’s policy for a telehealth model with wide coverage. The potential for the future provision of telehealth services at costs lower than those incurred when older people physically attend health institutions will be a key attraction for health systems internationally as will having clinicians who are champions of the telehealth model. The findings from this DCE study may be usefully incorporated into the design of future telehealth models of service delivery for older people. More generally, DCEs offer a promising approach for the systematic incorporation of older people’s preferences into the future design and delivery of service innovations in health and aged care.

Footnotes

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: The Flinders Telehealth in the Home trial that funded this study was an initiative funded by the Australian Government.

References

International Organization for Standardization (ISO). ISO/TS 13131: 2014. Health informatics – Telehealth services – Quality planning guidelines. 2014.

Fong

ACM

. Telemedicine Technologies: Information Technologies in Medicine and Telehealth, West Sussex: John Wiley & Sons, Ltd, 2011.

Taylor

Wade

Morris

. Technology support to a telehealth in the home service: Qualitative observations. J Telemed Telecare 2015.

Taylor

Morris

Pech

. Home Telehealth Video Conferencing: Perceptions and Performance. JMIR Mhealth Uhealth 2015; 3: e90–e90.

Satariano

Scharlach

Lindeman

. Aging, place, and technology: toward improving access and wellness in older populations. J Aging Health 2014; 26: 1373–1389.

Wagner

Basran

Dal Bello-Haas

. A review of monitoring technology for use with older adults. J Geriatr Phys Ther 2012; 35: 28–34.

Boissy

Choquette

Hamel

. User-Based Motion Sensing and Fuzzy Logic for Automated Fall Detection in Older Adults Telemed J E Health. 2007 13: 683–94.

Flodgren

Rachas

Farmer

. Interactive telemedicine: effects on professional practice and health care outcomes. Cochrane Database Syst Rev 2015; 9: Cd002098–Cd002098.

Steventon

Bardsley

Bilings

. Effect of telehealth on use of secondary care and mortality: findings from the Whole System Demonstrator cluster randomised trial. BMJ 2012; 344: e3874–e3874.

10.

Chumbler

Quigley

. Effects of telerehabilitation on physical function and disability for stroke patients: a randomized, controlled trial. Stroke 2012; 43: 2168–2174.

11.

Vitacca

Bianchi

Guerra

. Tele-assistance in chronic respiratory failure patients: a randomised clinical trial. Eur Respir J 2009; 33: 411–418.

12.

Saxon

Hayes

Gilliam

. Long-term outcome after ICD and CRT implantation and influence of remote device follow-up: the ALTITUDE survival study. Circulation 2010; 122: 2359–2367.

13.

Darkins

Ryan

Kobb

. Care Coordination/Home Telehealth: the systematic implementation of health informatics, home telehealth, and disease management to support the care of veteran patients with chronic conditions. Telemed J E Health 2008; 14: 1118–1126.

14.

Goernig

Doede

Brehm

. Ambulatory Disease Management in Cardiac Patients: 12 month follow-up of Home Care Telemedicine in Thuringia by the Management Program Zertiva^®. Phys Rehab Kur Med 2009; 19: 9–13.

15.

van der Pol

McKenzie

. Costs and benefits of tele-endoscopy clinics in a remote location. J Telemed Telecare 2010; 16: 89–94.

16.

Crotty

Killington

van den Berg

. Telerehabilitation for older people using off-the-shelf applications: acceptability and feasibility. J Telemed Telecare 2014; 20: 370–376.

17.

Cartwright

Wade

Shaw

. The Impact of Telehealth and Telecare on Clients of the Transition Care Program (TCP), Lismore: Aged Services Unit, Southern Cross University, 2011.

18.

Henderson

Knapp

Fernandez

. Cost effectiveness of telehealth for patients with long term conditions (Whole Systems Demonstrator telehealth questionnaire study): nested economic evaluation in a pragmatic, cluster randomised controlled trial. BMJ 2013; 346: f1035–f1035.

19.

Productivity Commission. Caring for Older Australians. Report No 53, Final Inquiry Report. Canberra: Australian Government Productivity Commission, 2011.

20.

Makai

Brouwer

Koopmanschap

. Quality of life instruments for economic evaluations in health and social care for older people: a systematic review. Soc Sci Med 2014; 102: 83–93.

21.

World Health Organisation. A health telematics policy in support of WHOs Health-For-All strategy for global health development: report of the WHO group consultation on health telematics, 11–16 December, Geneva, 1997, Geneva: World Health Organization, 1998.

22.

Ware

. Improving the accessibility of health services in urban and regional settings for Indigenous people. Closing the gap, Canberra: Australian Institute of Health and Welfare, 2013.

23.

Jurgens

Clissett

Gladman

. Why are family carers of people with dementia dissatisfied with general hospital care? A qualitative study. BMC Geriatr 2012; 12: 57–57.

24.

Collier A, Morgan DD, Swetenham K, et al. Implementation of a pilot telehealth programme in community palliative care: A qualitative study of clinicians' perspectives. Palliat Med. Epub ahead of print 19 August 2015. DOI: 10.1177/0269216315600113.

25.

Heidbuchel

Lioen

Foulon

. Potential role of remote monitoring for scheduled and unscheduled evaluations of patients with an implantable defibrillator. Europace 2008; 10: 351–357.

26.

Crossley

Boyle

Vitense

Chang

. The CONNECT (Clinical Evaluation of Remote Notification to Reduce Time to Clinical Decision) trial: the value of wireless remote monitoring with automatic clinician alerts. J Am Coll Cardiol 2011; 57: 1181–1189.

27.

Khosravi

Ghapanchi

. Investigating the effectiveness of technologies applied to assist seniors: A systematic literature review. Int J Med Inform 2016; 85: 17–26.

28.

Vedel

Akhlaghpour

Vaghefi

. Health information technologies in geriatrics and gerontology: a mixed systematic review. J Am Med Inform Assoc 2013; 20: 1109–1119.

29.

Davis J, Morgans A and Stewart J. Developing an Australian health and aged care research agenda: a systematic review of evidence at the subacute interface. Aust Health Rev. Epub ahead of print 5 November 2015. DOI: 10.1071/AH15005.

30.

Huang

. Exploring the acceptance of telecare among senior citizens: an application of back-propagation network. Telemed J E Health 2011; 17: 111–117.

31.

Chen

Chou

S-W

. Measuring patients' perceptions and social influence on Home Telecare Management System acceptance. Int J Healthc Inf Syst Inform 2010; 5: 44–68.

32.

Cimperman

Brenčič

Trkman

. Older adults' perceptions of home telehealth services. Telemed J E Health 2013; 19: 786–790.

33.

Russell

Gillespie

Hartley

. Exploring the predictors of home telehealth uptake by elderly Australian healthcare consumers. J Telemed Telecare 2015; 21: 485–459.

34.

Chen

Chan

Acceptance of ICTs by older adults: A review of recent studies. In: Zhou

Salvendy

(eds). Human Aspects of IT for the Aged Population Design for Aging, Geneva, Switzerland: Springer International Publishing, 2015, pp. 239–249.

35.

Ryan

Gerard

Amaya-Amaya

. Using discrete choice experiments to value health and health care, Dordrecht: Springer, 2006.

36.

Street

Burgess

. The construction of optimal stated choice experiments: theory and methods, Hoboken, New Jersey: John Wiley and Sons, 2007.

37.

Lancaster

. A New approach to consumer theory. J Polit Econ 1966; 74: 132–157.

38.

Benning

Kimman

Dirksen

. Combining individual-level discrete choice experiment estimates and costs to inform health care management decisions about customized care: the case of follow-up strategies after breast cancer treatment. Value Health 2012; 15: 680–689.

39.

Shulver W, Killington M, Morris C, et al. Well, if the kids can do it, I can do it”: Older rehabilitation patients' experiences of telerehabilitation. Health Expect. Epub ahead of print 18 February 2016. DOI: 10.1111/hex.12443.

40.

Boyatzis

. Transforming qualitative information: Thematic analysis and code development, Thousand Oaks, London, & New Delhi: SAGE Publications, 1998.

41.

Rihoux

Ragin

. Configurational comparative methods. Qualitative Comparative Analysis (QCA) and related techniques (Applied Social Research Methods), Thousand Oaks and London: Sage, 2009.

42.

Richards

. Handling Qualitative Data: A Practical Guide, London: Sage, 2005.

43.

Coast

Al-Janabi

Sutton

. Using qualitative methods for attribute development for discrete choice experiments: issues and recommendations. Health Econ 2012; 21: 730–741.

44.

McCaffrey

Gill

Kaambwa

. What features of `consumer-directed' home-based support services are important to older Australians and their informal carers – development of attributes for a discrete choice experiment Health Soc Care Comm 2015; 23: 654–664.

45.

Venkatesh

Thong

JYL

. Consumer acceptance and use of information technology: extending the unified theory of acceptance and use of technology. MIS Quart 2012; 36: 157–178.

46.

Bliss

Katz

Wright

. Estimating proximity to care: Are straight line and zipcode centroid distances acceptable proxy measures? Med care 2012; 50: 99–106.

47.

Australian Government – Department of Health. MBS and telehealth: Telehealth areas, Canberra: Australian Government – Department of Health, 2015.

48.

Australian Government Department of Health. MBS Online, Medicare Benefits Schedule: Telehealth Statistics, Canberra: Australian Government Department of Health, 2014.

49.

Choice Metrics Pty Ltd. Ngene 1.1.1. 2012.

50.

Norman

Cronin

Viney

. A pilot discrete choice experiment to explore preferences for EQ-5D-5L health states. Appl Health Econ Health Policy 2013; 11: 287–298.

51.

McFadden

Conditional logit analysis of qualitative choice behavior. In: Zarembka

(ed.) Frontiers in Econometrics, New York: Academic, 1974, pp. 105–142.

52.

Gell

Rosenberg

Demiris

. Patterns of technology use among older adults with and without disabilities. Gerontologist 2015; 55: 412–421.

53.

Oppenauer

. Motivation and needs for technology use in old age. Gerontechnology 2009; 8: 82–87.

54.

Peek

Luijkx

Rijnaard

. Older adults' reasons for using technology while aging in place. Gerontology 2016; 62: 226–237.

55.

Swait

Louviere

. The role of the scale parameter in the estimation and comparison of multinomial logit models. J Marketing Res 1993; 30: 305–314.

56.

Greene

. Econometric Analysis, 7th ed. Upper Saddle River, NJ: Prentice Hall, 2012.

57.

McFadden

Train

. Mixed MNL models for discrete choice. J Appl Econom 2000; 5: 447–470.

58.

Fiebig

Keane

Louviere

Wasi

. The generalized multinomial logit: accounting for scale and coefficient heterogeneity. Market Sci 2010; 29: 393–421.

59.

Greene

Hensher

. Does scale heterogeneity across individuals matter? An empirical assessment of alternative logit models. Transportation 2010; 37: 413–428.

60.

Hole

Knox

. Estimating the generalized multinomial logit model in Stata. Stata Journal 2013; 13: 382–397.

61.

Flynn

Louviere

Coast

. Using discrete choice experiments to understand preferences for quality of life. Variance-scale heterogeneity matters. Soc Sci Med 2010; 70: 1957–1965.

62.

Breffle

Morey

. Investigating preference heterogeneity in a repeated discrete-choice recreation demand model of Atlantic salmon fishin. Mar Resour Econ 2000; 15: 1–20.

63.

Mentzakis

Ryan

McNamee

. Using discrete choice experiments to value informal care tasks: exploring preference heterogeneity. Health Econ 2011; 20: 930–944.

64.

Akaike

. A new look at the statistical model identification. IEEE T Automat Contr 1974; 19: 716–723.

65.

Hole

. Fitting mixed logit models by using maximum simulated likelihood. The Stata Journal 2007; 7: 388–401.

66.

Payne

Fargher

Roberts

. Valuing pharmacogenetic testing services: a comparison of patients' and health care professionals' preferences. Value Health 2011; 14: 121–134.

67.

Lancsar

Louviere

. Deleting ‘irrational' responses from discrete choice experiments: a case of investigating or imposing preferences? Health Econ 2006; 15: 797–811.

68.

Bross

IDJ

Critical levels, statistical language and scientific inference. In: Godambe

Sprott

(eds). Foundations of Statistical Inference, Toronto: Holt, Rinehart & Winston of Canada, Ltd, 1971, pp. 500–513.

69.

StataCorp. Stata: Release 13. Statistical Software, College Station, TX: StataCorp LP, 2013.

70.

Australian Government – Department of Health. Telehealth: Specialist video consultations under Medicare. Australian Government – Department of Health, 2015.

71.

Demiris

Oliver

Porock

. The Missouri Telehospice Project: Background and next steps. Home Health Care Technol Rep 2004; 1: 49–55.

72.

Oudshoorn

. Physical and digital proximity: Emerging ways of health care in face-to-face and telemonitoring of heart-failure patients. Sociol Health Ill 2009; 31: 390–405.

73.

Ritterband

Gonder-Frederick

Cox

. Internet interventions: In review, in use, and into the future. Prof Psychol-Res Pr 2003; 34: 527–527.

74.

McIntosh

Cairns

. A framework for the economic evaluation of telemedicine. J Telemed Telecare 1997; 3: 132–139.

75.

Australian Bureau of Statistics. 2071.0 – Reflecting a Nation: Stories from the 2011 Census, 2012–2013. Canberra: Australian Bureau of Statistics, 2012.

76.

Clemens

Begum

Harper

Whitty

. A comparison of EQ-5D-3L population norms in Queensland, Australia, estimated using utility value sets from Australia, the UK and USA. Qual Life Res 2014; 23: 2375–2381.

77.

Janssen

Pickard

Golicki

. Measurement properties of the EQ-5D-5L compared to the EQ-5D-3L across eight patient groups: a multi-country study. Qual Life Res 2013; 22: 1717–1727.

78.

Milte

Walker

Luszcz

. How important is health status in defining quality of life for older people? An exploratory study of the views of older South Australians. Appl Health Econ Health Policy 2014; 12: 73–84.

79.

Australian Bureau of Statistics. 8146.0 – Household Use of Information Technology, Australia, 2012–13. Canberra: Australian Bureau of Statistics, 2014.

80.

Koch

Hagglund

. Health informatics and the delivery of care to older people. Maturitas 2009; 63: 195–199.

81.

Wade

Eliott

. The role of the champion in telehealth service development: a qualitative analysis. J Telemed Telecare 2012; 18: 490–492.

82.

Wade

Eliott

Hiller

. Clinician acceptance is the key factor for sustainable telehealth services. Qual Health Res 2014; 24: 682–694.

83.

Muttitt

Vigneault

Loewen

. Integrating telehealth into Aboriginal healthcare: the Canadian experience. Int J Circumpolar Health 2004; 63: 401–414.

84.

Singh

Mathiassen

Stachura

. Sustainable rural telehealth innovation: a public health case study. Health Serv Res 2010; 45: 985–1004.

85.

Magnusson

Hanson

Borg

. A literature review study of Information and Communication Technology as a support for frail older people living at home and their family carers. Technol Disabil 2004; 16: 223–235.

86.

Loh

Flicker

Horner

. Attitudes toward information and communication technology (ICT) in residential aged care in Western Australia. J Am Med Dir Assoc 2009; 10: 408–413.

87.

Liu

Yam

CHK

Huang

OHY

. Willingness to pay for private primary care services in Hong Kong: are elderly ready to move from the public sector? Health Policy Plann 2013; 28: 717–729.

88.

Johnson

. Age differences in decision making: a process methodology for examining strategic information processing. J Gerontol 1990; 45: P75–P78.

89.

Johnson

. Thinking about strategies during, before, and after making a decision. Psychol Aging 1993; 8: 231–241.

90.

Kaambwa

Lancsar

McCaffrey

. Investigating consumers' and informal carers' views and preferences for consumer directed care: A discrete choice experiment. Soc Sci Med 2015; 140: 81–94.

91.

McNamara

Chen

George

. What factors influence older people in the decision to relinquish their driver's licence? A discrete choice experiment. Accident Anal Prev 2013; 55: 178–184.

92.

Milte

Ratcliffe

Chen

. Cognitive overload? An exploration of the potential impact of cognitive functioning in discrete choice experiments with older people in health care. Value Health 2014; 17: 655–659.

93.

Moffatt

Eley

. The reported benefits of telehealth for rural Australians. Aust Health Rev 2010; 34: 276–281.