Cost–benefit analysis of orientation and mobility programs for adults with vision disability: a contingent valuation study

Abstract Purpose The purpose of this study is to conduct a cost–benefit analysis of orientation and mobility (O&M) programs from three perspectives: the general public, the experienced, and the potential users of O&M programs. Methods Willingness-to-pay (WTP) for O&M programs was collected via a contingent valuation survey using a double-bound dichotomous choice approach. WTP was estimated using interval regression analyses, accounting for study arm, sex, occupation, income, and self-rated health. The cost data were estimated from a service provider’s perspective. The net present value (NPV), variation if delivered by tele-O&M, was investigated. Results The adjusted mean NPV of O&M programs was $3857 (95% CI: $3760–$3954) per client, with highest NPV from the general public ($4289, 95% CI: $4185–$4392), followed by the experienced users ($3158, 95% CI: $2897–$3419) and the potential users ($2867, 95% CI: $2680–$3054). The NPV reached break-even for tele-O&M. Conclusions There was strong community support for investment into O&M programs considering benefits for clients over and above the cost of providing the services. Implications for rehabilitation This study demonstrates the feasibility of using cost–benefit analysis with a contingent valuation approach to economically assess a rehabilitation intervention, where its multi-dimensional benefits cannot be fully captured by a conventional appraisal technique such as cost-effectiveness analysis. The high willingness-to-pay (WTP) values amongst the general public suggests that Australians perceive government’s investment in orientation and mobility (O&M) rehabilitation as value for money and that individuals would be prepared to contribute to its costs. The lower WTP for O&M partially delivered via tele-practice (tele-O&M) indicates a lower acceptance of this innovation in comparison with the traditional face-to-face O&M. The estimated net present values of O&M programs, positive for traditional O&M and break-even for tele-O&M, can be of assistance to service planning and investment decisions within the Australian context.


Introduction
Of all the people living with disability in Australia, 5.8% are suffering from low vision or blindness [1], a health condition that leads to emotional and psychological distress, depression, sadness, loss of independence and self-esteem, isolation, loneliness, fear of injury, and vulnerability [2,3]. Vision disability services and supports, such as orientation and mobility (O&M) programs could enhance an individual's ability to perform everyday activities and optimise their social and economic participation [4,5].
Provided by trained O&M specialists, O&M programs aim to enhance visually impaired individuals' awareness of their environment and equip them for safe, independent travel through instructions in the use of mobility aids, public transport, road crossing strategies, and consultancy in modifying environments [6,7].
Historically in Australia, O&M programs have been provided free-of-charge by two large charitable organisations. However, the rollout of the National Disability Insurance Scheme (NDIS) in 2016 has transformed the vision disability market. On the demand side, Australians with vision disability who participate in the NDIS can now use their NDIS funding to purchase goods and services from their chosen providers. On the supply side, the incentives provided by the NDIS have attracted sole traders and small business operators to the vision disability market alongside the two large charitable organisations, while those two large organisations adjust their business model and start to charge NDIS participants fees for services. Price of goods and services in the NDIS marketsis currently set out by the National Disability Insurance Agency. It is anticipated that the need for price control will end once the NDIS markets are fully developed [8].
Service providers in the vision-disability sector not only embrace NDIS funding as an additional revenue source, but also become more innovative towards service delivery models. After delivering O&M programs face-to-face for more than 60 years, tele-practice (which can be conducted using teleconferencing facility and assistive technology) has now been considered as an alternative medium to conduct initial O&M consultation and monitor trainee's progress. Despite tele-practice having the potential to minimise travel costs, reduce waiting time and improve service accessibility (especially during COVID19 pandemic), its acceptance by people with vision disability remains unclear.
This study provides the first cost-benefit analysis (CBA) of the O&M programs from three perspectives: "experienced users" (vision-impaired individuals who have undertaken O&M programs), "potential users" (vision-impaired individuals who might undertake O&M programs in the near future), and the general public across Australia. The key objective of this study was to evaluate cost-benefit of the traditional O&M programs (delivered face-to-face). The second objective was to estimate how much Australians value O&M programs delivered partially via tele-practice (tele-O&M).
Cost-benefit analysis assesses the benefits of an intervention in monetary terms, which enables comparisons to a wide variety of interventions within the disability sector or across sectors (e.g., health, housing, and educational sector) [9]. CBA adopts multidimensional outcome measurement [10], accounting for not only health gains but also the non-health benefits that commonly exist in social care interventions, such as choices, dignity, independence, opportunity for social/education/economical participation. Frequently used in cost-benefit analyses, contingent valuation (CV) asks individuals directly in a hypothetical survey the maximum amount they would be willing to pay (WTP) for a commodity; or conversely the minimum amount they would be willing to accept (WTA) in compensation to forego use of a commodity [9]. The commodity in question usually cannot be bought and sold in the marketplace [11], such as O&M programs in Australia, and thus the objective is to ascertain a "shadow price" for it. As a type of the "stated preference" tool, CV is theoretically founded in welfare economics where rational consumers would maximise their expected utility given their income constraints [10]. Through a CV survey and a CBA, this study presents net present values (NPVs) of O&M programs from the general public, potential, and experienced users' perspectives to inform service planning and investment decisions within the Australian context. The NPV of an intervention equals the present value of benefits minus the present value of costs. If the NPV is positive, the proposed intervention is considered to be socially efficient and thus worth investment; and vice versa if the NPV is negative.

Participant and sampling strategy
This study comprised three research arms. Arm 1, the general public, were drawn from all states and territories of Australia via stratified random sampling with representation proportional to the population of each state. Arm 2, the experienced users of O&M programs, and arm 3, the potential users of O&M programs, were drawn from a major O&M service provider's database and two eye clinics via convenience sampling. All participants must have been aged 18 years or older and could understand and speak English. This study was approved by the Human Ethics Committee at the University of Sydney (2019/274).

Survey design and data collection
A working group was established to design the survey questionnaire, comprising three vision-impaired individuals, an O&M specialist, three health economists, an optometry/public health researcher, and a clinical psychologist. This working group ensured the relevance, understanding, and acceptance of the questionnaire, as well as mitigated any risk of bias. The monetary unit used in this study was 2020 Australian dollars, which was equivalent to 0.76 United State Dollar or 0.62 Euro. Two WTP questions were included in the questionnaire (Supplementary material 1). The first question, using a double-bound dichotomous choice (DBDC) approach, asked participants if they were willing to pay $7000 out-of-pocket to access O&M programs that would help them to travel safely and independently to the nearby shopping centre, assuming that they were blind (if they weren't already) and charity or government funding was not available. If participants accepted the first bid, they were presented with the higher price of $9000; otherwise, they were presented with the lower price of $5000. The first bid of $7000 was guided by the NDIS price. The second WTP question, using an open-ended approach, asked participants the maximum amount they were willing to pay if part of this O&M program, such as initial consultation and monitoring progress, was delivered via tele-practice.
Reasons for refusing to answer the abovementioned questions were recorded by the interviewers. The final section of this survey included a series of demographic questions regarding gender, age, education, income, occupation, living arrangement, postcode, eyesight, experience with O&M programs and low vision services, whether or not an NDIS participant or a disability support pensioner and self-rated health.
The survey was administered via telephone by the Survey Research Centre at Edith Cowan University on behalf of the research team during the period of 23 October 2019 to 22 June 2020. Telephone numbers (landline or mobile phone) of the general public were randomly selected from the Electronic White Pages, whereas the telephone numbers of the experienced and the potential O&M users were obtained from the service providers. The survey administered on the first day served as a pilot survey to check on the understanding and acceptance of the survey questions. No changes were made to the survey questions following the pilot and all 22 completed pilot surveys were included in the final analysis.

Data analysis
The study population was compared to the Australian general population and population with a disability using recent survey data (2018 Survey of Disability, Ageing and Carers, ABS [12]). The combinations of the responses to the first WTP question created four intervals: >$9k, $7k-$9, $5k-$7k, and $0-$5k. We hypothesised that the mean WTP value is a linear function of certain individual characteristics of participants such as study arm, age, gender, income, education, occupation, self-rated health, type of government supports received (the NDIS or Disability Support Pension), remoteness of residential location, and Socio-Economic Indexes for Areas (SEIFA). The associations between the explanatory variables and outcome variables were first assessed using univariate interval regression analyses. The interval regression method has been used for DBDC data in the CV literature [13,14].
Variables that emerged as significant (p < 0.30) were then evaluated in a multivariable interval regression model. The final model was derived using backwards stepwise selection techniques requiring 0.05 level of significance. The adjusted mean WTP for the O&M programs was obtained from linear predictions of the final model. Rather than interval regression, linear regression was used to assess the relationship between WTP for tele-O&M (continuous data) and explanatory variables. Similar model fitting strategy and prediction method were applied. A paired t-test was used to investigate the difference in the WTP for tele-O&M and the WTP for face-to-face O&M.
Cost data were collected from a service provider's perspective, including labour costs (hourly rate with a 30% overheads), travel costs (car and fuel expenses), and occupancy costs. An average of 40 h was allocated to the traditional O&M programs, including time spent on face-to-face assessment (one session), training (three sessions), progress review (two sessions), as well as administrative work in the office and travelling to and from client's residential location. Costs such as wage, overhead costs, and population dispersion (potential underutilisation of services in areas of low population) are affected by the service location. In our analysis, total costs of each observation were multiplied by a location weight to account for increased cost of service delivery due to remoteness of the location. Location multipliers were calculated as a weighted contribution of wages, building repairs and maintenance and consumables, multiplied by the underutilisation rate [15]. Location factor of major cities of Australia was set as 1.00, 1.11 for inner regional Australia, 1.26 for outer regional Australia, 1.73 for remote Australia, and 1.75 for very remote Australia [15]. There were no additional costs incurred for tele-O&M as it utilised mainstream devices and tele-conferencing facility already available to service providers.
The NPV was estimated by subtracting the total costs of O&M programs from the WTP values. Staff wage was the key component of the costs. A one-way sensitivity analysis was carried out by varying the hourly rate from $43.2/h (awards level 1, year 3) to $66.3/h (awards level 4, year 1) [16] to assess the impact of labour costs on the NPV.
Descriptive statistics were used to explore frequencies, mean/ median values of the key variables, standard error of mean and standard deviation.

Sample characteristics
Demographic characteristics of the 704 participants and comparison with the general Australian population where available are provided in Table 1. Arm 1 (the general public) included 471 participants (60.7% female) with a mean age of 64.6 years (range 18-100); arm 2 (the experienced user) included 96 participants (59.4% female) with a mean age of 61.9 years (range 19-97); and arm 3 (the potential user) included 137 participants (59.9% female) with a mean age of 74.2 years (range 21-97). The survey response rates were 38.3%, 99.0%, and 92.6% for arm 1, arm 2, and arm 3, respectively. Approximately, 15% (n ¼ 106) participants did not provide income information. The pattern of missing data appeared to be "missing at random" and was imputed using regression imputation method. Arm 1 sample (general public) was consistent with the general Australian population in terms of state of residency. Overall, there were more female and older participants in comparison to the Australian population (Table 1). Figure 1 shows the number and percentage of participants whose WTP for O&M programs lies within each bound. The mode of all three arms was in the "$0-$5k" category, with 39.3% for arm 1, 55.2% for arm 2, and 58.4% for arm 3. Results of univariate and multivariate interval regression analyses can be found in Supplementary material 2.

Discussion
To our knowledge, this study provides the first economic analysis of O&M programs. Our cost-benefit analyses estimated that O&M programs have an NPV of $3857 (95% CI: $3760-$3954) per client, indicating investment into these programs generates benefits for clients in excess of the cost of providing the services, and thus is deemed socially efficient investment.   In this cost-benefit study, we adopted the CV method to estimate the WTP for the O&M programs of which the price has yet to be determined by a free market in Australia. A feature of the study is that survey respondents defined the benefits of O&M, rather than relying on researcher guidance. The amount a respondent was willing to pay for the O&M programs represents the maximum utility/benefits they anticipate receiving from undertaking the programs. Therefore the anticipated utility/benefits, potentially extend beyond managing safe travel, may include a wide range of presumed functional capabilities and well-being outcomes that were understood and important to the respondents. While some respondents may focus on the ability to travel safely, efficiently, and fluently, others may appreciate the opportunity for education and employment, the ability to build and maintain networks of families and friends, or an increased sense of dignity, confidence, wellbeing, and life satisfaction in general.
In the disability sector, WTP can be considered as a superior outcome measure over other frequently used measures in economic evaluations such as quality adjusted life year (QALY), as it imposes no restrictions on which dimensions of benefits people are permitted to express their preference [10]. Moreover, a question about how much money an individual is willing to pay for a service appears more intuitive and realistic than the questions that underpin the measurement of QALYs such as what trade-off in life expectancy, or risk of immediate death, would an individual accept to avoid a particular health state [21].
However, the "all-embracing" nature of the CV method may limit its potential to offer practical policy guidance as the notion of "utility" is broad and potentially more difficult to grasp than health outcome indicators (such as QALYs) more conventionally used to evaluate health sector programs. Another drawback of CV method is its underlying assumption that individuals are the best judges of their own utility ("consumer sovereignty") and they make rational choices which lead to utility maximisation [22]. In reality, this assumption may not be fully realised due to information asymmetry, endogenous preferences, or lack of motivation to make consumption decisions [23]. Moreover, the CV method is based on hypothetical scenarios, which means it is not possible to tell with certainty how the person will react in a real-life situation [24]. As Drummond et al. pointed out, the gold standard of WTP is what consumers actually would pay [9].
In addition, there are potential methodological limitations associated with the CV approach in terms of elicitation format, sampling strategies, survey administration, data analysis and reporting. In order to address these, this study adhered the following recommendations by previous studies [11,25]: (1) pre-testing and piloting the survey questionnaire to ensure that the scenarios and the outputs of the O&M programs could be easily understood by the participants; (2) measuring willingness to pay rather than willingness to accept, to produce more conservative estimates; (3) reasons for "refusal to respond" were explored to identify any protest responses and underlining issues of the survey design; (4) the survey was administered via telephone interview, rather than mail or online self-administered questionnaire to minimise sample selection bias (as those who return the mail or online survey are typically more interested in the intervention in question than those who do not); (5) participants' demographic and socio-economic characteristics were collected and adjusted for in the regression analyses.
To estimate the WTP amount for O&M programs, respondents were asked two yes/no questions that narrowed the bounds on WTP, the so called DBDC approach. DBDC is generally a preferred elicitation format when compared to an open-ended question as it is simple and cognitively manageable [11]. DBDC minimises the zeros or implausibly high value responses and the probability of strategic response bias -a type of bias that occurs when respondents intentionally try to manipulate the survey outcomes by providing untruthful WTP [26]. However, DBDC is prone to starting point bias, that is, respondents take the proposed bid as a cue for a reasonable WTP amount [27]. In this study, the first bid of $7000 was purposefully chosen to reflect the NDIS price. Currently, the NDIS prices are set by the National Disability Insurance Agency, not by a competitive market. It has been suggested that the current NDIS prices set for the O&M services are "very generous" and may have supressed the price signal and hindered allocative efficiency. The findings of this study may provide an indication of how the market price of O&M services may look like once the National Disability Insurance Agency abolish its pricing regulation and allow supply and demand to determine the market prices.
In the follow-up question, a higher bid of $9000 was offered to respondents who accepted the first bid; and a lower bid of $5000 was offered to those rejected the first bid. The $2000 increment was chosen for the follow-up questions to ensure the WTP estimates were within a plausible range and to minimise any floor or ceiling effects. DBDC is more efficient than single bounded dichotomous choice as more information is elicited about the respondent's WTP, therefore decreasing the number of observations needed for a given level of statistical precision [25].
The mean WTP for O&M programs was only slightly lower than the current NDIS price. The high WTP from the general public (arm 1), especially students and those with better perceived health, may be a result of hypothetical bias or the exaggerated anxiety over the proposed scenario as they may have lacked information about the lived experience of vision disability; whereas people with vision impairment (arm 2 and 3) may be less concerned about independent travelling once their vision condition was stabilised and psychological adaption and adjustment had taken place. Studies show that healthy members of the general public tend to focus more on the transitional loss in utility associated with a change in health state (e.g., loss of sight) and ignore the adaption that takes place; whereas the individuals with lived experience tend to focus on the adapted levels of wellbeing and downplay transitional loss [28]. A previous study reported that when being asked to value utility associated with blindness, people with good vision will imagine the moment of becoming blind and focus on the peak and immediate transitional loss (i.e., "Peak-Start Rule") [28], which may lead to shock reaction or exaggerated fear of blindness. People with good vision not only fail to anticipate the adaption process but also assume the quality of life will worsen the longer the blindness lasts. As for people with vision impairment, their perception of quality of life may evolve through mechanisms of adaptions, coping and adjustment. Their judgement of utility associated with vision impairment may be affected by whether they were born blind, how fast they progressed to blindness, how long they have been blind, the comparison with other visually impaired people, and/or adoption of lower standards for the intensity of positive affect. In this study, adaption may explain the lower adjusted WTP for O&M services amongst respondents with vision impairment, after controlling for income and other socio-demographic factors. The higher WTP values from the general public also indicates strong community (and by implication, taxpayer) support for O&M programs; this is one of the key factors to the sustainability of the NDIS funding. The NDIS is a large-scale social reform in Australia, which replaces the welfare approach with a universal insurance scheme. It was argued that the financial sustainability of the NDIS is a function of the perceptions of the NDIS participants and the contributors (taxpayers), where the NDIS participants must perceive that they are receiving sufficient support to live a fulfilling life and the contributors must perceive that the cost of the NDIS is affordable, under control, and represents value for money and therefore remain confidence to contribute what is needed [29].
Participants were asked an open-ended question about how much they were willing to pay if part of the O&M programs were delivered via tele-practice. While the open-ended question was easy to convey and did not result in any starting-point bias, it did suffer a higher non-response rate (24.3%) and a higher proportion of zero WTP amount (22.9%) of those who replied. The possible cause of non-response is the cognitive challenge imposed to the respondents by this question [25,26]. Respondents may find it too difficult "to put a number on it" especially when they were not familiar with tele-O&M. Three main explanations for zero WTP were identified through a follow-up question: (1) some did not believe tele-O&M can achieve the desirable outcomes; (2) some could not afford/or were not willing to acquire access to internet or tele-practice equipment, such as computers, camera, and assistive technology, this mirrors barriers for delivering telehealth in rural Australia [30]; and (3) some strongly felt that government should pay for it and disregarded the hypothetical nature of this survey question.
Although the low WTP values suggest a lower acceptance in tele-O&M compared to face-to-face O&M, the benefits of this innovative model of service delivery may not yet be recognised. The benefits may include the opportunity to improve access to O&M programs in rural and remote areas, reductions in waiting time (this is particularly significant when the vision disability sector is facing a workforce shortage [31]) and the prevention of potential decline in functional ability over the waiting period. One of the challenges in providing O&M programs to rural and remote populations in Australia is the lack of critical population mass required to sustain viable "on the ground" and already in shortage O&M workforce to meet the needs of people with low vision or blindness. A small number of studies show that tele-O&M may be a cost-saving alternative to address this issue [32][33][34]. The intention of tele-O&M is not to replace face-to-face delivery of O&M programs, but rather to complement traditional O&M services and enhance service accessibility in rural and remote areas. For example, in a pilot study investigating the viability of delivering O&M service remotely via video conferencing, 25 sessions amongst 11 participants were delivered by a research team in Western Australia [32]. During these sessions, an office-based O&M specialist provided instruction, guidance, and training as required throughout sessions. Participants with vision impairment were wearing a chest harness carrying a smartphone. The angle of the smartphone was adjusted to suit a variety of O&M assessment and training purposes. A support person (e.g., a family member, friend, or support worker) accompanied the participant to provide additional information about the environment if required or assist the participant use the technology if needed. While results were promising with regards to long cane skills and travel to their chosen routes independently [32], tele-O&M needs to be evaluated in prospective trials.
Our survey was conducted before the COVID19 outbreak. Due to the pandemic, many Australians, including people with vision disability, have changed their way of accessing health care (as well as education and employment) towards an increasing utilisation of video conferencing. It is suspected that the acceptance of tele-O&M may have increased since the pandemic.
Some limitations were identified in this study. First, the representativeness of our data. Data from experienced and potential users may not be reflective of the broader Australian population with visual impairment as these participants were entirely from New South Wales. Although participants from the general public were recruited via random sampling stratified by state of residency, our samples comprised more female and older Australians and their representativeness for the Australian population may be compromised. Second, the potential bias due to a high nonresponse rate amongst general public. Survey response rates have been declining in developed countries to the point where many cross-sectional surveys now have response rates below 50% [35,36]. The decline has been even more drastic in telephone survey possibly due to unauthorised release of personal information from the internet being used in fraud and marketing activities [36,37]. In this study, low response rate amongst general public was resulted from non-contact and refusal: (1) almost half of the fixed-home phone owners and one in five mobile owners were not reachable by the interviewers as they had their phone/mobile phone number listed on the Do Not Call Register [38] and (2) for those who were reached by the interviewers, 61.7% refused to take part in the survey. While conventional demographic wisdom suggested the low response rates may lead to bias, survey research literature [36,39] argued that "low response rates produce bias only to the extent that there are differences between responders and non-responders on the estimate(s) of interest, and then only if such differences cannot be eliminated or controlled for through the use of observable and available characteristics of responders and non-responders." [36] Findings of Rindfuss et al.'s study suggested that low response rates need not necessarily lead to biased results, and bias is more likely to be present when examining a simple univariate distribution than when examining the relationship between variables in a multivariate model [36]. In this study, the potential non-response bias was addressed through multivariate regression analyses, controlling for a variety of background variables. Finally, the 10 -15-min telephone survey did not allow the inclusion of a detailed description of the tele-O&M procedures in the script. While some participants questioned how someone with vision impairment could undertake O&M training via teleconference, most may have kept their doubts to themselves. Uncertainty and apprehension about this innovation may have affected participants' WTP for tele-O&M.
A potential topic for future research would be to conduct a CBA including comparators that are competing for the same source of budget, from a societal perspective. Therefore, the opportunity costs, in terms of benefits forgone from the displaced disability services, can be accounted for. At present, published studies of telerehabilitation for people with vision impairment focus on assessing the feasibility and/or efficacy of the innovation. Future studies, such as randomised controlled trials and implementation studies, are required to investigate the replicability, adaptability, effectiveness, and scalability of the innovation.

Conclusions
This study reported a CBA of O&M programs that aim to equip a vision-impaired person to travel safely and independently to the nearby shopping centre from home. Overall, the NPV of O&M programs was $3857 (95% CI: $3760 -$3954) per client, indicating that investment into O&M programs generates substantial benefits for clients over and above the cost of providing the services. The high WTP values ($6716, 95% CI: $6373 -$7060) from the general public suggest a strong community support for O&M programs, hence, taxpayers' willingness to contribute.