*Healthcare* **2019**, *7*, 54

*Healthcare* **2019**, *7*, x 7 of 17

**Figure 2.** Post-session responses of all participants to the statements relating to issues raised in the juries. Differences between the groups were not significant for any of the statements. **Figure 2.** Post-session responses of all participants to the statements relating to issues raised in the juries. Differences between the groups were not significant for any of the statements.

#### *3.3. Analysis of Discussion During the Citizen's Jury Sessions*

The deliberations that took place at the two citizen's jury sessions were audio-recorded and analysed through exploring two perspectives. Personal experiences as well as hopes and fears about how technology may affect individuals was interpreted as reflecting the life-world. Participants discussed the potential efficiencies or improvements that the digital system could achieve; they also expressed concerns about surveillance of citizens and other risks, and these were interpreted as reflecting the systems-world. These two perspectives enabled a more nuanced interpretation, rather than a polarised interpretation of positive or negative outcomes (for the individual). Initial themes emerged from the groups of the open-space engagement session. The discussions within citizens' juries then added weight and resonance to these (see Table 3).



#### *3.4. Concept of Smart Health*

There was much discussion about the meaning of the term "Smart Health". Our assumption was that the term relates to digital technologies that may improve or affect health and healthcare, and much of the discussion resonated with that concept. Different interpretations were that SMART was an acronym for something or that smart meant healthy living, or equivalent to good health literacy.

" . . . it's what you eat. Now then isn't that an education process where we're talking about being smart with our health? It's nothing to do in essence we've got a gizmo on the table; it's whether or not we've got the capability to understand what in fact smart health is." (Group A, male respondent)

Whilst this quote initially appears to be discussing a different concept; it highlights a need to understand health literacy as well as digital devices. Having considered this range of concepts of the neologism "Smart Health", we will focus our interpretation on the meaning that many participants touched upon. This was very clearly described in the following quote from one participant:

" . . . about using devices like your mobile phone, your computer, an iPad-kind-of-thing, anything digital like that. And then using like little programmes that you might call apps with some computers to help you manage your health long term of your life. So that if you've got a health condition like diabetes or something, you can manage it yourself and take control and be independent, but I would only say that as an abstract concept, not as a living position." (Group B, female respondent)

For the main part of the discussion, we interpreted views about a number of topics, and we have attempted to contrast two perspectives that were voiced by participants; views about personal experiences, or life-world, and views about the system or citizens as a whole.

#### *3.5. Theme 1: Control and Privacy versus Mis-Trust in Purpose of Data Use*

Discussion about errors and fraud were voiced as a way to demonstrate concerns about control and privacy. One participant describes the GP software system being offline, possibly due to an error, and this preventing transfer of case notes. This may be frustrating at a personal level, due to inconvenience, but it may prompt general concerns about risks of data, due to error or fraud.

"At the moment the software at my GP place is—to use a technical term—buggered up, because I've got some other thing and they won't transfer electronically. (Group A, male respondent)

One participant had concerns about the Council using or sharing data in ways that were not in the interests of the individual. Concerns were raised about whether data was being collected in order to develop a marketable database of personal data. This indicates an awareness of the high value of personal data and also a lack of trust in the purpose of the system collecting this data.

" . . . I have a comment on the technology of this. That is, I think our approach is entirely wrong. The technology is being introduced so as to accumulate a large databank which is sellable; it's not got anything to do with our health." (Group A, male respondent)

There was discussion of governance and suggestions of additional regulation to reassure individuals. There was also an acknowledgement that there may be a diversity of views from individuals about the level of concern about sharing data.

"I have no problem personally with sharing my data, but I do understand other people do. And it's a matter of choice. For me the solution to this would be actually regulation. So, if people abused access to your data and information that there were penalties that they would pay." (Group A, female respondent)

Concern about private multinational companies collecting medical data.

" . . . Google are now wanting to set up a website to do with smart health. They want access to your medical records, and I'm against that, some people who agree with it, that's entirely up to them, but with me my information will stop with the people who I want to have my information." (Group A, male respondent)

In this section, experience of digital technology in the personal life-world may be a feeling of invasion of privacy of data, especially if an individual's data is being used or shared in a way that was not clear or transparent. Furthermore, digital technology may enable an individual to have a greater sense of control of their GP appointment, for example, but when an error occurs, this might spark concerns about a lack of control of their personal medical data. On the other hand, the weaknesses of the systems are revealed when a computer (ICT) problem occurs, which leads to loss of control. Where the system shares data, there may be concerns as to the purpose. A concern about the systems-world is that it gathers data, almost as an inherent characteristic. Beliefs about motivation for collecting data were because large datasets are seen to be valuable or because data could be used to control or surveil the individual.

Within this theme, the life-world perspective may be described as the convenience of using online systems, for example booking appointments or sharing data with different professionals. Whereas the systems-world perspective highlights a concern that personal data is being amassed, and this may be associated with risks of accidental breach of confidentiality, or purposeful selling of data. There was also a concern that data could be shared with a motivation of controlling aspects of people's lives (maybe welfare benefits) or services. Responses to these concerns were at both the systems-world level (regulation and sanctions) and the life-world level of acknowledging that people opt-out or refuse to share their personal data.

#### *3.6. Theme 2: Choice or Self-E*ffi*cacy versus Standardisation*

Fears were voiced that with an increasing implementation of digital systems, in the future it will not be possible to opt-out or use non-digital processes. This may be interpreted as the efficiencies of standardisation of the systems-world; that bureaucracies aim for a standard process rather than flexibility to individual preferences. Participants suggested that some individuals may not want to use digital technology; which indicates that there is an expectation within modern discourse that everyone will adapt to digital technology (given time and opportunity). The views expressed questions of whether some people may not accept digital technology, and whether their views and rights should be respected. This led to an expression of concern that a group of people may have their rights infringed upon in the future, and that they will be disadvantaged if they do not accept the use of digital technology. One participant used the analogy of online shopping:

"It's like people who buy things online now and get a better deal. But not everybody wants to do that, and not everybody should be forced to do it. So, it might be . . . based on individual need and the individual willingness to do it." (Group A, female respondent)

This description of buying goods online as an analogy to accessing welfare services indicates an acceptance of the discourse, in media and policy, about welfare services being conceptualised as commodities to be bought by, or given to, individuals, rather than as public goods to which citizens have a right to access. This is exemplified by the phrasing of this quote: " . . . manage for yourself; your health, your wellbeing over a long time" (Group B, female respondent).

The systems-world perspective is often about standardisation and efficiency of processes and services. Thus, there was a view that, in the future, older people would not have a choice, but would have to use digital technology to access health and care services.

"I think there is a certain section of society upon which it will be imposed. They won't have any choice, mainly for cost reasons. Services can only operate if we have a system working and everyone is included in it...the point will come when they cannot be cared for adequately without this system, without wearing something on their arm. And that will come with our 87-year-old [persona]. If she hasn't taken her chance to learn basic technology when younger, when she is older and very dependent, she's so confused she doesn't know how to use it, and she hasn't a position to say no I resist any longer. It will be forced on her; she will have to accept it. So, it will be unfair, it will be undemocratic, but that is the way it is likely to go." (Group B, male respondent)

This respondent makes a clear link between the systems approach of standardising care processes and the risk that this may mean that some individuals will have to accept technology with which they do not feel comfortable. At a personal, life-world level, this indicates a constraint in choice of care or treatment, while at a systems-level this becomes about democratic choice in investment in services and technologies.

#### *3.7. Theme 3: Data Sharing Enables Continuity of Care versus Cross-Checking between Agencies*

One participant described data-sharing in a positive way; this participant is describing telehealth.

" . . . if you're wearing or having some device, then the information you provide or is provided by you, or your piece of equipment, then goes back to a centre. So, it goes to your health worker, whether it's your GP, the hospital, district nurse or whatever they call them today, and that saves time, energy, money." (Group A, female respondent)

The participant implies that through sharing data between all members of the healthcare team, it will improve efficiency of communication, and hence improvement of continuity of care.

However, another participant had a very cynical view of how organisations could use personal data.

" . . . if you ever have a problem with [organisations] like I do, they can access your whole life near enough at the click of a mouse button. And I don't want them to have that." (Group A, male respondent)

This participant mentioned that he had previously had conflicts with the Council, so this may have shaped his mistrust in the digital information. This demonstrates how views about digital or smart technology are overlaid on previous relationships with institutions; these might be new technologies, but they are embedded in existing bureaucracies and systems.

These two respondents demonstrate how this interpretation may open new discourses about data use and trust in data-sharing. Whereas the first quote is about personal care and improving continuity, that is where the individual may gain benefits from opting into the system. The second quote shows how the individual is thinking about how the system works at a bureaucratic level, and what the implications might be for control of personal data. Development within smart cities should acknowledge these two discourses in order to improve governance and processes as well as communications about these with stakeholders and public.

*3.8. Theme 4: Systems-World Reach into Personal Devices; Convenient Reminders or Over-Reach?*

" . . . the appointments, notifications on the phone. Which I think is great, it's a good idea." (Group A, female respondent)

This participant is describing the healthcare system's use of efficient scheduling and digital communication to reach into the domain of personal communication, the mobile phone. The participant welcomes this, presumably from a perspective of convenience and preventing forgetting the appointment. However, this may be an area of tension, where other individuals may feel that reminder notifications on their mobile phone may invade their personal space and life-world. Another participant had had phone and skype consultations with the doctor and this participant had a similar view; that this was convenient and saved the doctor's time.

"I very often don't need to go down to the doctor. I've had one phone appointment with the doctor, but I would quite like a Skype for the next time appointment; to save me going down sometimes and to save them time." (Group A, female respondent)

Again, receiving a phone call from the doctor at home and conducting a medical consultation over the phone could be perceived as the systems-world accessing the personal space of home, and carries the risk of communications being unsecure. People may become concerned that organisations or systems can reach into their personal space to communicate or monitor their activity.

" . . . Even though I've got a laptop, I treated myself to a [Smart TV] . . . it frightens me to death. I've got this thing that somebody's watching me." (Group B, female respondent)

For individuals with limited cognition or communication, it may be difficult to understand their view on health monitoring and use of data; and yet this may be a situation where monitoring an individual's health status is a priority. One participant described the importance of understanding the individual's wishes before cognitive decline.

"I know my husband and I have talked about people having power of attorney at various time about care, about finances. People have got to make those kinds of decisions before they . . . [deteriorate]." (Group A, female)

#### *3.9. Theme 5: Ownership versus Collecting Population-Level Data*

Participants from one session mentioned ownership of health records, comparing the situation in Britain with France. Her experience in France was that individuals have ownership of their records and take them to the doctor, whereas Britain was perceived to be behind the times in not enabling people to own their records.

" . . . why Britain is one of the few countries in Europe that people don't keep their own records. I mean I know that when I'm in France if someone goes to the doctor, they take their records with them. And I don't see why I'm not grown up enough to know what's wrong with me . . . in Britain, it's always been the doctor's always the way; that knows the answer, and you're there listening to the great God doctor." (Group A, female respondent)

This participant is indicating that the lack of access and ownership to personal health data indicates an entrenched paternalistic relationship between healthcare professionals and patients. This is a description of the systems-world, and a frustration that the personal health information cannot be owned and co-located within the life-world of the individual.

" . . . I think it is important that the individual is in charge of it." (Group A, female respondent)

Ownership of data could lead to individuals checking the validity of data and correcting errors. Another respondent indicates that they would be willing to share personal data, as long as an appropriate regulatory framework was in place, with appropriate sanctions.

" . . . I have no problem, personally, with sharing my data, but I do understand other people do. And it's a matter of choice. For me the solution to this would be, actually, regulation. So, if people abused access to your data and information, that there were penalties that they would pay." (Group A, male respondent)

An exchange between two participants highlighted the difference between personal data for care of the individual compared to the same data being aggregated and used for population intelligence. The first participant starts by introducing the idea that information is provided by the individual, phrasing which may indicate a sense of ownership. This information then "goes back" to a centre which coordinates professional activity; this phrasing suggests a spatial distance between the personal and professional (systems) worlds. The outcome of these processes is that "your GP . . . district nurse" is notified of the issue and can respond in an efficient and timely manner, indicating a personal and convenient response. These savings may refer to the system, and the mention of money suggests efficiency for the system rather than savings for the patient (as there are no out-of-pocket fees for health professionals' time in the health service in England).

" . . . if you're wearing or having some device, then the information you provide or is provided by you, or your piece of equipment, then goes back to a centre. So, it goes to your, so your health worker, whether it's your GP, the hospital, district nurse or whatever they call them today. And that saves time, energy, money." (Group A)

In responding to this participant, another participant takes the "indirect" perspective of the systems-world. He argues that although there has to be potential to benefit the individual patient, there also has to be a benefit for the health system; this phrasing—"has to benefit the health service"—suggests a "business case" type of argument. Personal data collected by various devices is interpreted by analysts to yield population data in order to improve decision-making for future health service planning. This latter perspective is an objective argument which also has potential to benefit the individual in the long term, and is a strong contrast to the personal benefits of arranging multi-disciplinary care in a timely way to meet the needs of an individual (person-centred care).

"It has to be for the benefit of the patient. I fully accept that. But, also, there is an indirect benefit to the patient in that it has to benefit the health service itself. The collection of data about the community—and that will ultimately help you. It may not give you an immediate assistance, but down the line, people who are able to interpret it will know more about the population and be able to make more intelligent decisions about healthcare." (Group A, male respondent)

Taking a systems-world perspective, the participant argues that aggregated data can inform health planning. This is a complex argument and indicates a high level of knowledge and consideration by this particular participant.

#### *3.10. Theme 6: Co-Design for Older People*

Different perspectives may be characterised as "why do we have to use digital technology to access services that we had for years". This contrasts with the systems-world assumption that older people should use technology in the same way as younger people (already) do.

"Now does in fact Gladys [persona] want somebody to call in to her who can remind her how to in fact access a part of a computer programme? I forget, and I'd spend more time trying to remember how to do it, purely and simply because I only need to do that particular problem on an infrequent period of time. So, I get frustrated." (Group A, male respondent)

With this perspective in mind, participants were keen that technology developers should involve older people into the design of products and systems.

"But the technology companies have to employ people like Gladys [persona] and say right, we've got this thing, does it work for you? And I'm not sure the extent to which they use people like that when they're designing their products." (Group A)

Thus, at the systems-level, data might identify that a proportion of people are not accessing technologies or services delivered in a technological context; however, we need to understand how individuals interact and find meaning in digital technologies, in order to improve design to be accommodating of all older people.

#### **4. Discussion**

#### *4.1. Key Findings*

This paper compares the opinions and attitudes about smart cities and the impact on health and well-being. We held two citizens' juries, where the difference between the two juries was that one group had previously been involved in the co-design of the content of the session (B), whereas the other group were new to the project (A). The results revealed that there were no differences between the juries in existing levels of knowledge, opinions, and in attitudinal change. The pre-session survey was implemented to gauge the existing level of knowledge and opinions. Whilst the post-session survey was implemented to measure attitude change and measure opinions on the issues discussed.

The survey completed before and after the jury session can be linked to the topics discussed at the juries. Discussions revealed participants' deliberations about the benefits and risks of smart health technologies and system. During the pre-session survey, 44.4% (Jury A) and 14.3% (Jury B) of participants expressed concerns towards technology. This result highlighted the differing welcoming attitudes to smart health. Whilst discussing attitudes, participants voiced scepticism and resistance towards smart health technologies. Concerned participants expressed a preference for face-to-face support. In the post-session statements, a majority of participants did not agree it was a good idea to replace humans with technology. However, the participants did express that technologies can help reach those who live alone and aid in social interactions, mentioning benefits to health problems in older adults such as dementia. This was reflected in the over half of the participants agreeing that smart city initiatives can help reach more people.

Although a majority of participants suggested in the pre-session survey that they often use technology, in the discussion, participants made recommendations of training in technology for older adults. They also mentioned issues of the digital divide, which was expressed in rating in post-session statements. Despite identifying a digital divide, pre-session results suggest that participants do believe it is important that older adults use new technologies. Furthermore, responses to the post-session questionnaire which suggests that individuals will try and use health technologies more often, although responses were mixed. This should be an incentive on the potential of greater use of health technologies, provided technologies are accessible, simple, and affordable for the target population.

Participants recognise the importance of sharing information through health technologies and how it can potentially benefit their navigation in healthcare (such as making appointments). A majority agreed in the post-session statements that the benefits of health technologies exceed the risks. The group did, however, raise issues relating to the regulation of data sharing and their part in controlling the information. Generally, in the pre-session survey, when asked about ethical consequences, a majority of participants in both groups had some concern. This trend continued in the responses in post-session statements related to ethical consequences, where even after discussion, a majority disagreed with having minimal concern.

Qualitative analysis used a Habermasian approach of exploring perspectives on life-world and systems-world. The advantage of this approach is that the personal experience can be investigated and separated, to a degree, from the qualities of the emergent system. This is particularly important with integrated systems and data; it may not be the individual piece of technology which has a positive or negative outcome, but rather the technology within multiple interrelated systems (digital and process, i.e., bureaucratic systems). We applied this approach to explore six themes which were prompted or emerged during the citizens' jury sessions.

While raising some scepticism and concern, participants generally want to be more engaged in the design and implementation of health technologies. The participants stressed the importance of testing technologies on older adults, echoing ideas that technologies need to be simple and accessible. This collaborative approach reduced concerns of being forced to engage with technologies that are not wanted and allowing older adults to regain control.

#### *4.2. Internal Validity—Strengths and Weaknesses*

A range of views were expressed from male and female respondents and across the group which had previous involvement in the project and the group which was new to the project. No specific patterns were detected across these groups. Furthermore, we did not detect a self-selection issue.

These were small groups (*n* = 9 and *n* = 12) from one city in England. The views were likely to be influenced and contingent on the public discourse within the city. However, this approach was important to recognise for a city-based initiative such as "Smart City Nottingham" because processes and public communications should be adapted to local contexts.

We took a very broad approach to digital technology, rather than focusing on a particular platform or device. The weakness of this approach was that various comments may not relate, and there may be a lack of depth of discussion. However, the advantage was that the analysis gained a "bigger picture" interpretation of concerns which may be important to understand at an overarching level.

#### *4.3. External Validity—How Does It Compare to the Literature*

Our broad approach relates to many different disciplines, from healthcare to data-systems design. This approach is consistent with "lifeworld-led healthcare" and the previous body of work on patient-centred care [23,24]. We have built on a Habermasian analysis of the medical encounter where the intermediary between doctor and patient, a language interpreter, implicitly negotiates between life-world and systems-world [25]. Whereas, in our study, digital technologies and systems act as intermediaries between citizens and health professionals and the city bureaucracy. This approach has enabled a detailed interpretation of complex interrelationships which are often conceptualised as a "wicked problem" of the "digital divide" [26].

#### *4.4. Future Work*

Whilst some discourses perpetuate the view that older people respond in a passive way to innovative technology, our study has found a desire of individuals to be consulted and participate in the co-design of smart systems. There is a growing awareness of potential inequalities that may emerge as older people find it difficult to access services due to technological barriers. From a human rights perspective, older people have a right to be involved in the design and implementation of technologies and systems where they are the main beneficiary. Further work is needed to explore the two elements of health literacy and digital literacy and how these interact at a personal level and at a city-wide level.

#### **5. Conclusions and Recommendations**

Our study took a co-design approach in developing citizens' jury sessions to explore the views of how a smart city may affect people's health and well-being. Using a persona to discuss several dilemmas enabled exploration and deliberation on a number of common themes of data control, privacy, and convenience of technology. Surveys before and after the jury sessions captured the range of perspectives within the group and could counter any claims that these groups of participants represented any particular interest. Participants expressed concerns about the risks of data sharing and use of data; however, the convenience of booking appointments or accessing online healthcare records was valued. Participants were aware of the benefits of digital systems to the health and care sector, especially for efficiency and collection of data. Our interpretation of life-world and systems-world perspectives enabled a nuanced understanding of these tensions or trade-offs within the implementation and experience of a smart city for older people.

We recommend further research in the following topics that were found to resonate with participants: data-sharing and trust in use of data; personalisation or standardisation; and surveillance in the home. Many of these topics relate to trust between citizens and the organisations involved in the system (especially health and social care providers). Co-production may facilitate trusting relationships, and citizens' juries are one method to achieve this with a rights-based deliberative consultation. Further research is required to explore how statutory, private, and third-sector organisations can best respond and incorporate these views in strategy and implementation.

**Supplementary Materials:** The following are available online at http://www.mdpi.com/2227-9032/7/2/54/s1, Pre and post survey.

**Author Contributions:** Conceptualization, N.H.C., E.P.V.; Data curation, N.H.C., K.B., N.H., and E.P.V.; Formal analysis, N.H.C., H.C., N.H., O.A., and E.P.V.; Funding acquisition, N.H.C., N.H., and E.P.V.; Investigation, N.H.C., K.B., N.H., L.D., and E.P.V.; Methodology, N.H.C., H.C., N.H., and E.P.V.; Project administration, N.H.C. and K.B.; Resources, N.H.C., K.B., N.H., L.D., and E.P.V.; Writing—original draft, N.H.C., K.B., H.C., N.H., and E.P.V.

**Funding:** This research received no external funding.

**Acknowledgments:** Elvira Pérez Vallejos acknowledges the resources of the NIHR Nottingham Biomedical Research Centre. Liz Dowthwaite and Helen Creswick acknowledge the resources of Horizon Digital Economy Research (UKRI grant EP/G065802/1) and the UnBias project (UKRI grant EP/N02785X/1).

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


© 2019 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

## *Article* **Technology to Support Aging in Place: Older Adults' Perspectives**

#### **Shengzhi Wang <sup>1</sup> , Khalisa Bolling <sup>2</sup> , Wenlin Mao <sup>3</sup> , Jennifer Reichstadt <sup>4</sup> , Dilip Jeste <sup>5</sup> , Ho-Cheol Kim <sup>6</sup> and Camille Nebeker 7,\***


Received: 28 February 2019; Accepted: 3 April 2019; Published: 10 April 2019

**Abstract:** The U.S. population over 65 years of age is increasing. Most older adults prefer to age in place, and technologies, including Internet of things (IoT), Ambient/Active Assisted Living (AAL) robots and other artificial intelligence (AI), can support independent living. However, a top-down design process creates mismatches between technologies and older adults' needs. A user-centered design approach was used to identify older adults' perspectives regarding AAL and AI technologies and gauge interest in participating in a co-design process. A survey was used to obtain demographic characteristics and assess privacy perspectives. A convenience sample of 31 retirement community residents participated in one of two 90-min focus group sessions. The semi-structured group interview solicited barriers and facilitators to technology adoption, privacy attitudes, and interest in project co-design participation to inform technology development. Focus group sessions were audiotaped and professionally transcribed. Transcripts were reviewed and coded to identify themes and patterns. Descriptive statistics were applied to the quantitative data. Identified barriers to technology use included low technology literacy, including lack of familiarity with terminology, and physical challenges, which can make adoption difficult. Facilitators included an eagerness to learn, interest in co-design, and a desire to understand and control their data. Most participants identified as privacy pragmatics and fundamentalists, indicating that privacy is important to older adults. At the same time, they also reported a willingness to contribute to the design of technologies that would facilitate aging independently. There is a need to increase technology literacy of older adults along with aging literacy of technologists.

**Keywords:** retirement community; co-design; privacy; research ethics; artificial intelligence; robots

#### **1. Introduction**

The preference of older adults to "age in place," or to live independently at home, rather than in an assisted living facility, is widely recognized [1–4]. Individuals who develop disabilities and are no longer able to age in place are likely to be institutionalized in assisted living facilities. These declines, which can occur with advanced age, are key barriers to one's ability to maintain an independent lifestyle [5,6]. This often leads to more significant mental and physical decline, as well as quality of life decline and increased cost of care, compared to older adults who continue to live independently [7]. As a result, effective means of providing support for older adults are of central public health and ethical significance. In many cases, external aids provide crucially needed assistance that can prolong independent living.

Technologies, such as internet of things (IoT), Ambient/Active Assisted Living (AAL) robots and other artificial intelligence (AI), have been shown to have great potential in fostering independent living, improving mental and physical health, and increasing quality of life [8–11]. At the same time, they can also reduce caregiver burden, which can lead to more targeted and better quality care [12]. However, despite playing a significant part in successful interventions, adoption of these technologies has been limited [13,14]. One key barrier to wider adoption has been the "top-down" design process that is often used in creating technology for older adults. This process is based on technologists', or at best geriatricians', preconceptions of the needs of older adults with little consideration of user perspectives and preferences or their real-world constraints.

While there are a number of studies that have indicated the need for well-designed technologies that meet the needs of older adults [15], few have addressed user-related issues in the design process of these technologies. It has been recognized that effective technologies are those that prioritize the needs and wishes of older adults, general acceptance of potential users, and suitable preconditions for its adoption [16], but this is often difficult to achieve with a top-down design methodology that fails to engage users in the design process. This has frequently created significant mismatches between the needs and preferences of the users and the products that are developed to fulfill their needs. Areas of concern for users include visual appearance, functionality, affordability, platform sustainability, privacy concerns, and interaction complexity [17–19]. These mismatches can hinder meaningful adoption and sustained usage, and risk leaving priority needs of end-users unmet. Employment of user- or human-centered design (HCD) involves the end user in the early planning phases to better understand the needs of individuals for whom a product is being developed and to ensure relevant safety, access, and utility are built in [20]. A design process involving end users can reveal untapped areas for improvement, which can lead to improved user satisfaction and lower adoption barriers, and ultimately to much improved support for individuals who wish to age in place [21,22].

The goals of this study were to: (1) involve residents of a local continuing care senior housing community (CCSHC) in conversations about technologies that might facilitate their continued independent living status [23]; (2) assess their privacy attitudes and preferences; and (3) identify whether residents would be interested in co-designing technologies moving forward and if so, how to foster next steps.

#### **2. Materials and Methods**

#### *2.1. Recruitment*

Two focus groups were convened at a local retirement community located in San Diego, California, to explore barriers and facilitators to technology adoption as well as interest in participating in a "tech" co-design process. This study was carried out in conjunction with a longitudinal, observational study involving over 100 residents. Both focus groups were held in August 2018. Residents of the retirement community were invited to participate via an IRB-approved flyer. Community leadership helped distribute the flyer and placed copies in the community lobby. Input was solicited from community leadership to determine appropriate time slots for the focus groups. Residents interested in participating were directed to sign up for one of two time slots offered. Selection criteria included any resident of the independent living facility with an expressed interest in the study and willingness to participate in a 90-min group discussion. Residents requiring assisted living or skilled care were

excluded from the study. To confirm attendance, an email reminder was sent to those who signed up for a time slot. On the day focus group sessions were held, no individuals were turned away. The study was verified as exempt by the UC San Diego Institutional Review Board. Focus group attendees were provided with an introduction to the project purpose and they gave verbal consent to participate. Each participant received \$30.00 as compensation for their participation.

#### *2.2. Data Collection*

Data were collected via focus group and survey methodology, both of which are described below:

#### 2.2.1. Focus Group Semi-Structured Interview Guide

The focus group protocol was developed with a goal of obtaining the perspectives and guidance of older adults over 65 years old regarding their: (a) use of and interest in technology and connections to personal health; (b) preferences for involvement in participatory design of AI assistive aids; (c) familiarity with terms, concepts, and processes associated with the design of AI aids for aging in place; and (d) advice and preferences for how technology development experts should most effectively communicate such information to enable an authentic and informed participatory design process (see Appendix A). An initial draft of the semi-structured interview protocol was reviewed by a resident leader of the retirement community and revised to incorporate comments.

Focus groups were conducted in a conference room located on-site in the CCSHC's main building. Each group session was allocated 90 min and was led by a trained focus group facilitator (CN). A student research assistant took notes and managed logistics while two residents volunteered to record individual comments on poster paper (SW). The moderator asked open-ended questions and participants were asked to answer asynchronously. Some questions prompted the participants to reflect on their answers and write down their thoughts on a 3 × 5 card before further questions were asked. Group discussions were digitally recorded (audio) and professionally transcribed. The focus groups aimed to deepen the understanding of the topic as participants built on one another's discussions and viewpoints. At the end of the focus group session, participants were asked to complete a written survey to gather demographic information as well as to assess privacy attitudes.

#### 2.2.2. Survey

A survey was used to obtain demographic characteristics of the participants with the four scales designed to measure privacy attitudes described below:

#### Westin Privacy Segmentation Index

The Westin Privacy Segmentation Index (PSI) is used to classify participants into three groups: Privacy Fundamentalists, Privacy Unconcerned, and Privacy Pragmatists [24]. The PSI consists of three questions that assess whether consumers trust businesses and regulations to safeguard their privacy.

#### Westin Privacy Concern Index

The Privacy Concern Index (PCI) is another scale that was developed by Westin et al. in the 1990s [24]. Scoring is used to classify participants into three groups: high, medium, and low with respect to the level of privacy concern. The PCI used for this study included five questions from the original six-item version. One item regarding the creation of a privacy commission by the government was omitted as it was unrelated to the present study. The survey also asked participants about their concerns on current threats to their personal privacy. While this question is not part of the PSI or PCI, it was also developed by Westin and Louis, and was used in prior privacy research [24].

#### Internet Users' Information Privacy Concerns (IUIPC)

The Internet Users' Information Privacy Concerns (IUIPC) scale by Malhotra et al. consists of ten questions with three subscales (Cronbach's alpha = 0.92) [25]. The first three questions (1–3) comprise the Control subscale and focus on the extent to which participants believe that control is the key issue with consumer privacy. Questions 4–6 are labeled the *Awareness* subscale and are used to rate the importance of disclosure and awareness of information collection. The last four questions (7–10) are labeled the *Collection* subscale and focus on the amount of information that is collected by companies. Along with these ten questions, two additional questions developed by Malhotra et al. were also included. One asked whether participants falsified their personal information during registration on websites and the other focused on how frequently participants heard about the potential misuse of the information collected from the internet [25].

#### Data Sensitivity

The last section of the survey assessed participant perspectives of the sensitivity of different kinds of personal information. The 12 different types of personal information were rated using a ten-point scale with 1 = "Not sensitive at all" to 10 = "Very sensitive." The majority of the 12 types of personal information focused on health information (e.g., electronic health records and present fitness), with three types being more general (e.g., internet search terms, GPS).

#### *2.3. Data Analysis*

The transcribed audio recordings were analyzed using a methodology of "Coding Consensus, Co-occurrence, and Comparison" and rooted in grounded theory (i.e., theory derived from data and then illustrated by characteristics examples of data) [26,27]. Each transcript was independently coded by two project investigators (JR and KB) at a general level in order to condense the data into analyzable units. Segments of transcript were assigned codes based on a priori (i.e., questions in the focus group guide) or emergent themes. In a few instances, the same text segment could be assigned more than one code. The investigators subsequently met to discuss and refine the codes and to develop a final list of codes, construed through consensus, consisting of a numbered list of themes and sub-themes, issues, and opinions. With the final coding structure, interrater reliability was assessed for a subset of 10 pages of each transcript (about half of the total pages), with a kappa value of 0.97, indicating a high degree of concordance between raters.

Quantitative data were analyzed using SPSS.

#### **3. Results**

Participants included 31 older adults between the ages of 67 and 94 years (mean = 80.0, SD = 6.2) with 20 females and 11 males, and 70% having a college or graduate degree. Most (97%) were White, with 60% reporting an annual income of over \$100 K (see Table 1). Two individuals did not complete the demographic and privacy survey.

The qualitative analysis revealed several key barriers toward adoption of technologies and digital platforms, namely: (1) technology usability, (2) technology literacy, (3) data management and privacy, and (4) technology co-design. Below, each theme is defined and characterized by participant comments and survey results.


**Table 1.** Demographics of the Sample.

#### *3.1. Technology Usability*

The theme of "technology usability" was defined by how accessible a product is to those attempting to use it. Sub-themes point to barriers around user interface making intuitive navigation of a product difficult, and physical challenges that become obvious when technologies are not designed for the older demographic.

#### 3.1.1. Lack of Unified Frameworks and User Interfaces

Because of the fragmentation of different digital platforms and services, there are many cases where the lack of a unified user interface can lead to increased user frustration and compromise usability. For example, one participant was trying to set up an email service and expressed frustration with navigating what should have been a fairly routine task.

*"To get email hooked up, to get this done, to get the keyboard* . . . *all that stuff, and a lot of folks just give up."*

Another purchased a backup system for her computer, but had no idea how to set it up.

*"I'm going 160 days without backup because even the geek group that we have here can't answer my question in order for me to get it set up."*

Frustration appeared to be a significant barrier, which led to a lack of self-confidence and motivation to pursue using the technology.

*"I think technology can, for some people, get to the point where life is more difficult than it was before we had that technology."*

*"I have a smart phone. My kids think I can use this, but I don't really know how to use this, and they bought it for me thinking it would be a great tool, but I don't really know what to do with it."*

Devices that were easy to use due to having simple features, such as the "on/off" switch for an electronic toothbrush, or plugs that worked regardless of how you inserted it (i.e., the Apple Lightning Connector [28]), were noted as being supportive technologies. Lastly, concerns were voiced about technologies that introduce hazards to older adults, such as the use of power cords. Since having power cords around the house creates a tripping hazard and fall risk, several participants advocated for more wireless functionality.

3.1.2. Increased Mobility and Visual Challenges When Using Technologies

Older adults often face challenges in accessing hardware features or digital content due to a lack of accommodation of their limitations in mobility and decreased visual capacity. In some cases, the technology is designed to keep the battery charge connection out of sight—perhaps for aesthetics or weather proofing. However, when concealed, connectors can be more difficult for an older adult to access. One participant described his experience trying to charge the electric scooter that he uses to get around. Specifically, the connection was underneath the seat, requiring that he either bend over or get on his hands and knees to locate and connect the plug for charging.

*"One of the biggest frustrations that I've seen was folks with power carts and a lot of them, to charge the battery, the plug is way down underneath your seat* . . . *. Couldn't bend over to put it in, can't see it* . . . *"*

There are also times when a simple solution, such as replacing a battery, proves challenging as evidenced by this comment:

*"I have had more calls from people who say, "I just put a brand-new battery and it doesn't work." They put the battery in backwards (laughter) and, at times, it burns out the unit."*

Participants noted that the difficulty in replacing a battery was related to the inability for many to visually see the positive and negative symbols.

#### 3.1.3. Recommendations for Improving Technology Usability

Participants were pragmatic in their recommendations for improving usability. Simple instructions, fewer buttons, larger fonts, and speech-activated tools were noted during the discussion.

*"Why don't they have a senior version or an app that can get to on, off, volume up/down, channel, and make it kind of simple?"*

Of interest, but not surprising, were also suggestions such as having a universal remote to operate the television and peripheral devices—technologies that are often already on the market as finished products, but plagued by a lack of awareness of their existence.

*3.2. Technology Literacy*

Technology literacy is a theme defined by having sufficient knowledge to independently understand the instructions to facilitate use of a technology. Sub-themes point to knowledge barriers, need for resources, and data management.

#### 3.2.1. Knowledge Barriers

A lack of understanding of modern technologies and digital platforms was identified as a barrier resulting in underutilization of technology and dependence on others to operate basic features. Participants mentioned purchasing services (e.g., Netflix) they did not use, because they could not understand how it worked.

*"I know I'm looking for this connection, but I don't know what it's called, I don't know what the things are, and so there is no* . . . *terminology, you know* . . . *um."*

Another participant commented that in order to use her smart phone, she needed guidance from her granddaughter. Moreover, many of the participants left the workforce before technology was integrated into the daily work flows in a significant way, leaving them without the vocabulary or basic skills needed to function in the digital age. One participant described this tech literacy gap as follows:

*"* . . . *I retired 20-something years ago, so I didn't have the opportunity to work with them [technology] at work. So we got less work-based training on them and I don't understand the language of it. Trying to hook a printer up to my laptop—they said to put in the IPP [sic IP] address. Uhh, I can't find it* . . . *you know, and things like that, I don't know what they are talking about."*

#### 3.2.2. Recommendations for Improving Technology Literacy: Need for Resources

The tech literacy problem could be addressed with the "how to" manuals that accompany technology devices; however, the "Getting Started" instructions were described by participants as too technical due to the unfamiliar terminologies that were used. Most people relied on family members to help with setup, but this did not always result in the type of help they needed.

*"Show me, slow down, and it's hard to get 'em to slow down. And you know, and I feel like I'm being a burden or they just don't think Nana is smart enough. Maybe I'm not, but I could try to be if they were a little more patient."*

One participant asked if university students were being trained to help older adults learn to use the technologies. From a technology perspective, user interface is optimal if fewer, rather than more, support personnel are needed. The fact that older adults need assistance in using technologies is indicative of suboptimal designs for this demographic.

#### *3.3. Data Management and Privacy*

The theme of data management and privacy is characterized by three sub-themes: (1) how data can be collected and used; (2) whether knowledge gained is shared in a form that results in value to the participant; and (3) privacy attitudes.

#### 3.3.1. Data Collection and Use

While highly educated, most participants lacked understanding of the granularity of data that can be captured with pervasive sensing technology and the associated analytics used by digital platforms to identify patterns. The mystery of AI, including what it is and how it works, contributed to fears of data loss or being harmed from decisions made from their personal data.

*"If they're [the technology] so sensitive, they know three weeks before we know what's going wrong with our bodies. It seems to me that that kind of information could really be compromised, and seniors could, uh, who are very vulnerable, could really be hoodwinked more easily."*

#### 3.3.2. Return of Value

Moreover, the idea that data could be collected about them without a return of value was problematic. Participants expressed a desire for more instantaneous and understandable feedback, especially when participating in health research. The lack of feedback could potentially hamper enthusiasm for research study participation.

*"You need to talk to your doctor about X, Y and Z. Um, but if you just keep gathering data and nothing happens to that data* . . . *um* . . . *except that you can look at it and* . . . *and you can't really interpret it* . . . *"*

#### 3.3.3. Privacy

There was widespread desire by participants to understand how to use different technologies and how to control personal data. In addition, in order to better understand participants' privacy attitudes, this issue was discussed during the focus group, and participants' attitudes were measured via a survey.

3.3.3. Privacy

3.3.2. Return of Value

*interpret it…"* 

enthusiasm for research study participation.

Westin Privacy Concern Index Westin Privacy Concern Index

For this index, three questions are used to classify a person as low, medium or high with respect to their concern about privacy in the context of trust that business and law will protect their privacy. For this index, three questions are used to classify a person as low, medium or high with respect to their concern about privacy in the context of trust that business and law will protect their privacy.

*Healthcare* **2019**, *7*, x FOR PEER REVIEW 8 of 19

*"You need to talk to your doctor about X, Y and Z. Um, but if you just keep gathering data and nothing happens to that data … um … except that you can look at it and … and you can't really* 

There was widespread desire by participants to understand how to use different technologies and how to control personal data. In addition, in order to better understand participants' privacy

Moreover, the idea that data could be collected about them without a return of value was problematic. Participants expressed a desire for more instantaneous and understandable feedback, especially when participating in health research. The lack of feedback could potentially hamper

A majority of participants (66.7%) reported a medium privacy concern compared with 20% reporting a low concern, and 13.3% reporting a high concern (see Figure 1). A majority of participants (66.7%) reported a medium privacy concern compared with 20% reporting a low concern, and 13.3% reporting a high concern (see Figure 1).

**Figure 1.** Westin Privacy Concern Index Results. **Figure 1.** Westin Privacy Concern Index Results.

Westin Privacy Segmentation Index Westin Privacy Segmentation Index

Between 1979 and 2001, Westin randomly selected U.S. citizens to gauge privacy attitudes across a variety of domains, including health information, consumer and e-commerce [29], and identified three key privacy categories: pragmatists, fundamentalists or unconcerned [24]. Results from Westin's "Privacy On and Off the Internet" survey revealed that 25% of those surveyed were fundamentalists, 55% were pragmatists, and 20% were unconcerned [24,30]. Fundamentalists were described as having a high value for privacy, believing they own their information, and supporting strong laws and regulations to secure privacy rights. Pragmatists were characterized as open to information disclosure if to a trusted entity providing a personal benefit; and unconcerned were described as not having a high need for privacy and control of information [30]. While there has been some criticism of Westin's scale, it is a potentially useful baseline for understanding privacy attitudes. For the purpose of this study, we used this scale to compare our sample with national survey results. Nearly half of our older adult participants (46.7%) were categorized as "privacy pragmatist", compared to 55% from Westin's sample. Only 13% of our older adult sample was considered Between 1979 and 2001, Westin randomly selected U.S. citizens to gauge privacy attitudes across a variety of domains, including health information, consumer and e-commerce [29], and identified three key privacy categories: pragmatists, fundamentalists or unconcerned [24]. Results from Westin's "Privacy On and Off the Internet" survey revealed that 25% of those surveyed were fundamentalists, 55% were pragmatists, and 20% were unconcerned [24,30]. Fundamentalists were described as having a high value for privacy, believing they own their information, and supporting strong laws and regulations to secure privacy rights. Pragmatists were characterized as open to information disclosure if to a trusted entity providing a personal benefit; and unconcerned were described as not having a high need for privacy and control of information [30]. While there has been some criticism of Westin's scale, it is a potentially useful baseline for understanding privacy attitudes. For the purpose of this study, we used this scale to compare our sample with national survey results. Nearly half of our older adult participants (46.7%) were categorized as "privacy pragmatist", compared to 55% from Westin's sample. Only 13% of our older adult sample was considered "privacy unconcerned", with 40% categorized as "privacy fundamentalist", compared with 20% and 25% of Westin's sample, respectively (see Figure 2). When asked about the level of concern regarding threats to personal privacy in America nowadays, a majority of the participants (58.1%) reported being "somewhat concerned", with 29.0% being "very concerned". Compared to national averages, our sample of older adults scored lower in the privacy pragmatic and unconcerned categories and much higher in the privacy fundamentalist category.

Results of the Westin Privacy Concern Index showed that a majority of older adults in our sample had a medium or high privacy concern (80%) with 40% categorized as privacy fundamentalist using the Privacy Segmentation Index. These results indicate that the older adults we sampled are less willing to share information about themselves with others. However, we learned during the focus group discussion that participants were willing to share information if they received something in return, which is more aligned with the privacy fundamentalist classification where people weigh sharing information based on what they get back. For example, with respect to sharing personal information, one participant stated:

*"That's fine, you can take all the data you want, I mean* . . . *but is it gonna be of benefit to me?"*

Another participant liked the idea of getting personalized feedback from artificial intelligence tools as noted here: "privacy unconcerned", with 40% categorized as "privacy fundamentalist", compared with 20% and 25% of Westin's sample, respectively (see Figure 2). When asked about the level of concern regarding

*Healthcare* **2019**, *7*, x FOR PEER REVIEW 9 of 19

*"Well, I think if you can get some sort of readout that is, you know, available from the unit in your apartment, the status of where you are today, to be interactive in a sense, broadcasting the information that is* . . . *is collected about you and be analyzed by the artificial intelligence obviously to give you some kind of status, you know, you* . . . *you're doing okay today or* . . . *or you ate too much yesterday."* threats to personal privacy in America nowadays, a majority of the participants (58.1%) reported being "somewhat concerned", with 29.0% being "very concerned". Compared to national averages, our sample of older adults scored lower in the privacy pragmatic and unconcerned categories and much higher in the privacy fundamentalist category.

**Figure 2.** Westin Privacy Segmentation Index Results. **Figure 2.** Westin Privacy Segmentation Index Results.

Results of the Westin Privacy Concern Index showed that a majority of older adults in our Internet Users' Information Privacy Concerns (IUIPC):

adults had heard or read some information on this topic.

Sensitivity to Personal Information

sample had a medium or high privacy concern (80%) with 40% categorized as privacy fundamentalist using the Privacy Segmentation Index. These results indicate that the older adults we sampled are less willing to share information about themselves with others. However, we learned during the focus group discussion that participants were willing to share information if they received something in return, which is more aligned with the privacy fundamentalist classification where people weigh sharing information based on what they get back. For example, with respect to sharing personal information, one participant stated: *"That's fine, you can take all the data you want, I mean … but is it gonna be of benefit to me?"*  Another participant liked the idea of getting personalized feedback from artificial intelligence tools as noted here: The IUIPC is a 10-item scale with a high internal consistency (Cronbach's alpha = 0.90). The level of internet privacy concerns was high among participants with an average rating of 6.1 out of 7 (SD = 1.3). The *awareness* subscale score was high with an average rate of 6.5 out of 7 (SD = 1.2). This subscale showed that 70% of the older adults were aware of the issue of personal information collection online and strongly agreed that disclosure of information usage was important. Additionally, the *control* subscale was moderate with an average rate of 5.8 out of 7 (SD = 1.6). In fact, a majority (50%) of the older adults strongly agreed with the idea that control is the key issue with consumer privacy. These older adults also reported a moderate level of concern on the *collection* subscale with an average rate of 6.0 out of 7 (SD = 1.6). Specifically, 60% of participants felt offended about the amount of information that is being collected by companies (see Figure 3). *Healthcare* **2019**, *7*, x FOR PEER REVIEW 10 of 19

Additionally, the *control* subscale was moderate with an average rate of 5.8 out of 7 (SD = 1.6). In fact, **Figure 3.** Internet Users' Information Privacy Concerns (IUIPC) Results. **Figure 3.** Internet Users' Information Privacy Concerns (IUIPC) Results.

a majority (50%) of the older adults strongly agreed with the idea that control is the key issue with consumer privacy. These older adults also reported a moderate level of concern on the *collection* subscale with an average rate of 6.0 out of 7 (SD = 1.6). Specifically, 60% of participants felt offended about the amount of information that is being collected by companies (see Figure 3). When asked about the percentage of time older adults falsify their personal information during registration on a website, 86.7% of the older adults reported either never falsifying their personal information or falsifying their information less than 25% of the time. This suggests that older adults When asked about the percentage of time older adults falsify their personal information during registration on a website, 86.7% of the older adults reported either never falsifying their personal information or falsifying their information less than 25% of the time. This suggests that older adults

are less likely to take certain online privacy protection methods to protect themselves. When asked

health records (EHR) with an average sensitivity rating of 7.7 (SD = 3.4) out of 10.

To better understand privacy attitudes, our survey asked the participants to rate the sensitivity of 12 different types of personal information (see Figure 4). The results indicate that participants regarded their bank account information as the most sensitive data type with an average sensitivity rate of 9.7 (SD = 1.1) out of 10 (Very Sensitive). Across all 12 sensitive information types, participants rated present fitness and addictions as having the lowest sensitivity, with an average rating of 6.4 (SD = 3.0 and SD = 4.0, respectively) out of 10 for both. Next to bank account information, smartphone GPS data and internet search history ranked among the most sensitive types of data, both with an average rating of 8.1 (SD = 2.8) out of 10, suggesting that older adults generally consider online information as more sensitive. The highest sensitivity rating on health information was the electronic

**Figure 4.** Sensitivity of Personal Information Analysis.

are less likely to take certain online privacy protection methods to protect themselves. When asked about how frequently they heard about the potential misuse of the information collected from the internet, the mean score was 4.5 (SD = 1.6) out of 7 (Very Much), indicating that most of the older adults had heard or read some information on this topic. are less likely to take certain online privacy protection methods to protect themselves. When asked about how frequently they heard about the potential misuse of the information collected from the internet, the mean score was 4.5 (SD = 1.6) out of 7 (Very Much), indicating that most of the older adults had heard or read some information on this topic.

information or falsifying their information less than 25% of the time. This suggests that older adults

**Figure 3.** Internet Users' Information Privacy Concerns (IUIPC) Results.

*Healthcare* **2019**, *7*, x FOR PEER REVIEW 10 of 19

#### Sensitivity to Personal Information Sensitivity to Personal Information

To better understand privacy attitudes, our survey asked the participants to rate the sensitivity of 12 different types of personal information (see Figure 4). The results indicate that participants regarded their bank account information as the most sensitive data type with an average sensitivity rate of 9.7 (SD = 1.1) out of 10 (Very Sensitive). Across all 12 sensitive information types, participants rated present fitness and addictions as having the lowest sensitivity, with an average rating of 6.4 (SD = 3.0 and SD = 4.0, respectively) out of 10 for both. Next to bank account information, smartphone GPS data and internet search history ranked among the most sensitive types of data, both with an average rating of 8.1 (SD = 2.8) out of 10, suggesting that older adults generally consider online information as more sensitive. The highest sensitivity rating on health information was the electronic health records (EHR) with an average sensitivity rating of 7.7 (SD = 3.4) out of 10. To better understand privacy attitudes, our survey asked the participants to rate the sensitivity of 12 different types of personal information (see Figure 4). The results indicate that participants regarded their bank account information as the most sensitive data type with an average sensitivity rate of 9.7 (SD = 1.1) out of 10 (Very Sensitive). Across all 12 sensitive information types, participants rated present fitness and addictions as having the lowest sensitivity, with an average rating of 6.4 (SD = 3.0 and SD = 4.0, respectively) out of 10 for both. Next to bank account information, smartphone GPS data and internet search history ranked among the most sensitive types of data, both with an average rating of 8.1 (SD = 2.8) out of 10, suggesting that older adults generally consider online information as more sensitive. The highest sensitivity rating on health information was the electronic health records (EHR) with an average sensitivity rating of 7.7 (SD = 3.4) out of 10.

**Figure 4.** Sensitivity of Personal Information Analysis.

#### **Figure 4.** Sensitivity of Personal Information Analysis. *3.4. Co-Design of Technology*

Participants favored the idea of participatory design and were eager to participate in a co-design process. The consensus was that as people who have lived experiences being older, they brought a perspective to the tech development process that might not be present otherwise.

*"* . . . *ethics and morality and seeing further from having lived longer that I think collective wisdom of the elderly might be extremely important in the checks and balances put in place."*

*"I think it's a deal of bioethics. Sometimes the 85 and above have more human knowledge than the people working in the industry."*

Participants also expressed ideas around what an ideal design project might include to better meet their needs.

*"Why can't there be a feature on the TV that I can get the sound to come straight to my hearing aid electronically?"*

*"If they would come up with a universal remote that worked for the TV, for the iPhone* . . . *for everything so that you could program into it that you could remotely operate* . . . *I think that would solve some of the problems we all have."*

Clearly, there is a significant desire to participate and contribute to the ideation and development process, which could lead to technology better designed for adoption by older adults.

#### **4. Discussion**

The growth of general purpose and healthcare-related technologies has created the potential to help more older adults to age in place. Living independently is preferred by older adults and smart technologies like IoT, AI, and AAL can provide the necessary assistance. Due to improvements in communication and remote data gathering capabilities by healthcare providers and researchers, operationalizing smart communities will become more dependent than ever on sensors and predictive analytics of collected data. The results from this study reveal barriers to the adoption of technologies and facilitators that could foster increased access to and usability of technologies to support independent living. Factors identified through this study were: (1) technology usability, (2) technology literacy, (3) data management, (4) privacy attitudes, and (5) co-design.

#### *4.1. Technology Usability*

Older adults in our study tended to associate adoption of new technologies with a lack of confidence in their ability to understand or access them. A significant source of frustration in their interactions with digital products lay in the inadequacies in software and hardware interfaces that permit access to different functionalities. Participants gave examples of technologies they interacted with every day and identified specific examples of problems with their access to these technologies.

The physical decline that can occur as people age creates physical access barriers in technologies. These can be attributed to the dimensions and locations of certain components and how they interface with power sources and other technologies to conduct data transfer or data input. Another noticeable concern in physical access is the existence, or lack thereof, of visibility enhancement features. Visual aids are often inadequate or poorly designed for common use cases that can allow for easier access to content displayed in a visual medium.

It has been suggested that a key motivation in technology adoption by older adults is the presence of a significant perceived benefit [31]. Despite the fact that many participants in the study indicated ownership of a diverse set of modern devices, many of the features that participants wished for in future technologies already existed on the personal devices they already possessed. The lack of knowledge of the existence of these functionalities can vastly diminish the perceived potential of many technologies, affecting adoption or continual usage.

#### *4.2. Software Interface*

The software interface is also a potential source of friction for older adults. Lack of familiarity with and understanding of technology can make it difficult for older adults to be at ease while operating user interfaces. Because older adults were not introduced to modern digital work environments until later in life (or in many cases not at all), their ability to adapt to changes is hindered by a lack of fundamental knowledge in how digital infrastructures operate and how data is utilized.

While the lack of understanding in operating a device could be alleviated somewhat by instructional material documenting steps to access functionalities, older adults tend to rely on static content, such as printed manuals, to fulfill this need. Few participants in the study were capable of effectively utilizing online instructional and troubleshooting materials. In some cases, the barrier was a small font size that was difficult to read, even with prescription glasses. The vocabulary was often unfamiliar (e.g., Bluetooth) and proved meaningless when trying to understand instructions. For many, this meant an increased reliance on assistance from younger and more "tech savvy" family members. While relying on family is a possible solution for those who are fortunate enough to have younger and helpful family members who are more knowledgeable, it is less effective in helping older adults solve future problems.

Modern software and internet platforms have also adopted the model of constant incremental updates and iterations to adapt to user preferences. This has created fluid interfaces that change without warning, quickly rendering previous usability knowledge and documentation obsolete. This unpredictability is especially problematic with the fragmentation in interface philosophies on different digital platforms, necessitating repeated familiarization processes to keep pace with the latest changes. How to make these incremental updates while considering the impact on the digital novices is important if we are to design for needs of an older demographic.

#### *4.3. Data Control and Privacy*

A large majority (87.1%) of the participants indicated that they were concerned with data privacy in their day-to-day usage of technology. This high level of concern with controlling their personal data suggests a hesitancy in adopting a technology or submitting personal data to a digital platform. This mistrust and misunderstanding of the handling of data can be an especially serious obstacle to the adoption of technology that requires large amounts of personal data to be effective, such as machine learning algorithms.

Another concern voiced by the participants was a lack of feedback from data collected by digital devices or researchers. The awareness subscale in IUIPC showed that 70% of the participants were aware of the online information collection and strongly agreed that disclosure of information usage is important. However, the lack of feedback could potentially lead to decreased enthusiasm to participate in studies or an unwillingness to provide personal data. This is particularly an area of interest for digital health research, where the data collected and their analysis can be of high interest to the participants who are concerned about their health. At the same time, it is often unclear how to return the data back to the research participants in a manner that is meaningful to them. In traditional clinical research, research data are rarely returned to the participants. As such, there is a lack of a clear pathway to determine what would be meaningful (e.g., raw data or a short report) nor how often to provide feedback. The answer is likely to be person-specific—for instance, a person who requests and can process complex information versus someone who is satisfied with a very brief summary. In keeping with the growing focus on personalized medicine, there should be a framework for providing personalized data feedback.

#### *4.4. Implications for Creating Age-Friendly Communities*

As the number of older adults increases, the World Health Organization has initiated a movement to establish age-friendly communities [32]. An important component of this initiative should be identifying technologies that support aging in place. Our early stage HCD research sheds light on important issues that are unique to older adults specific to privacy and technology literacy. Engagement of older adults in the design of technologies is often overlooked or an afterthought. Technologies that are commonly used by older adults are often developed without consulting them at the early stage of product conception. This top-down design model means that user input is only received by the product developer after it is completed, making it much harder to alter in order to fit user needs. Our study showed that older adults are experts in their lived experiences and can identify the potential barriers to technology adoption and use.

In this study, participants voiced their concerns about technologies they interacted with daily, albeit with varying levels of success, and offered ideas for how to improve these products. One issue was their lack of understanding of fundamental technology concepts. A common barrier to the participatory design process involving older adults is the lack of expertise in product development and programming [33]. Because of this technology literacy gap, there is significant potential value in providing an educational component in the co-design process to overcome this issue. While impractical to educate older adults on more complicated topics in computer science and human computer interaction, basic knowledge about current technologies and how they interact with each other would be immensely valuable. For instance, one participant commented that the facility personnel spend a

lot of time letting people into their apartments because residents often misplace or forget their keys. An eye scanning or finger print sensor that could be used to unlock the door of the residence, or a system that mimics the proximity-based keyless lock system on modern cars, was suggested by a participant. By gaining a high-level understanding, the resulting ideas and concepts generated by older adults can be more meaningful, particularly in the prototyping stage of the participatory design process, where practical knowledge is needed [34].

In addition to the understandability and usability of different technologies, concerns were raised by participants over the use of data and the importance of privacy and control. This feedback is especially useful when designing technologies for older adults, who may have a very different perception of data and expectations around privacy than younger generations. Many indicated their willingness to provide more sensitive data if it meant getting meaningful feedback on the status of their heath. At the same time, they were also reluctant about sharing data of other categories due to hacking or data loss concerns. This indicates the importance of addressing privacy concerns in different scenarios for different technologies. A participatory design process that values privacy could be a key factor in improving user adoption.

By including residents in this formative research, we were able to identify what would be needed to engage older adults in the design process in a meaningful way and what they would like to receive in the form of feedback. With a better understanding of the technology that they are using, older adults can shape the design philosophy to better serve their needs as users. A next step in this research is to develop a co-design process that incorporates technology education as a component with a goal of increasing "tech literacy." We anticipate this education will facilitate identifying and prioritizing problems that can be addressed with a technological solution that residents help to co-design.

#### *4.5. Limitations*

The results presented here are part of a larger study to determine how AI can be used to assess individual cognitive and physical status through the use of traditional means and sensor technologies. Due to the parameters of the larger study, the sample drawn for this study involved people residing in a CCSHC, which is not a random nor representative sample of older adults. These results are also based on a relatively small sample of 31 participants. However, all participants contributed to the discussion and a data saturation point was reached. Finally, the PSI, PCI and IUIPC scales were developed for testing consumer and internet privacy.

#### **5. Conclusions**

This study demonstrates the significant gap that exists between the potential benefits offered by technologies such as AI and other AAI and the barriers that plague older adults in the adoption of these technologies. Education is critical not only for older adults, but also for technologists. While increasing "technology literacy" of older adults can provide meaningful improvements in helping these users interact more successfully with technology, we also must address the need to educate technology creators about older adults—i.e., increasing "aging literacy" of technologists. This education can occur through pragmatic exercises that involve partnering with older adults to design future technologies. Through co-design partnerships, we can create technologies that are useful and capable of reducing barriers at the design phase. Rather than intervening after a product is in the market place, we can preempt the problems introduced by low technology literacy and fundamentalist privacy attitudes. Moreover, feedback loops can be built in that will help older adults to better understand their data and how these data are used to predict their healthcare needs.

**Author Contributions:** Conceptualization, C.N.; methodology, C.N., S.W.; formal analysis, K.B., J.R., W.M., S.W.; writing—original draft preparation, S.W., C.N., K.B., W.M.; writing—review and editing, D.J., H.-C.K.; supervision, C.N.; project administration, C.N.; funding acquisition, C.N., D.J.

**Funding:** This study was supported by the UC San Diego Health Science Pilot Grant Program (Nebeker, PI) and IBM Research AI through the AI Horizons Network (Jeste, PI). The content is solely the responsibility of the authors and no conflicts of interest are reported.

**Acknowledgments:** We acknowledge the contributions of Carolyn Neuhaus, and Mary Anne Stro, who informed the initial research plan and the retirement community for hosting our focus groups and sharing their insights.

**Conflicts of Interest:** The authors declare no conflict of interest. A representative of the funding agency contributed to the writing of the manuscript; however, did not contribute to the study design; collection, analyses or interpretation of data; or in the decision to publish the results.

#### **Appendix A**

Focus group protocol CO-DESIGN TECHNOLOGY FOR AGING IN PLACE—HEALTHY AGING August 9, 2018

#### *INTRODUCTION*

**Welcome**: Hello everyone! I'd like to thank you for accepting our invitation to participate in this focus group.

**Confidentiality**: Your contributions to our discussion are important and Shengzhi will be taking notes to help us remember what we discuss today. To make sure we don't forget anything, we are audio recording each session. In our transcript of the recording, we will not identify you by name and your responses will be confidential. Only members of our research team will have access to the recording, transcripts and our notes.

**Purpose**: We are conducting this research as part of a UC San Diego Health Sciences project on technology-enabled health research. We've asked you here today to talk about how technology might be used to facilitate living independently. We are also interested in knowing how your think about your health information and privacy. In addition to asking you to respond to questions, I will also ask that you complete a survey which will take about 10 min.

**Consent:** As with any research, you are free to ask questions at any time and your involvement is completely voluntary. If you have any concerns about being recorded or decide you don't want to participate in this focus group, please know that you can leave at any time and there will be no hard feelings. To acknowledge your time and contributions, we will give you a \$30.00 script that can be exchanged for cash at a local bank.

So, to get started, I would like mention some basic instructions for how this works . . . This session will last about 90-min. Near the end of our time we will ask you to complete a survey. After that, we will give you a check that's called "script" that you can take to the bank to exchange for cash. We are interested in your thoughts and opinions—there are no right or wrong answers here. We expect that there will be some areas where most people agree but, there will also be times when there is disagreement. That's not a problem—we want to be respectful of each other's opinions. Differences of opinion are expected so feel free to share yours even if it's different. Please make sure everyone has a chance to chime in and please don't interrupt each other!

#### **A. Curiosity about technology and connection to personal health**

**1. Digital technologies you currently use** There is growing interest in how digital technologies can be used to support healthy living and aging-in-place. To get started, I would like you to think about what "digital technologies" you use in your daily life. Feel free to use the 3 × 5 cards in front of you to jot down any thoughts you have. *<Give 1-2 min to think>* Okay, what are some of the technologies that you're using? *<Write key words identified by participants on butcher paper>* Think back to a time when you felt that you'd like to use a technology like a smart phone app or Facebook and were not quite sure

how to do it? Can you tell us about that? Think back to a time when you felt that using a technology or a smart phone app has made you feel great. Can you tell us about that?


What do you think about the wearable camera? Would you wear it for a day or a week? What concerns would you have? What would motivate you to wear it?

What do you think about the wrist worn devices? Would you wear it for a day or a week? What concerns would you have? What would motivate you to wear it?

What do you think about the waist worn devices? Would you wear it for a day or a week? What concerns would you have? What would motivate you to wear it?

What do you think of the overall design of these devices? Can you get it on/off easily? Is the information that is collected useful to you?

Home Sensors

What do you think about having sensors placed in your apartment that could alert you or a friend about how you're doing?

*Show examples of data produced.*

## **B. Preferences for involvement in participatory design of intelligent assistive technologies**

*Healthcare* **2019**, *7*, x FOR PEER REVIEW 17 of 19

**1. Interest in co-design** Some researchers are trying to make activities such as tracking your health, keeping you safe, taking care of your home, communicating with your family even easier with new technologies and tools. *< Show figure of technologies >*

**Figure A1.** Technology by Domains. **Figure A1.** Technology by Domains.

Now, researchers, and especially engineers and technology developers, might think they know how to best help you, but they also need your input to make sure that they create products that actually fulfill *your* goals and are easy for *you* to use. Would you be interested in having conversations with technology makers to guide the design process of products that are specifically geared toward people in retirement communities? If you could develop technology to improve your life in any way, what would you Now, researchers, and especially engineers and technology developers, might think they know how to best help you, but they also need your input to make sure that they create products that actually fulfill *your* goals and are easy for *you* to use. Would you be interested in having conversations with technology makers to guide the design process of products that are specifically geared toward people in retirement communities? If you could develop technology to improve your life in any way, what would you develop? Do you have an idea of what it would look like? Could you draw a picture of it?

#### develop? Do you have an idea of what it would look like? Could you draw a picture of it? **2. Familiarity with terms, concepts, and processes associated with the design of assistive technologies for aging in place Term 1: Aging in Place**

**2. Familiarity with terms, concepts, and processes associated with the design of assistive technologies for aging in place Term 1: Aging in Place**  What does "aging in place" mean to you? What kinds of technologies do you think could support aging in place? What would "success" mean for the design of aids for aging in place?

#### What does "aging in place" mean to you? **Term 2: Participatory Design**

What kinds of technologies do you think could support aging in place? What would "success" mean for the design of aids for aging in place? **Term 2: Participatory Design**  Participatory design means that stakeholders are involved in the design of new products and devices. What do you think this means in practice? Participatory design means that stakeholders are involved in the design of new products and devices. What do you think this means in practice? What can you contribute? How would you think of your role on the design team? What type of process would you want to feel like you a part of it?

#### What can you contribute? **Term 3: Assistive Technologies/Intelligent Assistive Technologies**

**Term 3: Assistive Technologies/Intelligent Assistive Technologies**

How would you think of your role on the design team? What type of process would you want to feel like you a part of it? *<Show videos or prototypes, such as: https://www.smithsonianmag.com/innovation/how-willartificial-intelligence-help-aging-180962682/>*

**3. Advice and preferences for how technology development experts should most effectively communicate such information to enable an authentic and informed participatory design process** We want to know how we could maximize your involvement in a participatory design process. How much time would you want to spend? Do you want to do it at your home, or do you want to meet elsewhere? How often can you fit this in? Would you be willing to test out devices while they are still in development? Why or why not? That's the end of our session. Thank you for participating!

## **References**


© 2019 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

## *Article* **Living Alone Among Older Adults in Canada and the U.S.**

### **Sharon M. Lee \* and Barry Edmonston**

Department of Sociology and Population Research Group, University of Victoria, Victoria, BC V8W 3P5, Canada; be@uvic.ca

**\*** Correspondence: sml@uvic.ca

Received: 11 March 2019; Accepted: 3 May 2019; Published: 7 May 2019

**Abstract:** Increasing proportions of people, including older adults, live alone. Studying living arrangements of the elderly is important because these affect and reflect general well-being of the elderly and inform communities' response to elderly housing needs. We analyze data from the 2006 Canadian Census and the 2006 American Community Survey to examine living alone among non-married older adults aged 55 and older in Canada and the U.S. The paper has two parts. First, we compare native- and foreign-born elderly to see if immigrants are less likely to live alone. Second, we examine factors associated with living alone among older immigrants. While older immigrants in both countries are less likely to live alone, the large differences are substantially reduced once various explanatory variables are considered. Comparisons of four gender/country groups of older immigrants show the positive role of economic and acculturation factors on living alone among older immigrants. With few exceptions, predictors of living alone are similar for older immigrants in Canada and the U.S.: living alone is mainly explained by a combination of economic and acculturation factors, taking demographic variables into account. Findings underline the need for age-friendly housing with innovative design and technology that can accommodate older people who live alone, including older immigrants who may have different needs and cultural preferences.

**Keywords:** living alone; older adults; older immigrants; Canada; U.S.; older age-friendly housing

#### **1. Introduction**

Studying the living arrangements of older or elderly populations is important for several reasons. First, living arrangements affect and reflect family type and household structure among older people. These are in turn related to social support, inter-generational relations, health status, social isolation, satisfaction with life, and general wellbeing [1–5]. An older person living alone has different family and social relations from another living with a spouse or partner, or co-residing in a multi-generational family with an adult son or daughter and grandchildren, or co-residing with non-relatives.

Second, another reason for studying elderly living arrangements pertain to the idea of "age-friendly communities" that incorporate suitable physical environments, including housing, transportation services, and home modification programs with innovative assistive technology and designs, with a supportive social environment that promotes positive social relations for older residents [2,6]. Older adults may not need as much space as that provided by the usual single-family home, so smaller housing units would be more appropriate. Declines in physical mobility and health may mean that older adults are less able to climb stairs, for example, or bend low or reach high to access shelves, kitchen cabinets, and other storage areas. Housing designs that minimize such potential barriers, and innovative technologies, for example, voice recognition software for operating doors and appliances, can be part of age-friendly communities.

Factors that influence living arrangements of the elderly include preferences and resources that people have, and health and other constraints as they age. Residential options for older people include living alone if not married or partnered; living with spouse or partner only if married or partnered; co-residence with family members or extended family living; co-residence with non-family members; and institutional living, including retirement homes and assisted living facilities. Researchers often refer to the first two types of living arrangements (that is, living alone if not married or partnered, and living with spouse or partner only if married or partnered) as residential independence or independent living arrangements [7–9]. In the following review, the majority of references will be Canadian or U.S.-based, given the paper's focus.

#### *1.1. Rise in Independent Living Arrangements among the Elderly*

There has been a rise in independent living arrangements, as defined above, among older people in many countries, particularly in the west and more developed countries [10–13]. For example, a comprehensive report on living arrangements of older people around the world by the United Nations Population Division [13] shows marked differences between the elderly in more and less developed countries in independent living arrangements: 68 percent in the former and 20 percent in the latter live alone or with a spouse only. Conversely, 27 percent of elderly in more developed countries co-resided with a child or grandchild compared with 75 percent in less developed countries. Similar trends of independent living arrangements among the elderly are observed in Canada and the U.S. [7,8].

Over the past fifty years, there have been absolute and relative increases in the number of Canadian elderly people in independent living arrangements, mainly for married or partnered couples to live with spouse or partner only [1,5,14]. This contrasted with declining proportions in co-residential living arrangements, including living with other family members or with non-relatives. Recent data from the 2011 National Household Survey [15] show that among the population aged 65 and older, the majority (56.4 percent) lived as part of a couple and another 25 percent lived alone. In other words, over 80 percent of the population aged 65 and older were in independent living arrangements. The prevalence of living alone increases after age 50 for women and after age 70 for men, with a sharper increase for women [15].

Living arrangement patterns for the U.S. population aged 65 and older are fairly similar. Data for 2012 show that about 59 percent lived with a spouse or unmarried partner only and another 28.5 percent lived alone. Together, almost 88 percent of the population aged 65 and older were in independent living arrangements [16] (Table 3). The percent of older adults living alone was about 40 percent in 1990, but decreased to around 36 percent in 2000 and 2010 [17] (Table 72). As in Canada, living alone is higher among women, and increases with age, with sharper increases for women. For example, 47 percent of women aged 65 and older lived alone compared with 22 percent of men, in 2010 [17] (Table 72).

#### 1.1.1. The Special Case of "Living Alone"

While living alone is not a new form of living arrangement, Klinenberg [10] describes the increased trend of "going solo" as a new "social experiment" that is fundamentally at odds with much of human history. Using the term "singleton" to refer to a person who lives alone, Klinenberg [10] documents a global increase in singletons, driven by increased economic prosperity and social security, "cult of the individual", greater geographical mobility, greater job mobility, and several "revolutions", specifically, in gender relations (leading to improved economic and social status of women), communications, mass urbanization, and longevity.

The global rise in living alone occurs across the age range, but for this paper, we focus on "aging alone" [10], or the increase in older people living alone, a trend that is more pronounced in developed countries in the west and some parts of Asia [18]. Statistics on elderly living alone among developed countries include 38.4 percent in France, 32.7 percent in England and Wales, 30.8 percent in the U.S., and 22.5 percent in South Korea [18]. More elderly people are also living alone in China [19] and Japan [20]. For example, the 2015 census of Japan reported that the percent living alone among adults 65 and older was 12.5 percent among men and 20 percent among women [20]. Increased longevity is the main demographic reason for the rise in elderly people living alone. As people live longer, the risk of other life-course events increases, such as divorce and widowhood, which changes living arrangements, including a change to living alone [21]. As noted earlier, living alone is more frequent among elderly women [3,10,15,16,18,20] because of the gender gap in longevity and the common pattern of women marrying older men, which increases the risk of widowhood.

Conventional beliefs about elderly people living alone have some truth. Many are widows, and experience poverty, social isolation, poorer physical and mental health, and lower life satisfaction and quality of life [1,3,4,10,21]. Turcotte and Schellenberg [22] report that poverty is highest among female seniors who live alone. However, despite the distinctive challenges of aging alone, many elderly people who live alone express a strong preference for this over other living arrangements, including living with adult offspring and grandchildren, if this option were available, preferring "intimacy at a distance" [10]. Some persons who live alone may be in a stable relationship with a partner who also lives alone. However, these "living apart together" (LAT) couples are more common among young adults. For example, only about 2 percent of people over 60 in Canada are in a LAT relationship [23]. Many elderly people who live alone value their independence and privacy, and would not willingly change their independent living arrangements, and especially fear losing their ability to reside independently [3,4,10].

#### 1.1.2. Older Immigrants

While statistics on living arrangements show an increased trend to independent living arrangements among the older population, several U.S. and Canadian studies show that immigrants, including elderly immigrants, are more likely to live in extended family living arrangements, and by implication, less likely to reside in independent living arrangements, including living alone [8,24–26]. The preference for extended living arrangements among immigrants has been explained by several factors, including economic factors (co-residence as an immigrant economic coping strategy) or cultural and acculturation factors (immigrants from some cultural backgrounds have stronger family values that encourage co-residence and less acculturated immigrants retain traditional customs including those about extended living arrangements).

Still, older immigrants may be exposed to similar demographic forces such as increased longevity, gender gap in longevity, age gap between spouses, divorce and widowhood, as well as changing social norms and values regarding individualism, privacy, and independence, although the influence of these factors may vary between immigrant and native-born elderly. Older immigrants, particularly those who are more acculturated, may prefer independent living arrangements, including living alone if not married or partnered.

#### *1.2. Research Questions and Contributions*

This paper consists of two parts to addresses two research questions. First, are older immigrants less likely than Canadian- or U.S.-born elderly to live alone, once appropriate factors are considered? Second, we conduct additional analysis of older immigrants and examine the main factors associated with living alone among older immigrants in each country. Statistics and previous studies suggest that age and gender, and economic and acculturation factors will be particularly important. We examine similarities and differences in factors related to living alone among older immigrants across the two countries by comparing four groups by country and gender: Canada/female, Canada/male, U.S./female, and U.S./male.

We recognize that population aging has become an important demographic trend in many parts of the world [13,27]. Many countries including European and Asian countries are responding to changing social, health, and housing needs with population aging [18–20]. We chose to compare the U.S. and Canada mainly because of our focus on older immigrants as these two countries have long histories of immigration and have relatively large immigrant populations (more details are provided in Section 1.3).

This paper makes three contributions to existing research on living arrangements among the elderly. First, the focus on living alone highlights this form of living arrangement among older adults, with additional focus on immigrants. Much previous research on living arrangements of older immigrants had examined co-residence or extended living arrangements [24–26]. Living alone as a form of living arrangement has not received much attention. Although we do not directly examine the implications of living alone for older immigrants' well-being, identifying the factors associated with living alone among elderly immigrants furthers our understanding of the demographic, economic, acculturation, and other characteristics of elderly immigrants who live alone.

Second, this paper contributes to research on elderly immigrants in Canada and the U.S., a population which has not received much attention, as noted by Gelfand [28] and Wilmoth [9], compared to extensive research and discussion of the elderly in general. Aging immigrants are a growing part of the aging population in countries such as Canada and the U.S.; for example, 2006 Canadian census data show that more than one-quarter of the population aged 65 and older in Canada are foreign-born [22]. In the U.S., 12.7 percent of the population aged 55 and older are foreign-born [29].

Third, findings from this research provide evidence to support housing and other policy initiatives to respond to growing populations of older adults who may want to live alone but are unable to do so because of lack of appropriate housing designs and types. Most housing units in Canada and the U.S. are single-family units with front and/or backyards that become less appropriate over the lifecycle as children leave and parents age or become widowed. If we can document that more older adults are living alone (with potentially more who would have preferred to live alone but for lack of appropriate housing), public and private sectors can use this information to promote age-friendly housing and communities (for example, smaller housing units with close by amenities such as shops and healthcare facilities). The concept of "environmental gerontology" highlights the need for a multi-disciplinary approach to designing neighborhoods and housing that facilitate "mobility, independence, and quality of life of older people living in the community" [6] (116).

Finally, this study is the first comparative analysis of living alone among older immigrants in Canada and the U.S., offering a useful comparison of two large immigrant-receiving countries that have older immigrants from many different countries of origin. While Canada and the U.S. are distinctive in many ways, both countries have long histories of immigration with relatively high levels of current immigration from diverse source countries. In the following section, we discuss why a comparative study of Canada and the U.S. can be especially productive in advancing understanding of living alone among older immigrants.

#### *1.3. Comparing Canada and the U.S.*

While there are distinctive challenges in cross-country research, including the need for comparable variables and sensitivity to historical and contextual differences, a comparative analysis can advance knowledge in many unique ways. A comparative analysis has the advantage of allowing researchers to conduct similar analyses using different data from the countries selected to identify similarities and differences in how various factors and characteristics affect the outcome being examined. If the influence of particular factors is similar, this increases confidence in the validity and reliability of the findings.

Comparing Canada and the U.S. for this analysis is not simply based on the fact that they are North American neighbors with a long mostly amicable joint history, and that the two countries have been strong allies in modern times. Canada and the U.S. are far from "two peas in a pod" [30], but besides being North American neighbors and close economic trading and foreign policy partners, there are other similarities and differences between Canada and the U.S. that make for a fruitful comparative study of living alone among older adults, including immigrants. There are also differences, for example, Canada's population and economy are much smaller than that of the U.S.: Canada's population is about 33.5 million in 2011 [31] compared with 308.7 million in the U.S. in 2010 [32]. However, Canada and the U.S. share several sociodemographic trends. We describe similarities as well as differences to show the value of such a comparative study of living alone among older adults, including older immigrants.

#### 1.3.1. Role of Immigration

First, immigration has always been a major factor in both countries, historically and today. Canada and the U.S. are among the leading destinations for global migration streams in recent decades [33]. Both countries have long histories of immigration and generally take pride in their immigrant heritage.

However, we should note differences in immigration policies and systems. Canada has a selective points-based immigration system whereby potential immigrants are screened based on such human capital characteristics as age, education, English and/or French language proficiency (Canada's two official languages), and adaptability whereas the U.S. immigration system is primarily based on family reunification.

While the U.S. continues to admit more immigrants than other major immigrant destination countries [34], immigration has a larger influence on Canada's population. The percent foreign-born of Canada's national population stands at 24 percent compared with 13 percent for the U.S. [34]. Immigration has been the main source of Canada's population growth since 1993/1994 [31]. For the year ending June 30, 2012, net international migration accounted for two-thirds of Canada's population growth [31]. Population growth increasingly stems from the contribution of immigration because fertility levels are below replacement. In contrast, immigration accounts for a lower percent of U.S. population growth, at about one-third of U.S. population growth in recent decades [34].

#### 1.3.2. Population Aging

Second, population aging is another demographic trend shared by both countries. The Canadian population is aging, indicated by increased median age of the population from 26.2 in 1971 to 40.0 in 2011 [31]. Elderly immigrants are a growing segment of Canada's aging population, with more than one-quarter of the population aged 65 and older being foreign-born [22].

Similar population aging trends are observed in the U.S. [35]. The median age of the U.S. population has steadily increased, from 30.0 in 1980 to 37.2 in 2010 [17] (Table 7). The percent of the U.S. population aged 55 and older has also increased, from 20.8 percent in 1980 to 24.9 percent in 2010 [17] (Table 7). Of the population aged 55 and older, 12.7 percent are foreign-born [29].

Aging-in-place of younger foreign-born cohorts and the immigration of older immigrants contribute to the growth of the elderly immigrant population in each country. While both the Canadian and U.S. populations are aging, and aging immigrants are part of this demographic trend, statistics cited above show that Canada's population is older and elderly immigrants are a larger proportion of its elderly population.

#### 1.3.3. Racial and Ethnic Diversity

Third, closely related to the role of immigration is the expanded racial, ethnic, and cultural diversity of the Canadian and U.S. populations. Mainly because of immigration from Asia and other non-traditional (that is, non-European or North American) sources in recent decades, Canada's population has evolved from one dominated by the two founding peoples (British and French) and the indigenous (Aboriginal) population to the current situation where over two hundred ethnic origins were reported, and thirteen different ethnic origins had one million or more responses [36].

In 2011, close to 20 percent (19.1 percent) of Canada's population identified as members of "visible minority" groups, that is, racial minority groups other than Aboriginal peoples. The *Employment Equity Act* of Canada defines visible minorities as 'persons, other than Aboriginal persons, who are non-Caucasian in race or non-white in colour'. The visible minority population consists mainly of the following groups: South Asian, Chinese, Black, Filipino, Latin American, Arab, Southeast Asian, Korean, and Japanese [36].

The U.S. population has also been transformed by increased racial and ethnic diversity, also closely related to immigration. The main group, non-Hispanic White, has been slowly declining as a percent of the total population, to 63.7 percent in 2010, down from 69.1 percent in 2000. This means that other racial groups and Hispanics have been increasing in numbers and proportions, and together comprise about 36 percent of the total U.S. population in 2010 [37].

Asians were the fastest growing racial minority between 2000 and 2010, but account for just 5.6 percent of the population in 2010. The most notable change has been the growth of the Hispanic or Latino population to become the largest minority population in the U.S. since the 1990s. Hispanics are now 16.3 percent of the U.S. population compared with 13.6 percent Black and 5.6 percent Asian, the other main racial minority groups [37].

While both Canada and the U.S. have become more racial and ethnically diverse, there are important differences, including the larger share of racial/ethnic minority populations in the U.S. (36 percent), compared with 19 percent of visible minority groups in Canada, and the large presence of Hispanics in the U.S.

#### **2. Data and Methods**

We analyze data from two data sets. Data for Canada are from the Public Use Microdata File (PUMF) on individuals in the 2006 Census of Canada (see [38] for detailed technical and data documentation). These data are a 2.7 percent representative sample of the population enumerated in the census. The microdata sample for individuals is selected using a three-phase sampling plan. The first sampling phase consists of the sample of one-fifth of the population (20% sample data). This is a cluster sample. It consists of all households who completed the long questionnaire in the census. This sample was divided into two parts that are representative of Canada in order to create two sampling frames used to select the microdata samples. The first frame was used to select microdata from the individuals file. The second frame was used to select microdata from the hierarchical file. The third phase consisted in selecting records from the individuals file. The final sample contains 844,476 records, representing 2.7% of the target universe, which is the Canadian population.

U.S. data are from the 2006 American Community Survey (ACS) Public Use Microdata Sample (PUMS) (see [39,40] for detailed information on the ACS and technical documentation for the 2006 ACS PUMS file). During previous decennial censuses up to the 2000 census, most households received a short-form questionnaire, while one household in six received a long form that contained additional questions and provided more detailed socioeconomic information about the population (this was the long-form census). The 2010 census was a reengineered short-form only census, counting all residents living in the United States and asked for name, sex, age, date of birth, race, ethnicity, relationship and housing tenure, taking just minutes to complete. The more detailed socioeconomic information once collected via the long-form questionnaire is now collected by the ACS. The 2006 ACS PUMS consists of 1,277,561 housing unit records (1 percent of all housing units) from which 2,923,336 person records were sampled.

We chose these two data sets because in 2005, the United Nations published a pioneering piece on living arrangements of older people around the world [13] and in the following year, the World Health Organization released its guide to global age-friendly cities in response to global population aging [27]. This motivated us to conduct a comparative analysis of living arrangements of the elderly in Canada and the U.S. (see previous Section 1.3 for why we focus on Canada and the U.S.), using comparable nationally representative data from each country from around the time of the U.N. and WHO publications to provide baseline findings for future research on the subject.

For statistical analysis, we define the study sample as persons 55 years and older. The meaning and definition of "aging" and the "elderly" are increasingly open to question. Researchers studying the "elderly" or the "aged" recognize that using a particular age to define the elderly is arbitrary. We recognize that the "elderly" are a very heterogeneous group and reaching a particular age (be it 55, 60, 65 or 70) does not always imply declining economic or health status. Many statistics on the elderly refer to persons aged 65 and older, or persons aged 55 and older. In this study, we use age 55 in order to show more clearly potential differences that occur between age groups and to cover a wider age range at the "older" ages.

Given the outcome variable—living alone—we exclude persons who are married or living common-law (the latter status is officially recognized in Canada and often treated as equivalent to being married) or co-habiting. We include men and women, who are never married, separated, divorced, or widowed. We exclude older persons living in group quarters. Recent statistics show that for the population aged 65 and older, over 92 percent in Canada live in private households [41], and the comparable figure is 95 percent in the U.S. [42] (Table 35). For our study population of persons aged 55 and older, the percentages would likely be higher.

We identify Canadian-born or U.S.-born elderly and immigrants from responses to the questions on citizenship at birth and place of birth. Persons who are Canadian or U.S. citizens at birth are considered native-born while persons who are not Canadian or U.S. citizens at birth are considered immigrants. This avoids including persons born abroad to Canadian or U.S. citizens as immigrants (based on foreign place of birth) as these persons are not considered immigrants in Canada or the U.S., respectively.

Variables included in the analysis are as directly comparable as possible across the two data sets. We note where it is not possible to develop directly comparable categories for some variables.

#### *2.1. Outcome Variable*

The outcome variable, living alone, is coded as a binary variable (1 = live alone; 0 = don't live alone) based on responses to questions on household type, family structure, and individual family status.

#### *2.2. Explanatory Variables and Expected E*ff*ects*

We include explanatory variables that previous research had shown to influence living arrangements of older adults. Expected results are based on previous research and published statistics.

#### 2.2.1. Demographic Variables

Age is coded in seven age groups, 55–59, 60–64, 65–69, 70–74, 75–79, 80–84, and 85 years or older for descriptive analysis. We use 5-year age groups because 2006 Canadian census public-use microdata are limited to five-year age groups. Age is recoded into a continuous variable in the multivariate analyses. We use the mid-point of each 5-year age category to assign age values for the Canadian sample (for example, persons aged 55–59 were assigned an age of 57.5).

Statistics on living alone from Statistics Canada [41] and the U.S. Census Bureau [42] suggest that living alone increases with age but these statistics do not take into account marital status, health, and other factors that may make it more difficult for older adults to live alone. It is possible that living alone increases with age, but it is equally likely that once additional factors are considered, age may have negative or only modest influence on living alone.

Gender is a binary variable (0 = female; 1 = male). Living alone is expected to be more likely among women because of women's longer longevity. However, this may not be the case once other factors are considered.

Marital status includes four non-married categories: divorced, separated, widowed, and never-married. Living alone may be more likely among widowed adults.

#### 2.2.2. Economic Variables

The influence of economic variables is expected to be positive, as living alone requires adequate income to pay for a housing unit that is inhabited by only one person.

Education is coded in five categories: less than high school, high school graduate, some post-high school education but less than a Bachelor's degree, Bachelor's degree, and post-Bachelor's degree. Education is expected to have a positive influence on living alone.

Individual income is coded in six categories: less than \$10,000; \$10,000–19,999; \$20,000–39,999; \$40,000–59,999; \$60,000–99,999; \$100,000 and over for descriptive analysis. Income and other monetary variables are measured using Canadian dollars in the Canadian sample and U.S. dollars in the U.S. sample. In 2006, the Bank of Canada exchange rate was around US \$1 equals CAD 1.14.

In the multivariate analyses, individual income is a continuous variable. Living alone is expected to increase with income.

Government retirement income is a binary variable (0 = received less than \$100 in government retirement income during the past year or 1 = received \$100 or more in government retirement income during the past year). In Canada, government retirement income refers to benefits from the Canada or Quebec Pension Plan. In the U.S., government retirement income refers to payments and benefits from the Social Security Administration. Having government retirement income increases the likelihood of living alone.

Guaranteed retirement income is a binary variable (0 = received less than \$100 in retirement income from a private or personal pension plan during the past year or 1 = received \$100 or more in retirement income from a private or personal pension plan during the past year). Guaranteed retirement income refers to regular income received from being a member of an employer's pension plan, payments from individual annuities, private pensions paid to widows or widowers, pensions of retired civil servants, and other annuities paid to individuals by a private insurance company. Having guaranteed retirement income is expected to increase the chances of living alone.

Homeownership is a binary variable (0 = does not own home or 1 = own home). Homeownership's influence is expected to be positive on living alone because owning a home implies having sufficient economic resources to own a home. In addition, homeownership facilitates living alone, removing the need to look for alternative housing in the event of marital dissolution or widowhood.

#### 2.2.3. Cultural and Acculturation Variables

Culture and acculturation are closely related but distinct concepts. Ethnic origin, language background, religion, and other characteristics are usually used to indicate cultural background. Given differences across ethnic groups on other characteristics such as marriage, fertility, and family patterns, ethnicity has frequently been used as a sociodemographic variable to indirectly measure these differences. We describe ethnic origin as a demographic variable in the descriptive analysis. For immigrants, acculturation is usually indicated by duration of residence in the host country and proficiency in host country language [24,43,44].

Ethnic origin is coded using fifteen groups. These are "American" (in the U.S. sample) or "Canadian" (in the Canadian sample); British; French; Other European; Arab or Middle Eastern; South Asian/Asian Indian; Chinese; Filipino; Korean; Vietnamese; Other Asian; Latin American/Latino/Hispanic; African, Black, or Caribbean; Other single ethnic origins (including persons reporting Aboriginal only in the Canadian sample and Native American or Alaskan Native or Native Hawaiian/Pacific Islander only in the U.S. sample); and Multiple ethnic origins. Persons reporting ethnic origins that are culturally closer to mainstream American or Canadian culture (that is, American, Canadian, and various European groups) are more likely to live alone.

For immigrants, knowledge of official languages (in Canada) or proficiency in English (in the U.S.) is coded using four categories. Besides being an indirect indicator of acculturation, language knowledge or proficiency implies an ability to communicate and navigate social and other situations and understanding of broader societal norms.

The four categories range from excellent to poor competence in Canada's two official languages (English and French) or in English (for the U.S. sample), although the specific definitions differ for Canada and the U.S. For the Canadian sample, the four categories are (1) English or French mother tongue, and English or French home language; (2) other mother tongue, and English or French home language; (3) other mother tongue, and other home language, knows English or French; and (4) other mother tongue, other home language, and does not know English or French. For the U.S. sample, the four categories are: (1) only speaks English; (2) speaks English very well; (3) speaks English well or not well; and (4) does not speak English. Cultural closeness to the host country and/or acculturation decreases from category 1 to 4 for both the Canadian and U.S. samples, and living alone is expected to decrease from the first to the fourth category of the language variable.

Duration of residence for immigrants measures how many years immigrants have resided in Canada or the U.S. It is coded in five categories for descriptive analysis, from 0–9, 10–19, 20–29, 30–39, and 40 years or more. In the multivariate analyses, duration of residence is a continuous variable. Duration of residence is expected to have a positive influence on living alone as increased duration implies greater acculturation.

#### 2.2.4. Other Control Variables

Place of residence indicates metropolitan and non-metropolitan residence, and residence in several specific Canadian and U.S. cities. Metropolitan categories for this variable include three Canadian and five U.S. cities with the largest number of older immigrants. We include only three Canadian cities because immigrants in Canada are highly concentrated in them: 63.4 percent of Canada's immigrants reside in these three cities [36].

The codes for place of residence are (1) Montreal (Canada) or Chicago (U.S.); (2) Toronto (Canada) or Los Angeles (U.S.); (3) Vancouver (Canada) or Miami (U.S.); (4) New York City; (5) San Francisco; (6) other metropolitan areas; and (7) non-metropolitan areas. Living alone is expected to be higher in non-metropolitan areas because of lower cost of housing which facilitates independent living arrangements, including living alone.

#### *2.3. Methods of Analysis*

We begin with descriptive analyses to describe and compare the study samples. For multivariate analyses, we use Stata 12 statistical software [45] to estimate several logistic regression models because the outcome variable is coded as a binary variable. For the first research question on nativity differences in living alone, we estimate two regression models (Models I and II). Model I is estimated separately for Canada and the U.S., for all non-married elderly, aged 55 and older. Each equation includes dummy variables for nativity and gender, and other explanatory variables described above (except for duration of residence because it is collinear with nativity). Second, we estimate a logistic regression of living alone for all non-married elderly, aged 55 and older (Model II), for four groups: females in Canada, males in Canada, females in U.S., and males in U.S. Each equation includes a dummy variable for nativity, and explanatory variables described above (except for duration of residence because it is collinear with nativity, and gender). Results from Models 1 and 2 address our first research question.

For the second research question on predictors of living alone among older immigrants, we limit analysis to older immigrants only, and estimate a logistic regression model of living alone (Model III) for four groups: females in Canada, males in Canada, females in U.S., and males in U.S., to identify and compare predictors of living alone among older immigrants. Model III includes duration of residence, in addition to other explanatory variables.

For interpreting the logistic regression results, we calculate predicted probabilities for each explanatory variable using the margins command in Stata 12 [45]. The predicted probabilities provide a useful interpretation of the net effect of each categorical variable on living alone, evaluated by holding constant the effects of all other variables in the model [46]. Multiplying predicted probabilities by 100 converts them into percentages or proportions, which facilitates presentation and discussion of results. We include the logistic regression results from which the predicted probabilities are calculated in the Supplementary Tables.

#### **3. Results**

#### *3.1. Descriptive Results*

Selected characteristics of the Canadian and U.S. study samples are shown in Table 1. Non-married immigrant elderly in both Canada and the U.S. are much less likely to live alone than native-born elderly. In Canada, 54.8 percent of non-married older immigrants live alone, compared with 70.7 percent of Canadian-born non-married elderly (a difference of 15.9 percent). In the U.S., 51.7 percent of non-married older immigrants live alone, compared with 73.2 percent of U.S.-born non-married elderly (a difference of 21.5 percent). The gap is larger in the U.S.


**Table 1.** Descriptive statistics for non-married elderly, Canada and U.S. (in percents).


**Table 1.** *Cont.*

<sup>a</sup> For Canada, based on responses to the ethnic origin question. For the U.S., based on responses to the ancestry question. Two responses are allowed in the U.S. question while multiple responses are allowed in the Canadian question. <sup>b</sup> 'Canadian' ethnic origin in Canada, 'American' ancestry for the U.S. The latter is recorded if 'American' is the only response. <sup>c</sup> Includes persons reporting single Aboriginal origin in Canada, and single Native American or Native Alaskan or Native Hawaiian/Other Pacific Islander origin in the U.S. <sup>d</sup> The categories are not directly comparable. For Canada, (1) English or French mother tongue or home language; (2) other mother tongue, English or French home language; (3) other mother tongue and home language, knows English or French; (4) other mother tongue and home language, does not know English or French. For the U.S., (1) speaks English only; (2) speaks English very well; (3) speaks English well or not well; (4) does not speak English.

#### 3.1.1. Demographic Characteristics

Gender: There are more females in both samples, about 69 percent. There are also more women in the immigrant samples, at around 73–74 percent in both the Canadian and U.S. samples.

Age: The distribution across age categories is as expected, with higher percents in the younger age categories. The U.S. immigrant sample has higher percents in the younger age categories.

Marital Status: Marital status refers to non-married categories only. Being widowed is the most common marital status for both Canadian and U.S. samples, with a higher percent widowed among immigrants in Canada. The percent divorced is higher among the native-born in both samples.

Ethnic Origin: For describing the sample, ethnic origin is considered a demographic characteristic. Most native-born Canadian and U.S. elderly report European or multiple origins but immigrants are distributed over a wider range of ethnic origins compared with the native-born. There are two striking differences between the Canadian and U.S. immigrant samples. First, older immigrants in Canada have higher percentages of people reporting European origins (72.8 percent—this percent includes the 25 percent reporting multiple origins), compared with 33.8 percent (including the 5.9 percent reporting multiple origins) in the U.S. sample. Studies of ethnic origin in Canada show that persons reporting multiple origins are mainly reporting "Canadian" in combination with other European origins, and persons reporting "Canadian" used to report European origins, particularly French or British [47–49]. Second, almost one-third (32.1 percent) of older immigrants in the U.S. report Latin American/Hispanic/Latino origin while no single ethnic group dominates the immigrant sample in Canada (the largest three are British at 13.6 percent, Chinese at 8.7 percent, and South Asian at 6.2 percent).

#### 3.1.2. Economic Characteristics

Education: On average, older immigrants in Canada have more years of schooling compared to Canadian-born elderly (a mean of 12.1 years versus 11.5 years) whereas in the U.S., older immigrants have fewer years of schooling, with a mean of 10.3 years versus 12.3 years for the U.S.-born. The distribution across levels of educational attainment of Canadian-born and immigrant elderly is generally quite similar but higher percents of immigrants in the U.S. sample are in the lower educational categories.

Income: In both samples, older immigrants have lower mean incomes, and the percent of immigrants in the two lowest income categories exceeds that of the native-born in both countries. Homeownership: Homeownership is higher for the U.S. sample (70 percent are homeowners compared with 62 percent for the Canadian sample). However, immigrants in Canada are more likely to own their homes (67 percent, compared with 60 percent for Canadian-born elderly), while immigrants in the U.S. sample are less likely to be homeowners (62 percent, compared with 71 percent of U.S.-born elderly).

#### 3.1.3. Other Characteristics

Metropolitan Residence: Notably higher percentages of elderly immigrants in both samples reside in metropolitan areas. The metropolitan concentration of older immigrants in the U.S. is higher, at 94 percent, compared with 86 percent in Canada. U.S.-born elderly are also more likely to reside in metropolitan areas (73 percent) compared with Canadian-born elderly (at 59 percent).

#### 3.1.4. Immigrant-Specific Characteristics

There are two immigrant-specific characteristics in Table 1: duration of residence in the host country and host country language proficiency.

Duration of Residence: On average, older immigrants in Canada have resided in Canada for 37.7 years compared with 34.8 years for older immigrants in the U.S. Higher percentages of elderly immigrants in the Canadian sample have resided in Canada for 40 or more years (43.8 percent) compared with 36.4 percent of immigrants in the U.S. sample. More immigrants in the U.S. sample are recent arrivals: 10.7 percent have been in the U.S. for less than 10 years, compared with 5.5 percent of immigrants in the Canadian sample.

Language Proficiency: As noted earlier in the section describing variables and in Table 1, categories of the language proficiency variable are not directly comparable between the two samples. However, there is a similar pattern for interpreting its effects, that is, acculturation (indirectly indicated by language proficiency/knowledge) decreases from category 1 to category 4. About two-thirds of the Canadian sample are in the first two categories and would be considered highly acculturated but 14.6 percent are in the fourth category (considered the least acculturated). 44.7 percent of the U.S. sample are in the first two categories while 19 percent are in the fourth category.

#### *3.2. Logistic Regression Results*

Logistic regression results from Models I and II focus on the role of nativity on living alone. We begin with results from Model I, which was estimated separately for Canada and the U.S. Complete logistic regression results for Model I are in Supplementary Tables S1A and S1B. Table 2 compares observed (or descriptive) and adjusted results by nativity. Given the large amount of statistical results, we do not show predicted probabilities for all the explanatory variables included. The complete tables of predicted probabilities are available upon request. We highlight differences by nativity, as this is the main focus for this part of the analysis.

While the role of nativity is statistically significant (foreign-born older adults are less likely to live alone), the observed large gaps in living alone between native- and foreign-born elderly are substantially reduced once other factors in the equation are taken into account. Specifically, the observed difference of 15.9 percent between Canadian-born and immigrant elderly living alone is reduced to 3.4 percent, and the observed difference of 21.5 percent between U.S.-born and immigrant elderly is reduced to 1.6 percent. This shows that differences in living alone between older native-born and immigrants are modest, once all other factors in Model I are considered.


**Table 2.** Model I: observed and predicted (adjusted) percents, living alone, by nativity, non-married elderly, Canada and the U.S. <sup>a</sup> .

<sup>a</sup> Model I was estimated separately for the Canadian and U.S. samples. It includes a dummy variable for nativity. Adjusted or predicted percentages control for age, gender, marital status, individual income, government pension, private retirement income, homeownership, education, ethnic origin, language proficiency, and place of residence. Duration of residence for immigrants was not included because it is collinear with nativity. Predicted probabilities were multiplied by 100 to show the predicted (adjusted) percent living alone.

Results for Model II, estimated for four groups, Canada/female, Canada/male, U.S./female, and U.S./male, are shown in Table 3. Again, we highlight the role of nativity. Complete results from the logistic regressions for Model II are shown in Supplementary Tables S2A–D. While the influence of nativity remains statistically significant (except in the logistic regression for the U.S. female sample—Supplementary Table S2C), fairly large observed differences in living alone between nativeand foreign-born females and males in Canada and the U.S. are greatly reduced.

**Table 3.** Model II: observed and predicted (adjusted) percentages living alone, gender and nativity comparisons, Canada and U.S. <sup>a</sup> .


<sup>a</sup> Model II was estimated for all non-married elderly, for each of these four groups: Canada/female, Canada/male, U.S./female, U.S./male. It includes a dummy variable for nativity, and all the other explanatory variables included in Model I.

Model III is estimated for older immigrants only and addresses the second research question: what factors are associated with living alone among older immigrants? Complete results from estimating Model III for Canada/female, Canada/male, U.S./female, and U.S./male are shown in Supplementary Tables S3A–D. Table 4 shows predicted probabilities for categorical variables from estimating Model III for the four gender/country groups and results for three continuous variables—age, individual income, and duration of residence—are shown in Figures 1–3 (in the figures, predicted probabilities have been converted to proportions to facilitate presentation and description).

**Table 4.** Model III: predicted probabilities of living alone for categorical explanatory variables, non-married older immigrants in Canada and U.S. <sup>a</sup> .


<sup>a</sup> Model III was estimated for four groups of older immigrants: Canada/female, Canada/male, U.S./female, and U.S./male. <sup>b</sup> Language proficiency in official language(s) is not comparable for Canada and the United States. See text and notes for Table 1 for description of how this variable is coded. <sup>c</sup> The first city listed is for Canada and the second city is for the U.S.

0.9 1

3.2.2. Economic Factors

*Healthcare* **2019**, *7*, x 16 of 22

*Healthcare* **2019**, *7*, x 16 of 22

**Figure 1.** Predicted proportions living alone by age: Canada and U.S., female and male older immigrants. **Figure 1.** Predicted proportions living alone by age: Canada and U.S., female and male older immigrants. Individual income: Individual income results are shown in Figure 2. alone among older immigrants. However, ethnic group differences in living alone do generally support cultural expectations:

0 0 10 20 30 40 50 60 70 80 90 100 Income (in \$1,000s) **Figure 2.** Predicted proportions living alone by individual income: Canada and U.S., female and male older immigrants. **Figure 2.** Predicted proportions living alone by individual income: Canada and U.S., female and male older immigrants. The increase is sharper among female immigrants in both Canada and the U.S., and immigrants in Canada (both females and males) have higher proportions living alone compared to their U.S. counterparts at all values of duration of residence.

**Figure 3.** Predicted Proportions Living Alone by Duration of Residence: Canada and U.S., Female and Male Older Immigrants. **Figure 3.** Predicted Proportions Living Alone by Duration of Residence: Canada and U.S., Female and Male Older Immigrants.

#### 3.2.1. Demographic Characteristics

Age: Age is an important factor in living arrangements among older adults because of age-related health declines which can be expected to influence the ability to live alone [7,11,18]. The results for age are shown in Figure 1. Age differences are statistically significant for female immigrants in both Canada and the U.S., but not for males. Living alone increases with age for older immigrants in Canada, with sharper increases for females. The influence of age is negative for both sexes in the U.S., is more pronounced for females, and as noted, not statistically significant for males.

Marital Status: The results for marital status are shown in Panel A, Table 4. We highlight the role of marital status because of conventional views that most elderly who live alone are female widows. Logistic regression results show that the influence of marital status were not statistically significant for most groups and differences in predicted probabilities of living alone are relatively modest. Around 54–57 percent of female immigrants in Canada live alone across different marital status categories. Among male immigrants in Canada, widowed males have the lowest proportion living alone while separated males have the highest proportion. Differences by marital status are larger than those for females but are still modest.

Among female immigrants in the U.S., those who are separated are least likely to live alone while divorced female immigrants are most likely to live alone. Among males, widowed males have the lowest proportion living alone, while divorced and never-married males have the highest proportions living alone. In both samples, divorced older immigrants have higher proportions living alone.

#### 3.2.2. Economic Factors

Individual income: Individual income results are shown in Figure 2.

As expected, the proportions living alone increase with income, with the sharpest increase observed for females in the U.S. The influence of other economic factors are shown in Panel B, Table 4, and are all also in the expected positive direction.

Government and private retirement income: Older immigrants who have government or private retirement income are more likely to live alone. These results hold across all groups, but the impacts are larger for males in both Canada and the U.S.

Home ownership: Older immigrants who are homeowners are more likely to live alone, a pattern observed for all four gender/country groups. Older male immigrants in Canada who are homeowners have the highest proportion living alone.

Educational Attainment: The proportions living alone increase for all four groups as educational attainment increases. The proportions living alone are higher among males in both countries at each level of educational attainment.

#### 3.2.3. Acculturation Factors

The results for ethnic origin and language proficiency are shown in Panel C, Table 4.

Ethnic Origin: The influence of ethnic origin is not statistically significant for almost all ethnic origins. We interpret the influence of ethnic origin as a largely cultural and acculturation variable, but acknowledge that ethnic origin relates to other characteristics that can also influence living alone, such as ethnic group differences in marriage and fertility patterns, which relate to family size and availability of family for elderly co-residence. The lack of statistically significant results for all but a few ethnic origins suggests that this may not be an adequate proxy for characteristics that affect living alone among older immigrants.

However, ethnic group differences in living alone do generally support cultural expectations: older immigrants of European backgrounds are culturally closer to "mainstream" Canadian and U.S. culture, and higher proportions of these groups live alone. In contrast, lower proportions of older immigrants reporting Asian, Latin American, and other non-European origins live alone. Older female immigrants of all ethnic origins in both Canada and the U.S. are less likely to live alone than male

co-ethnics (there are two exceptions to this pattern: Korean and Latin American female immigrants in Canada have higher proportions who live alone compared to co-ethnic males).

Language Proficiency: This variable indicates high to low linguistic and related acculturation (from Category 1 to Category 4) and its influence is as expected. The proportions living alone decrease from Category 1 to Category 4 for both samples and for both genders. For example, 60 percent of older female immigrants in Canada in Category 1 live alone, compared with 43 percent of female immigrants coded Category 4. The difference by linguistic acculturation is larger among males: 72 percent of male immigrants in Canada classified in Category 1 live alone versus 53 percent of males classified in Category 4.

Duration of Residence: Acculturation is also indicated by duration of residence, increasing as years of residence increase. As expected, living alone among older immigrants increases with duration of residence, shown in Figure 3.

The increase is sharper among female immigrants in both Canada and the U.S., and immigrants in Canada (both females and males) have higher proportions living alone compared to their U.S. counterparts at all values of duration of residence.

#### 3.2.4. Other Controls

Place of Residence: Older immigrants who live in non-metropolitan areas are more likely to live alone, a pattern that is similar for all four sub-groups. Lower proportions of older immigrants in Canada who reside in one of Canada's three largest immigrant destination cities (Montreal, Toronto, and Vancouver) live alone, compared to immigrants who live in other metropolitan and non-metropolitan areas. For older immigrants in the U.S., a similar pattern holds, except for those who reside in Miami where the proportion living alone (69 percent) is quite close to the percent living alone in non-metropolitan areas.

#### **4. Discussion**

We return to our two research questions in this section. The first question focuses on the role of nativity and asks whether non-married older immigrants are less likely than Canadian- or U.S.-born non-married elderly to live alone. Descriptive results show that older immigrants in Canada and the U.S. are much less likely to live alone than native-born elderly. The difference is larger in the U.S. This finding is consistent with other research showing lower rates of independent living arrangements, including living alone, among immigrants, including older immigrants [9,24,43].

However, once appropriate factors are taken into account, nativity differences, while still statistically significant, are substantially reduced. Differences by gender and nativity are also reduced or become modest once various factors are considered. These findings suggest that aggregate differences in living arrangements between older immigrants and native-born elderly are largely due to differences in demographic, economic, and acculturation characteristics between the older native-born and immigrant populations. Findings show that notable proportions of older adults, including immigrants, in both countries live alone, reinforcing the need to consider these groups when discussing age-friendly communities.

Our second research question is directed at older immigrants and examines predictors of living alone among non-married older immigrants in Canada and the U.S. The main findings show higher levels of living alone for older male immigrants in both Canada and the U.S., across different characteristics, including age, marital status, income, education, and duration of residence. With some exceptions, the proportion living alone is higher among immigrants in Canada across different characteristics.

Factors influencing living alone are generally similar for older immigrants in Canada and the U.S., suggesting that living alone among older immigrants is mainly explained by a combination of economic and acculturation factors, after taking demographic variables into account. More acculturated older immigrants, and immigrants with more economic resources, are more likely to live alone, findings

that are consistent with previous studies on extended living arrangements among older immigrants: the predictors of living alone are opposite to those for extended living arrangements where less acculturated older immigrants with fewer economic resources are more likely to co-reside [26,50].

Aggregate statistics on older people living alone contribute to widespread beliefs and images that older women are more likely to live alone. It is of course correct that higher proportions of older women live alone, as shown in statistics from many countries [13,18,20], as well as Canada and the U.S. However, when appropriate demographic, economic, and acculturation factors are taken into account, this study of older immigrants shows that male older immigrants are more likely to live alone. Therefore, being male is a stronger predictor of living alone among older immigrants, once additional appropriate factors are considered. Perhaps the inclusion of several key factors in this analysis such as acculturation measures (host language proficiency and duration of residence) and economic resources (indicated by not just individual income but access to private and government pensions) permitted a more comprehensive examination of the role of gender on living alone among older immigrants. Other factors such as stronger male preference for living alone and greater social acceptability of males living alone could also be implicated. Different research using different data with information on availability and type of kin, social networks and relationships, and gender differences in preference for and acceptability of living alone would be useful to further explore these findings.

Another widespread image of elderly people who live alone is that of elderly widows living alone. Again, this is not entirely wrong, given women's longer longevity and the common age gap between spouses. However, once appropriate factors are taken into account, older immigrants who are widowed are not the most likely group to live alone, compared to other marital status groups. It is possible that widowed older immigrants have adult children with whom they can co-reside following widowhood, an option that may be unavailable to divorced, separated, and never-married older immigrants. Divorced and separated older immigrants are more likely to live alone than the widowed, and in the U.S., older male immigrants who are never-married are as likely as divorced males to live alone. Marital disruptions due to divorce is therefore a better predictor of living alone among older immigrants than widowhood. Marital disruptions can also be associated with other forms of disruptions such as moving away, which also disrupts previous family and social networks, thereby increasing the chances of living alone. Unfortunately, we are unable to examine the role of geographical mobility as well as other factors such as gendered differences in cultural norms about living alone and subjective preference for living alone with the data examined in this paper.

This paper contributes to the literature on living alone and housing in two ways. First, we show that notable percentages of older adults, including older immigrants, in two large countries with aging populations, live alone. This trend is expected to continue and reinforces the need for more private and public policies to design and build age-friendly communities that allow older adults to continue to live independently and participate fully in their community. Second, the findings show that once appropriate factors are taken into account, there are only modest differences between native-born and immigrant elderly's likelihood of living alone. This suggests that elderly immigrants should be included in housing initiatives that include more units geared towards elderly living alone, instead of mistakenly assuming, based on aggregate statistics, that elderly immigrants are somehow different, and are less likely to live alone and do not need to be included in these efforts.

Many countries have already implemented initiatives on age-friendly communities in response to population aging and the rise in independent living arrangements among the elderly, including living alone [6,18,20]. Such initiatives would need to include community services such as home care services, senior community centers, transport services, housing designs such as greater availability of smaller housing units, for example, one or two-bedroom single-level apartments, and other factors such as support for innovative technology that may make it easier for non-married older adults to continue to live alone at older ages.

Additionally, older immigrants may have different cultural preferences in housing design and use of technology. For example, Chinese immigrants may place great importance on the role of feng shui in housing alignment and design, immigrants may be less familiar with advanced technology, and some immigrants' accented speech may pose challenges for voice-recognition software. These potential differences would have to be considered in elderly housing designs and use of technology in planning age-friendly housing.

While this study has produced some new and useful findings, we note several limitations. First, this is a cross-sectional analysis, and findings refer only to the period when the data were collected in 2006. We do not know if the living arrangement recorded at time of data collection is temporary or permanent, and the findings cannot speak to trends in factors related to living alone.

Second, the sampling frame for both data sets are private households and individuals and families in private households. This misses the population in group or institutional living quarters, an important limitation for studying the elderly. As health declines accelerate with increased age, the oldest old are less likely to live alone in private households and more likely to be in group housing such as retirement or assisted living housing. This limitation may be implicated in the finding of age's negative influence on living alone among U.S. female older immigrants.

Third, the outcome, living alone, poses some conceptual challenges. Living alone is one type of living arrangement, and living arrangements are inherently dynamic and may be recursive. This means that an individual can transit through different types of living arrangements over her/his life (for example, living at home with parents → living alone as a young adult → living with spouse upon marriage → living with spouse and children → living alone upon divorce → remarriage, living with new spouse → widowed, living alone). In this example, living alone occurs at different stages over the life course, and has different determinants and implications. The study of living arrangements has therefore to be particularly sensitive to age, gender, and life course influences, including marital status.

Fourth, there are measurement challenges for studying living alone as a form of living arrangement. The data examined in this study do not tell us whether the person living alone is in a relationship with another person (the "living apart together" couples noted earlier). It is likely that the predictors and implications of living alone for such individuals would differ in important ways from others who live alone and are not in a relationship.

Finally, while the census and ACS data used are appropriate for identifying and comparing sociodemographic, economic, and acculturation factors on living alone among older immigrants, there is no information on other factors that influence older immigrants' living arrangements, including the key role of health status (the ACS includes a question on disability but there is no comparable information in the Canadian census). Other unmeasured factors include availability of family or friends to share housing, and community characteristics that either facilitate or discourage living alone (for example, community support for innovative housing designs and technology, and availability and affordability of housing units for older singletons).

#### **5. Conclusions**

We began our analysis by making no assumptions about whether living alone is the "best" living arrangement for non-married older adults. The increased social trend to elderly residential independence suggests that most elderly prefer independent living arrangements [10], but we recognize that for some older adults, particularly immigrants, co-residence may actually be preferable and more advantageous, and lowers the risk of social isolation [1,4,9].

However, as we reflect on our findings, a picture emerged suggesting that living alone is associated with characteristics that can only be described as advantageous. Results from examining older immigrants show that those who live alone have higher income and education and are more acculturated. These characteristics may be related to other dimensions of wellbeing, such as more extensive social ties and support because of being more acculturated, and better health, given the well-known socioeconomic status-health gradient [51,52]. Still, we cannot conclude that living alone is the optimal living arrangement for all non-married older immigrants, given study limitations noted above. However, this comparative research provides a reference point for additional research on living alone among older adults, including older immigrants, in other aging societies, particularly where aging immigrants are part of the aging population. The findings also provide useful information for planning and designing age-friendly communities to include older adults who live alone.

**Supplementary Materials:** The following are available online at http://www.mdpi.com/2227-9032/7/2/68/s1, Table S1A–B: Logistic Regression Analysis Predicting Living Alone for Elderly Adults in Canada and the United States, 2006, Table S2A–D: Logistic Regression Analysis Predicting Living Alone for Elderly Adults in Canada and the United States, by Nativity and Gender, 2006, Table S3A–D: Logistic Regression Analysis Predicting Living Alone for Elderly Immigrants in Canada and the United States, 2006.

**Author Contributions:** Conceptualization, S.M.L. and B.E.; methodology, S.M.L. and B.E.; formal analysis, S.M.L. and B.E.; writing original draft preparation, S.M.L.; writing review and editing, S.M.L. and B.E.

**Funding:** This research received no external funding.

**Acknowledgments:** Revised version of a paper presented at the 2014 Annual Meeting of the Population Association of America and the 2015 Annual Meeting of the British Society for Population Studies. The authors thank the two reviewers whose comments and suggestions helped to improve the paper.

**Conflicts of Interest:** The authors declare no conflicts of interest.

#### **References**


© 2019 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

*Article*
