Choosing between staying at home or moving: A systematic review of factors influencing housing decisions among frail older adults

Background Most older adults wish to stay at home during their late life years, but physical disabilities and cognitive impairment may force them to face a housing decision. However, they lack relevant information to make informed value-based housing decisions. Consequently, we sought to identify the sets of factors influencing the housing decision-making of older adults. Methods We performed a systematic literature search for studies evaluating any factors influencing the housing decisions among older adults over 65 years old without cognitive disabilities. Primary research from any study design reported after 1990 in a peer-reviewed journal, a book chapter or an evaluated doctoral thesis and written in English, French or Spanish were eligible. We extracted the main study characteristics, the participant characteristics and any factors reported as associated with the housing decision. We conducted a qualitative thematic analysis from the perspective of the meaning and experience of home. Results The search resulted in 660 titles (after duplicate removal) from which 86 studies were kept for analysis. One study out of five reported exclusively on frail older adults (n = 17) and two on adults over 75 years old. Overall, a total of 88 factors were identified, of which 71 seem to have an influence on the housing decision-making of older adults, although the influence of 19 of them remains uncertain due to discrepancies between research methodologies. No conclusion was made regarding 12 additional factors due to lack of evidence. Conclusion A wealth of factors were found to influence housing decisions among older adults. However, very few of them have been studied extensively. Our results highlight the importance of interdisciplinary teamwork to study the influence of a broader range of factors as a whole. These results will help older adults make the best possible housing decision based on their unique situation and values.


Results
The search resulted in 660 titles (after duplicate removal) from which 86 studies were kept for analysis. One study out of five reported exclusively on frail older adults (n = 17) and two on adults over 75 years old. Overall, a total of 88 factors were identified, of which 71 seem to have an influence on the housing decision-making of older adults, although the influence of 19 of them remains uncertain due to discrepancies between research methodologies. No conclusion was made regarding 12 additional factors due to lack of evidence. PLOS

Introduction
The proportion of older adults around the world is increasing dramatically. By 2050, the number of adults aged 60 years and over will nearly double and countries in Europe and North America will see the proportion of older adults increase by 30% [1]. In Canada, estimates from 2015 show that for the first time, there were more older adults in the country than children under 15 [2] and the number of adults aged 65 and older will represent more than 25% of the population by 2036, with 32% of them being over 80 years old [3]. As their age increases, older adults inevitably experience a progressive loss of their capacities and autonomy. Indeed, 57% of Canadians aged 85 and older report functional limitations, compared to 12% of adults between the age of 65 and 74 [4]. In addition, 30% of Canadians aged 75 years and over are receiving home care services [5]. Functional limitations among older adults will increase the pressure on healthcare services, social services and on the provision of housing as aging Canadians consider whether to stay at home or relocate.
Choosing between staying at home and moving to a more supportive environment is a complex decision for older adults facing autonomy loss. Most of them would prefer to stay in their home [6,7]. Indeed, 90% of Canadians aged 65 years old and over still live in their homes. Two-thirds of them live in a private house [8] and over 70% have not moved in the past five years [9]. The residential mobility rate of older Canadians even decreases with age: adults of 85 years or older living in private dwellings are 30% less likely to have moved than adults aged 65-69 years old [10]. However, half of adults aged 85 years or older living at home rely on caregivers or on home care services to help them perform their daily activities [5]. When this help becomes insufficient, frail older adults are likely to consider their housing options: either stay in their home and adapt it to their needs, or move to an already adapted dwelling, with or without additional care.
Over the years, studies investigating housing decisions have used several conceptual models. Three main theory families on living arrangements in old age have received more extensive empirical testing than the others [11]: migration theory [12][13][14], environmental press theories [15] and health behaviors theories [16]. However, none of these theories considered the residential experiences of older adults as well as the social and emotional meanings attached to these experiences. The role these factors play in older adults' decision-making about housing options is therefore still largely unspecified [11]. A framework proposed by Després and Lord [17] encompasses these lesser explored factors, looking at older adults' housing decisions through a new lens, the meta-concept of home. They suggest six main dimensions that best account for the meanings and experiences of home (Table 1).
To date, many factors have been taken into account in research on older adults' housing decisions, including health and social factors. However, not all potential factors have been identified, especially those related to the built environment and what it represents for older adults. Thus, our objectives were: 1) to identify all the factors that influence decision-making about housing options among older adults with loss of autonomy; 2) to classify them according

Eligibility criteria
All or some participants in the included studies had to be aged over 65 years and we excluded those with cognitive disabilities. If the age range of participants was not specified, we included studies in which participants' mean age was over 65 years or participants who were recruited in housing designed for older adults, with or without additional care. We included studies of any kind of intervention aimed at reporting or measuring factors influencing the housing decision. We included studies both with and without comparison groups. Study outcomes could be any objective or subjective measures of factors influencing housing decisions as reported by experts or as self-reported by participants. There was no restriction on study design. We included all articles in peer-reviewed journals, book chapters in books with editorial committees or doctoral theses with thesis committees. We excluded studies published before 1990 because the important developments in environmental gerontology and around the meta-concept of home occurred after that date [18].

Study selection
Two of the authors (NR and RD) combined search results and independently checked for duplicates. A pre-test screening using a Kappa k calculation was performed on 60 randomly selected titles and abstracts to check concordance between the two authors. The coefficient of Kappa k was 0.8691, corresponding to "excellent" agreement between the authors. This pretest allowed the authors to discuss the abstracts they disagreed about and to adjust their screening accordingly. Then they individually evaluated the remaining titles and abstracts and discussed in person all studies for which inclusion and exclusion criteria were not clear from the title or abstract. Any remaining disagreements were resolved through discussion with a third author (CD). Full-text copies of all studies that might be relevant and had not been excluded through screening were retrieved. All full-texts were reviewed by the authors (NR and RD) and again discussed to check agreement that they met the pre-established inclusion and exclusion criteria.

Data collection process
Two authors (NR and RD) extracted data independently from eligible studies using a data extraction sheet. General characteristics (e.g. publication year, country of study, authors' discipline of study retrieved from their curriculum vitae), study characteristics (e.g. study objectives, study design, data collection, nature of reported issues), participants characteristics (minimum age included, mean age, sample size, autonomy level, type of dwelling and neighborhood, tenure status) were extracted, as well as factors reported as associated with the decision to relocate or not, whether the factors were identified as statistically significant or not in quantitative study designs, or narratively reported in qualitative study designs. The extraction grid was inspired by the framework proposed by Després and Lord [17] to which was added a fifth dimension to include the socioeconomic and health-related factors of influence on the experience and meaning of home. After discussion with team members, the space-time dimension and the temporal dimension were also combined into one time and space-time related dimension due to their similarities, the psychological dimension was extended to include psychosocial factors, and the material dimension became the built and natural environment dimension. The authors subsequently classified all factors influencing housing decisions, as extracted from the studies, into the resulting six dimensions of the new framework. Each author (NR and RD) reviewed the other's extraction and resolved doubts or disagreements. Any remaining disagreements were adjudicated by CD.

Quality appraisal
The authors (NR and RD) appraised the quality of studies using the Mixed Methods Appraisal Tool (MMAT) [19]. The MMAT is a validated checklist for appraising the quality of quantitative, qualitative, and mixed methods-studies included in systematic reviews [20]. For quantitative randomized controlled studies, we assessed randomization, allocation concealment or blinding, completeness of outcome data, and withdrawal/drop-out rates. For quantitative nonrandomized studies, we assessed selection bias, appropriateness of measurements, comparability of groups, completeness of outcome data and response or follow-up rates. For quantitative descriptive studies, we assessed sampling strategies, sample representativeness, appropriateness of measurements and response rates. In qualitative studies, we assessed the relevance of the data sources, the relevance of the analysis process, context consideration and consideration of researcher influence. In mixed-methods studies, we assessed the quality of both qualitative and quantitative components.
After discussing their appraisals, the two authors (NR and RD) resolved any remaining doubts or disagreements through discussion with a third author (CD). Missing information was sought either by searching the website of the research project (if available) or contacting the authors.

Synthesis of results
Given the high level of methodological heterogeneity across studies, the authors conducted a qualitative synthesis of the studies. They also compared the results according to their study design (qualitative, quantitative and mixed method). The factors were classified by the level of agreement between studies that found an effect on the housing decision. Factors studied by fewer than three quantitative studies or fewer than five studies of any design method were treated as exploratory factors. The level of agreement between studies was therefore not calculated.

The quality of the studies
Following the quality appraisal of all studies, 59 scored as of high quality [10, 29- [22,23,[25][26][27][28], and two quantitative sub-studies within a single research project had different quality evaluations [21,24]. No lowquality studies were identified. All 86 retrieved studies were thus considered for analysis.

Synthesis of results
Factors influencing the housing decision. A total of 88 potential factors of influence on older adults' housing decisions were extracted from the 86 studies. Of these 88 potential factors, 78% were individually addressed in less than one quarter of the studies and 42% in less than one out of ten. Our previous study [111] reported on a total of 55 influential factors linked to the meaning and experience of home. By adding a dimension to the initial model associated with the socioeconomic and health-related factors, as well as refining the extraction, this paper brings the total number of factors up to 88. The effect of each of these factors of influence on the housing decision was also assessed. Table 3 reports the number of studies reporting on each of the 88 factors found to influence the housing decision of older adults, push and pull factors combined. The reported effect of each factor is recorded globally but also according to the study design (quantitative, qualitative, mixed).
Among all 88 potential factors of influence investigated, having a mortgage or reverse mortgage was found to have an effect on older adults' housing decisions in 33% of the assessed studies, gender in 37% of them, education, employment and traffic and car facilities in about half of them. For 12 additional factors, evidence was insufficient to discuss any trend since they were addressed in fewer than three quantitative studies or fewer than five studies of any design method. The remaining 71 factors were found to have an overall effect on older adults' housing Factors influencing housing decisions among frail older adults: A systematic review Factors influencing housing decisions among frail older adults: A systematic review decisions in at least 60% of the studies in which they were considered, although 19 of them show discrepancies between quantitative, qualitative and mixed methods study designs.
Of the 71 factors found to have an overall effect on the housing decision, 21 of them had an overall level of agreement among studies ranging from 90% to 100%. Thirty-two additional factors were found to influence older adults' housing decisions, with levels of agreement of 75% to 89% between studies, and the remaining 18 factors identified as having an effect on the housing choice had degrees of agreement ranging from 60% to 74%.
Discarding the 19 factors with discrepancies between study methods, 52 factors of influence identified remain as having an effect on the older adults' housing decision in at least 60% of the studies assessing them. Those factors cover all six dimensions of the meaning and experience of home. Eleven were related to the built or natural environment of the dwelling, nine Number of independent studies addressing each factor are not mutually exclusive. Ã n = number of independent studies. ! Factors for which a discrepancy was identified between studies with different methods. https://doi.org/10.1371/journal.pone.0189266.t003 Factors influencing housing decisions among frail older adults: A systematic review were of a psychological or psychosocial nature, nine to do with the economic dimension, nine were of a social nature, and nine to do with time-and space-time related dimensions of the experience of home. Five factors were related to socioeconomic and health-related dimension. Interestingly, most of the factors showing an effect belonged to the dimensions of the meanings and experience of home that were not related to older adults' socioeconomic profile or health status, while these were the dimensions most commonly considered by the studies under review. The effect of thirty-one factors remains unclear, either because the types of research design in which it was identified did not concluded of the same results (effect identified or no effect identified) or because it has not been sufficiently studied using any design. Surprisingly, mortgage and reverse mortgage, as well as gender seemed to be the only factors with little effect, according to our review, but this needs to be confirmed by further studies, given the large number of female older adults and the known difference between women and men such as their respective roles in caregiving [112].

Discussion
This systematic literature review provides an overview of the factors influencing housing decisions among older adults experiencing loss of autonomy. It shows the multiplicity of considerations involved in older adults' housing decisions and demonstrates the strength of our theoretical framework for organizing a diversity of interdisciplinary scientific evidence. Our results lead us to make several observations. First, our results are interesting in light of the Canadian experience where the housing decisions of adults experiencing loss of autonomy are described in the research as decisions about "location of care" [113]. Professional teams helping older adults make housing choices are mostly composed of health and social service professionals (occupational therapists, nurses, physicians, social workers), and their perspective is therefore one that focuses on where the person will receive care. Our results suggest that older adults address their housing desires and the care they need to remain independent as long as possible as one and the same question. The location-of-care perspective may have led to consideration of a reduced group of factors, focusing more on the socioeconomic and health profile of older adults and on their social supports than on what else they care about in a home. The broad reach of this literature review has brought to light a diversity of other factors, suggesting that the complexity of this decision and its multidimensional nature is still underestimated [114]. Interestingly, the same perspective question came up in our decision to use the term "staying at home" as opposed to "aging in place". The literature is not clear on the distinction between the two, as they are usually used synonymously. However, we chose to use "staying at home" because in general it reflects the perspective and preferences of the older adult himself/herself to remain in its current dwelling, while "aging in place" is a term that reflects the professional, bureaucratic or policy perspectives on the efforts to keep older people out of institutions, which could involve a move to another independent housing or not (similar to the perspective difference between the terms "housing options" vs. "location of care"). This subject of terminology choices and how they impact research would benefit from further study.
Second, to extract and analyze factors influencing the housing decision, we used the Després and Lord (2005) theoretical framework based on the meta-concept of home, designed to analyze the experiences of dwelling and neighborhood as well as the social and emotional needs of older adults [17]. This gave our analysis a new perspective and complemented the frameworks more commonly used for this purpose [11]. The diversity of factors identified showed that none of these frameworks by itself was adequate for understanding the factors that influence housing decisions. We thus created a new framework, adding a "socioeconomic profile and health" category to the Després and Lord framework for factors such as health status and age. This modified framework will allow for a fuller appreciation of the multiple dimensions of the housing decision and provide a tool for building bridges between various research domains [114]. It will also guide the updating of existing decision guides intended for older adults [115] to include consideration of the meaning and experience of home.
Third, this literature review shows a lack of diversity in studies addressing factors influencing housing decisions in old age in terms of the academic disciplines involved. More than half of the reviewed studies were written by only one author or by authors in the same research domain, while only a third benefited from experts in at least two research domains, with health sciences and social sciences being the most common combination. Even though almost 25% of the significant factors of influence were associated with the built and natural environment, less than 8% of the researchers involved in all studies were trained as geographers, planners, architects or designers. Built environment experts need to be more involved in research addressing older adults' housing needs to contribute their knowledge about these important factors in housing decisions and provide a more complete and accurate picture of what is involved. This also highlights the importance of training researchers in architecture and urban planning [116]. A more transdisciplinary perspective is clearly needed [117][118][119] to inform policy and have a real impact on the quality of life of the frail elderly. However, this type of research is still rather rare and hard to finance [120].
Fourth, studies using quantitative methods focused mostly on economic, socioeconomic and health-related factors. However, quantitative methods may have a limited capacity to grasp people's feelings, emotions and values, as well as their daily routines and social networks. Qualitative methods are more likely to be used to assess the social, psychological/psychosocial and time/space-time dimensions, as these factors are more subjective and more complex to assess using quantitative methods. Indeed, most such factors are closely linked to the meaning of home, which is the subjective heart of the housing decision. However, the effects on housing decisions of both emotional attachment to one's dwelling and the number of years lived in the present dwelling/neighborhood remains unclear according to the results of this review, as strong quantitative studies found no effect of these factors, while strong qualitative and mixedmethod studies agreed they had an important effect. While most economic and socioeconomic/health-related factors are more easily assessed with quantitative methods, studies investigating factors in the other four dimensions (psychological/psychosocial, social, time and space-time -related, built and natural environment) could also greatly benefit from more quantitative and mixed-method approaches to complement their qualitative results.
Last but not least, the effects of specific population characteristics on the housing decision, as well as several other factors identified as influential, are understudied. Very old and frail older adults were surprisingly little studied, even though we know that these are the people who suffer most autonomy loss and are most at risk of moving into long-term care [4]. Indeed, only five studies focused on very old adults and 20 specifically on frail older adults. This could be due to the difficulty of investigating this population where dementia, cognitive disorders and severe autonomy losses could limit their participation compared to younger or less frail older adults. Moreover, it may be more difficult to distinguish the very frail from the overall population of older adults, as few studies have attempted to assess frailty using validated instrument or scales and no clear definition has yet emerged in the literature [121]. Another important understudied characteristic in association with their housing decisions is the tenure status of very old adults. Only four of the 30 reviewed studies that recorded the tenure status of older adults compared the influence of being a renter or an owner on the housing decision. Yet owners and renters have been shown to have different residential mobility patterns [122,123]. Older adults with a renter profile might move more often, and this may decrease their attachment to home, which in turn appears to be an influential factor in housing decisions. Some factors identified through this review also lack supporting scientific evidence, such as having a pet, and the experience of falls. For instance, pet ownership has been shown to have an important influence on the health of older adults in other contexts but has been mostly ignored in the context of housing decisions. The experience of falls has also been investigated in other contexts as it is a leading cause of injury-related hospitalization among older adults [124] and is the cause of most hip fractures [125], but its impact remains almost unexplored in the housing decision context. This may be because older adults seriously injured by a fall are often directly discharged into a long-term care facility [124] without having had the opportunity to participate in a proper decision-making process.

Limitations
Our search strategy had some limitations. First, it mostly targeted databases of English-language publications. Search strategies in other languages such as French and Spanish may have found more local publications which could also have been relevant. The strategy also oriented the search results towards literature on relocation and less on staying at home. In the future, the search term "aging at home" and its synonyms could provide a broader understanding of the decision to stay at home.
Second, we did not perform all screening steps in duplicate which could have introduced a selection bias during the screening stage. However, the kappa k calculated during the pre-test suggested an excellent agreement between the two authors.
Third, the results of this review also suffer from an ethnocentric bias, as most of the studies reviewed come from Anglo-Saxon majority and higher-income countries. The proportion of these populations aged 60 years or older was greater than 20% in 2015 and is projected to be higher than 25% by 2050 [1]. This may not be surprising, as in Asian, African or Central and South American countries, for cultural reasons, families tend to keep their older relatives at home with them. Adding the perspective of other cultural approaches to housing in old age could be enlightening and suggest new housing solutions for older adults.

Conclusion
This systematic literature review reveals the diversity of factors influencing the housing decisions of older adults. It confirms that these decisions are complex and multidimensional, and that health, safety and functional autonomy are only a few of the factors that should be considered to understand what is at stake in this type of decision and to better support older people. Important influences relate to the built environment, as well as to the social, psychological, psychosocial, spatiotemporal and decisional contexts of older adults. Several gaps in the literature were identified, mainly regarding the housing decisions of very old adults, frail older adults and the different factors that affect renters and owners.
This review also highlights the fact that this field of research is still in its infancy in terms of embracing the transdisciplinary complexity of meeting an increased demand for care and services while taking into account the importance of feeling-at-home for older adults. That said, it is surprising, albeit worrisome, that with all policies and research funding on aging put forward in the last 20 years or so, a review of scientific evidence published since 1990 on this topic has identified so few that explore decision-making about housing options, and even fewer that have identified a comprehensive collection of relevant factors. Our analysis underlines the different directions taken by each discipline and the consequences of their different methodological approaches. It brings to light the importance not only of engaging all the concerned disciplines in this field of research but of putting together multisectoral teams with complementary methodological perspectives and developing collaborative methodological approaches. Knowledge exchange is also needed so that each discipline is aware of knowledge emerging in the others. The proposed framework presented herein is a first step to bridge-building between different disciplines interested in housing decisions among older adults. Our results will guide the future development of decision guides to support healthcare professionals, older adults and their caregivers in making housing decisions.
This review emphasizes the importance of adapting dwellings and communities to older adults wishing to stay at home in the residential environment that they know and value. It also pushes us to reconsider how we design alternative housing for frail older adults. In addition to safety considerations, alternative housing should also integrate meaning-of-home considerations that could help older adults adapt to their new dwelling and rebuild their feeling of being-at-home. We hope that our results will also provide housing and healthcare professionals, policy makers, housing authorities, relocation counsellors, real estate agents and developers with the evidence they need to adopt a holistic approach in addressing the needs of older adults, not only in making housing decisions but also in providing them with alternative housing that is suitable for them.