University of Birmingham A game of two halves? Understanding the process and outcomes of English care home closures: qualitative and quantitative perspectives

With care services increasingly delivered via a market there is always a risk that care homes could fail financially or struggle in terms of quality, ultimately having to close. When this happens, the received wisdom is that subsequent relocation can be detrimental to the health and well ‐ being of older residents (possibly even culminating in increased mortality). However, there is very little formal evidence in the United Kingdom (UK) or beyond to guide policymakers and local leaders when undertaking such sensitive work. Against this background, this article reports findings from an independent evaluation of what is believed to be the largest care home closure program in the UK (and possibly beyond). This consisted of qualitative interviews with older people, families, care staff, and social work assessors during the closure process in one case study care home and one linked day center, as well as self ‐ reported health and quality of life data for older people from 13 homes/linked day centers at initial assessment, 28 days after moving and at 12 ‐ month follow up. The study is significant in presenting public data about such a contested topic from such a large ‐ scale closure process, in its focus on both process and outcomes, in its mixed ‐ methods approach, and in its engagement with older people, families, and care staff alongside the use of more formal outcome measures. Despite significant distress part ‐ way through the process, - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - the article suggests that outcomes either stayed the same or improved for most of our sample up to a year after moving to new services. Care homes closures may thus be a “ tale of two halves ” , with inevitable distress during the closure but, if done well, with scope for improved outcomes for some people in the longer term. These findings are crucial for current policy and practice given that the risk of major closures seems to be growing and given that there is virtually no prior research on which to base local or national closure processes. While some of this research is specific to England, the underlying issue of care home closures and lessons learned around good practice will also apply to other countries.

developments are specific to England, there remains a risk that care homes could be forced to close in other systems internationally, whether through market failure in countries where this is a feature of care provision, or for practical reasons such as a care scandal or a disaster such as a flood/fire.

| The evidence
Despite the importance of these issues, there is little formal evidence about what happens when care homes close (Le Mesurier & Littlechild, 2007;Robinson, Glasby, & Allen, 2013;Scourfield, 2004). Holder and Jolley's (2012) international review of forced relocation between nursing homes identified a possible 3,500 initial results , but no studies about the outcomes of care home closures in the UK, and only one previous UK research team exploring closure process (Williams & Netten, 2005;Williams, Netten, & Ware, 2007). Worldwide, the review found only 10 studies: seven from the United States and one each from Canada, Australia, and Sweden. While all included quantitative measures of the impact of relocation, the outcome measures used were typically clinical in nature (e.g., mortality), rather than more socially orientated outcomes that older people say they value. Moreover, there were very few, if any, meaningful attempts to engage older people or care staff, with Holder and Jolley (2012) explicitly calling for greater understanding of these perspectives and describing a very limited example of such engagement from Sweden as "illuminating and thought provoking" (Holder & Jolley, 2012, p. 313). Even here, the study was only able to combine quantitative outcomes data with a small number of resident interviews, with no apparent data on the experience of families or care staff. The move concerned was also only temporary as a result of a rebuilding project and staff/residents moved as a single unit, making this potentially very different to a situation where a home closes and residents are dispersed to other services.
Prior to this, Smith and Crome's (2000) review of 40 years of the resettlement literature 2 identified only a handful of UK studies in the 1980s/1990s-all relating to the closure of long-stay geriatric/psycho-geriatric hospitals, where people with long-term needs may have been living for many years, and where often large, outmoded services were giving way to more community-based care. This seems very different to current debates about care home closures, where people may only have been residents for a relatively short period of time and may well be relocating to similar services. The only UK study identified which specifically explores care home closures appears to be a fourpage article from the early 1990s, where 59 residents moved from one large care home and from one psychiatric service initially opened to rehouse long-stay hospital patients (Hallewell, Morris, & Jolley, 1994). This reports quantitative data collected from care staff (on behalf of older people) and the experiences of general practitioners, but does not seem to have engaged directly with older people/their families or to have explored the experiences of care staff or social workers. Overall, Smith and Crome (2000) argue that a key methodological limitation is the tendency to rely on small sample sizes, and most of the broader literature seems to focus on quantitative outcome measures (with no mention of approaches which seek the views of older people themselves).
One of the probable reasons for this is that researching such topics can be so complex and sensitive. With emergency closures, localities are understandably focused on finding alternative services in a crisis. With planned closures, there can be significant public, media, and legal controversy, making it difficult to publicize lessons learned. As a result, there is almost no formal evidence about what happens when care homes close, with local services having to approach potential closures on an ad hoc basis (Scourfield, 2004).
When care homes close, the received wisdom is that older people's health suffers, with a risk of increased mortality. As but one example, a recent tabloid article cited quotes by key stakeholders (Fifield, 2017), "If people have to move abruptly from the place they know as home it will obviously cause huge distress-and could also seriously undermine their health" (Age UK), and "We know people die when providers fail and residents have to be removed" (Local Government Information Unit). However, others suggest that poorly conducted closures may cause significant damage, but that a well-managed process might achieve different results (see Holder & Jolley, 2012 for further discussion).

| Aims
Against this background, this article sets out results from what appears to be the only in-depth UK study of the process and outcome of care home closures. Following longstanding debates, Birmingham City Council (BCC) embarked on the modernization of older people's services across the city in the late 2000s. This involved the closure of all council care homes for older people and any day centers physically co-located on the same site (referred to hereafter as "linked day centers"), reassessing all service users and finding alternative services. Given BCC is believed to be the largest council in Europe and had a relatively high proportion of in-house services, this was felt locally to be the most substantial closure process in the UK and quite possibly in Europe, involving 29 care homes and six linked day centers. As part of this process, a series of new "special care centers" were opened, providing specialist placements for people with dementia and short-term rehabilitation for people coming out of hospital. Additional capacity was also developed in extra care sheltered housing and independent sector residential care.
Aware of the complexity and sensitivity of these changes, BCC commissioned an independent three-year evaluation in order to understand the potentially different perspectives of older people, their families, care staff, and social workers on the closure process, and to understand the impact on older people's quality of life. Crucially, the closure process was organized in two phases, with this evaluation taking place during phase one. This was intended to allow BCC to make changes if the research recommended improvements to the process and/or if negative outcomes were observed. In this way, the research was both summative (understanding the impact of the closures on older residents) and formative (with scope to change the methodology used during the closures in phase two if negative results were identified in phase one). Theoretically, it is best characterized as "action research" in its orientation, since-although conducted by independent researchers-the aim for BCC was to make changes to future processes if these were found to be needed.
The two main research questions were: 1. How did older people, families, care staff, and social work assessors 3 experience the closure process? 2. What impact did closures have on the health and quality of life of older people?
As discussed above, this study is significant for a number of reasons. First, it is the only study of its kind in the UK since 2000 (Holder & Jolley, 2012), quite possibly the only study of its kind in the UK in over 50 years (Smith & Crome, 2000), and probably only one of a handful of studies in the last 15 years internationally. Second, this was quite possibly the largest care home closure process in Europe and thus unique in scale and significance. Third, it presents public findings around a topic that is usually kept private at local level due to the legal and media sensitivities involved. To our knowledge, having such direct access to older people, families, care staff, and social workers involved in the closure process is unprecedented in the UK and probably beyond. Fourth, the current study focuses on both the process and the outcomes of closures. Lastly, it includes a unique combination of qualitative research with key stakeholders, an internationally recognized/validated quantitative measure of health and well-being, and survey data based on research into what older people say they want from care services. At a time of austerity when additional care homes may close in the future, this is important learning (particularly if media warnings of the risk of increased mortality are to be believed).

| METHODS
This study used a mixed methods approach to gain an insight into the closure process from the perspective of key stakeholders, as well as capturing service user outcomes before, during, and after closure. Ethical approval was granted by both the National Health Service (NHS) research ethics system and the BCC internal ethics process.

| The closure process
We used semi-structured interviews to ask older people, families, care staff, and social work assessors how they felt about the closures; the impact of these; information and support provided; overall views on the process adopted; and future areas for improvement. For this part of the research we based ourselves in one case study care home and one linked day center, making sure that these services were not first to close but still early on in the process. This ensured that any early "teething problems" were ironed out, whilst nonetheless conducting the research early enough to suggest improvements for subsequent closures if appropriate. While this might reduce the generalizability of findings, locating ourselves in two specific services enabled us to get to know care staff and residents informally, so that we could minimize the potential disruption and anxiety of "strangers" asking additional questions at a time when people might already be very nervous. We also chose two services which had a broad mix of service users from different ethnic backgrounds, and made sure that these were not "outliers" in any way; for example, in terms of high/low rates of staff turnover, disciplinary cases, or industrial unrest. Although this does not guarantee that our findings would be replicated elsewhere, we had no reason for believing that this care home and day center were anything other than fairly typical for existing BCC provision (see below for further discussion).
To ensure that interviews were conducted as sensitively as possible, we made sure that one of the research team was a qualified social worker specializing in working with older people. Interviews took place as participants were preparing for services to close, but did not yet have a new service identified. This was designed to capture people's perceptions at what was arguably the most stressful and uncertain point in the whole process (effectively a worst case scenario).
Preliminary analysis of all interview transcripts took the form of "conventional content analysis". As Hsieh andShannon (2005, p. 1279) explain: This type of design is usually appropriate when existing theory or research literature on a phenomenon is limited. Researchers avoid using preconceived categories … instead allowing the categories and names for categories to flow from the data. Researchers immerse themselves in the data to allow new insights to emerge … also described as inductive category development.
Thus, our approach focused on indexing data that felt the most significant and relevant to the research questions.
Working without a specific coding frame or conceptual framework at this stage maximized the extent to which the findings were data-derived, as opposed to theory-driven. As we sought to establish the real impact of changes on people involved and discover practical lessons for managing this, it seemed appropriate to analyze in a way that remained as close as possible to the accounts of participants. Following an axial coding approach, indexed data were then organized into themes with shared central characteristics, constantly checking back over the data to refine emerging themes and to check that these continued to represent the indexed data (Miles & Huberman, 1994). This was undertaken by two members of the research team who worked independently and then met to agree emerging themes. These were then reviewed by all three members of the research team, before agreeing the four axial themes presented below with quotations drawn from the indexed data. These quotes have been selected both to illustrate the data underpinning the four themes, and to display the range of data across care settings and types of participant.
Where quotes are presented, they are accompanied by an identifier based around the participant (SU: service user, CS: care staff, F: family member, SWA: social work assessor) and service setting (DC: day care, CH: care home). Thus, SU06DC is the sixth service user from a day care setting.
Once overarching themes were established, a "summative analysis" (Hsieh & Shannon, 2005) was undertaken to interrogate how the different themes were linked to the participant types (i.e., older people using services, their relatives, care staff, and social work assessors). This additional comparison activity aimed to discover whether the participant sub-groups made a distinctive contribution to specific themes. Questions of whether certain themes were more prominent or expressed differently by different groups are explored below.

| Longer-term outcomes
Quantitative methods were used to explore the impact of closures on the health and quality of life of older people in the first 13 care homes and linked day centers to close (a total of 314 potential participants). Older people being resettled were asked to complete two questionnaires at three points in time: at baseline when they were first assessed by the BCC Assessment and Resettlement Team; at 28-day review; and at 12-month follow up (a review approximately 12 months after moving to new services). The questionnaires included: • The EQ-5D, an internationally validated measure of self-reported health and quality of life for older people (see e.g., Kaambwa et al., 2015). This study was interested in exploring how participants valued their quality of life following closures. Due to responder burden we decided that it was not appropriate to ask participants to undertake the additional valuation exercise that allows the individual EQ-5D profile to be converted into a summary index. Instead, we used a conventional mapping exercise that allows each individual EQ-5D profile to be assigned to existing population-based values (see Dolan, Gudex, Kind, & Williams, 1995;EuroQol, n.d.;NICE, 2013). Using population values also allows us to understand how the population would value living in the health states described by the participants in our study. This is an important consideration given services are publically funded.
• A second questionnaire based on a national review of the literature on what older people say they value about care services (Glendinning et al., 2006). This was developed into a questionnaire by the current research team, taking each outcome from the review and using a 5-point Likert scale for each. This comprised 15 questions covering outcomes involving change; maintenance/prevention; and service processes. At 12-month follow-up, participants were asked an additional question to explore whether they felt that the new services they had received since the closure of their care home/day center had helped them to have a better life (or not). This element of the research was not based on pre-existing surveys (the EQ-5D outlined above was felt to be sufficient for this purpose), but instead sought to make sure that at least part of the research focused directly on the factors that older people using care services say they value. Given the tendency of previous research to measure medicallyorientated outcomes, this inclusion of a more user-focused approach was a key element of the design.
Both questionnaires were completed by the older people being resettled, but were given out and collated/ returned to the researchers by the Assessment and Resettlement Team (social work assessors) who were working with older service users as part of the closure process. In doing so, assessors were asked to tick one additional box to say whether the person had a history or a diagnosis of mental ill health (such as dementia or depression), which was felt to be the best and most sensitive way of collating some background information on people's mental health (always recognizing that there tend to be significant unmet needs and that the true level of mental ill health may be much higher than recognized by social work assessors).
Questionnaires focused on participants' overall quality of life and not on specific services or issues such as the role of the resettlement team, thus reducing any potential conflict of interest. Building data collection into the standard assessment and review process also enabled us to involve a much broader group of older people in the research than might otherwise have been possible, and also minimized additional disruption for residents at a potentially traumatic time. While older people could complete the questionnaires themselves, this meant that anyone who wanted practical help could be supported by professionally qualified workers, and that this process could be tailored to individual needs, including for people who did not speak English as a first language or people with a cognitive impairment. We also ran training sessions with the Assessment and Resettlement Team exploring any practical or ethical issues that collecting such data might raise. An experienced senior practitioner was available to assessors if they were experiencing any dilemmas, and the research team members were contactable by email, phone and face-to-face if there were any practical queries. Where care staff or assessors felt that an older person lacked the mental capacity to consent to taking part, they identified a "consultee" 4 -often a family member or carer who knew the person well-who was able to comment on whether the person would wish to take part and answer questions on their behalf (in the same way that a social worker would involve a family member in his or her assessment of the older person's need). Surveys were coded and entered in an Access database, where they were analyzed using Excel and SPSS.

| Participants
For our qualitative interviews around the closure process, we interviewed 49 people (18 older people, 11 family members, six care staff, and 14 social work assessors). While the assessors worked across both day care and care homes, there was a relative dominance of participants from day centers (in terms of older people and care staff), which may well reflect particularly strong feelings about day care discussed below.
For insights into longer-term outcomes, we summarize data from the two questionnaires administered at "initial assessment" (IA), "28-day review" (28DR) and "12-month follow-up" (12MFU). Of the 314 potential participants, 131 people (42%) suggested they wanted to participate in this aspect of the study; of these, 108 people completed both questionnaires at IA. This figure reduced to 93 at 28DR and 74 at 12MFU. This article reports data for the 74 individuals who completed all three stages (31 care homes (CH), 43 day centers (DC)). Table 1 provides a summary of our approach/participants, while 3.2 | The closure process

| Opposition to the initial decision to close services
There was significant opposition to the closures (particularly within day care) from older people and care staff, with social workers more positive about the changes taking place. Many older people and care staff found it difficult to move beyond their shock at the initial decision to close services. Few were interested in the closure process, and were much more concerned about the initial decision, which was perceived to have occurred without consultation: "[They should] listen to us first and not just drop it on you like they did." (SU04DC) Perhaps the strongest feelings came from people accessing day care, with many feeling that the main focus was on care homes, with linked day centers having to close primarily as an "after-thought" (SWA09). For older people and care staff alike, day centers were important in facilitating meaningful relationships and tackling social isolation; providing greater continuity of care, with the potential for better outcomes and a more preventative approach; and offering a break for family carers. This was poignantly voiced by one day care user: "[The day center] keeps me alive. They're [other older people and care staff] like brothers and sisters." Overall, participants' understandings of the closures differed, with older people in particular seeming to be uncertain about what was happening and why. This was often bound up with a sense of sadness and loss: "I don't know why they want to close it...I feel very sad. I don't want it to happen...I don't want to go anywhere else." (SU01CH) Unlike some older people, care staff and many relatives seemed to know the main reasons for the closures, but were not always convinced by the arguments put forward. Thus, several families did not believe that these changes were really designed to benefit older people in the long run, instead seeing the changes as a way of reducing costs.
Some were also surprised by the speed of the changes. In contrast, a number of other interviewees had known about the closure plans for a long time, but never really believed they would actually happen. Similarly, care staff were critical, seeing the changes as a way to privatize care and reduce services: "I just can't see any future for care of older people in [the city] now, that's the scary thing." (CS01DC) In contrast, assessors tended to be positive about the changes, feeling strongly that current services and buildings were not good enough. Financially, assessors saw the changes as a significant additional investment in new services and higher quality physical surroundings, not as cuts to services. Assessors described their work as a "privilege", highlighting the difference that resettlement was already making to older people: "This…is an opportunity to significantly improve quality of life. All the users I've worked with have settled well so far and been really happy with the results." (SWA03) While this view contrasted significantly with that of the older people, assessors were clear that they were commenting on real-life, recent closures where they had seen the initial changes once older people had moved to a new service. This was different to the older people we interviewed, who were commenting on their fears ahead of moving.

| Value conflicts
Throughout our interviews, older people and care staff seemed to value different things to assessors. One of the opportunities described by assessors was the replacement of outmoded buildings with brand new, state-of-the-art facilities. Some older people also moved to sheltered accommodation, and assessors expressed pride in people having their own flats and their own front door: "On a personal level [it's] the most rewarding part of the job. To be able to give them back their independence, their own space." (SWA10) In contrast, many older people and care staff rejected the assumption that current services were "unfit-forpurpose", arguing that the new special care centers were not "homely": "Some people have said they look more like hospitals-not homely at all." (CS03DC) "They're not like home-it's so clinical." (SU15CH) "We went to one place but it was so posh I wouldn't want to go." (SU09DC) When talking about what they valued about current services, no one spontaneously mentioned the quality of current buildings, instead highlighting issues such as the attitude of care staff, the underlying ethos of the service and relationships with other service users and staff. Indeed, one of the most common codes within older people's accounts was "friendship" with other service users and staff, coupled with a fear that this could be undermined by the closures. that most of their information had come from care staff, who they described as always doing their best to make sure they were kept informed.

| Challenges of communication
Feeling they were getting inaccurate information or that proposed closures were being delayed was a key concern for care staff and some families: "They keep changing the goal posts.
[They] give you one date then another." (F03DC) Some families expressed concern at being left "in the dark" (F02DC) or "in limbo" (F11CH), especially if they did not yet know which services the person they cared for would be attending in future. In contrast, other families felt that they had received regular information and that assessors had been respectful and helpful in trying to find new services. This was described by one relative as not just "doing the job-more than that … they have tried really hard to connect" (F01CH). Families were also divided on the speed of the closure process, with some feeling that it had been too quick and others that it had "dragged on" and prolonged the uncertainty for older family members.

| Emotional loss
A prominent theme for all interviewees was the emotional impact of closures, with older people and care staff expressing a particularly strong sense of shock, anger and sadness. For one older person, it felt like "being hit with a brick" (SU05DC), while another described it as "like the Titanic sinking and there's no escape" (SU10DC). For care staff, emotional concerns were typically expressed in relation to the impact on service users, who were described as "extremely anxious and upset" and "desperately sad" (CS03DC).
Although broadly supportive of the closures, some assessors expressed concerns about the immediate impact on current residents, with one person describing a situation in which a resident in their 90s had been constantly packing and repacking their things since being told that the home was closing, despite the fact that it was not due to close for a number of weeks. While this assessor understood that community-based services might work better for some people, they also wondered how the changes would impact on residents with complex needs. While concerns here were typically framed in terms of outcomes for residents, the use of the word "immoral" in the quote below suggests a personal toll on the individual assessor as well: "[The] experience is very traumatic for individuals, they are disorientated and confused. Sometimes it felt, well, immoral." (SWA14) Relatives' use of language was often less emotive than that of older people and care staff, with the changes described as "a shame" (F04CH) or "just one of those things" (F06CH). In contrast, staff reported high levels of stress caused by lack of clarity about their jobs. One member of day care staff explained that "empty promises" (CS05DC) had been made initially that all care staff would retain jobs locally, while others felt that decisions around whether employees would have new jobs and where these would be had come too far late. With only a month to go, some participants were still not convinced that services would actually close: "I'm not 100% sure. I don't really believe...not until it happens. Not until I see it." (CS06DC)  ranks suggests there is no significant difference between valuations at IA and 12MFU. These findings suggest that there were no significant longer-term impacts on participants' self-reported quality of life following the closure program.

| Self-reported health and quality of life (EQ-5D)
In order to explore how typical or not the health-related quality of life of our sample was in relation to the "normal" population, we compared our data against normal population values for similar age groups. The comparator group used data drawn from the UK general population survey conducted by Dolan et al. (1995). It was appreciated that caution is needed when making comparisons with other studies due to the small sample size here, and due to the fact that some of our sample were care home residents and others were accessing day care.  (Windle et al., 2009). The quality of life scores in our study are similar to those presented in other studies set in residential care (Devine et al., 2014;Gordon et al., 2014).

| Outcomes important to older people
As demonstrated in Table 4, the majority of participants rated their health and physical well-being positively or neutrally (that is, "neither good nor bad") over time. However, there was a positive shift in the number of participants  who rated their physical health as "good" or "very good" after resettlement. Perceived quality of life also improved in both CH and DC, with a particular increase for DC users.
Forty-two per cent of CH residents suggested life got better, and a further 35% suggested life had stayed the same.
A smaller proportion (19%) suggested that life had got worse. On closer inspection of the data, around half of those who reported life as getting worse at 12MFU suggested this was due to deterioration in their health, rather than due to their new service. Results demonstrate that for the majority of DC users (47%), life had stayed the same at 12MFU. A similar proportion (42%) suggested that life had got better, with only 12% of respondents suggesting that life had got worse. For three of these five participants, it was their health rather than their service which was making life worse at 12MFU.

| Possible limitations
We believe that this is a significant contribution to a crucial but neglected debate-and yet we are equally aware of potential limitations. As suggested above, our qualitative data derive from one care home and one linked day center, thereby giving us an in-depth perspective and enabling us to minimize disruption to the older people concerned, but potentially reducing generalizability. To counter this, we have described how we selected these sites in such a way as to minimize the risk that our findings would be heavily influenced by key contextual factors, for example if we had inadvertently chosen services with high staff vacancies/turnover or very difficult staff relations. Moreover, this  To some extent 35% (11) 32% (10) 38% (12) 44% (19) 37% (16) 12% (5) Neither yes or no 26% (8) 23% (7) 26% (8)  5% (2) 5% (2) 12% (5) No 13% (4) 3% (1) 6% (2) 2% (1) 5% (2) (1) How far do you feel that your cultural and religious preferences are respected?
Definitely 29% (9) 29% (9) 48% (15) 49% (21) 53% (23) 77% (33) To some extent 26% (8) 48% (15) 19% (6) 33% (14) 26% (11) 9% (4) Neither yes or no 35% (11) 23% (7) 29% (9) 14% (6) 16% (7) 14% (6) No 3% (1) 0 0 5% (2) 2% (1)  0 Definitely not 3% (1) aspect of the study was intended as case study research (Yin, 2013), based more on in-depth understanding of people's experiences and on theoretical rather than statistical generalization. In an ideal world, we would have been able to repeat our interviews at 12-month follow-up to gain additional insight into the process and outcomes of care home closures. However, this was not feasible within the confines of the study (given that residents were likely to move from the case study site where we interviewed them to a series of different services across the city, care staff would have changed employment, and the social work assessors would have moved to new roles). It could also have raised potential ethical issues if participants had experienced the closure process as distressing at the time and yet were now enjoying good outcomes (with a risk that the research team could bring back unpleasant memories).
A second set of limitations concerns our quantitative sample: • Outcomes data only focus on the 74 people who took part at all three stages of data collection, and not on the 314 older people eligible for the study. Although we can only comment on outcomes for people who chose to participate, participants' profile in terms of age and mental health do not seem to suggest that this was an unusually "young" or healthy group of older people. Additionally, internal BCC monitoring made them confident that mortality rates and subsequent hospital admissions did not seem to be any higher for the older people being resettled than might be expected. To reduce the risk of potential bias, we built our data collection into standard social work assessment processes so as to ensure that individual workers could tailor the invitation to take part and subsequent data collection to the needs of individual older people, including those with dementia or who did not speak English as a first language. It also enabled consultees to take part in situations where the older person lacked capacity to consent. As a result, our sample is comparable to other studies of older people's social care (Devine et al., 2014), and we have no evidence of systematic bias.
• Despite the scale of these closures, our quantitative sample remains limited and can only provide descriptive statistics. Given this may have been one of the most significant closure processes in the UK to date, it seems unlikely that this is a topic that could ever generate a sufficient sample size for data to be statistically generalizable to the whole population of older people in care. Given practical/ethical constraints, it also seems unlikely that a trial-based approach could be adopted, making our current study a pragmatic way of trying to understand the issues at stake in a challenging environment to be conducting research.
• Closures affected care home residents and people attending linked day centers, and these two groups may have different needs. While we have broken down our findings down by CH and DC, overall outcomes were positive for both groups despite what might have been anticipated in advance. For example, people receiving DC were unhappy at being included in a process that seemed focused primarily on care home re-provision-and yet even here the results were positive.
A third potential limitation is that our initial quantitative data came as older people were being assessed ahead of moving to a new service. Our results at IA may actually be worse than "normal" baseline levels, as people were already negatively affected by the prospect of closures. If this were the case, then improved outcomes might simply be a return to "normal" levels. In practice, this does not seem to have been the case, given that social work assessors in particular talked about the positive changes they had witnessed in the older people with whom they were working.
The possibility of closures had also existed for some time in BCC, and a number of older participants still did not really believe that the proposed changes would actually take place when they took part in our study.

| Implications
We have called this article "A game of two halves" 5 , as our analysis of the process and outcomes of care home closure seem to reflect such different results. Part-way through the process, when older people, families, and care staff still did not know exactly what would happen next, there was a strong sense of anger and loss. Although BCC seemed to be taking significant care over the closure process, most older people and care staff in particular felt disempowered and let down by what they saw as: • poor communication; • a lack of involvement in initial decision-making; and • an unthinking inclusion of linked day centers in the closure process.
In contrast, families tended to be more ambivalent about the closure process, while social workers were predominantly positive about the long-term potential. In the end, the social workers' view was borne out by outcomes after 28 days and especially one year later, which were much more positive-running counter to the perceived wisdom that relocation is detrimental to older people's health. One year on, a number of participants felt that their health and well-being was either the same or better, despite the fact that they were a year older. This suggests that, if planned well and if existing services are less than optimal, it may be possible to assess and resettle older people without making things much worse for them in the medium-term, and possibly to improve outcomes for some people.
This is of major significance given the risks of future closures, and we have distilled lessons learned from this study in a national good practice guide (Glasby, Robinson, & Allen, 2011). Above all, the key issues appear to be creating the time and space to plan the closure process and to work at the pace of individual residents. It is thus crucial that future service failure regimes build in sufficient time; for example if a future care home operator goes bankrupt, it should not be possible for a receiver to sell on the properties concerned until councils have had appropriate time to resettle existing residents. Equally, councils which fear that the current quality of services or of the built environment is insufficient should not necessarily feel that they cannot consider closure for fear of damaging the health of current residents.
Explaining the apparent gulf between our initial interviews and subsequent outcomes is challenging, but we believe that at least three main lessons can be drawn. First is the importance, but also the difficulty, of communication. Within BCC, the research team saw detailed communication plans, interviewed assessors based in the buildings that were due to close, heard positive things about the assessors from other participants, and met an independent advocacy team commissioned to provide additional support. To us, this seemed an example of good, principled communication. However, some residents genuinely could not recall what they had been told, by whom, or when. There were also conflicting messages around how communication should be timed. In some cases it all seemed to be happening too fast, with participants expressing their shock and sadness. In other cases, especially for relatives, the process was sometimes seen as unnecessarily drawn out, with closures hanging over people for several months, without any real indication of the detail of subsequent moves. Perceptions of being given the "wrong" information were to some degree related to the scale of changes: with such a complex process, some changes to initial plans were probably inevitable. When information seemed to change, this could either be seen as having received "wrong" information initially or the result of a genuine commitment to providing everyone with the latest news and thinking.
Second, while ongoing communication is crucial, it will still not be enough to prevent significant anxiety. Certainly, strong resistance to the closure of public services and related questioning of motivations behind local policy are well documented, especially in relation to hospital closure (Brown, 2003;Oborn, 2008). In particular, the perceived lack of involvement in the initial decision to close services formed narratives of disempowerment that correspond strongly to classic treatments of public participation (Arnstein, 1969). Participants described involvement in activities that Arnstein would view as "tokenism", for example as subjects of information giving, yet having no access to real power or the ability to engage and negotiate about key decisions. Although this resistance seems to some extent unavoidable when it comes to decommissioning public services, the experience of being on the receiving end of such decisions should not be overlooked.
As older people and care staff reflected on the decision to close services, they revealed a series of unresolved differences in values and perceptions that conflicted with the direction of travel of the closure process and with the views of some assessors. These conflicts focused on interpretations of what constitutes "fit-for-purpose" in the physical environment, the importance of "homeliness" and competing models of care, such as the importance of day care compared to other forms of delivery. Other policy areas such as housing and urban renewal provide a long legacy of debate around these issues. Critiques of policies such as slum clearance (Young & Willmott, 1957) and housing market renewal (Minton, 2009) illustrate the disconnect that can occur between policymakers and local people, with concerns from residents that the most unique, positive aspects of their environments had neither been understood nor valued by the policymakers disbanding their communities. As with these examples, BCC was faced with a complex task of designing provision not just for current but also future generations of older people. A good example here is the potential tension between the desire of assessors for state-of-the-art buildings, compared to the emphasis of current service users on "homeliness".
Above all, participants expressed a profound sense of loss. Some studies liken the stress of relocation to that of losing a loved one, with a cycle of emotions including, "Feeling of painful loss, continued longing, general depressive tone, frequent symptoms of psychological, social, and somatic distress, sense of helplessness, occasional direct and displaced anger, and a tendency to idealize the lost place" (Fried, 1963, p. 151). Heller (1982, p. 482) suggests that, "Post-relocation adjustment depends on the degree to which residents are 'psychologically prepared' for the change".
Certainly, a number of our participants expressed significant sadness and loss, and the notion of closures not just as a physical relocation but as a form of bereavement or psychological trauma may be helpful.
Lastly, things might get worse before they get better, and those planning closures may need to hold their nerve en route. In our experience, success can often look like failure part way through, and BCC could easily have talked to older people, families, and care staff during the closures and concluded that it was going to be too detrimental to continue. In fact, the opposite seems to have happened: without understating the sense of anxiety and distress caused part-way through the process, many people's lives either stayed the same or improved-and this seems a significant achievement.