Discussion

Introduction

Is there enough research to answer the question of whether dispersed or cluster housing is better? This review found 19 papers based on 10 studies including nearly 2500 people in four countries presenting data comparing dispersed housing with some kind of clustered housing (village communities, residential campuses or clusters of houses). The studies covered all eight domains of quality of life, providing information about 80 different aspects of these domains. Five studies presented data on different aspects of service design and operation and three presented comparative costs. Almost all of the studies used quantitative methods with robust approaches to measurement.

It is perhaps noteworthy that no studies of clustered housing were found in the research literature from the USA and Canada. It seems that in these countries the movement from institutions to dispersed housing has not led to research on clustered settings.

The problem faced by most of the studies identified is that the populations of people living in different kinds of setting are not generally comparable. People living in village communities are in general much less disabled than people living in dispersed housing. People living in campus or cluster housing are generally more disabled than people living in dispersed housing. The more sophisticated studies addressed this problem through statistical methods but in addition the ten studies include six good matched group comparisons (Hatton et al., 1995, Janssen et al., 1999, Emerson et al., 2000b (includes two matched groups), Emerson et al., 2000c, Young, 2006). Four of the six included people with severe and profound intellectual disabilities and complex needs.

Compared with many areas of public policy in most countries, this is a sizeable body of research. It is drawn from different countries (Australia, Ireland, Netherlands and the United Kingdom). The studies have all been carried out in the last 20 years. The services studied are provided by many different agencies, including governmental and non-governmental organisations.

There are three significant gaps in the available research. There is no research on village communities - intentional communities where support workers live communally with disabled residents - that serve people with severe and complex needs. If such communities exist they might be expected to have much higher staffing ratios than the village communities studied by Emerson et al (2000b) and issues of cost and quality could be usefully studied. Second, there is no research data on clustered settings for disabled people other than those with intellectual disabilities. Insofar as village communities, residential campuses or cluster housing exist for people with physical or sensory disabilities or long-term mental health problems then evaluation of their cost-effectiveness would add to the overall picture. Third, most studies are point-in-time comparisons and therefore do not address the question of whether services get better or worse (or more or less expensive) over time. The longest follow-up study of people moving out of institutions in England (Cambridge et al., 2001) showed maintenance of quality with reduced costs over time, but this study did not discriminate between residential homes, small clusters and dispersed housing.

Main findings

There is much less evidence comparing clustered settings with dispersed housing than comparing other congregate care settings (such as institutions) with dispersed housing. The results should therefore be treated with caution. They do, however, present a broadly consistent picture.

The body of research reviewed here presents a clear picture. Generally, campus and cluster housing provides poorer outcomes than dispersed housing for people with intellectual disabilities. In terms of the quality of life domains of social inclusion, material well-being, self-determination, personal development and rights there are no studies reporting benefits of clustered settings. In the physical well-being domain, campus or clustered settings have been found to be superior in hours of recreational activity, contact with dentists, psychiatrists and psychologists, some health screening, some aspects of safety, contact with family and friends, visitors to the home and satisfaction with relationships. However, in many of these cases the better results refer only to village communities and not to campus housing or cluster housing.

In terms of the areas identified in the introduction as those where proponents of cluster housing argue that it would perform better, these studies find that safety is no different between campus or cluster housing and dispersed housing but people in village communities are less likely to have been the victims of crime or verbal abuse. In terms of relationships with members of the wider community, two studies find no difference between dispersed and cluster housing. The one that does (McConkey et al., 2007) may have included relationships with other cluster residents. This is also a study of clustered supported living, rather than clustered group homes, and may therefore be different from other studies in the review.

In terms of costs, the commonest finding is that cluster housing is less expensive than dispersed housing. However, this cost difference appears to be due to differences in staffing levels - ie fewer staff are provided to support people in cluster housing than in dispersed housing. In two out of the three studies which examined costs controlling for this variable there was no statistically significant difference in costs - the comparison by Hatton et al (1995) of a specialised campus-based time-limited further educational service with specialised dispersed housing, and the comparison by Emerson et al (2000b) of matched groups of people in village communities and dispersed housing and campus and dispersed housing.

Thus, cluster housing is usually less expensive than dispersed housing because it provides fewer staff. It achieves less good outcomes for its residents on almost every indicator measured. The best performance of cluster housing is seen in village communities, but these only serve a less disabled population and they depend on a supply of people willing to live communally with disabled residents. They are therefore an important part of the spectrum of service provision but they are only ever likely to occupy a niche in the market for care. They are unlikely to be a feasible option across the board for disabled people.

Interpreting these findings

Are these findings robust - are the differences found due to the model of care or might they be due to weaknesses in management and organisation? Is it possible to have cluster housing that would achieve results as good as dispersed housing? What are the implications over the longer term? The studies reviewed, together with the wider literature on residential care of people with intellectual disabilities, provide some insight into these questions.

Are differences due to the model of care or weaknesses in management and organisation?

As noted in the Introduction, it might be argued that poorer outcomes reflect weaknesses in management and organisation rather than the service model itself. However, the weight of evidence - from different countries and different types of service system - might be thought to indicate that poorer outcomes in clustered settings represent something more than weak management and organisation. If exemplary clustered settings, comprehensively out-performing dispersed housing, did exist it seems likely that they would by now have appeared in the research literature.

Arguing that differences between types of service are due to management and organisation rather than the service model also applies to both dispersed and clustered settings. If it was possible to provide campus housing that achieved outcomes as good as dispersed housing in the many areas in which these studies show it does not currently achieve them, then it could be argued that dispersed housing could be improved on the handful of indicators where it does not out-perform cluster housing. In the research literature generally, there is substantial variation in the outcomes achieved within the same type of residential model, but the best-performing dispersed housing appears to be better than the best clustered settings (Emerson and Hatton, 1994, Mansell, 2006).

Is it possible to have cluster housing that would achieve results as good as dispersed housing?

The finding that the lower cost of clustered settings reflects lower staffing levels is important. Studies of the relationship between management and organisation of the residential setting and the outcomes experienced by people living there consistently show that the key factor is what staff actually do to support the people they serve.

"Through the provision of help and encouragement, staff members mediate access to, and use of, the opportunities presented by the home and community. They control access to many materials and activities directly (e.g., by opening or locking rooms) and indirectly (by setting out and preparing materials so that the people they serve can take part in an activity). They make it more or less likely that clients will experience the reinforcement intrinsic to the task by the level of assistance they provide. Through the disposition of their social interaction, they reinforce either client engagement in meaningful activity or passivity and inactivity. They shape client behavior by the feedback and reinforcement they provide."
(Mansell and Elliott, 2001)


As dispersed housing has become the dominant model of support for people with intellectual disabilities in some countries, variation in the quality of staff support has been intensively studied. What these studies show is that staff ratio is a weak predictor of staff performance: that it is possible to have high staff ratios and yet poor performance. These studies also show that better use is made of higher staffing ratios if staff are given clear guidance on how to support residents (Mansell et al., 1982) and if, instead of adding staff to a large group of residents to improve the ratio, the number of residents in the group is reduced (Felce et al., 1991). In settings where in general these conditions are met, Mansell (2008) recently showed that larger dispersed housing services had lower staff ratios (even when resident needs were taken account of); that these lower staff ratios translated into less staff support for residents; and that less support for residents led to worse outcomes for them.

Thus it seems unlikely that it would be possible to achieve lower costs in cluster housing without reducing the quality of outcomes of people living there. As a recent analysis of cost-effectiveness studies of residential services for people with intellectual disabilities in Europe (Mansell et al., 2007a) concluded:

"In a good care system, the costs of supporting people with substantial disabilities are usually high, wherever those people live. Policy makers must not expect costs to be low in community settings, even if the institutional services they are intended to replace appear to be inexpensive. Low-cost institutional services are almost always delivering low-quality care."
(Executive summary p7)


What are the implications over the longer term?

The point is made above that most studies reviewed were not longitudinal and therefore cannot comment on how different models perform over time.

However, in relation to costs it is important to note that cost comparisons can vary over time. (Mansell et al., 2007b). Lower costs of congregate settings (particularly for people with high support needs) may simply reflect poorer quality care (in particular lower staff ratios) which cannot be sustained over time as demands to improve quality take effect. Stancliffe and Lakin (2005) show that costs of institutional care in the United States of America increased to exceed those of community-based services over the period from 1970 to 2000, due both to the drive to increase quality and to the strategy of moving the least disabled people out first. The DECLOC study (Mansell et al., 2007a) concluded that

"There is no evidence that community-based models of care are inherently more costly than institutions, once the comparison is made on the basis of comparable needs of residents and comparable quality of care. Community-based systems of independent and supported living, when properly set up and managed, should deliver better outcomes than institutions."
(Executive summary p7)


The same argument may well apply to clustered settings compared with dispersed housing. A second consideration is that recent research shows that dispersed housing in the form of supported living, tailored more closely to individual need, achieves better outcomes in some quality of life domains at lower cost than group homes for people with low or moderate support needs (Stancliffe and Keane, 2000, Stancliffe, 2004, Felce et al., 2008). There may therefore be scope for refining dispersed housing models to ensure that staff allocation (and therefore costs) more closely reflects support needs of individual residents.

Thus, although dispersed housing for people with high support needs is likely to be as expensive as congregate settings of equivalent quality (if it was in fact possible to provide such settings), dispersed housing for people with low or moderate support needs is likely to be less expensive. This would require moving from relatively institutional models of group home organisation (with a large proportion of highly-qualified professional care staff, night staffing, constant attendance of residents in spite of their needs) to more person-centred models of care, where each individual receives only the level of support they need rather than the same level being provided to all residents whether they need it or not.

There is also some suggestive evidence about the longer-term outlook from policy and practice in England. Early in the deinstitutionalisation process in England (in the early nineteen-eighties), various kinds of congregate care setting were developed for people with severe and profound intellectual disabilities whom it was judged it would not be possible to support in dispersed housing. Many of these services were residential campuses of the kinds studied by Emerson et al (2000b). They were set up in response to the same sorts of concerns identified in the introduction about the proposed benefits of campus living.

In 2004, complaints about the quality of care in some of these services in Cornwall, and subsequently in south London, led to Inquiries which found major problems in local services provided by the NHS (Healthcare Commission and Commission for Social Care Inspection, 2006, Healthcare Commission, 2007a). A national audit was then undertaken of all NHS residential services for people with intellectual disabilities which confirmed that poor quality of care was widespread (Healthcare Commission, 2007b). The UK Department of Health responded to these problems by announcing the complete closure of all NHS residential campuses (Department of Health, 2007).

Although it is not possible to say whether the problems experienced in these services were due to the model of care or due to management by the NHS, it is relevant to note that this large development of clustered settings has, after 20 years, been deemed to have failed.

Conclusion

The results of this review show that dispersed housing is superior to cluster housing on the majority of quality indicators studied. The only exception to this is that village communities for people with less severe disabilities have some benefits; this is not however a model which can be feasibly provided for everyone. Cluster housing is usually less expensive than dispersed housing but this is because it provides fewer staff. There is no evidence that cluster housing can deliver the same quality of life as dispersed housing at a lower cost.