Results

Introduction

Of the 19 papers included in the review, 14 papers related to studies conducted in the UK, one in the UK and Ireland, two in Ireland alone, one in the Netherlands and one in Australia. There were no studies identified which were carried out in the USA or Canada. The majority of the papers used quantitative methodology (18 of 19) and one was a small scale qualitative study. 15 papers reported studies with a cross-sectional design and 3 were longitudinal studies (with a cross-sectional element also included). The size of studies ranged from 11-910 participants. All 19 papers reported findings relevant to people with intellectual disabilities - no other user groups were identified in the literature.

The majority of the papers report on studies which included large samples (11 of 19 papers refer to studies which had 500 participants or over). However, nine of the papers drew data from the same study - the study of 500 people in village communities, National Health Service (NHS) residential campuses and dispersed houses in the community funded by the English Department of Health. The two papers by Hatton et al (1995 and 1996) were also based on a single study. Therefore the papers included in the review reported findings from 10 different studies. Only six papers reported on costs and these referred only to three studies.

The following terms referring to residential accommodation were used in the studies included in the review:

  • For cluster or campus settings
    Village communities, residential campuses, campus style further education facilities, campus home on the grounds of an old hospital, campus bungalows, houses on the edge of care facilities, clustered supported living, campus settings, intentional village communities, cluster housing, hospital homes1, specialist units on the site of previous mental handicap hospitals1.
  • For dispersed community-based settings
    dispersed housing (which include group homes and supported living), community based, specialised community based group homes, ordinary community based group homes, group homes in the community, dispersed supported living, small group homes, residential homes, staffed house or apartment, community based residential support, residential care homes in the community1.
  • Other service types
    Residential care homes in the community1, hospital, hospital homes1, specialist units on the site of previous mental handicap hospitals1.
  1. Terms that could be considered in more than one category - eg small residential community homes could be under dispersed housing, larger homes could be more like nursing homes/hospitals. Some units or homes within the groups of hospital could be small and part of a campus or cluster, others could be large and just one or two of them on the site.


Quality of the studies included

In terms of design, 17 of the papers studied included in the review were rated as evidence level 3a, because they compared outcomes between participants in different kinds of setting. Two studies were level 4 because they were observational without a control group comparison. Most of these studies are natural experiments - exploring differences that occur because of the situation people are in rather than the manipulation of an independent variable. Experiments and quasi-experiments (levels 1 and 2 of the CRD classification) are less likely in social care, because the researcher is not usually in a position to assign participants to different groups. Five papers reported data from matched samples of participants across the settings (one paper including two matched groups) although in other studies the authors did attempt to control for these differences in participant characteristics by statistical means, using analysis of variance and regression analysis. Thus the majority of these studies are relatively rigorous comparisons given the subject matter.

Domains covered by each study

Box 4 summarises the number of papers addressing each domain of the quality of life framework together with quality of care and costs. The most common domain explored was physical well-being. The least commonly reported domain was that of material well-being. Only six papers reported on cost related data and these only referred to three studies.

Appendix 3 provides a detailed mapping of the domains covered by the papers reviewed.

Box 4 Number of papers addressing each domain

Domain

Number of papers

Social inclusion

7

Interpersonal relationships

11

Material well-being

4

Emotional well-being

6

Physical well-being

12

Self-determination

7

Personal development

6

Rights

5

Staffing/quality of care/staff support

5

Costs

6


The remainder of this section presents each of the 19 papers included in the review in detail. Each study is numbered so that they can be identified in the master tables (Table 19 and Table 20) presented on page 41.

1. Mansell and Beasley (1993)

Description of the study

Mansell and Beasley (1993) reports a longitudinal evaluation of the effectiveness of community based residential services, individually designed for people who, in addition to their severe or profound learning disability, had very serious challenging behaviour. It was carried out in south-east England. Observations were made over three years, tracking individuals in both pre- and post-move settings. The study focused on transfer from institutions to community living but includes some data on campus housing.

The study involved 18 participants. At the beginning of the study, 10 lived on ordinary hospital wards in two large institutions and eight lived in special units reserved for people with the most challenging behaviour in those institutions. At the time the paper was written, 11 people had moved to community based services. Each community service was individually designed for one or two people in the study group, usually living with one or two people with less severe disabilities in a staffed group home or apartment. Three of these 11 people moved to houses that were on the same site as other residential facilities on National Health Service campuses (one of approximately 70 and one approximately 40 places in urban areas of South London). In two cases this was a temporary move pending relocation to dispersed housing.

The design of the study was a multiple baseline design across individuals. The data collected included direct observations of resident and staff activity, staff allocation and aspects of the physical environment. Observations represented 'composite days' in the lives of each individual, being taken over three week days with a few days of each other to cover the period between 8am and 7pm so that at each data point there were 1980 observations of activity for each person. Information was collected on seven occasions over three years. Inter-rater reliability was assessed across codes, individuals and data points and was between 84 and 95% on each occasion.

In addition to the observations, staff allocation to each living unit was recorded on each visit by examining duty rotas. The number of staff and service users in the same room was recorded every 5 minutes. The material richness and space of the physical environment were recorded once for each living unit and included the space shared by people, the number of rooms locked and the amount and type of furniture.

Findings

Comparability

The eight people living in dispersed housing were five men and three women with an average age of 37.5 years (range 34-42). Their average score on Part One of the Adaptive Behavior Scale (ABS) (Nihira et al., 1974) was 92 (range 54-154). The three people living in cluster houses included two men and one woman, aged on average 36 years (range 35-38) with average ABS Part One scores of 104 (range 49-211). Those remaining in hospital at the time the paper was written included five men and two women, average age 29 years (range 19-38) with average ABS Part One scores of 86 (range 45-110). No analysis of comparability is attempted in the paper.

Staffing ratios were higher in community based services. Staff were present in the same room as participants for more time in houses (91%) than in the campuses (86%) or hospitals (74%).

Results

People living in dispersed houses were less likely to be in the company of people who had competing needs to their own. They were more likely to be the only resident in the room (dispersed housing 63% time, campuses 42%, hospital wards 33%, special units 27%). The quality of staff performance also differed between the houses and other types of setting. People living in hospital types of service received staff contact for, on average, 7% (hospital wards) and 9% (special units) of the time, compared with 13% in the 'campus' houses and 26% in the dispersed houses. The contact from staff included direct assistance in the houses in 11% of observations, compared with 4% in the 'campus' houses and 2% in the hospitals.

The activity level of participants is presented but not analysed by setting type.

Table 1 Summary of relevant findings from Mansell and Beasley (1993)

Measure

Dispersed housing better

No difference

Campus/cluster housing better

No other resident in room

Y

  
Staff contact

Y

  
Staff assistance

Y

  


Strengths and weaknesses of the study

The study provides evidence at level 3a on the CRD scale. It is an observational study involving time-series and cross-sectional comparisons. The measures involved are quantitative, widely used and robust.

The study was not designed to compare campus and dispersed housing models, has a very small sample of people with extreme characteristics and the data are not suitable for between-group statistical analysis.

The campuses were relatively large and were only transitional placements for two of the three people concerned. One was closed shortly after this study and the other is being closed as part of the UK Government's decision to phase out campus provision.

Conclusion

The study suggests that dispersed housing is superior to campus housing on the measures presented but the limitations of the study design make this finding of limited use. The study did not address costs.

2. Dockrell et al (1995)

Description of the study

Dockrell et al (1995) reports an evaluation of a specialist assessment and treatment unit for people with mild learning disabilities and challenging behaviour (65% of whom had a diagnosis of mental illness). It was carried out in south-east England. It tracked 34 people (typically male, young, dangerous and on medication) and gives their destination/situation at 6 months. There were seven people whose destination was a campus home - these were homes for about six people located on campuses of between 40 and 100 people. There were three people in hospital homes - small to medium sized units in the grounds of a hospital. There were eight people in community homes and five people who went to their own home or a parental home. There were 11 people who at six months were still in institutional settings.

The design of the study was a point-in-time survey. Information collected included the level of adaptive and maladaptive behaviour (using the Vineland Adaptive Behavior Scale (Sparrow et al., 1984) and the ABS Part 2), though these data are not presented in this paper. In addition, keyworkers were interviewed and ratings made (low, medium and high) of each type of home on six criteria:

  • Personal privacy - having a single or shared bedroom
  • Choice/autonomy - freedom of access to facilities in the establishment
  • Participation in domestic activities - contributions to cooking, cleaning, etc
  • Freedom - whether the establishment is locked or not, freedom to make visits outside
  • Access to neighbourhood - the proximity to local shopping area
  • Use of community facilities - the frequency of use of cinemas, swimming pools etc


It is important to note that these ratings were of the opportunity for achievement in the domain, rather than actual performance:

"In this study no attempt was made to measure quality of life directly. Instead a comparison was made of quality of life opportunities offered in different care settings--hospital homes, campus homes and community homes. This is not the same as quality of life or quality of care, but is an attempt to assess the extent to which it is, in theory, possible for clients to achieve a reasonable quality of life" (p897).


Costs were calculated for two hospital homes, two community homes and one campus home. The cost estimates included both running costs and rent (or imputed rent) for buildings and equipment.

Findings

Comparability

No information is presented on the characteristics of people resident in each type of setting. Day-time staffing ratios were higher in dispersed housing (0.75 staff to 1 service user) than in campus homes (0.45:1) and hospitals (0.3:1).

Results

Dispersed housing was scored 'high' on all six domains; campus housing scored 'high' on choice, participation in domestic activities and use of community facilities, 'medium' on freedom/level of security and access to neighbourhood and 'low' on personal privacy. Hospital settings scored 'medium' on choice, participation in domestic activities and freedom/level of security and 'low' on personal privacy, access to the neighbourhood and use of community facilities.

The cost per place per year of dispersed houses was £64000; of campus houses £36000; and of hospital houses £27000. Differences in cost largely reflect differences in staff ratio. The authors comment that in the context of treatment outcome these may not be equivalent:

"While a campus home may prepare clients for a campus life, the community home may prepare clients for either low service intensity support or eventual 'discharge' from the health authority's services. This reinforces the point that it is important to consider care for this client group in a dynamic model" (p899).

Table 2 Summary of relevant findings from Dockrell
et al (1995)

Measure

Dispersed housing better

No difference

Campus/cluster housing better

Personal privacy

Y

  
Choice/autonomy 

Y

 
Participation in domestic activities 

Y

 
Freedom

Y

  
Access to neighbourhood

Y

  
Use of community facilities 

Y

 
Cost  

Y


Strengths and weaknesses

The study provides evidence at level 4 on the CRD scale. It is an observational study involving a cross-sectional comparison. The measure of 'opportunity for quality of life' is crude and not validated.

The study was not designed to compare campus and dispersed housing models, has a very small sample of people with extreme characteristics and the data are not suitable for between-group statistical analysis. The study is primarily focused on outcome of treatment rather than comparing service models.

The cost information presented does make an attempt at comprehensiveness and like-for-like comparison and the authors do comment on the possibility that costs might diminish over time. However the lack of comparability of the groups makes cost-effectiveness judgements impossible.

The campuses were relatively large. One was closed shortly after this study and the others are being closed as part of the UK Government's decision to phase out campus provision.

Conclusion

The study suggests that dispersed housing is superior to campus housing on the quality measures presented, but is more expensive. The limitations of the study design make this finding of limited use.

3. Hatton et al (1995) and 4. Hatton et al (1996)

These papers relate to the same study. Hatton et al (1995) explores the quality and costs of four different service models for 40 adults with multiple disabilities (especially sensory disabilities) in England. Hatton et al (1996) uses the same data to explore predictors of outcome.

The four service models were

  • Specialised institution-based units (15 people in two units)
  • Specialised campus time-limited further educational service (16 of 38 people in small, usually 5-person apartments)
  • Specialised dispersed housing (four residents living in three houses provided by one agency and 4 residents in one of three homes provided by another agency)
  • Ordinary dispersed housing (eight people living in houses managed by three agencies)


In addition, some data are presented for five people before their move from an institution.

The study used a group comparison design. Information collected included (i) non-participant observation of patterns of staff and resident activity over composite 12-hour days covering 0800-2000, (ii) resident:staff ratios in the setting and the room, collected during observations, (iii) the Disability Assessment Schedule (Holmes et al., 1982) by interview with care staff, (iv) scheduled activities collected by interview with care staff, (v) a shortened version of the Group Home Management Interview (Pratt et al., 1980), (vi) frequency and nature of outings and frequency of visits to residence from family or friends from daily records and (vii) the cost of services collected using the Client Service Receipt Interview (Knapp et al., 1992), cost and price information collected at facility and agency levels and costing of the capital value of buildings. Inter-rater reliability data were collected for the observational measure and calculated using product-moment correlations (campus) and Cohen's kappa (other units). Only data achieving acceptable levels of inter-rater reliability were presented.

Findings

Comparability

One-way analysis of variance on resident needs and characteristics showed that the groups were comparable in terms of level of intellectual disability, the proportion with visual and hearing impairments, gender, total and subscale scores on the Disability Assessment Schedule. Residents in the ordinary dispersed housing and in the specialised institution-based units were older than residents in the campus or the specialised dispersed housing.

Staff ratios were higher in the campus houses (0.69 staff to 1 service user) and ordinary dispersed houses (0.67:1) than in the institutions (0.38:1) but were not significantly different from each other or from the specialised dispersed housing (0.61:1) or the institution-based special units (0.5:1); staff ratios observed in the room were higher in the campus houses (0.69:1) than in the institutions (0.32:1), the institution-based units (0.43:1) or the specialised dispersed housing (0.49:1), but not the ordinary dispersed housing (0.52:1).

Results

In terms of social climate, the campus houses had higher levels of block treatment than either type of dispersed housing. Institutions had greater social distance than campus houses, specialised and ordinary dispersed housing, but not institution-based units; these units had greater social distance than either type of dispersed housing but were not significantly different from the campus houses. Institution-based units had greater levels of depersonalisation than campus houses, specialist or ordinary dispersed housing. Residents of specialised dispersed housing had more hours of scheduled activity (38.5) than institutions (8.0), institution-based units (16.2) and campus houses (15.7) but were not significantly different from ordinary dispersed housing (26.6).

Residents in campus houses had fewer contacts with health and welfare professionals (average 10.3 contacts in preceding four weeks) than those in other service settings (specialised dispersed housing 50.4, ordinary dispersed housing 43.6, institution-based units 64.0, institutions 48.0). No difference was found in the number of visits received from friends/family across service models. Residents of specialised and ordinary dispersed housing had significantly more outings than residents in institution-based units and campus houses (average number of outings per week: specialised dispersed housing 24.6, ordinary dispersed housing 29.7, institution-based units 4.6, campus housing 6.8).

Significant differences between service models were found in the contact received from staff. People living in specialised dispersed housing received a higher rate of positive staff contact than in any other service model (average rate per hour in specialist dispersed housing 23.3, institution-based units 8.0, campus houses 7.3, ordinary dispersed housing 1.9, institutions 1.9). They also received more overall staff contact than those in institutions, and more direct assistance than those in institutions or ordinary dispersed housing, but were not different from campus houses on these measures. People living in campus houses received more staff care - "providing care for residents without encouraging their participation (eg medical assessments, dressing the resident rather than using instruction to encourage resident participation in dressing)" than in institutions or specialised dispersed housing (average percentage time in campus houses 5.6%, institutions 1.1%, specialised dispersed housing 1.2%).

Table 3 Summary of relevant findings from Hatton et al (1995) and Hatton et al (1996)

Measure

Dispersed housing better

No difference

Campus/cluster housing better

Block treatment

Y

  
Social distance 

Y

 
Depersonalisation 

Y

 
Scheduled activity

Y2

Y1

 
Contacts with health and welfare professionals

Y

  
Number of visits from family/friends 

Y

 
Number of outings

Y

  
Total staff contact 

Y

 
Positive staff contact

Y2

Y1

 
Staff assistance 

Y

 
Staff care 

Y1

Y2

Resident constructive activity 

Y

 
Challenging behaviour 

Y

 
Stereotyped, repetitive behaviour 

Y

 
Cost

Y1

Y2

 
  1. Comparison between ordinary dispersed housing and campus/cluster housing
  2. Comparison between specialised dispersed housing and campus/cluster housing

There were no differences in involvement in constructive activity by people living in dispersed or campus housing. Residents of specialised dispersed housing engaged in higher levels of constructive activity than residents in institutions and institution-based units. There were no significant differences between models in terms of challenging behaviour or stereotypy.

Costs were higher in the campus houses and the institution-based units than in ordinary dispersed housing but not significantly different from those in specialised dispersed housing.

The authors note that all results show great variability within the same type of service. Hatton et al (1996) drew on the data from the same study and conducted path analyses using resident engagement in constructive activity and number of outings as dependent variables. They found that engagement in constructive activity was directly predicted by resident ability, positive staff contact and community, rather than campus or institutional, location. There were two indirect predictors - scheduled activity and specialised service orientation, acting through positive staff contact. Outings were directly predicted by more scheduled activity, less social distance, less block treatment and non-specialised orientation, with indirect predictors less block treatment and resident skills (through scheduled activity) and community location (through social distance).

Strengths and weaknesses

The study provides evidence at level 3a on the CRD scale. It is a cohort study involving a cross-sectional comparison. The measures used are quantitative, widely used and robust.

Another strength of the study is that it was designed to compare different housing models and the groups of residents were comparable on the characteristics studied. Overall staff ratios were also comparable. Cost information was comprehensive and the comparisons made on a like-for-like basis. The authors note the relatively small sample size, the fact that this is a study at a single point in time and that there was large variation between different services of the same type. Nevertheless it is a relatively comprehensive and effective comparison.

Conclusion

The study suggests that dispersed housing is superior to campus housing in terms of avoiding block treatment, scheduling of activities, contacts with health and welfare professionals, number of outings, positive staff contact and cost. Two of these advantages appear to be related to the presence of specialised knowledge in the staff. Campus housing is superior to dispersed housing in the provision of staff care. Staff care - that is, caring for residents without enabling their participation in the activity - is better than neglecting the needs of residents but arguably is not as good as enabling residents to participate in self-care and any other activities. Although specialised dispersed housing achieved higher average levels of engagement in constructive activity than campus houses, this difference was not statistically significant. However, path analysis identified community location as predictive of engagement in constructive activity and in this respect the dispersed housing is therefore likely to be superior to campus housing.

5. Janssen et al (1999)

Janssen et al (1999) was a study of 668 people with intellectual disability randomly selected from seven residential facilities in the Netherlands. Their average age was 38, more than 60% had mild to moderate intellectual disability, 59% were male and the average number of people per living unit was 9 (range 1-18). From these 668, two groups were selected: 80 people living in houses on the edge of the residential care facilities and 119 people in group homes outside the residential facilities. Over 90% of these groups were people with mild or moderate intellectual disability.

The study used a group comparison design. Information was collected using two specially developed measures of quality of life each with approximately 300 items, one completed by a member of staff and the other by a family carer or by the resident themselves. The authors refer to unpublished pilot studies demonstrating validity and inter-rater reliability and present data showing high internal consistency of the scale domains. The measure completed by staff had 17 domains and that completed by service users or their families had 20 (see Box 5). There were 14 domains that covered the same issues, three domains were rated only by staff and six only by residents or family carers, giving a total of 23 different domains (see Table 4).

Findings

Comparability

The authors report that the two groups were comparable in terms of their characteristics and the amount of staff support required but do not present any statistical test of this.

Results

Results are presented first for the ratings by staff, then by the ratings made by family members or residents themselves.

Quality of life for people living in dispersed housing was rated better than for those in campus housing in 15 of 17 domains rated by staff.

In respect of the physical domain 30% of people in campus houses were rated as not receiving adequate care compared with 6% in dispersed housing, because they have less freedom to participate in physical activities, to choose (eating/drinking), mobility to visit others, receive less training in personal care and in transportation and less integration related activities.

Box 5 Domains of the quality of life measure

Staff measure

Service user/Family measure

  1. Physical Domain
  2. Adapted Housing
  3. Acceptance, Sexual Activity.
  4. Identity
  5. Vision/Religion
  6. Housing
  7. Property/Belongings
  8. Housekeeping
  9. Relation Caretaker
  10. Relation Others
  11. Recreation
  12. Daily Activities
  13. Safety
  14. Freedom
  15. Training
  16. Integration
  17. General Care
  1. Care Physical Health
  2. Care Fitness
  3. Freedom in Eating/Drinking
  4. Care Mental Health
  5. Care Identity
  6. Religious Activities
  7. Housing
  8. Care Properties
  9. Independent Housekeeping
  10. Living Environment
  11. Treatment by Caretakers
  12. Relation with Family
  13. Freedom/Relations with Others
  14. Quality and Quantity Recreation
  15. Care Daily Activities
  16. Safety
  17. Freedom
  18. Training
  19. Integration
  20. General Care


In the identity domain 23% of people in campus houses were rated as not receiving adequate care compared with 3% in dispersed housing. This meant they had less structured living situations, received less preparation for changes and did not always have a special private place in the house. They received less training in handling emotions, handicaps and problems, had less chance to decorate their own room and to own a key to the house.

For housekeeping, 22% of people in campus houses were rated as not receiving adequate care compared with 7% in dispersed housing, for example having less self-determination and training in housekeeping.

In terms of relationships with others, 47% of people in campus houses were rated as not receiving adequate care compared with 17% in dispersed housing. They had fewer training opportunities in social skills and less freedom in relation to meeting others.

In recreation, 27% of people in campus houses were rated as not receiving adequate care compared with 11% in dispersed housing because they took part less frequently in recreational activities in and outside the house, had less freedom to participate in recreational activities, less training in and stimulation of hobbies and talents.

The proportion of residents in campus houses rated as not receiving adequate care in daily activities was 21% compared with 13% in dispersed housing. This reflected fewer training opportunities, lack of suitable guidance and less freedom of choice.

Table 4 Summary of relevant findings from Janssen et al (1999)

Measure

Dispersed housing better

No difference

Campus/cluster housing better

Physical Domain

Y1

  
Adapted Housing 

Y1

 
Acceptance, sexual activity 

Y1

 
Care Physical Health 

Y2

 
Care Fitness 

Y2

 
Freedom eating/ drinking

Y2

  
Care mental health 

Y2

 
Identity

Y1

Y2

 
Vision/Religion

Y1

Y2

 
Housing

Y1

Y2

 
Property/Belongings

Y1

Y2

 
Housekeeping

Y1

Y2

 
Living environment 

Y2

 
Relation Family 

Y2

 
Relation Caretaker

Y1

Y2

 
Relation Others

Y1

Y2

 
Recreation

Y1

Y2

 
Daily Activities

Y1

Y2

 
Safety

Y1

Y2

 
Freedom

Y3

  
Training

Y3

  
Integration

Y3

  
General Care

Y1

Y2

 
  1. Rated by staff
  2. Rated by service users or families
  3. Rated by both staff and service users or families


As rated by staff, significantly poorer results were also found in respect of vision/religion, housing, property/belongings, relations with caretaker, safety and general care. The two domains in which there was no significant difference were the provision of adapted housing and 'acceptance and sexual activity'.

Service users and families responding to the second questionnaire generally rated quality of life slightly higher than staff. In 16 of 20 domains there was no significant difference between campus or dispersed housing. The four domains where there was a difference were less freedom in eating and drinking (23% residents in campus houses rated as not receiving adequate freedom compared with 13% in dispersed housing), freedom in relationships (21% residents in campus houses rated as not receiving adequate freedom compared with 16% in dispersed housing), training (11% residents in campus houses rated as not receiving adequate care compared with 3% in dispersed housing) and integration (21% residents in campus houses rated as not receiving care compared with 9% in dispersed housing).

Strengths and weaknesses

The study provides evidence at level 3a on the CRD scale. It is a cohort study involving a cross-sectional comparison. The measures used are quantitative and reported as robust though evidence of this is not presented. The measure has not been widely used and it is therefore more difficult to compare findings from this study with others.

A strength of the study is that it was designed to compare different housing models and the groups of residents were comparable on the characteristics studied. Overall staff ratios were also comparable. The weaknesses of the study are that it focuses only on services for people with mild or moderate intellectual disability and it does not include any information on costs. Nevertheless it is a relatively comprehensive and effective comparison of many aspects of quality of life.

Conclusion

The study suggests that dispersed housing is generally superior to campus housing when ratings are made of many aspects of quality of life by staff, and that there is no difference in ratings made in most domains by service users or families. There are no domains in which campus houses were rated by staff or service users or carers as superior to dispersed housing.

6. Emerson et al (2000b), 7. Emerson et al (2000c); 8. Robertson et al (2000b); 9. Emerson et al (2000a), 10. Robertson et al (2000a), 11. Robertson et al (2001), 12. Gregory et al (2001); 13. Walsh et al (2001); 14. Hallam et al (2002)

There are nine papers published drawing on data from this study comparing quality and costs in English National Health Service campuses, village communities and dispersed housing (including small and large group homes and supported living) in England and Ireland. In the main study (Emerson et al., 2000b), information was available for 500 people with intellectual disabilities: 133 in five newly built residential campuses ranging from 94 to 144 places; 86 people in three village communities in England; and 281 people in 10 dispersed housing schemes.

The study was a cross-sectional group comparison. As well as reporting data for the whole group, Emerson et al (2000b) also present data on 81 matched pairs of people living in village communities and dispersed housing, and 121 pairs living in residential campuses and dispersed housing. Emerson et al (2000c); presents data on two sub groups of 20 people with severe and complex needs in residential campuses and dispersed housing. Emerson et al (2000a) present information about the management of challenging behaviour. Gregory et al (2001) present data on a subgroup of residents interviewed. Walsh et al (2001) included data from relatives of people living in village communities in Ireland in their paper. Hallam et al (2002) present data on costs of the different types of setting.

Information about the residential setting was collected by interviewing managers or key informants and through ratings by the research team and included the Residential Services Setting Questionnaire (Emerson et al., 1995), the Architectural Features Scale (Thompson et al., 1990), the Residential Services Working Practices Scale (Felce et al., 1995), the Group Home Management Interview (Pratt et al., 1980), information about participants and specific support (by questionnaire and interview with staff), Adaptive Behavior Scale Part 1 (Nihira et al., 1993), the Aberrant Behavior Checklist (Aman et al., 1995), the PAS-ADD (Moss et al., 1996, Moss et al., 1998), the Autism Screening Questionnaire (Howlin, 1996), the Index of Community Involvement (Raynes et al., 1987), the Social Network Map (Tracy and Whittaker, 1990, Tracy and Abell, 1994), selected items from the Health Survey for England (Bennett et al., 1995), the Thameside and Glossop Health Needs Survey (Turner, 1997), the Choice Scale and the Risks Scale (Emerson et al., 1999), views of participants obtained by semi-structured interviews based on Azmi et al (1997) and Mason et al (1997), and views of relatives obtained through a postal questionnaire based on that used by Hatton et al (1995). Observational measures of engagement in meaningful activity, stereotypy, challenging behaviour, staff contact and assistance were also taken.

Cost information was obtained using the Revised Client Service Receipt Inventory (Beecham, 1995, Beecham and Knapp, 1992) with each participant, the Residential Services Setting Questionnaire, description of service and facility accounts, direct staffing costs, non-staffing costs, administration of professional staff over site and central office, costs of buildings-based services and cost of services received independently of accommodation, using national unit cost data (Netten et al., 1998).

Generally, these studies used analysis of variance with Mann-Whitney posthoc tests to identify group differences and logistic regression to identify predictors of outcome. Where there were pre-existing differences in service user characteristics between groups being compared, the potentially confounding variables were entered first as covariates.

Findings

The findings are presented separately for each paper.

6. Emerson et al (2000b)

Comparability

Taking the whole sample, people in village communities had higher levels of adaptive behaviour than people in dispersed housing who in turn had higher level of adaptive behaviour than those in NHS campuses. People in NHS campuses had higher levels of challenging behaviour. The first matched group of 81 pairs of people living in either village communities or dispersed housing was matched on ABS Part 1 score but was a relatively able group (scoring nearly 70% of the maximum possible on the ABS). The second group of 121 pairs was matched on both the ABS Part One (scoring 39% of maximum possible) and the Aberrant Behavior Checklist (campus houses average score 29.3, dispersed housing 28.9). Data from the matched groups is only presented where the dependent variables in the analysis of variance were significantly associated with either adaptive or challenging behaviour.

After taking account of differences in characteristics of people living in the different settings, there were higher overall staffing ratios (of all staff employed over 24 hours) in dispersed housing (1.7 staff per resident) compared to campus houses (1.3:1) but the difference between dispersed housing and village communities (1.4:1) was not statistically significant. Taking only senior staff, dispersed housing had significantly higher staff ratios (0.4:1) than campus houses (0.3:1) or village communities (0.3:1).

Results

Results are summarised in Table 5. Dispersed housing was significantly more homely than campus housing or village communities. It was better than campus housing in terms of block treatment, depersonalisation, rigidity of routine and social distance between staff and residents. It was better than village communities in terms of block treatment and depersonalisation, the same in terms of social distance and worse in terms of rigidity of routine.

Village communities were better than dispersed housing which was better than residential campuses at person-centred planning, activity planning and staff training and supervision. Campus houses and village communities were better than dispersed housing at procedures for assessment and teaching. Settings did not differ on planning staff support to residents.

There were no differences between service types in contact with a community physician. People in village communities were more likely to have contact with dentists than people in campus or dispersed housing. People in campus housing were more likely than those in either other model to have contact with psychiatrists and psychologists. People in village communities were more likely to have had a general health check and a vision test than people in campus housing, who were more likely than those in dispersed housing. People in village communities were also more likely than those in dispersed housing to have had blood pressure and testicular checks in the last year. Differences between service models in cervical smears, hearing checks and mammograms were not statistically significant. More people received anti-psychotic and anti-depressant medication in campus houses than in village communities or dispersed housing. There was no difference in numbers of people using anxiolytics or hypnotics.

Although for the full sample, people living in dispersed housing had more choice than those in campus housing, the difference between them and those in village communities was not statistically significant for the matched sample. This implies that the difference in the full sample is likely to reflect differences in resident characteristics.

People living in dispersed housing and village communities had more extensive social networks than those in campus houses. Although people living in dispersed housing had more people with intellectual disabilities in their social networks than those in campus housing, the difference between them and those in village communities was not statistically significant for the matched sample. There were no differences between types of setting in the number of family members or staff in the social networks of residents, though people in dispersed housing had more local people in their social network.

People in dispersed housing took part in a greater number of recreational and community activities than those in campus houses or village communities. People in dispersed housing and village communities had more variety of activity than those in campus houses. People in village communities had more hours per week of scheduled activity than those in dispersed housing, who had more than those in campus houses.

Men in dispersed housing were less likely to be physically inactive than those in campus houses. There was no difference between dispersed housing and village communities, nor between any settings in terms of underweight, obesity, women inactive or number of people experiencing accidents.

People in dispersed or campus housing were more likely to have been victims of crime or experience verbal abuse by members of the public than people in village communities. There were no differences between settings on physical or sexual abuse in the home or on whether the home had been vandalised. People in campus housing were perceived to be more at risk of accidents in the home than those in village communities; people in dispersed housing did not differ from either group. People in dispersed housing were perceived to be more at risk of exploitation by people in the local community than those living in either campus houses or village communities. There were no differences between settings on other types of perceived risk.

Views of participants could be obtained from sufficient people in village communities and dispersed housing to permit analysis. There were no significant differences between the groups. Relatives' satisfaction with the quality of support did not differ between types of setting.

For the full sample, costs of dispersed housing were higher than for campus housing which were higher than for village communities. However, comparison of the matched samples found that the same pattern of differences was not statistically significant. This implies that cost differences in the full sample were due to differences in resident characteristics (see page 31).

Table 5 Summary of relevant findings from Emerson et al (2000b)

Measure

Dispersed housing better

No difference

Campus/cluster housing better

Homelikeness

Y

  
Block treatment

Y

  
Depersonalisation

Y

  
Rigidity of routines

Y2

 

Y1

Social distance

Y2

Y1

 
Person-centred planning

Y2

 

Y1

Activity planning

Y2

 

Y1

Staff training and supervision

Y2

 

Y1

Assessment and teaching  

Y

Planning staff support 

Y

 
Contact with community physician 

Y

 
Contact with dentist  

Y1

Contact with psychiatrist  

Y2

Contact with psychologist  

Y2

General health check  

Y

Blood pressure check 

Y2

Y1

Vision test  

Y

Hearing test 

Y

 
Mammogram 

Y

 
Cervical smear 

Y

 
Testicular cancer 

Y2

Y1

Anti-psychotic medication

Y2

Y1

 
Anti-depressant medication

Y2

Y1

 
Anxiolytic medication 

Y

 
Hypnotic medication 

Y

 
Choice

Y2

Y1

 
Number of people in social network

Y2

Y1

 
People with intellectual disability in network

Y2

Y1

 
Family members or staff in network 

Y

 
Local people in network

Y

  
Number recreational/ community activities

Y

  
Variety recreational/ community activities

Y2

Y1

 
Hours recreational/ community activities

Y2

 

Y1

Physical inactivity (men)

Y2

Y1

 
Physical inactivity (women) 

Y

 
Obesity 

Y

 
Underweight 

Y

 
Accidents 

Y

 
Victim of crime 

Y2

Y1

Verbal abuse 

Y2

Y1

Physical or sexual abuse 

Y

 
Vandalism 

Y

 
Perceived risk of accidents 

Y

 
Perceived risk of exploitation by members of local community  

Y

Other perceived risks 

Y

 
Participants' satisfaction 

Y1

 
Relatives satisfaction 

Y

 
Costs  

Y

  1. Comparison between dispersed housing and village communities
  2. Comparison between dispersed housing and campus houses


7. Emerson et al (2000c)

Comparability

This is a sub-study from the larger sample reported, which focused on costs and quality of community based residential supports and residential campuses for people with severe and complex disabilities. Two groups of 20 people with severe and complex needs were identified from dispersed and campus housing. People were selected if they were under 56 years of age, were nonverbal or nearly non-verbal and had low scores on a range of ABS items indicating severe and complex disability. The groups obtained matched well on age, adaptive behaviour, autism and mental health but those in campus houses had more challenging behaviour, had been living in the current setting for longer and were more likely to have lived in hospital.

Dispersed housing had higher staff ratios than campus housing overall and for senior and care staff separately.

Results

People in dispersed housing lived with fewer other people with intellectual disabilities than those in campus housing and their houses were more homely.

Dispersed housing was better than campus housing in respect of all four domains of social climate (depersonalisation, rigidity of routine, social distance and block treatment). There was no difference in person-centred planning or in procedures for assessment and teaching but dispersed housing was better at activity planning, planning staff support to residents and training and supervision of staff.

There was no difference between dispersed or campus housing in the amount of staff contact people received, either overall or separately as assistance, processing or other contact.

Table 6 Summary of relevant findings from Emerson et al (2000c)

Measure

Dispersed housing better

No difference

Campus/cluster housing better

Number co-residents

Y

  
Homelikeness

Y

  
Block treatment

Y

  
Depersonalisation

Y

  
Rigidity of routines

Y

  
Social distance

Y

  
Person-centred planning 

Y

 
Activity planning

Y

  
Staff training and supervision

Y

  
Assessment and teaching 

Y

 
Planning staff support

Y

  
Contact from staff 

Y

 
Choice

Y

  
Number of people in network

Y

  
Contact from other residents 

Y

 
Contact from visitors/others 

Y

 
Contact with family 

Y

 
Composition of network 

Y

 
Number community activities

Y

  
Variety community activities

Y

  
Hours of scheduled activity 

Y

 
Engagement in meaningful activity 

Y

 
Stereotypy

Y

  
Costs  

Y


People in dispersed housing had greater choice than those in campus houses and wider social networks. There were no significant differences in contact received from other residents or visitors or others during observations, contact with family in the last three months or composition of the social network. People in dispersed housing had a greater number and variety of community activities than those in campus housing, but there was no difference in the average number of hours of activity scheduled in the week. There was also no difference in the level of engagement in meaningful activity, but people in campus houses spent more of their time in stereotyped behaviour than those in dispersed houses.

The authors note that results show great variability within the same type of service and therefore conducted regression on the whole sample to identify predictors of engagement in meaningful activity. Higher engagement in meaningful activity was predicted by ability on domestic tasks and by assistance from staff.

Total costs of dispersed housing were higher than those of campus housing. This difference was entirely due to higher staffing ratios in the dispersed housing.

8. Robertson et al (2000b)

Comparability

This study focused on health outcomes for people in dispersed housing, campus housing and village communities. It used the whole sample (so the groups were not matched) but excluded people who were not physically capable of walking alone. This excluded 22% of people in dispersed housing, 28% in campus houses but only 2% in village communities. No information is presented on the characteristics of the sample actually used in this study.

Results

This study found that the only difference between settings in terms of diet was that fewer people in village communities met the criterion for adequate consumption of dairy products and fewer had a poor or fatty diet.

In contrast to the finding presented for the whole sample of 500 by Emerson et al (2000b), this paper did not find significant differences in the percentage of men defined as inactive. It did find that people in residential campuses were significantly more likely to be inactive than in village communities, where people were in turn more likely to be inactive than in dispersed housing. This difference was explained by the findings for women. There were significant differences in the percentage of people taking vigorous activity but post-hoc tests are not provided.

Table 7 Summary of relevant findings from Robertson et al (2000b)

Measure

Dispersed housing better

No difference

Campus/cluster housing better

Consumption dairy products

Y1

Y2

 
Poor/fatty diet 

Y2

Y1

Inactivity

Y

  
Drinking/smoking health risk 

Y

 
  1. Comparison between dispersed housing and village communities
  2. Comparison between dispersed housing and campus houses


People in campus houses were less likely to drink alcohol than those in either dispersed housing or village communities. Levels of alcohol consumption were notably lower than those of the general population and did not exceed recommended levels. There were no differences between settings in either smoking or drinking as health risks.

9. Emerson et al (2000a)

Comparability

This study focused on the treatment and management of challenging behaviour for people in dispersed housing, campus housing and village communities. It reports some data for the whole sample and then for people recorded as either having one 'moderately serious' or 'severe' form of challenging behaviour in the previous month. The number of people so described living in different types of setting was different, with most living in campus houses and least in village communities. When only these people were included there were no differences between settings in the form or nature of challenging behaviour shown. The paper uses multiple logistic regression on all the people identified as having 'moderately serious' or 'severe' challenging behaviour to identify predictors of various outcomes.

Results

The only findings where setting type was predictive were that people in campus houses were more likely to be subject to physical restraint; people in 'supported living' placements were more likely to have a written intervention programme; and people living in group homes were less likely to have an individual programme plan goal addressing challenging behaviour.

Table 8 Summary of relevant findings from Emerson et al (2000a)

Measure

Dispersed housing better

No difference

Campus/cluster housing better

Subject to restraint

Y

  
Written intervention programme

Y1

Y2

 
Individual programme plan goal 

Y1

Y2

  1. Comparison between supported living and campus/cluster housing only
  2. Comparison between group homes and campus/cluster housing only


10. Robertson et al (2000a)

Comparability

This study presents findings on the use of psychotropic medication in the different types of residential setting, using the total sample as in Emerson et al (2000b). The groups are therefore not comparable.

Results

In addition to the results reported in the main study, this study found that people in campus houses were more likely than those in dispersed houses or village communities to receive both antipsychotics and antidepressants and to receive more than one type of antipsychotic on a regular basis. Other drug combinations were not significantly different across settings.

Table 9 Summary of relevant findings from Robertson et al (2000a)

Measure

Dispersed housing better

No difference

Campus/cluster housing better

Receive both antipsychotics and antidepressants

Y2

Y1

 
More than one antipsychotic

Y2

Y1

 
  1. Comparison between dispersed housing and village communities
  2. Comparison between dispersed housing and campus houses


11. Robertson et al (2001)

Comparability

This study presents findings on the social networks of people living in the different types of residential setting, using the total sample as in Emerson et al (2000b). The groups are therefore not comparable. The paper uses logistic regression to identify predictors of various outcomes.

Results

In addition to the results reported in the main study, this paper reports some findings relating to type of residential setting. People living in dispersed housing for three or fewer people had larger social networks and were more likely to have a person who was not a staff member, a relative or person with intellectual disability in their social network.

Table 10 Summary of relevant findings from Robertson et al (2001)

Measure

Dispersed housing better

No difference

Campus/cluster housing better

Size of social network

Y

  
Member of public in network

Y

  


12. Gregory et al (2001)

Comparability

This study presents results of interviews with 45 residents of village communities and 51 residents of dispersed housing. There were no statistically significant differences between these groups in terms of resident ability.

Results

People living in village communities were more satisfied with friendships and relationships than those in dispersed housing. There was no difference between the groups in satisfaction with accommodation, choice, social and leisure activities, risks, day activities and the support received.

Table 11 Summary of relevant findings from Gregory et al (2001)

Measure

Dispersed housing better

No difference

Campus/cluster housing better

Satisfaction with accommodation 

Y1

 
Choice 

Y1

 
Satisfaction with social and leisure activities 

Y1

 
Satisfaction with friendships and relationships  

Y1

Satisfaction with risks 

Y1

 
Satisfaction with day activities 

Y1

 
Satisfaction with support received 

Y1

 
  1. Comparison between dispersed housing and village communities


13. Walsh et al (2001)

Comparability

This paper extends the study reported in Emerson et al (2000b) by including an additional 60 people with intellectual disabilities in dispersed housing and 65 in residential campuses in Ireland to explore the family views of the quality of residential supports. Only people who had had family contact in the previous six months were approached and data were obtained from the families of 81 Irish residents and 210 UK residents, 130 in dispersed housing and 161 in campus or village settings. The only data presented to judge comparability of the groups living in different settings is that there was no difference in distance from the family home.


Results

Families of residents in campus/village settings reported higher level of contact in the previous three months. There was no significant difference in satisfaction between settings over the whole sample.

Table 12 Summary of relevant findings from Walsh et al (2001)

Measure

Dispersed housing better

No difference

Campus/cluster housing better

Family contact  

Y

Family satisfaction 

Y

 


14. Hallam et al (2002)

Comparability

This study provides more detailed cost comparisons using the same samples as presented in Emerson et al (2000b). These samples are not, therefore, comparable in terms of resident characteristics.

Results

Total costs of village communities were less than campus housing or dispersed housing. This differs from the finding in Emerson et al (2000b) that the overall costs of village communities were less than those of campus houses which were less than those of dispersed housing. The reason for this appears to be that Emerson et al adjusted costs to take account of differences in adaptive and challenging behaviour.

Table 13 Summary of relevant findings from Hallam et al (2002)

Measure

Dispersed housing better

No difference

Campus/cluster housing better

Cost 

Y2

Y1

  1. Comparison between dispersed housing and village communities
  2. Comparison between dispersed housing and campus houses


Strengths and weaknesses

The strengths of this group of studies are that they are comparative observational studies using a large number of quantitative, widely used and robust measures, studying services identified as representing among the best practice of each type. The studies provide evidence at level 3a on the CRD scale. The study included comprehensive costing and like-for-like comparison.

The authors note that the services are not representative, because they were selected as examples of good practice; that they are drawn from relatively few organisations and that there are a large number of comparisons. A further weakness of the study is that the groups were not comparable and that it was necessary to try to control for this difference statistically, though three smaller matched comparison studies were carried out. Data from the Irish component of the project appears only to have been used to comment on family views (in Walsh et al (2001).

Conclusion

In general, these studies show that campus houses perform less well on the overwhelming majority of measures of inputs, processes and quality outcomes studied. These differences cannot be accounted for by differences in the characteristics of people supported. Differences between dispersed housing and village communities are more mixed, with different benefits for each model. Although there are distinct differences noted in terms of the profile of people living in village communities and dispersed housing (village communities serving people who, on average, have higher levels of adaptive behaviour and lower levels of challenging behaviour), the different benefits of each model cannot be explained by differences in the characteristics of the residents. However, it is relevant to note that village communities only served people with relatively high levels of adaptive behaviour and low levels of challenging behaviour.

Although overall dispersed housing was 15% more expensive than campus housing and 20% more expensive than village communities, this was only true for the whole sample (where adaptive behaviour and challenging behaviour differed between settings). When matched samples were compared in Emerson et al (2000b) there were no statistically significant differences in cost; Emerson et al (2000c) did find differences, but these were attributable to differences in staff ratio. Costs largely reflect staff ratios in the different kinds of service and if staff ratios match resident needs there appears to be no difference in cost in this study.

15. Emerson (2004)

Emerson (2004) reports a comparison between people with intellectual disabilities living in dispersed housing and cluster housing. From an audit of 1542 people living in residential services in the north west of England, full data was available on 910 people, of whom 169 were in cluster housing and 741 were in dispersed housing.

The information collected included participant and setting characteristics (Bliss et al., 1999), the Learning Disability Casemix Scale (LDCS) (Pendaries, 1997), an extended version of the Index of Community Involvement (Raynes et al., 1987), Body Mass Index (BMI) and the Physical Activity Scale from the Health Survey for England (Bennett et al., 1995).

The data were analysed using analysis of variance, controlling statistically for age, gender, ability, challenging behaviour, mental health status and prior residential history. The study also used logistic regression analysis and PLUM ordinal regression. Differences are reported here if they are statistically significant at p<0.05. The paper also reports several findings that, though not statistically significant, are very large (more than 50% more likely or frequent in one setting than another). For example, the odds of being in any form of employment were 1.7 times as great for people living in dispersed houses than for people in clustered housing. Further details of these results are given in the original paper.

Findings

Comparability

Compared with people living in dispersed housing, people in cluster housing were younger, more disabled and with more challenging behaviour; were more likely to have moved from home/school and less likely to have moved from a hostel or another cluster. Dispersed housing had higher staff ratios than cluster housing.

Results

People living in dispersed housing lived with fewer other people, were less likely to have their home used for short-term care of other people or be supported by casual staff. There was no difference between dispersed and cluster housing in terms of staff turnover.

People living in dispersed housing were more likely to have an individual plan, but people in cluster housing were more likely to have a keyworker.

There were no differences in health checks, vision and hearing tests or mammograms but people in cluster housing were more likely to have seen the dentist in the last year.

There were no differences in the proportion of people who were overweight or obese or physically inactive. People in dispersed housing were more likely to undertake moderate or vigorous activity (exercise) and people in cluster housing were more likely to be underweight.

There were no differences in whether people had a written treatment plan for challenging behaviour, use of antipsychotic, anxiolytic or antidepressant medication. People in campus houses had higher levels of hypnotic medication and of more than two antipsychotics (polypharmacy). They were also more likely to experience seclusion.

People in dispersed housing took part in more community-based activities, more and a greater variety of leisure, social and friendship activities than those in cluster housing. There were no differences in contact with family, or friends or in participation in social activities.

Strengths and weaknesses

The strengths of this study is that it offers a clear comparison between cluster housing and dispersed housing, using a number of quantitative, widely used and robust measures. The study provides evidence at level 3a on the CRD scale.

The authors point out that participants cannot be assumed to represent all adults with intellectual disabilities receiving support in England (participants were marginally older, living in slightly smaller units than national data). There are no within-study checks of reliability or validity of data collection. The study does not address costs of services.

Conclusion

As in other studies, the populations in the two groups were not completely comparable and it was necessary to control for these differences statistically. Once this is done, important differences emerge. People living in cluster housing were more likely to live in larger settings, with fewer staff, be exposed to greater inconsistency in living arrangements (their home also being used for short-term care for others and greater use of casual staff), be exposed to more restrictive behaviour management practices (seclusion, hypnotic medication, polypharmacy), lead more sedentary lives, be underweight, and participate in fewer and a more restricted range of leisure, social and friendship activities. The only benefits of cluster housing were in the provision of a keyworker and visits to the dentist.

16. McConkey, Walsh-Gallagher and Sinclair (2005)

McConkey, Walsh-Gallagher and Sinclair (2005) compared people's involvement with their families and the local community in campus settings and in community housing in Ireland. Participants were 55 people living in 11 bungalows on a campus setting and 51 people in 10 dispersed houses. All the services studied were run by the same organisation.

Information collected in the study included demographic information about the participants, Index of Social Competence (McConkey and Walsh, 1982), information on frequency of challenging behaviours, the Life Experiences Checklist (Ager, 1998) and information about the range and frequency of contacts the person had with their family. No information on validity or inter-rater reliability is presented in the study.

Analysis of variance was used to identify differences between groups, with binary logistic regression carried out on the whole sample to provide further information about relationships between variables.

Table 14 Summary of relevant findings from Emerson (2004)

Measure

Dispersed housing better

No difference

Campus/cluster housing better

Individual plan

Y

  
Keyworker  

Y

Number co-tenants

Y

  
Home used for short-term care

Y

  
Use of casual staff

Y

  
Staff turnover 

Y

 
Work experience 

Y

 
Voluntary work 

Y

 
Adult education 

Y

 
Day centre 

Y

 
Hours per week at day programme 

Y

 
Health check 

Y

 
Dental check  

Y

Hearing test 

Y

 
Vision test 

Y

 
Mammogram 

Y

 
Underweight

Y

  
Overweight 

Y

 
Obesity 

Y

 
Physical inactivity 

Y

 
Exercise

Y

  
Treatment programme for challenging behaviour 

Y

 
Antipsychotic medication 

Y

 
Antidepressant medication 

Y

 
Anxiolytic medication 

Y

 
Hypnotic medication

Y

  
More than two antipsychotics

Y

  
Seclusion for challenging behaviour

Y

  
Community-based activities

Y

  
Number of leisure, social and friendship activities

Y

  
Variety of leisure, social and friendship activities

Y

  
Contact with family 

Y

 
Contact with friends with intellectual disabilities 

Y

 
Contact with friends without intellectual disabilities 

Y

 
Participation in social activities 

Y

 


Findings

Comparability

The two groups of participants were quite different. The people living in campus houses were significantly more disabled: they had been in residential care for longer, had lower adaptive behaviour, had more epilepsy, had more challenging behaviour and had more medication.

Results

On all five domains of the Life Experiences Checklist (Home, Freedom, Opportunities, Leisure and Relationships), people in dispersed housing scored higher than those in campus houses. Scores for people living in dispersed housing were comparable and some were higher than for a sample of the British general population. Given that the groups were not comparable, binary logistic regression was used to explore which variables predicted scores on the Opportunities and Freedom domains of the Life Experiences Checklist. These were the only domains which distinguished the two groups of residents in the regression analysis - those living in dispersed housing and those in campus housing. The only significant predictor was that people taking a larger number of medications were more likely to have lower scores.

People in dispersed housing had more telephone contact and support from families and participated in more family events than those in campus houses. Binary logistic regression found only one significant predictor of family involvement: residents who were more dependent for personal care had less involvement with family members.

Table 15 Summary of relevant findings from McConkey, Walsh-Gallagher and Sinclair (2005)

Measure

Dispersed housing better

No difference

Campus/cluster housing better

Life experiences

Y

  
Family contact

Y

  


Strengths and weaknesses

The study provides evidence at CRD level 3a as a cohort study. The measures used have been validated. Costs are not addressed. That the services were all managed by one organisation removes this is a source of variability. However the major weakness in the study is that the groups were not comparable and no statistical or matching process was used to control for this. The results of binary logistic regression show that type of housing was not a predictor of the different results obtained.

Conclusion

Although this study found that people living in dispersed housing had better life experiences and family contact than those in campus housing, the logistic regression analysis shows that these differences are attributable to differences in participant characteristics.

17. Young (2006)

Young (2006) compared two matched groups of 30 people with intellectual disability living in dispersed and cluster housing in and around Brisbane, Australia. Both groups of people had moved out of hospital settings at the beginning of the study and so longitudinal data are also available at 12 months and 24 months after the move.

The study used a matched group comparison design. Information was collected on the adaptive and challenging behaviour of participants using the Adaptive Behavior Scale. Information was also collected on choice-making by participants, using the Resident Choice Assessment Scale (Kearney et al., 1995) and on quality of life using the Life Circumstances Questionnaire (Ashman et al., 1991). These are instruments of demonstrated validity and inter-rater reliability was checked during the study.

The data were analysed using analysis of variance.

Findings

Comparability

The groups were matched post-hoc individual by individual for age, gender, level of intellectual disability, other disabilities, challenging behaviour and length of time in institutional care. The sample included people with profound intellectual disabilities.

Results

This study reports three sets of comparisons: (i) changes over time for each group (clustered and dispersed housing), (ii) differences between the two groups at the middle and end of the study and (iii) differences between the groups in the amount of change over time.

People in both dispersed and cluster housing increased their adaptive behaviour over time after transfer from the institution. People in dispersed housing increased their adaptive behaviour in five domains ('Numbers and time', 'Domestic activity', 'Pre-vocational and vocational activity', 'Responsibility' and 'Socialisation'). People in cluster houses showed increases only in 'Pre-vocational and vocational activity' and improvements in 'Physical development'.

There was no change over time in either dispersed or cluster housing in the overall score for maladaptive behaviour. People in dispersed housing improved their scores in Trustworthiness and Sexual behaviour whereas people in cluster housing showed no change in any domain.

Both groups improved over time in choice making but people in dispersed housing improved at 12 and again at 24 months whereas those in cluster housing improved at 12 months but then showed a decline at 24 months.

Both groups improved their scores on the Life Circumstances Questionnaire over time at 24 months on the overall score and on all eight individual domains. Both groups increased contact and social interaction with family and friends.

In terms of comparisons between groups, there was no statistically significant difference in overall scores for adaptive or maladaptive behaviour at 24 months, though people in dispersed housing had significantly higher scores than those in cluster housing for 'Domestic activity' and 'Responsibility'. People in dispersed housing had significantly more opportunities for choice making than those in cluster housing. There were no differences between the groups in contact and social interaction with family and friends.


Comparing the extent of change over time, increases were less for the cluster group than for the dispersed housing group in the Pre-vocational and vocational and Responsibility domains of the Adaptive Behavior Scale. On the Life Circumstances Questionnaire, people in dispersed housing improved more on six domains: Community access, Routines, Self-determination, Residential well-being, General life improvements and Physical well-being. There was no difference in improvement in Material well-being or Social-emotional well-being. People in campus houses did not improve more in any area.

Table 16 summarises comparisons between groups at 24 months and comparisons of the extent of change over time.

Table 16 Summary of relevant findings from Young (2006)

Measure

Dispersed housing better

No difference

Campus/cluster housing better

Change in adaptive behaviour overall 

Y

 
Independent Function 

Y

 
Physical Development 

Y

 
Economic Activity 

Y

 
Language Development 

Y

 
Numbers and Time 

Y

 
Domestic Activity

Y

  
Prevocational/Vocational

Y

  
Activity 

Y

 
Self-direction 

Y

 
Responsibility

Y

  
Socialization 

Y

 
Change in maladaptive behaviour overall 

Y

 
Social behaviour 

Y

 
Conformity 

Y

 
Trustworthiness

Y

  
Stereotyped/hyperactive behaviour 

Y

 
Sexual behaviour 

Y

 
Self-abusive behaviour 

Y

 
Social engagement 

Y

 
Disturbing interpersonal behaviour 

Y

 
Choice

Y

  
Change in life circumstances overall

Y

  
Material Well-being 

Y

 
Physical Well-being

Y

  
Community Access

Y

  
Routines

Y

  
Self-determination

Y

  
Social-emotional Well-being 

Y

 
Residential Well-being

Y

  
General

Y

  


Strengths and weaknesses

This study provides evidence at level 3a on the CRD scale. Its strength is that it is a comparative cohort study with well matched groups using robust quantitative measures. It also includes people with severe and profound intellectual disabilities and is longitudinal over 24 months.

The authors note as weaknesses the lack of random allocation to the dispersed or cluster housing groups, practical delays and obstacles in doing the research and the possibility that some declines were age related for the cluster housing group. The study does not address costs.

Conclusion

People moving to dispersed housing increased their scores on more domains of the Adaptive Behavior Scale than those moving to cluster housing. At 24 months, the two groups differed on two domains (Domestic Activity and Responsibility). For clarity, only differences at 24 months (not 12 months) are reported in Table 16.

This study produces a robust finding that dispersed housing is superior to cluster housing in

  • Acquiring skills in domestic tasks (including cleaning, laundry, table setting, food preparation and other routine household chores) and responsibility
  • Choice-making,
  • Physical well-being
  • Community access
  • Routines
  • Self-determination
  • Residential well-being
  • General life circumstances

The benefits of change after moving out of institutional care were also more widespread for people in dispersed housing than in cluster housing. There are no measures on which cluster housing is superior to dispersed housing.

18. Owen, Hubert and Hollins (2007)

Owen, Hubert and Hollins (2007) present a small scale qualitative study exploring the experiences of 11 women who moved from a locked hospital ward. Eight moved to a new purpose-built home on the campus of the hospital and three to dispersed houses in the community. The women were aged between 29 and 72; all had intellectual disabilities, 'most' had severe intellectual disabilities, 'several' were on the autistic spectrum and 'some' had mental health problems. All had challenging behaviour.

The methodology involved over 300 hours of participant observation and in-depth interviewing, visiting the ward and campus house for about five hours at different times of the day and night every two weeks over 18 months and the three dispersed houses for five to seven hours at different times of the day every three months for a year. No account is given of procedures to check validity or reliability of the data. Analysis was a grounded theory approach involving identifying and coding themes and returning to collect more data to check assumptions.

Findings

Comparability

No information is presented on the comparability of the women living in the campus house versus those in dispersed housing.

Results

The women who moved to the campus house experienced more chaos and confusion, including an unexpected move to temporary accommodation (trans-institutionalisation) for 10 months. Staff known to residents from hospital left and new temporary staff were used. This resulted in a decrease in daily activities. The women who moved to dispersed housing had less traumatic experiences, but experienced more losses overall, including relationships with former staff (who understood communication, needs and preferences), relationships with other women that they previously lived with for many years, loss of comfort and aloneness in new surroundings. One woman lost her relationship with her advocate. Overall the experience was negative for all women - they were not involved in important aspects of moving with little preparation or information about the move. There were also problems with a lack of information transfer from staff on wards.

For the majority of the women, their quality of life changed very little in the new homes. One woman in dispersed housing was able to explore new places in the community, exercise more choice and learn new skills. The two other women in dispersed housing were able to take more control over several aspects of their lives but in general they were unable to explore their own interests/abilities and preferences as this was dictated by what others in the home were doing. Women in the campus house experienced least change for the better. They had fewer opportunities for daily activities and for relationships beyond staff than women moved to dispersed housing. Their lives continued to be restricted by rigid routines, strict rules, few activities and few opportunities for relationships.

Table 17 Summary of relevant findings from Owen, Hubert and Hollins (2007)

Measure

Dispersed housing better

No difference

Campus/cluster housing better

Chaos and confusion

Y

  
Loss of relationships, comfort  

Y

Control

Y

  
Daily activities

Y

  


Strengths and weaknesses

This study is not intended primarily as a comparison between campus and dispersed housing. Its strength is the detail of description of people's experience but, for the purposes of this review it is very small, there is no comparative data on individual needs and characteristics, and no basis for assuming representativeness or generalisability of the findings. There is also the problem of the confounding effect of the campus group having to move to temporary accommodation before the campus house was ready. The study is a case series and as such is CRD level 4.

Conclusion

Like some of the earlier small-scale studies this suggests that dispersed housing is superior to campus housing but the nature of the study make the finding of limited weight. The more important finding from the study is the very poor planning, organising and support of the people concerned through change in living circumstances.

19. McConkey et al (2007)

McConkey et al (2007) explores the variations in social inclusion of people with intellectual disabilities living in different types of residential accommodation in Northern Ireland and the Republic of Ireland. It included 620 people in five types of service:

  • Dispersed supported living (103 people)
  • Clustered supported living (132 people)
  • Small group homes (152 people)
  • Residential homes (138 people)
  • Campus settings (95 people)

McConkey et al describe the supported living settings studied as follows:

  • In dispersed supported living, the person holds a tenancy agreement for an ordinary house or apartment dispersed among other properties, either as an individual or shared with no more than two other persons. Support staff are provided according to assessed needs, and they visit on a regular basis.
  • In clustered supported living specially built groupings of houses or apartments on the same site have shared staffing across the houses. About 15 tenants may live in the same cluster with tenancy agreements, in either single-person or shared housing.

Information was collected on participant demographics and characteristics, including on 12 social competencies based on McConkey and Walsh (1982), on five defined challenging behaviours shown at least daily as well as serious behaviours that occur infrequently, on seven indicators of social inclusion as used in previous studies (Emerson and McVilly, 2004) and on use of community amenities in the past month (Raynes et al., 1994). The design was a group comparison. Analysis of variance with post-hoc tests was followed by binary logistic regression to identify predictors of outcome.

Findings

Comparability

There were no significant differences between the groups in terms of gender. People in both types of supported living schemes (dispersed and clustered) were broadly similar but had higher levels of social competence than those in residential homes or campuses. They also had fewer behaviour problems and less epilepsy. In campus houses people were younger, had lower levels of social competence and more epilepsy. People in group homes fell between supported living and campus housing in their characteristics.

Results

There were significant differences between settings on all seven social inclusion indicators (average number of community amenities visited in last month, visits to or from family, having friends outside accommodation, knowing neighbours by name, having had visitors in the past month, having stayed away from home and having had guests to stay) but there is no indication that dispersed or clustered supported living differed in any of them. People in campus housing used significantly fewer community amenities than those in supported living. No other information is given about the statistical significance of differences between campus housing and other models.

There were no differences between dispersed and clustered supported living in terms of use of individual community amenities.

Binary logistic regression showed that use of community facilities was predicted by social competence of residents and by type of residence. Taking campus housing at the reference point, people living in clustered supported living were just over four times more likely to use community facilities and people in dispersed supported living nearly six times more likely to use community facilities. Similarly, people in dispersed supported living were nearly nine times more likely than those in campus housing to have friends outside the home, and people in clustered supported living over 14 times more likely. People in dispersed supported living were nearly eight times more likely than those in campus housing to have visitors, and people in clustered supported living nearly 18 times more likely. Finally, people in dispersed supported living were just over three times more likely than those in campus housing not to be socially isolated (without contacts with family, friends or neighbours and to have had no visitors), and people in clustered supported living nearly six times more likely.

Table 18 Summary of relevant findings from McConkey et al (2007)1

Measure

Dispersed housing better

No difference

Campus/cluster housing better

Community amenities visited

Y

  
Visits to or from family 

Y

 
Friends

Y2

 

Y3

Knowing neighbours 

Y

 
Visitors

Y2

 

Y3

Stayed away 

Y

 
Guests to stay 

Y

 
Use of individual community amenities 

Y

 
Socially isolated

Y2

 

Y3

  1. In this table differences are reported where found through binary regression analysis, even when no difference is reported for group comparison
  2. Comparison between dispersed supported living and campus housing
  3. Comparison between dispersed supported living and cluster supported living


Strengths and weaknesses

The strength of this study is that it focuses on comparison of groups in dispersed and cluster housing that appear relatively well-matched. It is a cohort study at CRD evidence level 3a. The cluster housing is also of interest in that it is not a cluster of residential homes but a cluster of supported living placements. As such it may be more like housing dispersed over a small geographical area and supported by the same staff team, such as is found in Sweden.

The major weakness of the study is that the authors note that informants completing the questionnaire on social inclusion may have misinterpreted visitors and use of community facilities to include people and amenities in the cluster or campus. Costs are not addressed.

Conclusion

This paper shows that campus housing of the type provided in the Republic of Ireland produces worse outcomes for residents than other kinds of service. Cluster supported living shows benefits over dispersed supported living but it is impossible to know whether this is in part because other residents in the cluster were recorded as visitors. Nevertheless, the finding that residents in cluster supported living are more likely to have friends and visitors, even if these are from other houses in the cluster, is important.

Summary of the findings across studies

Table 19 presents all of the results of the studies reviewed, organised into the domains of the quality of life framework (Social inclusion, Interpersonal relations, Material well-being, Emotional well-being, Physical well-being, Self-determination, Personal development and Rights).

Table 19 Summary of results by domain

Quality of life domain/indicator

Dispersed housing better

No difference

Campus/ cluster housing better

Social inclusion
Social inclusion

5

  
Access to local community/ neighbourhood

2

  
Use of community facilities 

2, 19

 
Number of community amenities visited

19

  
Community activities and opportunities

15,16, 17

7

 
Residential well-being

17

  
Interpersonal relations
Interpersonal relations

16, 18

3, 17

 
Sexual activity 

5

 
Relationships with family, carers, others

56

57

 
Number of people in network

62, 7, 11

61

 
Composition of network 

7

 
Contact with family/family members in network 

6, 7, 15, 19

13

People with ID in network

62

61, 15

 
Local people in network

6

  
Contact with friends

192

15

195

Contact with neighbours 

19

 
Observed contact from others 

7

 
Stayed away/guest to stay 

19

 
Visitors to home

192

 

195

Material well-being
Material well-being

56

57, 17

 
Emotional well-being
Emotional well-being 

61

 
Challenging behaviour/stereotypy

7

3, 9

 
Satisfaction in all areas except friendships/relationships 

12

 
Satisfaction with friendships/relationships  

12

Chaos and confusion

18

  
Physical well-being
Physical well-being

56, 17

57

 
Physical activity

62 (men), 8, 15 (exercise)

61,62(women), 15 (inactivity)

 
Participation in domestic activities/engagement

17, 18

2, 7

 
Recreational/community activities

15, 16

  
Number of activities

6

  
Variety of activities

62,

61

 
Hours of recreational activities

62

 

61

Diet   
Poor/fatty diet 

82

81

Adequate dairy

81

82

 
Drinking/smoking health risk 

8

 
Underweight

15

6

 
Overweight/Obesity 

6, 15

 
Number of outings

3

  
Contact with health professionals

3

  
Contact with GP 

6

 
Contact with dentist  

61

Contact with psychiatrist  

62

Contact with psychologist  

62

Health Screening   
General health check 

15

6

Dental check  

15

Blood pressure check 

62

61

Vision test 

15

6

Hearing test 

6, 15

 
Mammogram 

6, 15

 
Cervical smear 

6

 
Testicular check 

62

61

Medication   
Anti-psychotic/anti-depressant medication

62, 101

61, 102, 15

 
More than two anti-psychotics

15

  
Anxiolytic medication 

6, 15

 
Hypnotic medication

15

6

 
Mental health care 

57

 
Safety   
Accidents 

6

 
Victim of crime 

62

61

Verbal abuse 

62

61

Physical or sexual abuse 

6

 
Vandalism 

6

 
Perceived risk of accidents 

6

 
Perceived risk of exploitation from community  

6

Self-determination
Self-determination

56, 62, 7, 16, 18, 17

2, 57, 61, 12

 
Personal Development
Personal Development 

5

 
Scheduled activity

34

33, 7

 
Constructive activity 

3

 
Opportunities to learn new skills

16

  
Change in adaptive behaviour over time (not domestic/responsibility) 

17

 
Change in domestic activity and in responsibility

17

  
Life achievements and changes

17

  
Education/employment

17

  
Work experience/adult education/day centre activities 

15

 
Rights
Privacy

2

  
Access/adapted environment 

5

 
Freedom

5

  
Exclusion/restraint, sedation used for challenging behaviour

9, 15

  
  1. Comparison between dispersed housing and village communities
  2. Comparison between dispersed housing and campus settings
  3. Comparison between ordinary dispersed housing and campus/clustered settings
  4. Comparison between specialised dispersed housing and campus/clustered settings
  5. Comparison between clustered supported living and other settings
  6. Rated by staff
  7. Rated by users or families
  8. Comparison between supported living and campus/clustered settings
  9. Comparison between group homes and campus/clustered settings


Table 20 presents summary information from the studies reviewed in relation to service characteristics and processes, and the cost of services.

Table 20 Measures of service provision and cost

Measure

Dispersed housing better

No difference

Campus/cluster housing better

Homelikeness/environmental quality/number of people sharing home

6, 7, 15, 16, 17

  
    
Social climate
Block treatment

3, 6, 7

  
Depersonalisation

6, 7

3

 
Rigidity of routines

62, 7

 

61

Social distance

62, 7

3, 61

 
    
Working practices
Person-centred planning

62, 15

7

61

Activity planning

62, 7,

 

61

Staff training and supervision

62, 7

 

61

Assessment and teaching 

7

6

Planning staff support

7

9

 
    
Written behavioural intervention programme/treatment programme

98

99, 15

 
IPP plan goal related to CB 

98

99

    
Staff ratio

1, 2, 62, 7, 15

3, 61

 
    
Staff contact and assistance

1

3, 7

 
    
Staff care (general)

56

31, 57

32

    
Family satisfaction 

6, 13

 
    
Costs

33

44, 610, 142

2, 611, 7, 141

  1. Comparison between dispersed housing and village communities
  2. Comparison between dispersed housing and campus settings
  3. Comparison between ordinary dispersed housing and campus/clustered settings
  4. Comparison between specialised dispersed housing and campus/clustered settings
  5. Comparison between clustered supported living and other settings
  6. Rated by staff
  7. Rated by users or families
  8. Comparison between supported living and campus/clustered settings
  9. Comparison between group homes and campus/clustered settings
  10. Matched groups
  11. Full sample