Inclusion criteria

A national cross-sectional survey was conducted of alternatives to standard acute psychiatric in-patient care. Ethical approval was obtained from the Metropolitan Multicentre Research Ethics Committee. Services were included if they aimed to serve adults aged 18–65 years who would otherwise be admitted to an acute ward and involved the patient staying overnight at the service. They also had to meet at least one of the following criteria:

1. • based outside hospital

2. • dedicated to a specific diagnostic or sociodemographic group

3. • had a fixed maximum length of stay

4. • had implemented a specific therapeutic model involving changes in the practice of more than one profession within the service.

Our aim was to prepare the way for further research in this underinvestigated field and to avoid missing any potentially important models by using a broad operationalisation. The risk associated with this strategy is of including services that may not in fact differ greatly from standard wards. However, we judged this risk less important than that of missing important innovative models, especially as Phase 2 of the Alternatives Study involves in-depth investigation of the main models identified, including examination of content of care. Models involving change in only one profession's practice were, however, judged unlikely to be far-reaching enough in their effects on the service's approach for inclusion. We therefore introduced the criterion that practice needed to change for at least two professions.

Identification of the alternative units

As no single method was likely to identify all relevant services, multiple methods were used. These included examination of the Mental Health Service Mapping for Adults of Working Age in England, ^{Reference Glover, Bennett, Bradley and Barnes9} telephone calls to all mental health trusts, Google internet searches, consultation with a variety of expert sources, including the national voluntary organisations MIND and Rethink, and snowball sampling with identified participants.

Data collection

All identified services were invited to participate in an interview with a researcher, using a questionnaire designed to cover the main clinical and organisational characteristics of services. Data were collected in the second half of 2005 and first half of 2006. Interviews were usually conducted over the telephone with the manager of the service, who had the opportunity to prepare answers to the questions in advance. Topics covered included location and facilities, types of care provided, referral criteria and pathways, funding and management, links with other services and staffing. So as to characterise the service's client group, participants were also asked for anonymised sociodemographic and clinical details of all residents in their service on the preceding night. To check how comprehensive the identification of alternatives had been, respondents were asked to name any other alternatives of which they were aware in the surrounding area; this yielded only two previously unidentified services.

Ascertainment of deprivation

Government office region and indices of multiple deprivation ^{Reference Jordan, Roderick and Martin10} were obtained for the local authority in which each alternative was situated. These indices are estimates of different forms of deprivation derived from multiple data sources, including the 2001 census ¹¹ and data-sets regarding local benefits, crime, education, housing and health. Six different summary measures make up the overall Index of Multiple Deprivation. Since there is no single best way of comparing area-level deprivation, no single measure is favoured above the others. We therefore explored the distribution of the alternatives according to each summary measure. We also recorded whether each alternative was situated in the centre of one of the ten most populous English cities.

Analysis

The typology of services was obtained using cluster analysis. This is a statistical technique that allows categorisation of a previously unclassified data-set, based on the principle that cases that resemble each other on a defined set of variables should be grouped together in the same cluster and those that are different on most variables should be in different clusters. ^{Reference Everitt12} Two-step cluster analysis ¹³ was the specific procedure used in this analysis, selected because it allows inclusion of both binary and continuous variables. Variables for the cluster analysis were initially selected from the questionnaire by the Alternatives Study steering group (a group of 19 relevant experts from university, National Health Service (NHS) and voluntary sector backgrounds) who rated each candidate variable for inclusion in the analysis in terms of its importance in characterising the alternatives and discriminating between them. In this way, we identified 18 variables that were rated overall as at least moderately important. The ten that were rated most highly were included in our initial version of the cluster analysis and we then investigated what happened when we added each of the other eight one by one. Our aim in conducting the cluster analysis was to obtain a model that: did not alter very much when small changes were made in the list of variables included; did not have clusters containing a very small or a very large proportion of the sample; and was clinically plausible, in the sense that it grouped the services in a way that could be convincingly interpreted. We tried out models that contained between five and nine clusters to see which fitted best.

Community-based service types

Hospital-based service types

Hospital v. community services

The typology produced a complete separation between hospital and community services, with no cluster containing both. Community services were characterised overall by lower severity of target group, fewer medical and nursing staff, and fewer services with on-site staff awake through the night. However, even among the less clinically oriented types of community service, there were indications of considerable collaboration with NHS secondary mental health services. Every service accepted referrals from NHS mental health staff and most named an NHS mental health service as their most prolific referrer. Apart from two clinical crisis houses and two non-clinical alternatives, all 42 community services reported that, in a situation of high risk, they would make a referral to NHS services (e.g. for a Mental Health Act assessment) even for an individual who had not consented to this.

Only five of the services described as situated in the community reported that they accepted admissions directly from the community of patients compulsorily detained under the Mental Health Act. These services had all acquired the legal status of hospitals, but were described by their managers as not being hospital services: they were typified by having small numbers of beds, not sharing premises with other in-patient services, being unlocked and not being recognisable as in-patient units from outside. Their legal status as hospitals suggests some blurring of boundaries between community and hospital services. The majority of the community-based services included in the survey accepted as transfers from hospital patients who were still legally detained but had been given leave from the hospital.

Table 2 describes the care reported to be available at each service. Not surprisingly, ‘medical’ interventions such as physical examinations and review of medication were more likely to be available in hospital-based services, but this also applied to most other interventions, including individual psychological treatment, groups, complementary and alternative treatments, and occupational therapy or other organised activity. Again, the clinical and specialist crisis houses were intermediate between the hospital services and the other community services, offering a programme of organised activity in every case, medication review and investigations in the majority, and cognitive–behavioural therapy in around a third of the services.

Table 2

Types of care delivered

	Hospital-based service types				Community-based service types
	General therapeutic wards	Wards for specific demographic groupsa	Therapeutic wards for specific groups	Short-stay wards and general wards for specific groups	Clinical crisis houses	Crisis team beds	Non-clinical alternatives	Specialist crisis houses
Number in group	35	20	4	8	13	13	11	5
Distinctive therapeutic model,b n (%)	35 (100)	3 (15)	4 (100)	2 (25)	0	0	0	5 (100)
Review of medication by a doctor routinely provided for everyone, n (%)	34 (97)	20 (100)	4 (100)	5 (63)	9 (59)	2 (15)	4 (36)	5 (100)
Physical examination routine part of admission procedure, n (%)	35 (100)	20 (100)	3 (75)	8 (100)	7 (54)	0	0	2 (40)
Blood test can be carried out at service, n (%)	34 (97)	20 (100)	4 (100)	8 (100)	7 (54)	2 (15)	1 (9)	4 (80)
Recognised type of individual psychological treatment available, n (%)	22 (63)	13 (65)	4 (100)	4 (50)	4 (31)	3 (23)	2 (18)	3 (60)
Cognitive—behavioural therapy available, n (%)	20 (57)	13 (65)	3 (75)	4 (50)	4 (31)	3 (23)	1 (9)	2 (40)
Complementary or alternative therapies available, n (%)	14 (41)	9 (45)	1 (25)	3 (38)	4 (31)	3 (23)	2 (18)	3 (60)
Programme of occupational therapy or other organised activities available, n (%)	35 (100)	19 (95)	4 (100)	3 (38)	13 (100)	2 (15)	4 (36)	5 (100)
Group programme held at service, n (%)	34 (97)	17 (85)	3 (75)	6 (75)	8 (77)	3 (17)	5 (45)	4 (80)

Table 3 shows the characteristics of the residents in the services on the census night. On the census night, women were in the majority in all models except the clinical crisis houses and specialist crisis houses. Overall, 76% of the residents of the alternative services were White British, 9% were Black Caribbean, British or African and 3% were Other White. Two of the community service types, the non-clinical alternatives and the specialist crisis houses, had especially high representation of ethnic minorities (one non-clinical alternative was a crisis house for people of Black Caribbean or African ethnicity). Except in the non-clinical alternatives, a majority of residents in every service type had a previous history of hospital admission and the clinical and the specialist crisis houses were the types with the highest proportions of residents with current psychotic symptoms (56% and 72% respectively). Comparison data from the Count Me In ¹⁸ 2006 national census of the England and Wales in-patient population are also shown in Table 3.

Table 3

Characteristics of patients in residence at 1-night census

	Male residents n/N (%)	White British n (%)a	Black Caribbean, British or African n (%)	Other White n (%)	Asian n (%)	History of previous admissionb n (%)	Compulsorily detainedc n (%)	Psychotic symptomsd n (%)
Hospital-based services
General therapeutic wards	259/571 (45)	445 (78)	26 (10)	18 (3)	22 (4)	395 (70)	207 (36)	285 (50)
Therapeutic wards for specific groups	10/44 (23)	30 (68)	10 (23)	2 (5)	1 (2)	24 (54)	25 (57)	20 (45)
Wards for specific demographic groupse	4/92 (4)	65 (71)	10 (11)	1 (1)	10 (11)	59 (64)	14 (15)	38 (41)
Short-stay wards and general wards for specific groups	37/84 (44)	71 (85)	5 (6)	0	2 (2)	68 (81)	33 (39)	44 (52)
Community-based services
Clinical crisis houses	48/93 (51)	80 (86)	6 (6)	3 (3)	3 (3)	80 (86)	9 (10)	52 (56)
Crisis team beds	13/48 (27)	38 (79)	7 (15)	1 (2)	1 (2)	33 (69)	0 (0)	14 (29)
Non-clinical alternatives	21/61 (34)	37 (61)	14 (23)	2 (3)	4 (7)	24 (39)	1 (2)	13 (21)
Specialist crisis houses	17/29 (59)	10 (34)	9 (3)	2 (7)	5 (17)	24 (83)	11 (38)	21 (72)
Total	409/1022 (40)	776 (76)	87 (9)	29 (3)	48 (5)	648 (63)	300 (29)	487 (48)
National in-patient populationf	55.3%	79.5%	7.8%	5.6%	3.2%	—	40%	—

Table 4 shows the geographical distribution of the alternatives, especially in relation to area social deprivation and region. Most alternatives were situated outside the centres of the ten most populous cities. Nonetheless, far more alternatives were available in local authorities falling in the most deprived quartile than in the least deprived quartile, a finding confirmed by use of several different indices of deprivation, all showing the same pattern (between 52 and 63% of services were in the most deprived quartile of the country, depending on which index of multiple deprivation was used). At least 70% of the alternative services were located in the more deprived half of the country. Regional distribution was highly uneven: two-thirds of the alternatives were in London and the South East (34%) or in the North East of England (32%), very few in the South West (3%), the North West (7%) and the East of England (6%).

Table 4

Geographical distribution of the alternatives

	Hospital-based service types					Community-based service types
	Total	General therapeutic wards	Wards for specific demographic groupsa	Therapeutic wards for specific groups	Short-stay wards and general wards for specific groups	Clinical crisis houses	Crisis team beds	Non-clinical alternatives	Specialist crisis houses
Government office region, n (%)
North West	8 (7)	0	2 (10)	0	3 (38)	2 (15)	0	1 (9)	0
North East and Yorkshire	35 (32)	17 (49)	4 (20)	2 (5)	2 (25)	5 (39)	3 (23)	1 (9)	1 (20)
South West	(3)	0	3 (15)	0	0	0	0	0	0
South East and London	37 (34)	11 (31)	6 (30)	2 (5)	2 (25)	2 (15)	7 (54)	6 (55)	1 (20)
Midlands (East and West)	19 (17)	6 (17)	3 (15)	0	0	3 (23)	2 (15)	2 (18)	3 (60)
East	7 (6)	1 (3)	2 (10)	0	1 (13)	1 (8)	1 (8)	1 (9)	0
In the inner city,b n (%)	34 (31)	12 (34)	7 (35)	2 (50)	2 (25)	1 (8)	2 (15)	4 (36)	4 (80)
Index of Multiple Deprivation,c n (%)
Average rank of local authority scores (overall score)
Services based in most deprived quartile	57 (52)	17 (49)	7 (35)	2 (50)	5 (63)	7 (54)	6 (46)	8 (73)	5 (100)
Services based in most deprived half	77 (71)	26 (74)	10 (50)	2 (50)	7 (88)	9 (69)	10	8 (73)	5 (100)
Rank of income scale
Services based in most deprived quartile	68 (62)	21 (60)	9 (45)	2 (50)	7 (88)	9 (69)	7 (54)	8 (73)	5 (100)
Services based in most deprived half	88 (81)	29 (83)	15 (75)	4 (100)	7 (88)	13 (100)	9 (69)	8 (73)	5 (100)
Rank of employment scale
Services based in most deprived quartile	69 (63)	22 (63)	11 (55)	2 (50)	6 (75)	9 (69)	7 (54)	7 (64)	5 (100)
Services based in most deprived half	89 (82)	28 (80)	15 (75)	4 (100)	7 (88)	12 (92)	9 (69)	9 (82)	5 (100)

Limitations

The major methodological limitation is that data were obtained solely from service managers, with no independent check on accuracy. Social desirability effects are especially likely in response to questions about interventions available, severity of patients' needs and service user involvement. Cluster analysis is rarely definitive – other methods might have led to different classifications; however, the typology that we obtained led to distinct subgroups which were interpretable and remained stable when small changes were made in the variables selected to define the clusters. Finally, a basic assumption was that there is an identifiable group for whom acute in-patient admission is appropriate: in reality, defining this group is difficult and thresholds are likely to vary between clinicians, teams and areas, and to be influenced by acute bed availability.

Availability of alternatives

Our findings indicate substantial previously unreported activity in the development of alternatives to standard in-patient care. Assuming similar numbers of beds for non-responders to our survey as for responders, we estimate that there are just under 1300 occupied beds in these alternative units, of which around 250 are outside hospital and 380 in the private sector. This compares with a national tally of around 12 400 acute beds for adults of working age. ¹⁹ Thus, these alternatives now represent an important, but so far largely undocumented, uncoordinated and unevaluated component of the national mental health economy, mainly within the statutory sector.

Types of alternative

A spectrum of alternatives was identified. In hospital settings, the most popular form of initiative has been the introduction of distinctive therapeutic models such as the tidal model ^{Reference Fletcher and Stevenson16} on general acute wards. An advantage of such a strategy is that no substantial reorganisation of the local mental health system is required, but a literature search found no evidence from randomised controlled trials or any other large-scale evaluations regarding any of these named models of in-patient care. Some of their components, for example giving priority to the service user's narrative and increasing staff–patient contact, sound like goals that one might expect any mainstream service aiming for high standards to adopt rather than distinctive approaches. Data at this stage of the study allow us only to identify approaches that aim to be innovative, not to establish whether the resulting care and environment are really different from standard services in important respects. Our study criteria classify the adoption of distinctive therapeutic models on general acute wards as an alternative to standard care. However, they might also be viewed as approaches intended to improve quality within the framework of standard care, arguably a goal that may yield more widespread benefits than development of alternatives. ²⁰

Brief-admission wards are one of the few forms of alternative to have any supporting evidence, though all the studies identified in a systematic review dated from before 1985. ^{Reference Johnstone and Zolese21} These early studies suggested that short hospital stays were as effective as standard ones, but little research or service development has followed on from them. Detailed evaluation of the two hospital units we found that reported a maximum length of stay of less than a week would thus be of considerable interest. The international literature also yields descriptions of services specialising in particular subgroups, based on demographic characteristics such as ethnic group ^{Reference Mathews, Glidden, Murray, Forster and Hargreaves22} or on diagnosis. ^{Reference Bonsack, Borgeat and Dubois23} This has some potential to allow development of more focused therapeutic models for in-patient care, but mother and baby units are currently the only type of specialist acute unit found on any scale within the NHS.

In the community, new service types with distinctive sets of characteristics are emerging around England. They range from community-based services that retain a substantial resemblance to hospital wards, through short-stay community beds very closely linked to, and sometimes directly managed by, crisis resolution teams, to more radical community alternatives. The close links between most community alternatives and (other) NHS secondary mental health services suggest that they function as components in local acute care systems and pathways rather than freestanding services that attract different client groups. However, we cannot assess from our data whether patients are in practice offered a choice between hospital admission and these community alternatives. Notable by its absence from current service provision in England was one model with a long history on both sides of the Atlantic: the family sponsor home ^{Reference Johnson, Gilburt, Lloyd-Evans and Slade24,Reference Polak, Kirby and Deitchman25} in which acutely unwell patients are placed with carefully trained and selected families, supported by a home treatment team.

Types of care

Some types of care reported to be available within the hospital services were less likely to be provided in community alternatives. Examples were psychological treatments, structured activity programmes, physical investigations and review of medication. We do not know how far the use of resources outside the units compensated for some of these apparent gaps (e.g. crisis resolution teams may be major providers of interventions such as medication review and supervision in some alternatives). Some of the ways in which alternatives may help their clients, such as through service user empowerment and peer support, may not have been well captured by our survey. If alternatives encourage service users to become involved in mainstream activities in the community, this may serve social inclusion better than organised activities in mental health service settings. However, medication review, physical health assessment and psychological treatments are important components of the management of disorders such as bipolar affective disorder and schizophrenia. ^26,27 Service planners thus need to guard against developing community alternatives in which there is less access to some important treatments for severe mental illness than in hospital settings. Lack of activity and of clear therapeutic models are among the recurrent criticisms of in-patient services, ^{Reference Szmukler, Holloway, Thornicroft and Szmukler28,29} and alternatives that do not offer organised activities, psychological treatments or a defined therapeutic model may not be addressing these problems. However, some community alternatives do provide a wide range of types of medical, psychological and social care, suggesting that this is feasible in a community setting given adequate resources.

Clinical populations served

The large proportions of service users who have current psychotic symptoms and/or a previous history of admission suggest that the alternatives are focusing on a group with severe mental illness. We cannot, however, say whether their severity of risk and disturbance at the time of admission was as great as in hospital services. The less clinically focused community alternatives – crisis beds and non-clinical alternatives – reported fewer residents with psychotic symptoms. This may well be appropriate: people with acute psychoses are likely to need a range of clinical interventions delivered by mental health professionals, as is offered in hospitals and the more clinical community alternatives, whereas people with other types of mental health problem (e.g. depression or personality disorder) may benefit from a markedly different approach. However, further evidence is needed as to whether the groups served by these non-clinical alternatives overlap with acute-ward patients or whether they are new groups who would not otherwise receive a residential acute service.

A significant limitation of many community services is that they manage mainly voluntary rather than compulsorily detained patients. Thus, any increase in choice about care will be available only to those in-patients who are, in any case, less subject to compulsion (though it is possible that some people admitted voluntarily to community alternatives would otherwise be detained in hospital). The fact that some community alternatives do accept detained patients, albeit in very small numbers, is of considerable interest, suggesting scope for further development of alternatives to meet the needs of such patients.

Equity of access

Men are in a slight majority among in-patients nationally, ¹⁸ but the reverse applies to alternative services. Intensive home treatment has also been found to prevent more female than male admissions ^{Reference Glover, Arts and Babu30} so that overall, women may have more access to alternatives to standard acute admission. This may reflect differences in actual or perceived severity or risks, or in willingness to engage: the imbalance warrants further investigation as equity of access to a range of types of acute care is desirable if feasible.

Availability of alternatives is especially salient for people from those minority ethnic groups, who are overrepresented in in-patient settings and for whom coercive pathways to care are especially likely. ^{Reference Bhui, Stansfeld, Hull, Priebe, Mole and Feder31} The proportion of White British residents and that of Black Caribbean, Black African and Black British residents were very similar to the national in-patient population; other groups also diverged relatively little from the national picture. Caution needs to be exercised in interpreting this, as adjustment is not made for the local ethnic mix, but it provides a preliminary indication that the full range of groups represented in hospital is also served by alternative services.

The uneven distribution between regions is potentially inequitable and probably reflects the origins of alternative services in local interest rather than central policy and planning. A more encouraging finding is their tendency to develop in deprived areas, suggesting that they may be a response to high levels of local need. ^{Reference Harrison, Barrow and Creed32} Given this, it is surprising that relatively few have developed within inner-city areas.

Further research

Given the pressing need to improve acute care, the substantial investment of public funds in these alternatives and the fact that some are relatively long established, they have been remarkably underinvestigated. Prior to the present survey, a few descriptions and, occasionally, evaluations ^{Reference Boardman, Hodgson, Lewis and Allen33,Reference Fenton, Mosher, Herrell and Blyler34} have appeared of model services, but these are the exception. Evaluative research is thus urgently needed to explore their effectiveness, which alternatives work best for which groups of service users, and whether, as intended, they resolve some of the problems in acute care systems. Such research is required both to assess whether provision of these alternatives is an effective use of public funds, and, if some alternatives are indeed an improvement on standard care, to secure their futures and allow dissemination of effective ways of working.

Variables used to obtain typology of services

1. (1) Number of beds

2. (2) Voluntary sector management

3. (3) Typical length of stay, reported by manager

4. (4) Proportion of staff who are nurses

5. (5) Whether outside hospital

6. (6) Whether care programme approach meetings (care planning meetings which are a statutory requirement within the NHS) are organised within the service

7. (7) Whether any programme of structured activity is available to residents

8. (8) Whether service is dedicated to people with a specific diagnosis

9. (9) Whether service is dedicated to a specific sociodemographic group

10. (10) Implementation of a specific therapeutic model

11. (11) Score for degree of integration with crisis resolution team, measured by a scale made up from six binary items (maximum score 6, Cronbach's α for scale 0.69):

    1. • whether crisis resolution teams can refer to the service

    2. • whether a crisis resolution team is among the top three referrers

    3. • whether crisis resolution teams are the only referrers

    4. • whether a crisis resolution team decides who can be admitted to the service (‘gate-keeping’)

    5. • whether a crisis resolution team manages the service

    6. • whether the service's premises are shared with a crisis resolution team

12. (12) Score for extent of hospital-like interventions, measured by scale with six binary items (maximum score 6; α=0.91):

    1. • whether staff hold and dispense all medications

    2. • whether blood tests can be carried out on site

    3. • whether physical examination is part of standard admission procedures

    4. • whether one-to-one supervision can be carried out for at least 12 h

    5. • whether residents on medication are normally reviewed by a doctor during their stay

    6. • whether the staff team includes a doctor

13. (13) Score for severity of target group, measured by scale including seven binary items (maximum score 7; α=0.73):

    1. • whether all admissions are for an acute crisis otherwise requiring standard acute ward admission

    2. • whether such admissions are the top priority

    3. • whether admissions are on day of referral when beds are available

    4. • whether compulsorily detained patients are accepted directly from community

    5. • whether detained patients are admitted on any basis, including on leave from hospital

    6. • whether manager judges at least 50% of residents similar in severity to general acute ward patients

    7. • whether history of violence is an exclusion criterion.