For healthcare professionals only


Relation between bed use, social deprivation, and overall bed availability in acute adult psychiatric units, and alternative residential options: a cross sectional survey, one day census data, and staff interviews

BMJ 1997; 314 doi: (Published 25 January 1997) Cite this as: BMJ 1997;314:262
  1. G Shepherd, head of researcha,
  2. A Beadsmoore, senior researchera,
  3. C Moore, researchera,
  4. P Hardy, statisticiana,
  5. M Muijen, directora
  1. a Research Section, Sainsbury Centre for Mental Health, London SE1 1LB
  1. Correspondence to: Professor Shepherd
  • Accepted 6 December 1996


Objectives: To examine the relation between bed use, social deprivation, and overall bed availability in acute adult psychiatric units and to explore the range of alternative residential options.

Design: Cross sectional survey, combined with one day census data; ratings by and interviews with staff; examination of routine data sources.

Settings: Nationally representative sample of acute psychiatric units.

Subjects: 2236 patients who were inpatients on census day.

Main outcome measures: Bed occupancy levels, judged need for continuing inpatient care, reasons preventing discharge, scores on the Health of the Nation outcome scales.

Results: Bed occupancy was related to social deprivation and total availability of acute beds (r =0.66, 95% confidence interval 0.19 to 0.88, F=8.72, df=2, 23; P=0.002). However, 27% (603/2215) of current inpatients (61% (90/148) of those with stays of >6 months) were judged not to need continuing admission. The major reasons preventing discharge were lack of suitable accommodation (37% (176/482) of patients in hospital <6 months v 36% (31/86) of those in hospital >6 months); inadequate domiciliary based community support (23% (113) v 9% (8)); and lack of long term rehabilitation places (21% (100) v 47% (40)). Scores on the Health of the Nation outcome scale were generally consistent with these staff judgments.

Conclusions: The shortage of beds in acute psychiatric units is related to both social deprivation and the overall availability of acute beds. Patients currently inappropriately placed on acute admission wards should be relocated into more suitable accommodation, either in hospital or in the community. A range of provisions is required; simply providing more acute beds is not the answer.

Key messages

  • This study shows that there is a shortage of acute psychiatric beds and that this is related to both social deprivation and the overall availability of beds

  • Over a quarter of inpatients were considered to be inappropriately placed on admission wards–for those with lengths of stay of more than six months this figure increased to 61%

  • Lack of suitable accommodation, home based support, and rehabilitation places was identified as the major reason preventing discharge

  • A range of hospital and community facilities is therefore required to address these problems of over-occupancy and shortages of acute beds


The policy to establish more community based mental health services is now well established.1 Both professional and public concern remains, however, that inadequate resources may have been allocated to carrying out the policy and that the reduction of beds in the old mental hospitals has proceeded much more rapidly than the expansion of new hospital and community services to replace them. Acute psychiatric units, especially those in inner cities with high levels of surrounding social deprivation, have been described as “in crisis,”2 with no available beds, an accumulation of difficult and disturbed patients, increased rates of extracontractual referrals, and apparently unsolvable problems of over-occupancy.3 This has led to calls for more acute beds4 5 and an end to hospital closures.6 It has also provided the impetus for a recent ministerial announcement about the importance of a “spectrum of care” that does not neglect the need for “24 hour nursed” beds to deal with some of the most difficult patients who need longer term residential care.7 The NHS Executive's guidance suggests a need for relatively small (12 bed) units, provided at an average level of 10 per 100 000, at an estimated overall running cost of £35 000-£50 000 per resident per year depending on size and location.

Despite passionate debate, the evidence on bed use in acute psychiatric units is sparse and is based on either small samples of districts8 9 or studies of conditions in inner London alone.2 3 It is therefore important to examine these problems in a national context and to explore the relation between bed use, indices of social deprivation, and overall availability of beds. We aimed to investigate why people are remaining in hospital and what alternative accommodation might meet their needs.



We drew up a list from the 1991 census of all district health authorities in England and Wales as of April 1994, together with their corresponding Jarman underprivileged area 8 scores10 as the principle stratifying variable. This resulted in a sampling frame of 119 authorities. We then ranked these into six “bands” (five containing 20 districts and one containing 19) and randomly selected seven districts from each band (42 in total). We obtained from each district information on the number of providers of acute psychiatric inpatient services. If there was only one provider (a trust) this was selected into the sample. Where there were multiple providers, one was selected randomly. In cases of refusal or non-response, a new district was randomly selected from the same band.


Recruitment of services took place between April and July 1995 and data collection between May and October 1995. Contact was established with each provider, initially through the chief executives, and researchers visited the units to train staff to use the data collection instruments. Because of the necessity to phase sites into the project, a single census day common to all participating districts was impractical, so sites chose their own census day. A locally appointed liaison officer supervised data collection on each site. A researcher was available on the census day to help.


Data were collected with a specially designed form to record basic information about numbers of available beds, staffing levels, and acute unit activity. Information was also collected on the range of locally based community services, such as community mental health teams, crisis services, and supported housing. A census day was selected for the acute wards in each provider, and senior nursing staff were asked to collect the information on all patients recorded as occupying a bed on that day in consultation with the named nurse responsible for that individual patient's care. This included patients on leave, absent without leave, and temporary transfers to secure facilities or non-psychiatric beds. Patients in specially designated beds for elderly functionally ill people, in mother and baby units, and in secure units were excluded. Staff completed a background data sheet covering age, sex, primary diagnosis, ethnic group, Mental Health Act status, and date of admission. They also completed the 12 item field trial version of the Health of the Nation outcome scales,11 which contains 11 items, each rated 0-4, covering common symptoms and functional problems (box).

Health of the Nation outcome scales

  • Overactive, aggressive, disruptive behaviour

  • Non-accidental self injury

  • Alcohol or drug misuse

  • Cognitive problems

  • Physical illness or disability problems

  • Problems associated with hallucinations and delusions

  • Problems with depressed mood

  • Other mental or behavioural problems

  • Problems with relationships

  • Problems with activities of daily living

  • Problems with living conditions

  • Problems with occupation and activities

For all scales 0=no problem; 1=minor problem, requiring no action; 2=mild problem but definitely present; 3=moderately severe problem; 4=severe to very severe problem. Total severity score is sum of all rated scales

Finally, members of staff were asked to indicate whether they judged that the individual should remain on an acute inpatient ward and to complete a standardised checklist of the reasons preventing discharge or transfer to a more appropriate setting.12


Means, standard deviations, and confidence intervals are quoted as appropriate. When the data are skewed, medians and ranges are given. A priori hypotheses were tested with two tailed tests on frequency and means. For categorical data, differences were analysed with χ2 tests. For continuous data, when approximately normally distributed, means were compared with t tests and one way analysis of variance. Otherwise, Mann-Whitney U tests were used. Bivariate correlations and regression analyses were used to investigate the relation between continuous variables. Linear trends for ordinal data were investigated with χ2 for trends. All analyses were performed with the statistical software spss version 6.1.


The sample

A sample of 38 providers (all but one having trust status) was achieved out of the original aim of 42. Three of the missing districts did not reply to repeated requests to supply data, and one withdrew for “internal reasons.” The participating services covered total catchment area populations of 11.1 million (10.4 million in England and 730 000 in Wales). This represents 22% and 26% of the total population of England and Wales respectively. The median population size served by the providers was 236 000 (range 95 000-857 000).

In all, 34 out of 38 of the services (89%) returned usable census data. The Jarman scores and geographical location (urban, rural, or mixed) of the 34 services that provided data were compared with the eight that did not, and there was no evidence to suggest a difference in Jarman scores (Mann-Whitney U=132, P=0.91). When services were compared by geographical location, there was insufficient data to perform statistical analyses, but there was evidence of an underrepresentation of mixed locations in the eight services not returning data. Representation of urban and rural settings was similar.

Information was provided on 2236 patients. Of these, 48.2% (1074) of patients were resident in acute units on general hospital sites, and 45.1% (1009) were resident in acute units on mental hospital sites; the remaining 6.7% (149) of patients were accommodated in acute wards in other locations. In all, 52.0% (1139/2129) of the census patients were female, the mean age was 41 (SD=15.07) years, 88.0% (1950/2217) of patients were white, and 69.9% (1562/2234) were “informal” (that is, voluntary patients not detained under a section of the Mental Health Act). Altogether, 74.5% (1652/2216) of patients had schizophrenia or mood disorder, and substance misuse, neurosis, and personality disorder accounted for 6.5% (143), 6.2% (130), and 6.2% (138) respectively. The median length of stay was 28.5 days (lower quartile, 11 days; upper quartile, 69 days). This kind of cross sectional sample is biased towards those with longer current stays; 6.7% (149/2208) of patients in the sample had current lengths of stay greater than six months and can therefore be regarded as “new” long stay patients.13 The prevalence of these patients in general hospital based acute units was almost double the prevalence in mental hospitals (9.3% (67/724) v 4.8% (25/516)).

Altogether, 28 services returned information on community services, of whom 21 reported having multidisciplinary community mental health teams. However, only three services provided a crisis team (defined as providing evening and weekend cover; taking referrals mainly from secondary or tertiary services; and used as an alternative to admission to hospital). Seventeen of the 25 services without specialist crisis teams relied on an on call system whereby medical staff could be accessed through the local accident and emergency department. Eight services provided an “extended hours” service, but not 24 hours–for example, a community psychiatric nurse on call 9 am to 5 pm seven days a week or a rota of community mental health team workers. Four services provided an out of hours helpline for clients known to the service, and one had a system of “crisis cards” for discharged patients. Three services provided an intensive support team (defined as being intended for severely mentally ill people; providing evening and weekend cover; having caseloads of less than 20; having the capacity to provide daily visits; and offering long term commitment). None of the services had both a crisis team and an intensive support team.

Bed occupancy levels and social deprivation

Bed occupancy levels for the past 12 months were provided by 26 of the 38 services. These were calculated as the proportion of occupied patient days to number of available beds and included patients who might be on leave, absent without leave, or temporarily transferred to secure facilities (hence occupancy levels could exceed 100%). Figure 1 shows the level for each provider, expressed as a percentage and arranged according to the Jarman score for each provider.

Fig 1
Fig 1

Distribution of percentage bed occupancy ranked by Jarman underprivileged area 8 score

There was a negative relation between the number of available adult acute beds and bed occupancy (r =-0.51, 95% confidence interval -0.75 to -0.15, F=8.27, df=1, 24; P=0.008). There was also a relation between Jarman scores, acute bed availability, and bed occupancy (r =0.66, 0.19 to 0.88, F=8.72, df=2, 23; P=0.002). The services with low bed availability and high social deprivation therefore appeared to have the highest bed occupancy levels.

When we analysed linear trends with interquartile ranges of Jarman scores, services in more deprived areas had a higher proportion of people with schizophrenia (χ2 trend=45.53, df=1, P<0.0001), people from non-white ethnic groups (χ2 trend=9.94, df=1, P<0.01), and people detained under the Mental Health Act (χ2 trend=7.71, df=1, P<0.01). Overall there was no evidence to suggest a relation between Jarman scores and numbers of admissions (Spearman r =-0.17, P=0.48), number of finished consultant episodes (r =0.09, P=0.69), or mean length of finished consultant episodes for past 12 months (Spearman r =-0.14, P=0.47). Numbers of admissions under the Mental Health Act per thousand population were related to Jarman scores in a curvilinear (quadratic) relation (r =0.97, 0.90 to 0.99, F=163.11, df=2, 20; P<0.0001). This held even when one particularly influential point with a high “leverage” value was removed (r =0.77, 0.36 to 0.93, F=14.05, df=2, 19; P<0.0001). This point simply exerted high leverage and was not actually an outlier. Thus, admissions under the Mental Health Act were higher in the more deprived areas (which were all located in the inner cities), then declined with decreasing deprivation, but increased again in some of the relatively more affluent districts. Figure 2 illustrates these data.

Fig 2
Fig 2

Admissions under Mental Health Act v Jarman underprivileged area 8 score (without leverage point)

One possible explanation for this very strong statistical relation may be that these districts were all located close to major urban centres (London and Birmingham) and were thus having to cope with an “overspill” from acute beds in the inner cities. Many of these patients were only covered by extracontractual referrals.

Judged needs for admission

In all, 27.2% (603/2215) of current inpatients were judged not to need continuing admission. This proportion was significantly larger (61% (90/148)) among those with stays of more than six months (difference=36%, 28% to 44%, χ2=86.28, df=1, P<0.0001). Table 1 shows the reasons preventing discharge in the two groups, excluding those who were on predischarge leave or for whom discharge plans were already in progress. As a check on the validity of these ratings, mean total severity scores on the Health of the Nation outcome scales were calculated for each category for all those patients for whom only one reason was given (91% of total) (table 1).

Table 1

Reasons preventing discharge for current admissions of more than or less than six months and Health of the nation total severity scores* for patients judged to be inappropriately placed

View this table:

The most common reason preventing discharge was the non-availability of various community and residential options, in particular the lack of suitable supported housing. A lack of specialised rehabilitation places was seen as a more important reason for the new long stay patients, who were generally much more disabled. The prevalence of these new long stay patients in acute units in general hospitals was higher than in units in mental hospitals. “Requiring higher levels of psychiatric supervision” (which usually meant secure accommodation) contributed to around 10% of the reasons, although this proportion was much higher (17%) among the new long stay patients. “Other” reasons were also important in both categories of lengths of stay, and this highlights the shortage of certain specialist placements (for example, for people with acquired brain damage, dual diagnosis of psychosis plus learning disability, drug and alcohol complicated psychosis, and eating disorders). The total severity scores are consistent with the staff judgments, indicating that the most severe problems were among the long term rehabilitation group, followed by those requiring higher levels of supervision, supervised accommodation in the community, and “other” special needs. Patients judged to be able to manage with intensive, domiciliary based support received the lowest total severity scores.


These data, collected from a carefully constructed nationally representative sample, support the conclusion that the shortage of beds in acute psychiatric units is related to both social deprivation and lack of available beds. They confirm the unacceptably high levels of bed occupancy (often more than 100%), which are particularly common in inner cities. However, they also suggest that a simple expansion of acute beds would not effectively address the problem.

Over-occupancy is the product of several factors. Firstly, acute beds are “blocked” by patients with relatively short stays who might be more appropriately accommodated elsewhere if suitable intensive community supports were available. Secondly, patients with longer stays need alternative provision of a more specialised and highly supervised kind, which mental hospitals used to provide but which is no longer available. Thirdly, patients with dangerous and violent behaviour need more secure facilities; this is a relatively small group of patients, mainly new long stay patients. Finally, there is a heterogenous group of patients with a range of special needs (for example, patients with acquired brain damage, dual diagnoses, and eating disorders), each subgroup of which are small in number, but who together comprise an appreciable proportion of those misplaced on acute admission wards. They often fall between different specialties (for example, neurology and psychiatry; learning disabilities and general psychiatry) or are simply not numerous enough for a single trust to be able to provide for them. If all the patients who are inappropriately placed on acute admission wards could be relocated then the problem of over-occupancy would be solved. Acute units could then look after only patients whose needs are for immediate treatment and stabilisation of symptoms.

Developing a spectrum of care

To achieve this a spectrum of care must be created, with several residential options for different levels of need and different kinds of problems. Thus, the concept of 24 hour nursed beds may be appropriate for some patients (those requiring longer term intensive rehabilitation in a domestic scale environment) but not for others. Considerable difficulties exist in achieving such a spectrum of care, particularly in the inner cities, with their background of poor housing, high unemployment, and other social problems. Traditionally, long stay patients were transferred out of inner city acute units to the old mental hospitals that ringed the conurbations; the budgets (and services) for mental health services in inner cities therefore never really adequately reflected the need to provide for the most disabled patients.8 Now that the mental hospitals do not exist or are much smaller, inner cities are struggling to develop specialised (and expensive) services for the most difficult patients, under the least favourable economic and social conditions. Our data also show that these problems are not confined to the inner cities, and there are examples of services in areas with relatively low levels of social deprivation that are in difficulty for precisely the same reasons as those in the inner cities–that is, they have reduced the availability of long term, highly supervised accommodation and have not replaced it with anything else.

These problems cannot be solved without some new resources. However, money will not be enough. Developing good community services means refocusing teams, retraining staff, and reorganising the service to provide access out of hours. At the heart of the service must be various kinds of sheltered and supported housing, and we were therefore concerned that many of our districts could not supply us with accurate information about the range and nature of housing provisions in their area. Similarly, although three quarters of the sample reported having multidisciplinary community mental health teams, we found only three examples of functioning crisis services and the same number of specialist, intensive support teams, specifically focusing on the most severely disabled people, with low caseloads and a commitment to long term care. Without this basic infrastructure, it is difficult to see how mental health services will ever be able to compensate for the loss of mental hospital beds.

Improving effectiveness

The disadvantages of traditional institutional care for long term mentally ill people far outweigh any potential advantages.14 Most people with long term mental health problems are more satisfied living in the community; they function better; and their care costs no more than it would do in large institutions.15 The care of patients with severe problems and complex needs may be expensive, but this will be the case wherever they are looked after. Intensive community support, combined with a brief inpatient stay when needed, is also superior to traditional inpatient care and follow up in terms of social outcomes, costs, and patient satisfaction.16 17 Specialist community teams, if properly resourced and organised, can maintain continuity of care, improve levels of patient satisfaction, and may even produce some clinical and social improvements.18

Community care is therefore possible, but all the elements in such a service–including the availability of acute inpatient beds–are interdependent. If community care is to succeed we cannot turn the clock back. We must find a new range of solutions, and this will require both money and ingenuity. We cannot afford to carry on with services and policies that are obviously not working. Neither can we afford to pin our hopes on any single, “magical” solution.


Funding: Grant from the Gatsby Charitable Trust.

Conflict of interest: None.


  1. 1.
  2. 2.
  3. 3.
  4. 4.
  5. 5.
  6. 6.
  7. 7.
  8. 8.
  9. 9.
  10. 10.
  11. 11.
  12. 12.
  13. 13.
  14. 14.
  15. 15.
  16. 16.
  17. 17.
  18. 18.
View Abstract