Numerous reports have highlighted the pressures on acute adult psychiatric beds, both in London (Reference Powell, Hollander and TobianskyPowell et al, 1995; Reference Johnson, Ramsay and ThornicroftJohnson et al, 1997) and nationally (Reference Shepherd, Beadsmoore and MooreShepherd et al, 1997). These have demonstrated associations between occupancy, deprivation and number of beds. In contrast, there has been no comprehensive national survey to determine the frequency and geographical spread, rather than the severity and causes, of such problems outside London.

Methods and analysis

Between November 1996 and May 1997, the chief executive of every NHS trust providing in-patient adult psychiatric services in the UK was sent an open-ended questionnaire requesting information on:

1. (a) the size of their catchment area population;

2. (b) the number of acute adult psychiatric beds (excluding specialist beds, for example those for eating disorders, puerperal disorders and intensive care);

3. (c) the frequency of problems with bed availability and over-occupancy;

4. (d) how often the trust used extra-contractual referrals (ECRs).

Trusts within Greater London were excluded. Nonresponders were followed up with two repeat mailings and telephone calls.

Responses were assessed and graded into three groups (rarely, sometimes and often) independently by D. H. and R. S. (see table footnote). The responses were rated as follows: rarely includes never, rarely, and very rarely; sometimes includes occasionally and sometimes; often includes often, frequently and continuously.

Trusts were grouped by country, region and health authority within which they were located, by Office for National Statistics area classification and by level of deprivation (Jarman UPA-8 score <1, 1-25, >25) for the host health authority, and by the number of beds per thousand population (<0.22, 0.22-0.42, >0.42). χ²-tests of significance were calculated to assess differences between groups of trusts and are presented as odds ratios with 95% confidence intervals.

Findings

One hundred and seventy-three (82%) of the 210 acute adult mental health trusts outside Greater London replied (64% Northern Ireland, 80% Scotland, 81% Wales, 84% England). Fifty-three (31%) trusts were rarely or never over-occupied, while 30 (17%) often had such problems. The findings for bed availability are similar. Eleven trusts frequently use ECRs, all except one were in England (see Table 1).

Table 1. Results of survey

	Country/Region	Frequency of problem			Total	OR (95% CI)1
		Never, rarely	Sometimes	Often
Beds over-occupied	England	38 (27.6%)	74 (53.6%)	26 (18.8%)	138 (100%)	1.0
	Wales	4 (44.4%)	4 (44.4%)	1 (11.1%)	9 (100%)	0.48 (0.12-1.86)
	Scotland	10 (58.8%)	5 (29.4%)	2 (11.7%)	17 (100%)	0.27 (0.09-0.75)*
	Northern Ireland	1 (11.1%)	7 (77.7%)	1 (11.1%)	9 (100%)	3.04 (0.37-25.13)
	Total	53 (30.6%)	90 (52%)	30 (17.4%)	173 (100%)	-
Beds not available	England	41 (29.8%)	76 (55%)	21 (15.2%)	138 (100%)	1.0
	Wales	4 (44.4%)	4 (44.4%)	1 (11.1%)	9 (100%)	0.53 (0.14-2.07)
	Scotland	10 (58.8%)	5 (29.4%)	2 (11.7%)	17 (100%)	0.30 (0.1-0.83)*
	Northern Ireland	1 (11.1%)	7 (77.7%)	1 (11.1%)	9 (100%)	3.38 (0.41-27.9)
	Total	56 (32.3%)	92 (53.2%)	25 (14.5%)	173 (100%)	-
Extra-contractual referrals	England	79 (57.2%)	49 (35.5%)	10 (7.2%)	138 (100%)	1.0
	Wales	5 (55.6%)	4 (44.4%)	0	9 (100%)	1.07 (0.28-4.16)
	Scotland	17 (100%)	0	0	17 (100%)	0.0
	Northern Ireland	5 (55.5%)	3 (33.3%)	1 (11.1%)	9 (100%)	1.07 (0.28-4.16)
	Total	106 (61.3%)	56 (32.4%)	11 (6.3%)	173 (100%)	-
Extra-contractual referrals used:	Southern four2	27 (44.3%)	26 (42.6%)	8 (13.1%)	61 (100%)	1.0
English regions	Northern four2	52 (67.5%)	23 (29.9%)	2 (2.6%)	77 (100%)	0.4 (0.2-0.8)*
	England total	79 (57.2%)	49 (35.5%)	10 (7.2%)	138 (100%)	-

No Scottish trust reported using ECRs, and problems were relatively infrequent in Wales and Northern Ireland. Over-occupancy occurs more often in English trusts (19% v. 11%) as does the use of ECRs (43% v. 23%) but the differences are only significant for Scottish trusts.

Trusts which often use ECRs were significantly more likely to have fewer than 0.22 beds per thousand population than others (odds ratio 5.8; 95% Cl 1.3-35.0). Those in England are significantly more likely to be located in one of the four southern regions (ever use ECRs: odds ratio 2.62; 95% Cl 1.24-5.57; often use ECRs: odds ratio 5.66; 1.06-56.14). There is no similar association for bed availability or over-occupancy.

Trusts reporting frequent problems with over-occupancy were significantly more likely to be situated in health authorities with Jarman scores above 25 (odds ratio 8.1; 1.7-41.8). There is no relationship between ECR use and deprivation, nor between any of the measures and area classification.

Five of the seven trusts reporting both frequent use of ECRs and frequent problems with bed availability were in one of the four most southerly English regions and none were outside England.

There is a significant correlation between bed density and the deprivation score (r=0.337, P=0.001).

Discussion

This survey was undertaken to quantify problems with bed availability nationally and to compare those previously reported for London. Hirsch et al (1998) have defined ideal bed occupancy as 85%, allowing a margin for the unexpected. However, Lelliott et al (further details available from author upon request) reported levels up to 130%; so this study started from the premise that, in practice, bed occupancy up to 100% did not constitute over-occupancy.

The methods used differ from previous studies in particular by using open-ended questions. However, despite the possibility of response bias (which could be expected to exaggerate the extent of problems) this simple approach was adopted to obtain a rapid overview of the situation and is justified by the high response rate.

We did not observe associations between bed occupancy or use of ECRs and the type of area served by the trust. This may be because the Office for National Statistics classifications used relate to the host health authority, which is not necessarily the same as the catchment area served by the trust. However, as reported previously (Reference Jarman, Hirsch and WhiteJarman et al, 1992; Reference Powell, Hollander and TobianskyPowell et al, 1995; Reference Shepherd, Beadsmoore and MooreShepherd et al, 1997), we noted relationships with deprivation and bed density. We therefore believe the findings are valid.

English trusts (particularly in southern regions) experience greater pressure on beds and consequently use ECRs more often than elsewhere. However, compared with reported mean four-year bed occupancy figures for London trusts of 98% (Reference Powell, Hollander and TobianskyPowell et al, 1995), and subsequent increases (Reference Hollander, Powell and TobianskyHollander et al, 1996), problems appear considerably less severe nationally than in the capital.

Nonetheless, we did observe frequent problems with bed availability in individual cities and trusts, suggesting the need for more detailed study using, for example, bed census models. In addition, certain rural areas face surprisingly frequent pressures on beds. The question raised is whether the development of community-based crisis teams, day hospital places and assertive outreach teams, as proposed by the government, will reduce these pressures. Certainly, these preliminary findings would not support further bed reductions in the absence of a greatly improved community infrastructure.