Presentations to accident and emergency (A & E) departments after deliberate self-harm (DSH) are common (over 150 000 annually in the UK) and becoming more so (Reference House, Owens and PatchettHouse et al, 1998). These patients suffer high rates of psychiatric disorder (Reference Morgan, Burns and PocockMorgan et al, 1975) and are 100 times more likely than the average member of the population to commit suicide in the year after presentation (Reference Greer and BagleyGreer & Bagley, 1971). Although patient assessments do not reliably predict the risk of future DSH (Reference Hawton and FaggHawton & Fagg, 1995), they can identify patients with treatable mental disorder (Reference House, Owens and PatchettHouse et al, 1998). The Department of Health recommends that every patient have a ‘specialist psychosocial assessment’ before discharge from hospital (Department of Health and Social Security, 1984). Assessments by A & E staff alone have been shown to omit important items such as continuing suicidal ideation (Reference Black and CreedBlack & Creed, 1988; Reference O'Dwyer, Dalton and PearceO'Dwyer et al, 1991; Reference Ebbage, Farr and SkinnerEbbage et al, 1994).

At present, the majority of screening assessments are carried out by junior doctors on emergency rotas (Reference House, Owens and PatchettHouse et al, 1998), although suitably trained nurses and social workers can perform assessments of similar quality (Reference Newson-Smith and HirschNewson-Smith & Hirsch, 1979; Reference Catalan, Marsack and HawtonCatalan et al, 1980). In October 1998, a team of part-time specialist nurses supported by a psychiatrist was set up at Kettering general hospital. This team now sees the majority of patients: specifically, those presenting to A & E between 5.00 p.m. and 1.00 a.m. and those admitted to a general hospital bed. We examined quality variables linked to DSH psychiatric assessment by repeating an audit conducted in 1994 (Reference Gordon and BlewettGordon & Blewett, 1995), when most assessments were by on-call trainee psychiatrists.

Method

Both the 1994 audit and this 1999 reaudit used the same reported criteria (Reference Black and CreedBlack & Creed, 1988). We set out to examine 50 consecutive sets of notes for DSH patients presenting to A & E from March 1999, aged 16 years or over. Cases were identified using the A & E computerised patient database. Four patients' case notes were untraceable; the 50 included patients were therefore drawn from the first 54 patients presenting after 1 March 1999. To avoid bias, staff in A & E and on the medical admissions unit, and psychiatric staff performing assessments, were not told when data were to be collected, although all were aware that an audit was planned.

The samples were analysed in checklist fashion according to 58 predetermined criteria, for example: “Had the A & E senior house officer (SHO) made any note, however brief, of mental state?” To ensure comparability, the 1994 data were reanalysed using the 1999 criteria. Data analysis was completed by S.W. A sub-sample was checked for agreement by A.B., with complete concurrence. The patients concerned were almost all unknown to S.W., and in most cases he did not know which staff member had performed the assessment. Measures of statistical significance were all based on a simple comparison of independent proportions, analogous to McNemar's test for paired proportions (Reference MouldMould, 1998).

Results

In the 1999 audit, 25 of the 50 study patients were admitted to a general hospital bed and seven were eventually admitted to a psychiatric bed (Fig. 1). Thirtynine were assessed by trained psychiatric personnel: 15 in A & E and 24 in a general hospital bed. One patient discharged himself after being admitted to a medical bed with a prior non-specialist A & E assessment and no further psychiatric evaluation. The projected annual DSH presentation rate rose from 521 in 1994 to 852 in 1999. Patient characteristics in both samples were similar for age group and method of self-harm and consistent with published studies (Reference Charlton, Kelly and DunnelCharlton et al, 1993; Reference McLoone and CrombieMcLoone & Crombie, 1996). There is a continuing predominance of analgesic poisoning as a method of self-harm and a low female to male ratio: 1:1 in 1994 and just under 2:1 in 1999.

Fig. 1 Assessment and admission patterns for patients presenting to the accident and emergency (A & E) department. One of the three patients admitted directly to a psychiatric ward in 1994 was a readmission of a patient who had discharged himself without permission. One of the patients admitted to a medical ward in 1999 discharged himself before assessment by the medical senior house officer (SHO)

The proportion of patients receiving a specialist assessment before discharge from hospital rose from 24/50 (48%) in 1994 to 39/50 (78%) in 1999. The change in service structure was also reflected in the type of specialist assessment. In 1994, 17 (71%) of the 24 specialist assessments were completed by on-call SHOs and the remainder by community nurses. In 1999, 34/39 (87%) were completed by a member of the DSH team and the remainder by on-call SHOs. The proportion of patients seen by a mental health worker who were referred on to a statutory sector agency other than the general practitioner fell significantly, from 16/24 (67%) in 1994 to 12/39 (31%) in 1999 (P<0.01 for difference). This is contrary to the suggestion that non-medical staff performing DSH assessments recommend psychiatric follow-up more often than doctors do (Reference Newson-Smith, Farmer and HirschNewson-Smith, 1988).

Bed utilisation did not significantly alter. The medical admission rate increased marginally, from 22/50 (44%) in 1994 to 25/50 (50%) in 1999 (P>0.5, NS), and eventual new admission to a psychiatric bed rose from 4/50 (8%) to 7/50 (14%) (P>0.1, NS). In contrast, the change in rate of eventual psychiatric admission following specialist assessment was unremarkable: 4/24 (17%) in 1994 compared to 7/39 (18%) in 1999 (P>0.5, NS).

The most consistent and significant change was of improved quality of specialist assessments in the second audit (Table 1). The quality of assessments by A & E and medical ward staff either remained poor or declined slightly. Data for assessments by medical ward staff not shown.

Table 1. Quality of specialist deliberate self-harm (DSH) and accident and emergency (A & E) mental state assessments

	Assessments on which parameter was documented (n (%))
Parameter documented on assessment	In 1994	In 1999	Test for difference
Specialist DSH assessments
Act of self-harm	11 (46)	39 (100)	P<0.001
Precipitating factors	17 (71)	37 (95)	P<0.01
Whether alcohol was used	3 (13)	38 (97)	P<0.001
Degree of planning	16 (67)	38 (97)	P<0.001
Social circumstances	14 (58)	39 (100)	P<0.001
Past psychiatric history	9 (38)	39 (100)	P<0.001
Family psychiatric history	5 (21)	36 (92)	P<0.001
Patient's mental state	14 (58)	38 (97)	P<0.001
Ongoing suicide risk	15 (63)	38 (97)	P<0.001
Plan of action	19 (79)	39 (100)	P<0.01
Total number of assessments	24	39
A & E assessments
Act of self-harm	50 (100)	50 (100)	P=NS
Whether alcohol was used	50 (100)	17 (34)	P<0.0011
Past psychiatric history	14 (28)	36 (72)	P<0.001
Patient's mental state	31 (62)	40 (80)	P<0.05
Ongoing suicide risk	24 (48)	21 (42)	P=NS
Total number of assessments	50	50

Comments

This paper replicates earlier published findings that a trained team of nurses can perform high-quality assessments, and shows a substantial and significant improvement in assessment quality after such a team was set up in Kettering. There was no evidence that this improvement generalises to non-specialist staff working in A & E and medical assessment units.

There are weaknesses intrinsic to the design of this audit. The two case series are relatively small, and we were not systematically blinded to the identity of patients or assessors. Failure to identify patients using the information system in A & E was a risk, although the 1994 sample was hand-checked and previous experience suggested that data entry and extraction by predetermined codes minimised errors.

The data presented are an accurate reflection of the original 58 criteria. The 14 criteria not described here include additional demographic variables. Further items are presented in a condensed form; for example, “was there a documented plan of action?” is a conflation of “was there an immediate management plan?”, “was there a follow-up plan?” and “was there a decision to admit?”.

We considered that improvement in quality of documented assessments was attributable to the development of a specialist team with dedicated staff and time. The team's cohesion and positive ethos, with a strong emphasis on training, monitoring of standards and mutual support, is experienced as very important.

There is still a paucity of firm evidence guiding interventions aimed at reducing repetition of DSH or suicide following it. This audit focused on the broader question of assessment quality, in the reasonable expectation that if treatable disorders are identified patients will be more likely to access appropriate help. Despite improvements, the current service arrangement in Kettering does not yet ensure that every patient has an adequate psychosocial assessment, which begs a question about whether its clinical activities should be expanded. A major concern is that consistent and satisfactory basic mental health evaluation and management of DSH patients by non-specialist staff have not yet been achieved.