Method

All attendances aged 16 and over at the Leicester Royal Infirmary A&E department with a diagnostic coding of “self inflicted injury” (n=1598), or “psychiatric illness” (n=330) were identified from the computerised database in the department. It is unlikely that episodes of DSH would have been given any other diagnostic coding. The A&E records were located for 1769 attendances and cases excluded if they were not episodes of DSH as defined by Morgan.⁵ A total of 1359 episodes of adult DSH were identified, 1277 from those coded as “self inflicted injury”, and 82 from those coded as “psychiatric illness”. The quality of psychosocial assessment was then examined using the same audit instrument as the first audit; this was based upon the Royal College of Psychiatrists statement on the standards of service for the general hospital management of adult DSH,³ and has been published previously.² The standards for assessment by non-psychiatric medical staff in the A&E department requires specific information to be collected and documented in the case notes, in particular concerning; conscious level, psychiatric history and mental state examination, social situation and events, alcohol and drug use, risk, and arrangements for any follow up. An experienced research associate, supervised by the psychiatrists on the audit team, was responsible for examining the case notes and using the audit instrument to assess whether the information required had been recorded. The results of the audit were then compared with those from the first audit (1994/5, n=934).² Case ascertainment in the first audit was similar, although only the notes of those with a discharge diagnosis of “self inflicted injury” had been scrutinised for DSH cases. A 12 month period was chosen for both audits so as to span three rotations of junior medical staff.

If any patient left the department before, or during the A&E doctors' assessment they were excluded from the analysis, even though some information may have been recorded from other sources.

Results

Table 1 shows the comparison of the information recorded in the case notes by A&E medical staff for the two audits. The assessment checklist was attached to clinical notes in only 549 episodes (40.4%) and was fully completed in 73% of cases. Even when excluding those cases with the checklist present there was still a significant improvement in documentation, in particular for: conscious level (χ² = 72.2, p < 0.001); psychiatric history (χ² = 201, p < 0.001); medical history (χ² = 455, p < 0.001); previous DSH (χ² = 9.6, p = 0.002); threats of DSH (χ² = 14, p < 0.001); chronic alcohol use (χ² = 22, p < 0.001); illicit drug use (χ² = 7.82, p = 0.005); suicide intent (χ² = 8.8, p = 0.003); and risk of further DSH (χ² = 13.2, p < 0.001).

There were significant changes in treatment between the two audits, with a higher proportion in 1997/8 (362, 26.5%) assessed by a mental health specialist (DSH team nurse specialist, or on call SHO in psychiatry) in the department than in 1994/5 (154, 16.5%; χ² = 33, p < 0.001). In 1997/8 an even smaller proportion of patients were admitted directly to medical and surgical wards than in 1994/5, but interestingly a greater proportion were admitted to psychiatric wards during the period of the second audit (table 2).

When comparing the quality of documentation for episodes admitted to medical or surgical wards with those not admitted, there were significantly higher frequencies of recorded information for those not admitted for all factors (p <0.01) apart from conscious level and medical history. The frequency of information recorded for those not admitted was particularly encouraging—especially for psychiatric history (97%), recent stresses (85%), previous DSH (80%), suicide intent (87%), and risk of further DSH (84%).

For DSH cases attending during the second audit (1997/8), we compared the quality of assessment for cases identified from a coding of “self inflicted injury” with those identified from a coding of “psychiatric illness” and found no significant differences in the documented psychosocial assessment. The improvement in the quality of documentation found in the second audit was therefore not in any way contributed to by the minor differences in case ascertainment.

Discussion

We have shown a substantial improvement in the quality of documentation of assessment in the second audit compared with the first. By studying the management of three rotations of junior doctor it was unlikely that the improvement was attributable to a cohort effect of one particular group of senior house officers. Considering the service initiatives that occurred it is probable that these interventions were an important factor in the improvement that was achieved. Encouraging staff to use a comprehensive checklist, derived from the Health Advisory Service review on suicide prevention⁴ proved particularly beneficial. It was disappointing that the checklist was attached to the notes in only 40% of cases. However, in some instances the case notes indicated that the sheet had been completed but this was not present in the patients' file. The fact that documentation still improved even when the checklist was absent might indicate that the seminar was having a positive impact and senior house officers had learnt the rudiments of a psychosocial assessment. The fact that the checklist was only fully completed in 73% of cases was again disappointing, but this may reflect the amount of time required to complete such a comprehensive form.

Crawford and colleagues have also showed an improvement in the quality of DSH assessment by A&E staff after a one hour teaching seminar.⁶ They encouraged staff to use a proforma, which include the SAD PERSON scale.⁷ The assessment was more likely to be rated as “adequate” if the proforma had been used. The SAD PERSON scale is a much less comprehensive proforma than the checklist used in the Leicester A&E department. Hockberger and Rothstein⁸ have modified the SAD PERSON scale and shown that by using this non-psychiatric medical staff can confidently identify those patients requiring a specialist assessment. Others have shown the usefulness of risk assessment checklists for house physicians on medical wards.^9,10

As previously² there was evidence of selective documentation with much higher rates of recorded information for episodes not admitted directly to medical or surgical wards. It is probable that a proportion of patients admitted directly to general hospital wards would have had a lowering of consciousness hampering their assessment, and A&E staff would also expect these patients to receive a comprehensive psychosocial assessment during the course of their admission.

There were clear changes in the patterns of management of DSH patients in the second audit compared with the first; the increased rates of assessment by mental health specialists and psychiatric admission during the second audit might indicate better detection of those at risk of suicide. However, in an increasingly litigious climate, it could also be that medical staff are aware of the importance of careful documentation and assessment, and are more likely to refer patients for a specialist assessment if they have any doubts. The reduction in the proportion admitted directly to medical or surgical wards may reflect the increased pressure on general hospital bed usage. Encouragingly, the number of patients who were referred to a mental health specialist but left the department before being seen halved in the second audit (from 46 to 24), and only six left the department before being seen by the A&E doctor. Crawford and Wessely showed that patients who discharge themselves before the completion of an initial assessment have a considerably increased risk of repeating self harm.¹¹

It is not known if improving the assessment of DSH patients in A&E improves outcome. Further evaluation of whether improved assessment leads to a reduction of repetition of DSH, or subsequent suicide is therefore required. A follow up study of patients receiving an “adequate” psychosocial assessment, and of those referred to mental health services may provide important information.