Assessment and management of risk is now considered an essential component of good psychiatric practice. The National Service Framework for Mental Health (Department of Health, 1999) states that all mental health service users on the Care Programme Approach should receive care that optimises engagement, anticipates or prevents a crisis and reduces risk. The Report of the Confidential Enquiry into Homicides and Suicides by Mentally Ill People (Steering Committee of the Confidential Inquiry into Homicides and Suicides by Mentally Ill People, 1996) noted in cases of homicide that previous convictions were frequently not documented in the mental health case notes. Vinestock (Reference Vinestock1996) wrote that the information gathering process, the maintenance of good medical records and good communication form the basis of the management of the risk of violence. With incomplete or inaccurate information the assessment is likely to be flawed. The Clunis Report (Reference RitchieRitchie, 1994) recommends increased research and audit into risk of violence; however we were unable to find any published audit in this area.

Approved standards

The audit form was based on the Assessment and Clinical Management of Risk of Harm to Other People (Royal College of Psychiatrists, 1996). This council report contains guidance in the form of a list of important risk factors in the history, mental state and environment (see Table 1). In conclusion, it is suggested that a formulation should be made based on these factors and others in the history.

Table 1.

Number of newly assessed patients who have specific items of risk addressed

	Before	Percentage	After	Percentage	χ2	P
Number of patients	47		50
History
Availability of collateral information	10	21	5	10	1.6	NS
Forensic history	9	19	27	54	11.1	0.001
History of violent thoughts	7	15	23	56	9.5	0.01
History of violent actions and their nature	9	19	21	42	4.9	0.05
Presence/absence of remorse of serious past events	4	8	10	20	1.7	NS
Alcohol history	25	53	43	86	10.9	0.001
Substance misuse history	19	40	35	70	7.4	0.01
Previous deliberate self-harm/deliberate risky behaviour	17	36	33	66	7.4	0.01
Personality assessment	28	60	26	52	0.3	NS
Impulsiveness	13	28	12	24	0.03	NS
Nature of fantasies	6	13	10	20	0.5	NS
Nature of compliance with treatment/services	31	66	23	46	3.1	NS
Recent loss — person, lifestyle, expectations	31	66	29	52	0.4	NS
Unresolved source of stress	28	60	34	68	0.4	NS
Rootlessness/‘social restlessnes’	19	40	34	68	6.3	0.02
Mental state examination
Violent thoughts	4	15	18	36	8.9	0.01
Homicidal thoughts	2	4	13	26	7.2	0.01
Emotions related to violence	4	8	12	24	3.6	NS
Acting on delusions	5	11	10	20	0.99	NS
Acting on hallucinations	4	8	10	20	1.7	NS
Risk assessment statement	12	26	17	34	0.5	NS
Risk management plan	5	11	9	18	0.6	NS

In addition, a number of other items were included on the audit form, such as whether collateral information was available, a history of violent thoughts, attitude to any violent incidents and a personality assessment, including nature of fantasies and interests and presence of impulsiveness (Reference VinestockVinestock, 1996).

The study

The study was carried out by a community mental health team (CMHT), which serves the general adult population of 18-75-year-olds in a multicultural inner-city area of South-West London. The team has a multi-disciplinary composition and adopts a flexible approach to the assessment of new referrals. A form was developed and agreed by team members to act as a guideline for history taking and mental state examination.

Initially we completed a baseline, retrospective audit of the assessments of 46 consecutive referrals to the team. The College standard for taking a history of risk was taught at an hour-long meeting. This was incorporated into the regular meeting time of the team and was led by M.C. No additional resources were necessary. All team members, including social workers (non-trust employees), attended and participated in data collection.

Following the training, a further 50 consecutive assessments were audited in a similar way. Chi-squared analyses were used to measure significant differences in the items included on the form. Results have been fed back to the team and a continuous cycle of audit will be maintained.

Findings

The recording of items in the history, mental state examination and risk assessment statement is presented in Table 1. There was an increase in recording in 17 of the 20 items relating to the history and mental state, of which nine (45%) reached statistical significance. Significant improvement occurred in the recording of forensic history, violent thoughts, violent actions, alcohol and substance misuse history and self-harm behaviour following the intervention. Relevant items in the mental state examination were recorded infrequently both at baseline and follow-up. However, there were improvements in recorded frequency of all items, in particular violent and homicidal thoughts. The recording of a risk assessment statement (e.g. low, medium and high) and risk management plan did not increase following the teaching, remaining at a low level.

Comment

The most obvious finding was the neglect of attention to risk of violence in the initial sample assessed. These scores were considered worryingly low by the team. However non-urgent, general practitioner referrals are probably at low risk of harm to others compared to inpatient or CMHT case-load cohorts. Many of the patients assessed had already been discharged back to primary care at the time of the audit. However, presumption of a low-risk of violence may lead to the ‘collusive denial’ described by Bowden (Reference Bowden1997). A percentage of these patients required ongoing treatment and/or had serious mental illness and may not have had an adequate risk assessment on their initial appointment. Alcohol and drug misuse are strongly associated with psychiatric illness and risk of violence, but were recorded in only 53% and 40% of assessments, respectively. Omission of this part of the history is probably best explained by the lack of a systematic approach of the team to assessment.

Our findings at baseline are supported by Sanders et al (Reference Sanders, Milne and Brown2000), who found that clinical enquiry regarding recent violent thoughts and behaviour and ongoing violent thoughts was rarely recorded in the case notes of psychiatric in-patient admissions. Aggressive ideation regarding damage to property and interpersonal violence was recorded in 2.6% and 13.2% of cases, respectively.

The audit has, however, demonstrated three important points. First, it is possible to make significant improvements in the recording of the relevant aspects of history taking with a minimal intervention and without the need of additional finance or team time. Second, the improvements were gained at a cost of no extra paperwork. And finally, the audit was prioritised, designed and carried out in a true multidisciplinary setting. The main goal of risk assessment audit, to demonstrate an effect on the actual incidence of violent incidents, needs to be studied in further research.

Recommendations

1. (a) Taking a full risk of violence history is necessary to manage risk appropriately. Risk assessment forms cannot be completed if the relevant information is not available.

2. (b) CMHTs should regularly audit the quality of their notes with regard to assessment of risk of violence.

3. (c) A structured approach to assessment will improve the comprehensiveness of risk assessment.