Background: Structured assessment of risk is a standard element of the core psychiatric assessment, particularly in respect of individuals, who are perceived to be at risk of violence or suicide. Many acute mental health trusts have introduced risk assessment tools, often without evidence of validity.

Aim: To evaluate a tool designed to predict the acute risk of violence by patients detained in a PICU.

Method: A risk assessment form was completed weekly for each patient on the PICU. Multivariate logistic regression analysis was used to measure the predictive value of particular variables in respect of violent behaviour during the next two weeks.

Results: Violent behaviour during the previous two weeks was a significant predictor of subsequent violence. Drug use was a negative predictor. None of the other variables tested were predictive of violent behaviour.

Conclusions: A complicated risk assessment tool probably adds little to the process of predicting acutely violent behaviour in PICU patients. The introduction of procedures such as structured risk assessment tools should be subject to the same validation process as any other clinical initiative.

Background: Psychiatric intensive care units (PICUs) have become a standard part of UK mental health services. They treat patients who may be very unwell and who are usually compulsorily detained. There is little published data about the clinical activity of such units to provide an evidence base for clinical governance.

Aim: To describe the socio-demographic characteristics, mental state and outcome of treatment for 332 patients admitted consecutively to seven English PICUs.

Method: Prospective, multi-centre case note analysis.

Results: PICU patients were predominantly Caucasian males, in their mid thirties, with complex needs and chronic psychotic illness often complicated by substance misuse. Most were admitted because of perceived risk of violence to others. Whilst most admissions appeared to be broadly in line with Department of Health guidelines some patients experienced an excessive length of PICU stay. Patients from particular BME groups were over-represented.

Conclusions: PICUs appear to deliver effective treatment but are not always the least restrictive environment as envisaged by the Department of Health. Further work is needed to evaluate treatment interventions and develop valid measures of quality of care.