Intended for healthcare professionals


Offending in psychiatric patients after discharge from medium secure units: prospective national cohort study

BMJ 2004; 328 doi: (Published 24 June 2004) Cite this as: BMJ 2004;328:1534
  1. A Maden, professor of forensic psychiatry (a.maden{at},
  2. F Scott, research worker2,
  3. R Burnett, research worker2,
  4. G H Lewis, professor of psychiatric epidemiology3,
  5. P Skapinakis, clinical research fellow4
  1. 1Academic Centre, West London Mental Health NHS Trust Headquarters, Southall, Middlesex UB1 3EU
  2. 2Institute of Psychiatry, London
  3. 3University of Bristol
  4. 4Department of Psychiatry, University of Bristol
  1. Correspondence to: A Maden
  • Accepted 8 March 2004


Medium secure units admit patients for the assessment and treatment of mental disorder associated with risk. We investigated how many patients offend after discharge.

Participants, methods, and results

We collected demographic and clinical data on all 959 patients discharged from medium secure units in England and Wales between 1 April 1997 and 31 March 1998 and used the offenders' index to find details of subsequent convictions.

A total of 145 patients (15%) were convicted within two years, including 60 (6%) who were convicted of violent offences (table).

Risk factors for conviction over two years for patients discharged from medium secure units in England and Wales (n=959)

View this table:


The rate of violent offending is low and the strongest association with offending was previous offending. Psychiatric variables were less important, with diagnosis and number of previous admissions showing no significant association. Substance misuse and sexual abuse were associated with increased offending risk, although patients were less likely to be convicted after a lengthy admission or if they had a history of self harm.

What is already known on this topic

Offending by psychiatric patients is of great public concern

What this study adds

In the two years after discharge from medium security, only 6% of patients commit a violent offence and the strongest association with reoffending is previous offending

Patients who lost contact with services were more likely to be convicted but this finding is ambiguous, as the sample included patients found to have no psychiatric diagnosis. It is consistent with the possibility that aftercare is effective in preventing reconviction, and we know that compulsory aftercare in restricted patients is associated with a rate of serious offending below 0.6% a year.1

Our study cannot establish causal relationships, but it shows that offending is uncommon. The rate of violent offending is so low that there is little scope for overall reduction and it would be better to concentrate on the identification of high risk patients. The American Macarthur study2 identified the “psychopathy checklist—screening version” (PCL-SV)3 as the best single predictor of violence in psychiatric patients, and we recommend further exploration of its use in medium security hospitals in England and Wales.

This article was posted on on 28 May 2004: //


  • Contributors AM designed and supervised the study and wrote the paper. FS and RB collected the data. FS, RB, and PS analysed the data and wrote the paper. GHL gave epidemiological advice on analysis, supervised the analysis, and wrote the paper. AM is guarantor.

  • Funding Department of Health.

  • Conflict of interests None declared

  • Ethical approval South Thames Multi-centre Research Ethics Committee.