Education And Debate

Dangerous patients with mental illness: increased risks warrant new policies, adequate resources, and appropriate legislation

BMJ 1996; 312 doi: (Published 13 April 1996) Cite this as: BMJ 1996;312:965
  1. Jeremy W Coid, professor of forensic psychiatrya
  1. a St Bartholomew's and the London Hospitals Medical College, London EC1A 7BE
  • Accepted 8 November 1995

British people have become increasingly concerned about the risk of violence from patients with severe mental illness who have been discharged into the community. A few incidents have received considerable media attention and left a strong impression of the potential dangerousness of psychotic patients. Subsequent inquiries have revealed serious shortcomings both in the clinical management and supervision of these patients, with failures in liaison and cooperation between agencies being coupled with inadequate service provision, particularly of secure inpatient beds.1 2 3 The government has become increasingly vulnerable to opposition or pressure groups that will seize any opportunity to criticise its programme of care in the community. Health ministers may now be questioning whether current policies are adequate to deal with a growing crisis of public confidence and whether the proposed new mental health legislation will deal with the problem.4

No evaluation of current policy for dangerous patients can be undertaken without knowing whether the general public is genuinely at increased risk of harm from mentally ill patients. Unfortunately, the Department of Health is not helped by lack of research in this subject and the recent shift of position by academics.5 6 It is widely feared that bed closures mean predisposed patients are spending longer periods at risk of offending in the community. But earlier research suggesting that cohorts of patients discharged from public mental hospitals showed increasing rates of offending over time is now seen as inadequate.5 7 The progressive reduction in beds has itself altered the picture. Inadequate data and poor methodology still continue to confuse debate on clinical management. For example, the risks that mentally disordered people will commit homicide have been claimed to be both higher and lower than the risks for the general population but without adequate data to support either assertion.8

Epidemiology of violence and mental illness

The few data available suggest that rates of homicides by mentally ill people may be remarkably similar in all societies, in stark contrast with overall homicide rates, which vary widely.9 This can lead to erroneous impressions of high risk in a country with an overall low homicide rate (such as the United Kingdom) and a negligible problem where the overall rate is high. As homicide remains comparatively rare, it is important to broaden the range of behaviours studied. Six major epidemiological studies carried out in the United States, Sweden, Switzerland, and the United Kingdom assessed whether the risks of other forms of violence are increased.10 11 12 13 14 15 Four were population based11 12 14 15 and the two others used a national case register.10 13 Three controlled for confounders and used operational diagnostic criteria.11 12 14 All but one15 found that people suffering from major mental illness are more dangerous than the general public in terms of self reported violent behaviour and official records of arrests and in terms of violent convictions. Risks of violence were increased about fourfold for men and were considerably higher for women in five of the studies. Risks of non-violent criminal behaviours were not increased for men. Factors that are associated with criminality and violence in the general population, such as younger age, male sex, low social class, and unemployment, also characterise violent mentally ill patients. However, statistical associations with major mental disorder still remained in studies that controlled for these factors.11 12 14 People with diagnosed drug and alcohol misuse are much more likely to be violent than those with a major mental disorder, but alcohol and drug misuse increases the risk in patients with a major mental illness. However, emerging findings of a strong association between violent behaviour and the times when symptoms of severe mental illness are active rather than in remission11 16 have major implications for future health care policy.

Identifying high risk patients

Research into the specific associations between individual symptoms and dangerous behaviour, along with other factors that might improve accuracy of identification of those who pose risks, still remains in its infancy.17 18 But three recent studies have indicated that violence is more likely in the specific context of perceived threat, sometimes leading to a pre-emptive strike, and when psychotic symptoms result in a weakening of self control mechanisms, such as thought insertion and mind dominance by outside forces (threat/control override).19 It is also becoming increasingly accepted that the true potential for dangerous behaviour may have been seriously underestimated. For example, a significant minority of patients show highly dangerous behaviour that is not officially recognised or processed through the criminal justice system, usually before being admitted.20 21 After admission, most patients are not violent, but a few “problem patients” cause a disproportionate number of incidents.22 Overall, one in 10 patients is estimated to commit violent assaults against staff. NHS staff are three times more likely than industrial workers to be injured, mainly through assaults.23

Most patients with severe mental illness do not pose a danger to themselves or the community. For example, even if the relative risk for mental illness and violence turns out to be as high as 4, it will account for only a small proportion of the annual number of violent acts in a population.24 Nevertheless, future mental health policy must come to terms with the emerging evidence that the overall risk of violence is still higher than that of the general population. And even if there was no proved increased risk to the public, it is still apparent that a small subgroup of mentally ill patients are manifestly dangerous and that in certain cases their dangerousness can be assessed and their behaviour predicted on the basis of past behaviour in given circumstances. Research is urgently needed to improve the accuracy of prediction for individual patients. However, research suggesting that the risks are greatest when symptoms are active11 16 has three clear implications for future policy.

Firstly, patients and their relatives should have ready access to skilled assessment and treatment, both as inpatients and in secure conditions.

Secondly, health care workers and social services employees in the community should receive better training in identifying and managing risk, as well as practical experience of dangerous patients.

Thirdly, patients who default from treatment, especially those who have behaved dangerously, should not be allowed to become ill in the community to the extent that they pose a risk to themselves or others.

Clinical indicators of risk


Previous violence

Restlessness or “social restlessness”

Poor compliance with treatment and aftercare

Substance misuse

Recent severe stress

Recent discontinuation of treatment

Evidence of planning, obtaining weapons


Loss of social support

Loss of accommodation

Access to potential victims identified in mental state abnormalities (see below)

Mental state

Threat/control override symptoms: persecutory delusions or delusions of passivity, or both

Emotions related to violence—for example, irritability, anger, hostility, suspiciousness

Specific threats made by patient

Risks are increased when:

  • Systematic assessment of risk not carried out

  • Risk indicators denied or minimised by responsible professionals

  • Information not passed from one professional to another

  • Clinical responsibility not clearly defined or transferred inappropriately

  • Inadequate community support (includes family and friends as well as community based services)

  • Carers unaware of services available locally

  • Provision of resources inadequate (includes inpatient beds and access to secure beds)

  • Management has failed to introduce a risk strategy appropriate to local circumstances. Includes policies and procedures for clinical risk asessment and management; induction training for new staff and continuous training for established staff; serious incident review; and clinical audit

Risk from patients or risks to professionals' careers?

To this end, clear instructions of good practice are now necessary to guide professionals on exactly when they should intervene, together with the minimum level of resources that should be provided to all local psychiatric services. The most dangerous patients do not readily accept supervision or treatment in the community and the care plan approach will reduce risk only in those who are willing to cooperate in its implementation. Similarly, merely placing such patients on a computerised register will not in itself ensure the public's safety. Assertive outreach programmes that aim to reduce violence by high risk patients25 have yet to be shown as effective in the United Kingdom and are unlikely to be adequately funded beyond a handful of model community services and districts where overall service demands are low. Recommendations have been made for the revision of current mental health legislation.2 At present, coercive treatment with recall to hospital is permitted only after the event of serious, dangerous behaviour and the imposition of a restriction order in a crown court. But recommendations of increased powers of supervision and recall to hospital of patients residing in the community1 26 were rejected by the government after advice that overtly forcing treatment in the community, other than under a restriction order, would contravene article 5 of the European Convention of Human Rights.27 Unfortunately, the Mental Health (Patients in the Community) Act 19954 contains few powers to effectively intervene in future and has been criticised as clinically unachievable when legal control of treatment and living arrangements cannot be practically extended to the community.28

Despite growing evidence of increased risk in a subgroup of severely mentally ill people, clinical intervention and the prevention of dangerous behaviour are hampered by vague policies, inadequate resources, and ineffective legislation. The Department of Health's guidelines for the care and treatment of discharged patients diverted attention from these issues.29 The introduction of mandatory inquiries after untoward incidents effectively displaces the blame downwards on to individual clinicians, neatly avoiding the wider context of psychiatric practice in the United Kingdom. The cost to morale and, in some cases, the careers of the professionals concerned has yet to be counted. Clinical and medicolegal responsibility for dangerous behaviour of mentally ill people in the community no longer has any clearly defined limits. Furthermore, the remit has been widened in the government's guidelines to include diagnostic categories where there is no convincing evidence one way or the other that psychiatric treatment is effective.29 30 This should no longer be passively accepted by psychiatrists. It may ultimately become unacceptable to ministers. It is they who will be repeatedly called on to justify their policies to an eagerly awaiting media after each inquiry is published.


  1. 1.
  2. 2.
  3. 3.
  4. 4.
  5. 5.
  6. 6.
  7. 7.
  8. 8.
  9. 9.
  10. 10.
  11. 11.
  12. 12.
  13. 13.
  14. 14.
  15. 15.
  16. 16.
  17. 17.
  18. 18.
  19. 19.
  20. 20.
  21. 21.
  22. 22.
  23. 23.
  24. 24.
  25. 25.
  26. 26.
  27. 27.
  28. 28.
  29. 29.
  30. 30.
View Abstract

Sign in

Log in through your institution