Suicide and homicide by people with mental illness

The national confidential inquiry into suicide and homicide by people with mental illness began in 1992 in response to concern about mental health services in the United Kingdom. The usefulness of the initial reports was limited by the disappointing case ascertainment rate.^1–3 Two papers in this issue (pp 1235, 1240)⁴,⁵ report the methods and results of Safer Services, the 1999 inquiry report.⁶ Case finding has now been much improved and the new report provides a valuable descriptive cross section of the characteristics of suicides and homicides in relation to the mental health services.

About 1000 people who commit suicide each year (a quarter of all UK suicides) and about 40 of those who commit homicide (about 8% of all UK homicides) have had some contact with the mental health services in the year before death. In patients committing suicide comorbidity, including substance misuse, and previous self harm are common. In people convicted of homicide, personality disorder and substance misuse are common; fewer than 10 homicides each year are committed by people with a primary diagnosis of schizophrenia.

In the BMJ papers⁴,⁵ the authors correctly emphasise that systematic reviews have found that no interventions have reliably been shown to prevent suicide or, indeed, deliberate self harm.⁷,⁸ However, the report itself makes 31 recommendations for changes in clinical practice.⁶ These include recommendations about training in risk assessment, documentation (including the introduction of “patient passports”), the use of specific drug and psychological treatments, reducing access to means of suicide, and changes in the Mental Health Act to allow compulsory community treatment. Policymakers should, however, be cautious about implementing these wide-ranging recommendations because there are substantial uncertainties, largely unacknowledged in the report, in our current knowledge about suicide prevention.

Although we have some information about risk factors for suicide, we have very little reliable knowledge about the accurate clinical quantification of risk, a prerequisite for effective risk assessment.^9–11 One of the main problems is that even in high risk groups suicide is rare. The report identifies the period after discharge from hospital as being a high risk period. Cohort studies show that the rate of suicide in the first 28 days after discharge is between about 1 in 500 and 1 in 1000 patients discharged.¹²,¹³ This low incidence rate, coupled with the limited sensitivity and specificity of current risk assessments, means that the positive predictive value is low and the number of false positives high.¹⁰,¹¹ For example, even if a risk assessment had a sensitivity and specificity of 80% (which probably exceeds those currently available), for every 20 000 patients discharged, 40 would commit suicide—32 of whom would be identified as high risk. However, in total 4024 patients would be considered to be high risk, 3992 of whom would be false positives. Thus recommendations for the clinical management of high risk groups will apply to large numbers of patients.

The report suggests that improving compliance by a community treatment order might prevent 30 suicides and two homicides. But even if there were evidence that such a strategy was effective, the number needed to treat to achieve this would be enormous. The humanitarian implications and opportunity costs of the recommendations will be substantial. Mental health services can be improved in many ways, and it would be wrong to focus all our training and service development resources on these important, but rare, events.

Furthermore, we should not miss this valuable opportunity to recognise the substantial uncertainty about this subject and to make recommendations about research priorities. Studies into risk factors for suicide and homicide, as in the rest of psychiatry, typically need to be at least an order of magnitude larger than at present. The sample on which the report is based should be used as the basis for case-control studies to develop possible risk assessment tools. Recognising that the low base rate of suicide means that many patients will need to be treated to prevent one suicide, we also need large randomised trials of widely practicable interventions. Realistically, we may never be able to use suicide as a primary outcome in randomised controlled trials. Trials will therefore need to recruit from high risk groups to increase event rates and use deliberate self harm as an outcome. They will need to be carefully designed in the knowledge that the results will need to be extrapolated to other groups.

The report also recommends that individual local inquiries into homicide should be discontinued. They should be discontinued, but not, as Safer Services suggests, because they perpetuate a “climate of blame.” It is right that mental health services should be accountable for failures of care, and all public services are increasingly subject to external review. They should be discontinued because they are inefficient and methodologically inadequate for making general recommendations about future UK mental healthcare provision.¹⁴ Being retrospective, they foster a simplistic notion of the preventability of homicides and suicides.

Papers pp 1235, 1240