THE SUICIDE OF PATIENTS IS TRAUMATIC FOR PSYCHIATRISTS
As Gitlin (Reference Gitlin1999) noted recently: “In the literature on suicide, among the least commonly discussed topics is the reaction of mental health professionals when one of their patients in treatment commits suicide”. Furthermore, this sparse literature has generally been anecdotal and based on single cases or on small samples. Chemtob et al (Reference Chemtob, Hamada and Bauer1988) obtained a 46% response from 643 psychiatrists in the USA. More than half of the respondents had experienced one suicide or more among their patients, and such events had a significant impact upon their professional and personal lives. In our survey of Scottish consultants and specialist registrars in psychiatry respectively (Reference Alexander, Klein and GrayAlexander et al, 2000; Reference Dewar, Eagles and KleinDewar et al, 2000), 67% of consultants and 47% of senior trainees reported experiencing one or more suicide of patients in their care. In response to their most distressing suicide, one-third of consultants reported symptoms suggestive of depression which endured for over a month in 60% of cases (Reference Alexander, Klein and GrayAlexander et al, 2000). Themes which commonly emerged included guilt, lack of support, the ‘blame culture’ and unrealistic expectations (from within and outwith the profession) that suicide was usually preventable. Particularly in general adult psychiatry, suicides were relatively common, with a broad expectation of between 10 and 20 during the course of one's career.
HEALTH OF THE NATION TARGETS
In 1992, targets for reducing suicide rates by the year 2000 were set in England (Department of Health, 1992). These comprised a reduction of 15% in overall suicide rates and a reduction of 33% among those with serious mental illness. A further target of reducing suicide rates by 17% from the 1996 baseline by the year 2010 was later proposed (Department of Health, 1998).
In 1998, we sought the views of consultants and senior trainees in Scotland on the 1992 targets, asking if they thought there should be such targets and, if they did, whether they considered that the proposed rate of reduction was appropriate. For reduction in overall suicide rates by 15%, of the 346 respondents only 17% agreed with both the existence and the level of the targeted reduction. For serious mental illness, only 14% agreed with both the existence and the level of the target.
When they were first invoked, the targets inspired lively debate, much of which focused on the predictability and preventability of suicide (see below). Rather less has been said about the intrinsic desirability and relevance of such targets. Hawton (Reference Hawton1998) has suggested that imperfect targets are better than no targets since at least they focus attention on the needs of people with mental health problems. The difficulties with this argument are twofold. First, there has been a recent slight decrease in suicide rates, for which psychiatric services may seek to claim credit (Reference McClureMcClure, 2000). But this attribution may be deemed equitable only if we also carry the responsibility/blame when they increase. However, if we are meeting our major target, does this indicate that mental health services are adequate? Second, to take the reductio ad absurdum, however inappropriate a target is, is it better than no target at all? If the success of meeting targets is determined by factors largely outwith our control, does this circumstance not commit psychiatry to being a hostage to fortune?
TO WHAT EXTENT CAN SUICIDE BE PREDICTED AND PREVENTED?
This paper cannot do justice to the literature on these topics, and a précis will be followed by the views of psychiatrists in Scotland.
Prediction
Beck et al (Reference Beck, Brown and Steer1999) purported to identify risk factors for suicide among out-patients with depression. However, the suicide rate among these patients was 0.2% per annum, so that even the higher-risk patients within this high-risk group (with five times the average risk) would have a 1 in 100 annual risk of suicide. Among psychiatric inpatients, Powell et al (Reference Powell, Geddes and Deeks2000) concluded that only 2% of patients with a suicide risk of 1 in 20 or higher could be correctly identified by the five best predictive factors. Thus, while psychiatric patients are at increased risk of suicide, suicide is relatively rare, and our predictive capabilities are very limited.
Prevention
If prediction of suicide is difficult, it follows that prevention is even more problematic. Gunnell & Frankel (Reference Gunnell and Frankel1994) concluded that no intervention has been shown to reduce suicide in a well-conducted randomised trial and this situation remains virtually unchanged (Reference Appleby, Shaw and AmosAppleby et al, 1999). From naturalistic studies and one controlled trial, there is evidence suggesting that lithium has protective effects against suicide and serious deliberate self-harm (Reference Muller-Oerlinghausen and BerghoferMuller-Oerlinghausen & Berghofer, 1999). A similar case may well be emerging for clozapine in schizophrenia, although it is difficult at present to evaluate the importance of regular patient contact mediated through clozapine monitoring (Reference MeltzerMeltzer, 1999).
If prevention of individual patient suicides by doctors is to be a realistic possibility, then doctors must see patients prior to the fatal act. However, only 1 in 4 of over 10 000 suicides studied by Appleby et al (Reference Appleby, Shaw and Amos1999) had had contact with psychiatric services in the preceding 12 months. In Bristol, it was found that only 20% of people who committed suicide under the age of 35 years had had contact with their general practitioner (GP) in the 4 weeks prior to death (Reference Vassilas and MorganVassilas & Morgan, 1993). Matthews et al (Reference Matthews, Milne and Ashcroft1994) found that 36% of 627 Scottish suicides had had contact with their GP in the 4 weeks prior to death, and that the final hospital contact of these patients was with a psychiatrist in only 29% of cases. Even if we did have sound preventive strategies, therefore, we could implement them in only the small proportion of people with whom we are in contact prior to suicide.
The powerful social determinants of suicide (e.g. Reference Johansson, Sundquist and JohanssonJohansson et al, 1997) should not be forgotten by psychiatrists since this helps to balance the equation of responsibility for prevention between government and the medical profession. Indeed, it may be salutary to reflect on whether issues which relate to social exclusion, changes in male roles and the growing gap between rich and poor are likely to have more to do with recent rises in suicide among men than have putative inadequacies in psychiatric services impinging upon this particular demographic group.
Views of psychiatrists in Scotland
Psychiatrists in Scotland were asked how predictable and preventable they deemed suicides to be in general, by means of the following response options: ‘not at all’, ‘to some extent’, ‘to a great extent’ and ‘completely’. The last category was very seldom endorsed and, thus, the last two response options were collapsed. The 207 psychiatrists who had experienced one suicide or more were asked to describe their ‘most distressing’ suicide in detail; the contrast with responses to the questions about suicides in general is shown in Table 1. For predictability, suicides in general were viewed very differently from the most distressing suicide (χ2=64.1, d.f.=2, P < 0.001). This also obtained for preventability (χ2=86.3, d.f.=2, P < 0.001). Given that one can infer causality of distress from these responses, it will be noted that for a few psychiatrists more predictable suicides (and to a lesser extent more preventable suicides) caused more distress. The major difference, however, is that the most distressing suicide was deemed less predictable and less preventable than were suicides in general. Given the implausibility of accurately predicting and effectively preventing suicide for individual patients (as described above) this distress is not ‘evidence-based’. It seems probable that onerous expectations of prediction and prevention, therefore, whatever their source, contribute to the distress which suicides cause psychiatrists. This relationship was reinforced by many respondents in the open text sections of the questionnaire (Reference Alexander, Klein and GrayAlexander et al, 2000).
Table 1 Comparison of the views of psychiatrists in Scotland on the predictability and preventability of suicides in general and of their most distressing experience of suicide
| Extent of predictability and preventability | Suicides in general | Most distressing suicide | ||
|---|---|---|---|---|
| Predictability | Preventability | Predictability | Preventability | |
| Not at all | 19 (9.4%) | 13 (6.4%) | 66 (32.8%) | 90 (44.8%) |
| To some extent | 180 (88.7%) | 183 (90.1%) | 106 (52.7%) | 97 (48.3%) |
| To a great extent or completely | 4 (2.0%) | 7 (3.4%) | 29 (14.4%) | 14 (7.0%) |
SUICIDE RESEARCH
Since the formulation of the Health of the Nation targets, suicide research in the UK has gained prominence. Our Medline search on suicide in four leading UK medical journals (British Journal of Psychiatry, Psychological Medicine, Lancet and British Medical Journal) over a 3-year period before the targets were conceived (1989-1991) identified 51 references on the topic. During the last three complete years (1997-1999), this number had risen by 71% to 87. While the numbers of all Medline suicide references increased over the same period from 1665 to 1776, the four UK journals increased their ‘suicide content’ disproportionately, with a rise from 3.1% to 4.9% of all suicide publications.
As occurs not infrequently in various fields of medicine, a relatively small number of researchers have attained national expert status in the field of suicidology, and generalist practising psychiatrists are wary of stating their views, which may not accord with the views of the experts. Much suicide research appears to be conducted, funded and published on the basis that research will lead to successful preventive measures. The generalist researcher or clinician may feel at risk of being regarded as uninformed and nihilistic if he or she questions this assumption. Meanwhile, policy-makers may feel they have the support of the practising psychiatric community in awaiting the appropriate data which will tell us how we ought to be taking forward our responsibilities for suicide prevention.
WHAT IS DESIRABLE FOR THE FUTURE?
The suggestions below are based on the premises that we cannot predict or prevent suicide with any reasonable level of accuracy among the individual patients we see and that, therefore, the only realistic way forward is to regard all of our patients as potential suicide risks and to aspire to clinical excellence in the care of our patients in this context (Reference MortensenMortensen, 1999).
-
(a) Research into suicide should focus less on epidemiological data collection, and more on large-scale multicentre trials with adequate statistical power to detect differences in the most promising preventive possibilities, such as lithium in recurrent affective disorders and clozapine in schizophrenia.
-
(b) We need a mutually supportive and responsive milieu available for all psychiatrists following patient suicides (see Reference Alexander, Klein and GrayAlexander et al, 2000; Reference Dewar, Eagles and KleinDewar et al, 2000). This could be appraised during visits to inspect training schemes, although such support networks would apply equally to consultants and to trainees.
-
(c) We need to review realistically our abilities as a profession to influence suicide rates, with an expectation of a reciprocal appraisal by politicians and health service planners.
-
(d) We need to accept that our predictive capabilities are decidedly limited and that all of our patients are at increased risk of suicide. Within this context, we should aspire to ‘suicide-aware excellence’ in the management of all of our patients.
ACKNOWLEDGEMENTS
The study was funded by Grampian Primary Care NHS Trust. We are grateful to all psychiatrists who responded to our questionnaire survey and to those who helped to develop the questionnaire at the pilot stage. The secretarial work for this paper was done by Lana Hadden.
eLetters
No eLetters have been published for this article.