Immediate responses

Suicides by patients cause considerable distress, including symptoms of bereavement and feelings of professional threat in psychiatrists, irrespective of their grade. These are unlikely to resolve until after the coroner's hearing. To ‘cope’ means to face and deal with responsibilities.

On hearing about a possible suicide it is crucial that you read the relevant policies and procedures of your trust. This may include recommendations regarding informing other members of the team, conducting urgent multidisciplinary case reviews (Box 1) and instigating local confidential inquiries.

Policy may suggest contacting a colleague or a mentor for support and also the trust's legal advisor or your medical defence organisation in the event of a contentious situation.

Discussing the case with the patient's general practitioner (GP) might not only serve communication, but may also give you new information and help (re-)establish contact with the family. The family will be in need of support because of the effects of bereavement and shock at the manner of death. Their grief may encompass anger towards staff members that they feel should have prevented the suicide. Contacting the family may make the inquest less adversarial.

On notification of a suicide, it is important to record all the relevant information about the death in the patient's notes, including the date, time, place and mode of death and how the death was notified. Time, date and sign each entry. It is illegal to alter any existing records, for example by adding ‘no suicidal ideas’ to an entry dated a month previously, even if you believe this to be true. However, if you think in retrospect that previous records are inadequate, it is advisable to add your additional information at the end of the notes, dated with the day on which you are writing.

It is often worth photocopying the complete set of notes in case the originals go missing or cannot be accessed. If the file is voluminous, copy at least a selection of the most important parts, such as notes, letters, risk assessment forms, discharge summaries, the last written entry and details of the last face-to-face contact.

A comprehensive review of the psychiatrist's role following a patient's suicide can be found in Reference Hodelet and HughsonHodelet (2001) and Reference Campbell and FahyCampbell (2002).

The coroner: notification, report and hearing

The report for the coroner

You may be asked to provide a report for the coroner. A good report may reduce the likelihood of being summoned to give evidence. It will certainly assist the coroner in deciding the relevant information for the inquiry. It can be an opportunity to demonstrate the quality of the care plan and the care provided. The report should be comprehensive, covering the relevant history up to the time of death in chronological order (Box 3). It is often worth getting a colleague to check the report.

The letter of instruction from the coroner may include specific questions in addition to the request for the report. You should answer these point by point at the end of your report. Do not anticipate the verdict and do not express an opinion as to whether this was suicide or not. That is the responsibility of the coroner.

Be aware that the relatives and their solicitors might have access to these documents. Issues of confidentiality should be carefully considered. Nothing relevant should be omitted, but it would be better to be circumspect when mentioning other people's personal details.

Going to court

Attendance at an inquest can be stressful. We recommend that all psychiatrists sit in on one, as it can be a very informative experience. If you are summoned to attend an inquest, having a trial run with a colleague, especially one who has recently been to an inquest, will be helpful.

The venue of an inquest may vary from a small office to a formal courtroom. After you receive notification of the venue, it would be worth visiting it if this is possible, in order to assess the setting. Remember that coroners follow different procedures, for example some like witnesses to stand and others are happy for them to sit.

Review the patient's medical records immediately before the inquest, especially if there has been a long adjournment, and take them with you on the day. A pre-inquest meeting within the trust can help everyone involved be better prepared.

Dress professionally and smartly for the court. Arrive early, allowing adequate time for parking and so on. On arrival at the venue, report to the receptionist, who will give you a ‘visitor’ identity badge. Inform the coroner's officer about the sort of affirmation or oath you wish to take. Be prepared for a long wait, depending on what other cases are being heard that day. It can add to everyone's stress if you have to sit wait for long in the same room as anxious, grieving or angry relatives. You could ask for a separate waiting-room. It might help to take a colleague with you, both for distraction in the waiting-room and for support throughout the proceedings.

In addition to the relatives, who might also bring lawyers, a number of other people may be present at the inquest (and in the waiting-room). These include police officers, pathologists, the GP and other staff from your trust. There may be advocates from user and carer organisations such as Rethink or MIND to support the family. There may be expert witnesses hired by the family or support organisations. They may work with the lawyers for the family in a cross-examination. In high-profile cases it is advisable to seek advice from your trust's public relations officer and your medical defence organisation regarding possible questions from the press.

On your way out, return your identity badge and collect a fee claim form from the coroner's officer.

Reference Calthorpe and ChoongCalthorpe & Choong (2004) provide a useful overview of the coroner's court and further information is available on the Department for Constitutional Affairs' website (www.dca.gov.uk/corbur/coronfrhtm).

Prediction and screening

Screening

Understanding the sensitivity, specificity and positive predictive value of screening tools and their applicability in different situations is essential. Potential problems of screening are well described by Reference WarnerWarner (2004).

The US Preventative Services Task Force (2004) found only limited evidence that the avail able screening tools, including tools to identify those at high risk, accurately identified people at risk of suicide in primary care settings. They found no or insufficient evidence that screening or treating those at high risk reduced suicide attempts or mortality.

The known risk factors for suicide include male gender, suicidal ideation on admission, diagnosis, outcome on discharge and the number of previous suicide attempts. Goldstein and colleagues assessed these risk factors and recorded suicides over a 10-year period in 1906 patients with affective disorder admitted to a psychiatric hospital (Reference Goldstein, Black and NasrallahGoldstein 1991). They were unable to identify with a 50% probability any one of the 46 patients who died by suicide during this period. They could identify only one of them with a 15% probability.

Previous child and adolescent psychiatry patients are known to have higher risk of suicide. One study (Reference Engqvist and RydeliusEngqvist 2006) followed up 1400 such individuals for between 12 and 33 years. There were 38 deaths, 34 of which were by suicide. However, only 2 of these 34 people had been admitted for attempted suicide.

A practical example: the Manchester Self-Harm Rule

The Manchester Self-Harm Rule was developed from the study of over 9000 patients who attended emergency departments with self-harm (Reference Cooper, Kapur and DunningCooper 2006). The instrument, for assessing suicide after self-harm, uses four clinical correlates: any history of self-harm; previous psychiatric treatment; benzodiazepine use in the current attempt; and any current psychiatric treatment. The study reported a sensitivity of 94% and a specificity of 25% in identifying high-risk patients. Of those identified, 17% re-attended and 22 died by suicide within 6 months.

Let us apply this instrument to screen a population of 100 000 people with severe mental illness and therefore an annual suicide rate 30 times the general population rate, i.e. 300/100 000. Table 1 shows the figures and the numbers of false positives and false negatives given the reported sensitivity (94%) and specificity (25%) of the instrument.

TABLE 1 Suicide prediction in a population (n = 100 000) of people with severe mental illness

	Suicide risk present	Suicide risk absent	Total
Screening test positive	a True positive 282	b False positive 74 775	a + b 75057
Screening test negative	c False negative 18	d True negative 24 925	c + d 24 943
Total	a + c 300	b + d 99 700	a + b + c + d 100 000

Among this group of 100 000 people, 300 will die by suicide within a year and 99 700 will not. With the 25% specificity of the Manchester Self-Harm Rule we will correctly identify 24 925 as not at risk of suicide. We will also identify 75% of the 99 700, i.e. 74 775, as at risk of suicide when they are not.

Among the 300 who will die by suicide, the instrument will identify 94%, i.e. 282. However, it will not identify 6% (18) of them.

The positive predictive value, i.e. the chance of having the condition in question given a positive test result, is calculated as PPV = a/a + b. Thus, from Table 1 the Manchester Self-Harm Rule in this example has a PPV of 0.38%. Its negative predictive value, i.e. the chance of not having the condition in question given a negative test result, is NPV = d/d + c = 99.9%.

Thus, this instrument will accurately identify non-suicidal patients 99% of the time. However, compared with 75 057 people not ruled out as not at risk of suicide, only 282 would really be at risk of suicide.

If we assume that hospital admission would prevent all suicides, one would have to admit 75 057 people in order to prevent 282 suicides and would still miss 18 suicides.

Causation

Causation is a central topic in the philosophy of science. We attempt to use science and controlled studies to liberate us from our biased misconceptions about causes. Yet if one considers the MMR vaccine autism scare, public misconception may be fuelled by ill-advised reporting of research. Moreover, generalisation of causes from epidemiological and/or group studies to the individual or vice versa is a problem for attributing or predicting causation. Both legal and medical approaches to causation emphasise integrity of the process of gathering and presenting information. Evidence then needs to be critically appraised according to the findings. There are rules of evidence and standards of certainty that should be used in both procedures. Epidemiological and research evidence (including that from the National Institute for Health and Clinical Excellence (NICE) and the National Confidential Inquiry into Suicide and Homicide by People with Mental Illness), as well as individual case history presented should be viewed in this light. Appropriate safeguards against rash opinions as distinct from an adequate investigatory process also are needed. It is important that after a coroner's hearing ‘justice’ is done and seen to be done.

The National Patient Safety Agency's (NPSA's) ‘seven steps to patient safety’ (Box 4) may be used as a model after serious untoward incidents. The NPSA admits that there is no single way to measure patient safety. However, paradoxically, an increase in reporting of patient safety incidents is a sign that organisations have implemented an open and fair culture in which staff learn from things that go wrong. For example, in the aviation industry, as reporting goes up, the number of serious incidents begins to decline (Reference MiyagiMiyagi 2005).

Strategies for prevention

Medication and suicidality

You may be questioned on the effect of medication on suicidality, i.e. suicidal thoughts and behaviour. Both lithium (Reference Burgess, Geddes and HawtonBurgess 2001) and clozapine (Reference MeltzerMeltzer 2002; Reference Hennen and BaldessariniHennen 2005) may have an antisuicidal effect. However, this effect may be related to the regular and frequent monitoring required for these drugs rather than their direct pharmacological action. Although selective serotonin reuptake inhibitors (SSRIs) and atypical antipsychotics are less likely than older drugs to cause death by suicide, especially in overdose, this is not the same as reducing suicidality. There have been suggestions that the SSRIs may actually increase suicidality, possibly by causing akathisia (Reference Fergusson, Doucette and GlassFergusson 2005; Reference Gunnell, Saperia and AshbyGunnell 2005b). However, akathisia also occurs in 20–75% of patients taking antipsychotic drugs (Reference King, Waddington and KingKing 2004), without necessarily increasing the risk of suicide.

In England during 1997–2004 there was a fall in the number of deaths from self-poisoning with tricyclic antidepressants. However, the number of suicides among patients with depression did not fall. Presumably they used some other means (Reference Morgan, Griffiths and MajeedMorgan 2004).

Hospital admission

For patients in crisis, it is generally assumed that admission to hospital or containment at home in the continuous care of home treatment or assertive outreach teams will reduce suicide rates. However, there have been no prospective randomised placebo-controlled studies of the efficacy of any of these measures and there is no evidence that temporary hospital or home containment ultimately reduces suicide rates among people thus treated. Nevertheless, if mental illness is making someone particularly suicidal or impulsive, containment with high levels of supervision may, in the short term, prevent suicide. It is good clinical practice to monitor patients closely when they are particularly at risk, disturbed or impulsive, although it should be noted that some individuals have personalities that make them enduringly impulsive, and their agitation and impulsivity may be worsened if they are under close observation. The assumption that the level of observation of people with acute mental illness can be reduced when symptoms improve is not evidence based: only hindsight can decide on the right or wrong of a decision to alter observation levels.

As mentioned above (Reference Appleby, Shaw and KapurAppleby 2006), during 2000–2004, 856 deaths by suicide, i.e. 14% of all suicides by people with mental illness, occurred while the individuals were in-patients. Of these, 81% were assessed to be at no or low immediate risk of suicide and 49% were considered to be at low or no long-term risk of suicide; 22% of those who died on the ward were under special observation. A further 14% of suicides by people with mental illness occurred within 3 months of discharge from an in-patient unit. Among these, 15% occurred within a week and 22% before the first follow-up appointment. The report noted that suicides by in-patients fell from 17% of all suicides by people with mental illness in 1997 to only 11% in 2004 (i.e. 67 fewer deaths), but this may simply be secondary to the reduction in psychiatric bed numbers.

Preventable suicides

In their second report on the findings of the National Confidential Inquiry (Reference Appleby, Shaw and KapurAppleby 2006), the authors tried to identify the preventable suicides, i.e. those who died in close proximity to services, especially those who were at evident risk but did not receive appropriate support. They highlighted three groups:

1. • 349 in-patients who took their lives while under close observation, within 7 days of admission, on home leave but living alone, or on absconding from the ward (41% of the 856 in-patient suicides)

2. • 255 patients who took their lives before their first follow-up (22% of the post-discharge suicides)

3. • 436 patients who were not on the enhanced care programme approach (CPA) following informal admission, or despite severe mental illness or recent self-harm, and 68 patients who, although on CPA, were non-adherent and/or missed appointments (12% of community suicides).

Thus, with hindsight, 1078 (22% of the 4942 suicides in England and Wales between April 2000 and December 2004) might be considered preventable. This means that they did not consider preventable the remaining 78% of suicides by people with possible mental health problems, which formed only 27% of all suicides.

Even though 49% of the patients who died by suicide had been in contact with services in the previous week, at the final contact immediate suicide risk was rated to be low or absent in 86% of cases. Moreover, among the 1271 people who died by suicide within 3 months of discharge from in-patient care, there was no clear change in risk associated with the post-discharge period.

It should be noted that the report's authors depended on the opinion of the psychiatrists involved in the care of the deceased. One cannot ignore the fact that there was no evidence base or scientific rigour to the process by which preventability was decided. It does not seem unlikely that the question regarding preventability tapped more into the frustration of the psychiatrist than into the empirical facts of the death.

The report noted that there has been a rise (proportionate or concomitant?) in suicides by patients outside the priority groups. Thus, when suicides in one area or by one means are reduced, they seem to increase in other areas or by other means.

Reference Masterton, Cavanagh, Johnstone, Cunningham Owens and LawrieMasterton & Cavanagh (2004) summarised the problems in preventing suicide. Suicide is a rare and unpredictable event even among those who are at high risk. Suicidal intent is not constant. It waxes and wanes, often suddenly and unexpectedly. Moreover, the risk can be greater when the patient is becoming ill, is recovering or has recovered, than when severely ill.

The report's recommendations (Box 9) mostly represent pearls of good clinical practice and common sense, albeit not systematically evidence based.

The National Confidential Inquiry uses the Suicide Questionnaire to enable systematic exploration of the facts in cases of suicide (National Confidential Inquiry into Suicide and Homicide by People with Mental Illness 2005). It is sent out to consultants of every patient who dies by suicide. It is worth looking at a copy, before you are faced with it. In our experience the questionnaire does not go into the distant root causes, which are unique to each case.

Surprises

As discussed earlier, a number of different verdicts are available to the coroner, and deaths that initially appear to have been suicide may not be recorded as such. To bring a verdict of suicide, the coroner must be satisfied beyond reasonable doubt that the deceased did the act that caused their death, that they knew that it would cause their death and that they really wanted to die. If the coroner believes that the act that caused a death was in fact a ‘cry for help’, they will often bring in a verdict of accidental death, because death was the unwanted consequence of the individual's action. If, on the balance of possibilities, the coroner cannot be satisfied that it was a cry for help but cannot be sure that the individual really wanted to kill themselves an open verdict may be brought in.

Sometimes mental health teams greatly distressed by the apparent suicide of a patient find at the inquest that an unsuspected physical disease caused the death. Moreover, certain physical events can cause accidents that look like suicide. Deaths are more likely to be due to natural causes than to suicide even in people with mental illnesses. It is therefore worth waiting for all the information, including the post-mortem and toxicology reports and the outcome of the inquest. There is one simple, though rarely used, measure that trusts can take to reduce everybody's distress and the chance of surprises: they can ask the coroner's office for a copy of the post-mortem and toxicology reports. At present, there is a statutory fee of £1.10 per page.

On occasions adversarial parties demand explanations of a psychiatrist's failure to predict and prevent a suicide, claiming that suicide prevention ‘is not rocket science’. One might agree with this assertion: suicide prevention is much more complicated and far less predictable. Human beings are not inert objects whose behaviour is simply predicted by Newtonian physics. People are able to deny or mislead others regarding their intentions, which complicates predictions.

In the aftermath of a verdict consider a review with the multidisciplinary team to look at the impact the case has had and whether further contact needs to be made with the family.

Conclusions

As no suicide-prevention intervention has been demonstrated to be effective in well-conducted clinical trials, Masterton & Cavanagh assert that the belief that medical interventions could prevent suicide not only has no evidence base but runs the risk of raising false expectations among the public and of encouraging grieving relatives to blame healthcare professionals inappropriately (Reference Masterton, Cavanagh, Johnstone, Cunningham Owens and LawrieMasterton 2004). Nevertheless, there is no excuse for blatantly poor clinical practice, and the inquiry process may reveal grossly negligent or dangerous practice in individual cases.

Looking at suicide prevention another way, one could argue that, to guarantee the prevention of a particular suicide, a health worker would not only have to successfully predict the if, when and where factors for the individual, but also have an effective and permanent solution or treatment. Again, not having 100% effective predictions or treatments is no excuse for not trying. Predicting suicide is like predicting fatal road accidents. It is probabilistic, not deterministic. Dangerous behaviours do not always result in death; conversely, deaths occur in people who are at no obvious increased risk. The fact that finite but evidence-based predictors and prevention strategies do not offer 100% certainty should not stop us from using them in both cases.

Suicide is epidemiologically relatively rare but clinically quite common. Given the number of patients at high risk of suicide, it is inevitable that sooner or later some of these individuals will die by suicide. It is therefore important that clinicians keep up to date with suicide risk assessment and prevention strategies, the difficulties of predicting and preventing suicides, and their own and their trust's responsibilities in relation to their patients.

MCQs

1. 1 In England and Wales:

   1. a about 1 in 1000 people has suicidal ideation in any one week

   2. b about 140 000 people attempt suicide each year

   3. c 3–5% of the population engage in self-harm sometime in their lives

   4. d 3–7% of those who self-harm eventually die by suicide

   5. e the suicide rate is 20 per 100 000 population over 10 years of age.

2. 2 As regards coroners' inquests:

   1. a a coroner's inquest is a trial

   2. b witnesses are banned from attending the inquest venue prior to the inquest

   3. c the possible coroner's verdicts include death due to industrial disease

   4. d only those who are invited can attend the inquest

   5. e the psychiatrist's report to the coroner is confidential to the coroner.

3. 3 Recommended measures to reduce suicides exclude:

   1. a strengthening observation procedures on wards and reducing absconding

   2. b strengthening the transition from community to wards

   3. c ensuring that high-risk patients receive enhanced CPA

   4. d responding robustly when care plans break down

   5. e developing services for dual diagnosis patients.

4. 4 As regards in-patient and post-discharge suicides:

   1. a less than 10% of all suicide victims are in-patients at the time of death

   2. b most individuals who die by suicide while in-patients were considered to be at high immediate risk

   3. c randomised controlled trials demonstrate that close observation prevents in-patient suicides

   4. d most post-discharge suicides occur within the first week of discharge

   5. e people rarely take their lives on the day of discharge.

5. 5 Regarding suicide by hanging:

   1. a 20% of suicides by hanging occur in controlled environments

   2. b fatality of attempted suicide by hanging is 70%

   3. c 40% of people brought to hospital alive after attempted hanging survive the attempt

   4. d strategies focused on restricting access to means of hanging are of significant value

   5. e little can be done to improve emergency management of ‘near hanging’.

MCQ answers

1		2		3		4		5
a	t	a	f	a	f	a	f	a	f
b	f	b	f	b	t	b	f	b	t
c	f	c	t	c	f	c	f	c	f
d	f	d	f	d	f	d	t	d	f
e	f	e	f	e	f	e	f	e	f