The Scottish Government

« Previous | Contents | Next »

Listen

CHAPTER FIVE: EVIDENCE FOR PRIORITY POPULATIONS, SETTINGS AND INTERVENTIONS

5.1 The previous sections of the report have been 'data driven' and have presented evidence taken, respectively, from the full range of studies identified and from the 'highest quality' studies identified. These data have provided an overview of what is currently known about intervention for suicidal behaviour and suicidal ideation and have outlined the most robust conclusions which can be drawn from the existing literature. This section of the report aims to briefly summarise study outcomes for the populations and types of intervention identified as priorities for Choose Life. Clearly, a number of studies will fall into more than one of the categories set out below. Since different priority groups may be of interest to people with distinct specialist interests reading this report, we replicate the relevant information within each category. Further details of the studies included in this chapter are given in Annex I.

Populations

Intervention by age group

5.2 For the purposes of the current report, age is categorised following the recommendations of the Research Advisory Group:

0-15 ('children', 10 studies included in the review)
16-25 ('young adults', 17 studies included in the review)
26-65 ('adults', 80 studies included in the review)
66+ ('older people', 2 studies included in the review)

Issues regarding the lack of reporting of participant age and the lack of substantial differences in outcome between age groups (where these have been specified) are discussed further in the context of outcomes from the Scoping Review presented in Annex G. The 'adult' age group includes the bulk of the general population. It is also the one age group from which study participants are regularly recruited. Consequently, it is fairly safe to assume that the majority of outcomes discussed elsewhere in this report relate primarily to this group. Here, we will therefore concentrate on outcomes specific to the 'minority' age groups (older age groups, children, young adults) identified as a priority for Choose Life.

Older people

5.3 As can be seen from the figures above, none of the 'priority' age populations are well served. However, the least well-served section of the population in terms of available intervention research is the substantial population of older people. Whilst outcomes for interventions evaluated in the context of other populations may well apply also to the elderly population, we were able to identify only 2 studies explicitly focussed on those aged 66+. However, both of these studies (Barak et al 2006, De et al 1995) were robust quantitative studies, supported by statistical evidence and both also reported positive outcomes in relation to a reduction in completed suicide. The first study focussed on treatment with SSRIs and also reported reductions in attempted suicide as a result of intervention. The second focussed on a community-based intervention involving telephone support services aiming to maintain contact and offer elderly people home assistance (for issues unrelated to suicidal behaviour).

5.4 Additional evidence can be drawn from studies which focussed on participants at the 'older end' of the 26-65 age category (Bruce et al 2004, Kugaya et al 1999, Lapierre 1991b, Oyama et al 2004, 2006a and 2006b and Ripamonti et al 1999). Four of these studies evaluated outcomes for suicide and three for suicidal ideation. With regard to suicide, two of the three studies evaluated community based support programmes for older people living in rural areas (Oyama et al 2004 and 2006a, discussed in greater detail in the context of interventions for rural populations). Both provided narrative support for a reduction in completed suicide, without accompanying statistical evidence. A third study (Ripamonti et al 1999) reported a reduction in suicide as the result of providing palliative care for older people with cancer. With regard to suicidal ideation, one study provided additional support for the effectiveness of palliative care for older people with cancer (Kugaya et al 1999). This study reported a reduction in suicidal ideation following anti-depressant treatment. The second study (Bruce et al 2004) reported a reduction in suicidal ideation following the introduction of improved treatment guidelines for the care of older people in primary care settings. The final study (Lapierre 1991b) reported reductions in suicidal ideation following treatment with sertraline for older people with major depression.

5.5 The above outcomes may be seen as promising, but cannot on their own provide unequivocal support for a particular strategy. Four of the eight studies cited (Kugaya et al 1999, Oyama et al 2004, 2006b, Ripamonti et al 1999) provided only narrative evidence of positive outcomes, with no statistical support, one cited statistical evidence but failed to give adequate details of this evidence (Lapierre 1991b) and one study (Oyama et al 2006b) did not report positive outcomes. It is essential that further 'purpose designed' intervention studies specifically focussed on older people are carried out. Given that the focus of the limited number of available studies reporting positive outcomes is, broadly speaking, on helping to resolve the particular problems faced by older people (via the provision of palliative care, anti-depressant and other support through primary care services and interventions to reduce the adverse impact of social isolation) similar provision of supportive services may be a good place to start in developing further initiatives specific to older people.

Children

5.6 At the other end of the age spectrum, a slightly larger number of studies focussed on evaluating interventions for children aged 15 and under . Outcomes for these studies are, unfortunately, not overwhelmingly positive. Four of the studies reported no reductions in either suicidal behaviour or in suicidal ideation. These studies evaluated treatment with escitalopram for children with major depressive disorder (Wagner et al 2006); interpersonal psychotherapy for depressed young adolescents (Mufson et al 2004), a psychoeducational intervention in schools involving a youth-nominated support team (King et al 2006) and a token for readmissions to hospital for suicidal young adolescents (Cotgrove et al 1995). The latter 2 studies did not give specific age ranges for their sample, each citing only a mean age of 15. Older adolescents are therefore also likely to have been included within the remit of these studies and it is not possible to separate out any outcomes specific to the younger age groups.

5.7 Three quantitative studies reported significant reductions in suicidal ideation, supported by statistical analysis. The focus of these studies was intervention with fluoxetine and fluoxetine combined with CBT (March et al 2004) and either staff training videos or family-oriented training videos designed to modify family expectations (both studies reported in Rotherham-Borus et al 1996). In addition to these comparatively robust outcomes, additional studies provide purely narrative support for a number of interventions. One of these (Rotherham-Borus 2000) also focussed on the use of a video-based educational initiative to educate families regarding the nature of self-harm. The authors report reductions both in suicide attempts and in suicidal ideation in response to this intervention. The final two studies reported reductions in self-harm based, respectively, on a case study of behaviour therapy using differential reinforcement with a nine-year old self-mutilating boy (Cowdery et al 1990) and on an RCT of developmental group psychotherapy for young adolescents with multiple episodes of repeat self-harm (Wood et al 2001). The latter study reported no concomitant reduction in suicidal ideation as a consequence of the intervention.

5.8 Additional evidence can be taken from five studies (Deykin et al 1986, Harrington et al 1998 and 2000, Toumbourou & Gregg 2002 and Valuck et al 2004) which provided few details of the age of their participants but which, by implication, focussed primarily on children. One of these studies (Deykin et al 1986) evaluated outcomes for suicide following a youth education programme, a second (Valuck et al 2004) evaluated outcomes for attempted suicide following treatment with anti-depressants. Neither reported positive outcomes. The remaining three additional studies evaluated outcomes for both self-harm and suicidal ideation. No impact on self-harm or suicidal ideation was found for an intervention involving empowerment-based parent education groups (Toumbourou & Gregg 2002) . Home-based family interventions with adolescents who had self-poisoned were found by two studies (Harrington et al 1998 and 2000) to be effective in reducing suicidal ideation but not effective in reducing further self-harm.

5.9 As the literature currently stands, it seems that there is no evidence to inform a targeted prevention strategy aimed at reducing suicide in children. There is limited evidence that some interventions, including pharmaceutical, psychotherapeutic, behavioural and staff or parent training initiatives may be effective in reducing attempted suicide, self-harm and, in particular, suicidal ideation. However, even in the context of these behaviours, the small number of studies, combined with the diverse modes of intervention evaluated, fails to provide a consistent body of evidence suggesting any clear way forward for intervention with children.

Young adults

5.10 Studies evaluating interventions for young adults (aged 16-25), although greater in number and generally more positive in outcome than those available for children and young adolescents, provide no substantially greater evidence to support a targeted intervention strategy. Taken together, the outcomes for studies focussed on children and young adults suggest that we currently have very little evidence of how to proceed in intervening with young people to prevent or reduce suicidal behaviour and, in particular, to reduce suicide. None of the available studies explicitly focussed on young adults within the specified age range reported outcomes for suicide and only one study including young adults did so. There are some limited pointers towards interventions which may be effective in reducing other suicidal behaviours and suicidal ideation in this age group, but currently the number of studies addressing any given intervention is again small and successful outcomes will need replication.

5.11 To summarise the available evidence supporting particular interventions for young adults: 6 quantitative studies with adequate statistical analysis have reported reductions in suicidal behaviours and/or in suicidal ideation following an intervention. These studies focussed on diverse interventions in a range of population groups. Two studies reported reductions both in suicide attempts and in suicidal ideation following school-based interventions involving, respectively, an emphasis on personal growth (Thompson et al 2000) or crisis intervention (Thompson et al 2001); one reported a reduction in suicidal ideation following treatment with fluvoxamine for older adolescents with obsessive-compulsive disorder (Apter et al 1994), one reported reductions in attempted suicide, self-harm and suicidal ideation for young adults with borderline personality disorder treated with DBT (Turner 2000) and one (Joiner et al 2001) reported a reduction in suicidal ideation in 'young adults' of unspecified age with anxiety and/or depression receiving a community-based problem-solving therapy. Finally, one study (Brent et al 1997) reported reductions in suicidal ideation following the treatment of depressed adolescents with either CBT, systemic behaviour family therapy or the provision of non-directive support.

5.12 Limited additional support is provided for an even broader range of interventions and populations based solely on the narrative report of authors. Specifically, these studies addressed the use of clozapine for self-mutilating behaviour in young adults with borderline personality disorder (Ferreri et al 2004), naltrexone for the prevention of suicide in a young heroin addict (Krupitsky et al 2001), psychoanalysis for the reduction of suicidal ideation in a young male university student (Maltsberger & Weinberg 2006) and hypno-behavioural therapy, including self-hypnosis, for the prevention of self-harm in a young woman (Orian 1989). Three of these four studies are single case studies only and outcomes from all four studies are unsupported by statistical analysis, so the reported outcomes should be treated with caution in the absence of further replication.

5.13 In contrast to certain of the above outcomes, three studies addressing school-based interventions for young adults failed to find any reduction in the target behaviours focussed on (attempted suicide and suicidal ideation).The first of these studies (reported in Eggert et al 1995 and Eggert et al 2002) evaluated skills training combined with either peer or adult support, the second (Randell et al 2001) evaluated brief counselling versus peer group coping and support training, the third (Vieland et al 1991) evaluated an educational intervention focussed on developing and maintaining social networks. Three further studies (all RCTs addressing suicidal ideation only) also failed to report any positive outcomes. These studies focussed, respectively, on outpatient problem-solving therapies for young adults (Rudd et al 1996, Wingate et al 2005) and on a writing therapy intervention for young adults with instructions designed either to produce positive cognitive changes in the response to an adverse stimulus or to allow exposure to the adverse stimulus only (Kovac & Range 2002). Finally, one further study (Hopko et al 2003) reported statistically significant increases in self-harm following treatment with mianserin in comparison to either nomifensine or placebo.

5.14 As previously, additional information can also be gleaned from studies which provide few details of participant age but which, by implication, seem to focus primarily on the 16-25 age group considered here. There were three such studies (Brent et al 1993, LaFromboise & Howard 1995 and Metha et al 1998). One (Metha et al 1998) evaluated legislative initiatives introduced in 50 US states to prevent youth suicide. Despite the broad range of initiatives evaluated, the authors found no significant change in rates of suicide. They concluded that in part this was due to poor implementation of the initiatives. A second study (Brent et al 1993) evaluated individual-level restrictions on the access to firearms in a community case-control study. This study reported a reduction in suicide and suicidal ideation as a result of the intervention, but no noticeable impact on attempted suicide. The final study evaluated a culturally tailored multi-component intervention for a minority ethnic group. This study reported reductions in attempted suicide and suicidal ideation, but specific details of the components of the intervention programme are limited and no attempt was made to identify which components were most closely associated with the observed outcomes.

5.15 As with studies focussed on interventions for children, the available evidence of effective interventions for young people is both limited and spread thinly across a broad range of interventions, giving little clear direction to future prevention strategies. There is very little evidence of effective interventions to prevent suicide (controlling individual access to firearms being the single exception here) and little robust evidence of interventions to prevent attempted suicide or self-harm (although school-based programmes, DBT and a culturally tailored initiative have all received support from at least one study). Suicidal ideation has been addressed by a larger number of studies, with equivocal outcomes overall, but some support for school-based programmes, treatment with fluvoxamine for young people with obsessive-compulsive disorder and a diverse range of psychotherapeutic approaches ( DBT, CBT, systemic behaviour family therapy and problem solving therapy).

5.16 In summary, the available evidence for older people and for children is very limited and provides little clear direction for prevention initiatives. The evidence for intervention with young adults is more extensive, but also equivocal, with evidence both in favour of and against comparable interventions which have been evaluated by more than one study (school-based interventions and either outpatient or community based problem-solving therapies). Also with regard to young people, the diversity of interventions and populations for which outcomes have been evaluated hampers the development of a strategic approach. Comparing across all three identified age groups, it can be seen that the literature is both sparse and lacking in any specific intervention focus. It is also worth noting that only one of the 'highest quality' studies identified evaluated outcomes for any of the above age groups. Directed programmes of high quality research addressing these priority population groups are needed to provide firm evidence of effective interventions. In the meantime, broad approaches to intervention which it may be justified to pursue, given the weight of available evidence, include support-based initiatives for older people and psychotherapeutic or similar 'personal development' initiatives for children and young people.

Intervention by setting

5.17 The evidence available readily differentiates into the following settings:

• Community (92 studies start and 106 studies end with participants living in the community)
• In-patient open-ward (21 studies start and nine studies end in this setting)
• Outpatient unit (17 studies start and 21 end in this setting)
• School or high school (eight studies start and eight end in this setting)
• A&E (eight studies start and two end in this setting)
• 'Other' (settings with few available studies, studies recruiting participants from across a range of settings and studies failing to identify setting).

5.18 Studies evaluating outcomes for participants living in the community are the single largest group of studies in the literature. However, few statistically significant differences were observed in respect of outcomes in this setting compared to all other settings combined. Since the bulk of the evidence already reported relates to community settings we will not replicate this here. Interventions specific to school settings are reported in a subsequent section within this chapter. The diverse range of 'other' settings provided insufficient studies for each alternative setting to be of value in terms of highlighting interventions potentially appropriate to these settings. Therefore, we will focus here only on the three main clinical settings for which there is at least a slightly more substantive amount of specific information, namely in-patient wards, outpatient units and A&E. Studies focussing on A&E are rare given the prominence of this setting, as pointed out earlier, but to provide additional information we combine outcomes here for studies either starting or ending in each setting.

In-patient open wards

5.19 Of the available studies evaluating intervention in in-patient open ward settings (N=22), 7 quantitative studies reported positive outcomes supported by statistical analysis. Six of these evaluated pharmaceutical interventions. Reductions in both suicide and self-harm were reported for treatment with SSRIs (Barak et al 2006) and reductions in both attempted suicide and suicidal ideation were reported for treatment with paroxetine in comparison to amitriptyline (Moller & Steinmeyer 1994). Treatment with clozapine was reported to significantly reduce self-mutilation and related aggression in people experiencing psychosis (Chengappa et al 1999). As previously, it should be noted here that outcomes for clozapine across the full range of studies as a whole are rather equivocal. Three studies (Apter et al 1994, Baker et al 2004ps, Kudoh et al 2002) reported a reduction in suicidal ideation following treatment with, respectively, fluvoxamine, olanzapine and ketamine ⁸. Finally, one of the seven studies reported outcomes for psychotherapeutic interventions (Patsiokas & Clum 1985). All three psychotherapeutic approaches (cognitive restructuring, problem solving and non-directive therapy) resulted in equivalent reductions in suicidal ideation.

5.20 As previously, a number of studies give limited additional support to these or other interventions via the narrative report of study authors, unsupported by statistical evidence. Specifically, there is additional narrative support for clozapine in reducing self-harm in people with borderline personality disorder (Ferreri et al 2004) and for lithium in reducing both suicide and attempted suicide in people with bipolar disorder (Thies-Flechtner et al 1996). The caveats regarding these two pharmaceutical interventions raised previously (concerns regarding the association of clozapine with agranulocytosis and reports of an increase in deaths by suicide following lithium treatment and an increase in risk of suicidal behaviour following discontinuation of treatment with lithium) should be borne in mind in evaluating overall outcomes for these interventions. In the context of non-pharmaceutical interventions, narrative support is provided for a reduction in self-harm following stress-inoculation training (Kaminer & Shahar 1987) and a reduction in suicidal ideation following short-term hospitalization in a crisis intervention unit (Yu-Chin & Arcuni 1990).

5.21 Finally, it is important to note that the remainder (one half) of the studies set in in-patient open wards failed to identify any impact, significant or otherwise, in respect of the interventions evaluated. These studies evaluated: DBT (Bohus et al 2004), Fluoxetine (Cornelius et al 1993), 'inpatient treatment' per se (Etzersdorfer 1993, treatment involved both pharmacological and psychotherapeutic approaches at different points in time), 'any psycho-pharmacotherapy' (Gaertner et al 2002), matching or mismatching treatments to patterns of cognitive impairment (Miller et al 2005), ECT (Milstein et al 1986), mianserin (Montgomery et al 1983), intensive psychosocial intervention (Van et al 1997), a psycho-educational initiative for young, initially hospitalized, adolescents reported on earlier in the context of interventions for children (King et al 2006) and, finally, two studies reporting treatment with unspecified 'anti-depressants' (Oquendo et al 1999 and 2002).

5.22 In summary, positive outcomes for interventions taking place in in-patient open ward settings are not wholly lacking, but it is again the case that the majority of interventions evaluated receive either no support or support which can only be regarded as equivocal. The range of distinct interventions reliably evaluated and found to result in reductions in suicidal behaviour or ideation are so diverse that further replication of individual studies would be required before any robust conclusions regarding efficacy could be reached. The over-riding focus of evaluations taking place in in-patient settings is on pharmaceutical intervention. However, the range of individual pharmaceutical agents evaluated is quite broad and this again results in an evidence base which fails to provide substantive support for any specific intervention. As the evidence currently stands, there is some support for pharmacological treatment of in-patient depression as a mechanism to reduce suicidal behaviour and ideation. However, further confirmatory studies are required and it would be helpful if the research base was expanded to address mental health problems other than depression. It would also be useful for future research to explore alternative non-pharmaceutical interventions which have some evidence of effectiveness in other contexts (e.g. DBT), in order to increase the range of options available for in-patient treatment.

Outpatient settings

5.23 Twenty-two studies evaluated outcomes for interventions focussed on people attending outpatient units at either the start or end of the study. A number of these studies are also referred to in other contexts, as the outpatient context is relevant to a number of populations and overlaps a range of circumstances. For example, three of the studies were educational interventions to train staff and/or to modify family expectations. These studies (reported in Rotherham-Borus et al 1996 and 2000) are referred to also in relation to interventions in the A&E setting and in relation to interventions for children. Although, strictly speaking, the primary intervention took place whilst the patients were still in the A&E setting, evaluation tracked the children through to an outpatient clinic and the outcomes are relevant in all three contexts.

5.24 As was the case for interventions relevant to the in-patient setting, fewer than half of the studies addressing interventions relevant to treatment in outpatient units (N=9) reported positive outcomes supported by statistical analysis. Also as previously, these 'successful' studies reported on outcomes for a diverse range of interventions. Six of the nine studies reporting positive outcomes focussed on suicidal ideation. Reductions in suicidal ideation were reported for evaluations of an A&E-based educational video intervention targeting either staff or the patient's family in A&E (two studies reported in Rotherham-Borus et al 1996) and for a range of pharmaceutical interventions. With regard to the latter, reductions in suicidal ideation were reported as a consequence of treatment with fluoxetine (Heiligenstein et al 1993), fluoxetine plus CBT (March et al 2004), nortriptyline (Papakostas et al 2003) and sertraline (Lapierre 1991b).

5.25 Three studies focussed on suicidal behaviour reported positive outcomes supported by statistical analysis (Evans et al 1999, Kleindienst & Greil 2000, Bateman & Fonagy 1999). The first of these was a study of manual-assisted CBT which reported a reduction in suicide attempts and, by narrative report only, also a reduction in suicidal ideation. The second study reported greater reductions in both suicide and attempted suicide for treatment with lithium in comparison with treatment using carbamazepine (this outcome should be set against the fact that other studies of lithium treatment in this setting reported no positive effect). The third study reported reductions in attempted suicide and self-harm following (unspecified) outpatient treatment supplemented by partial hospitalisation.

5.26 Additional narrative support for the interventions evaluated included a further report of the study evaluating an educational video for families (Rotherham-Borus et al 2000), which reported reductions in suicide attempts and suicidal ideation and studies reporting reductions in self-harm as a result of a behaviour therapy intervention based on differential reinforcement (Cowdery et al 1990, a single case study referred to also in the context of interventions for children); reductions in suicidal ideation based on informal social support for brain injured patients (Kuipers & Lancaster 2000) and on electromagnetic field therapy for patients with multiple-sclerosis (Sandyk 1996) and reductions in both suicide attempts and suicidal ideation based on treatment with naltrexone (Krupitsky et al 2001).

5.27 Interventions failing to find any support, narrative or statistical, for their impact on suicidal behaviour or ideation in outpatient settings included treatment with lithium (three studies: Nilsson & Axelsson 1989, Coppen et al 1991, Kleindienst & Greil 2000), problem-solving therapies (two studies, also referred to in the context of interventions for young adults: Rudd et al 1996, Wingate et al 2005), DBT (Bohus 2004), CBT (Hengeveld et al 1996) and a multidisciplinary collaborative initiative across services (Jobes et al 2005).

5.28 In summary, outcomes for a diverse range of interventions evaluated wholly or in part in out-patient settings provide little firm evidence for the effectiveness of any specific intervention. Purely on the basis that there has been a successful attempt to replicate outcomes, it may be of value to explore further the use of educational video training, for staff or family members. Similarly, there are some grounds for concluding that out-patient anti-depressant treatment may be of value, since a number of studies evaluating anti-depressants, albeit using distinct drug types, reported positive outcomes. However, beyond this, all evaluated initiatives require, at best, further validation. It should also be noted that the majority of successful evaluations addressed the relatively 'soft' target of suicidal ideation, rather than providing evidence of a successful outcome relating to a reduction in suicidal behaviour. The outpatient setting provides a unique point of contact between health service providers and people experiencing suicidal behaviour or suicidal ideation. Since such contact can easily be lost, it is of particular importance that the interventions experienced in the outpatient setting provide positive outcomes. Given that the structure of outpatient services differs quite widely in different areas, this is a context in which locally targeted evaluations may be of particular value in increasing the existing evidence base.

A&E

5.29 The A&E setting is the first point of contact for a substantial proportion of patients presenting with suicidal behaviour. It is also, in effect, a potential springboard to other services. Despite the importance of this setting in the prevention of suicidal behaviour and suicidal ideation, only two of the studies identified for the review addressed interventions specifically designed for the A&E setting (Rotherham-Borus 1996 and 2000). Outcomes from these studies have been presented in respect of follow-up in the outpatient setting and in relation to interventions for children. Both of these studies reported reductions in suicidal ideation following a video-based educational training intervention for staff and the families of patients respectively, with the latter study also providing narrative support of a reduction in attempted suicide. In addition to these two 'purpose-designed' studies, a further eight studies addressed interventions either evaluated in the A&E setting or which followed up patients on discharge from the A&E setting.

5.30 Of the eight additional studies, five reported positive outcomes. Of these, two studies (Kapur et al 2002 and 2004) evaluated the impact of A&E management practices. No difference in outcomes was reported for people who were or were not given a psychosocial assessment, but specialist follow-up resulted in a reduction in further self-harm in both studies. One study (Carter et al 2005) reported a reduction in self-harm as a result of maintaining regular contact with people discharged from A&E via brief postcards asking after their welfare. The two remaining studies are less directly relevant to the A&E setting, but report outcomes for different treatment options offered on discharge from A&E. One reported a greater reduction in attempted suicide, self-harm and suicidal ideation following DBT in comparison to a more generalised client-centered therapy (Turner 2000), the other reported a reduction in suicide attempts following treatment with paroxetine (Verkes et al 1998). Of the three studies failing to find any positive impact on suicidal behaviour or ideation, two evaluated a nurse-led case management approach (Clarke et al 2002, Congdon & Clarke 2005). The third compared general hospital admission following presentation to A&E with discharge home (Waterhouse & Platt 1990).

5.31 The available evidence for interventions specific to the A&E setting is very limited. This represents a significant gap in the research base in respect of a setting which is a critical point in the care pathway. There is therefore an urgent need for further research studies in this area. Of the most pertinent evidence available, it is worth noting that, as in other contexts, there is support both for maintaining ongoing contact and for providing specialist care.

5.32 In summary, we have very little evidence to suggest which interventions specifically focussed on the A&E setting are likely to prove effective in reducing suicidal behaviour or ideation. It is crucial to effective service delivery that this very significant gap in the evidence base is addressed, since A&E is the first point of contact with services for many people who self-harm or who are suicidal. Considering the limited evidence which is available, the approaches most pertinent to this setting which find support in the literature are purpose-designed training and educational videos for staff and family members; the maintenance of ongoing contact following discharge and the provision of specialist follow-up care. All of these approaches, however, would require further evaluation to confirm their effectiveness in reducing self-harm, attempted suicide and suicidal ideation. There are currently no interventions which have been evaluated in the A&E context and shown to prevent suicide. In addition to the evident need for additional evidence regarding interventions in the A&E setting, further studies directly comparing different treatment options post-discharge could be of value.

Intervention for people with mental health problems

5.33 Broad outcomes for those with and without a mental illness are reported in the Scoping Review presented in Annex G. Here we focus more specifically on those studies which evaluated interventions for particular psychiatric populations. The populations for which such specific information is available are limited to the following diagnostic categories:

Depression

5.34 Comparing across the full range of 'priority populations' of interest to Choose Life, people with depression are the best served, both in terms of the numbers of studies available and the quality of these studies. A high proportion of studies evaluating interventions for depression are RCTs and six of the 20 studies identified as of relatively high quality evaluated outcomes for participants diagnosed with depression. The proportion of studies reporting positive outcomes supported by statistical analysis (N=15, 45%) is, however, no greater than that found for other priority groups. Furthermore, only two of these studies identified a statistically significant reduction in suicidal behaviour, the remainder reporting reductions only for suicidal ideation or providing purely narrative support for the interventions addressed.

5.35 The two studies reporting a significant reduction in suicidal behaviour report a reduction in both self-harm and suicidal ideation following treatment with paroxetine versus amitriptyline (Moller & Steinmeyer 1994) and a reduction in suicide and attempted suicide following treatment with SSRIs (Barak et al 2006). Studies reporting statistically significant reductions in suicidal ideation following pharmaceutical intervention supported treatment with sertraline (Lapierre 1991a, 1991b), fluoxetine and fluoxetine plus CBT (Heiligenstein & et al 1993, March et al 2004), moclobemide (Gagiano et al 1995), duloxetine (Hirschfeld et al 2005), fluvoxamine (Kasper et al 1995), paroxetine (Smith & Glaudin 1992) and ketamine (Kudoh et al 2002).

5.36 Four studies report positive outcomes for non-pharmaceutical interventions. These provide support for the following interventions in reducing suicidal ideation: telephone counselling (King et al 2003); CBT, systemic behaviour family therapy and non-directive support (Brent et al 1997, all three interventions significantly reduced suicidal ideation, with no significant differences in outcomes between the three therapies); primary care treatment guidelines and care management (Bruce et al 2004) and a school-based support group to enhance coping skills (Houck et al 2002). The latter study also reported outcomes for attempted suicide, but no significant reduction in this behavioural measure of intervention outcomes was found.

5.37 Limited additional support is provided for pharmaceutical intervention by studies failing to provide statistical analysis. Specifically, there are narrative reports of a reduction in both suicide (Isacsson et al 1996) and suicidal ideation (Kugaya et al 1999) following treatment with anti-depressants and of a reduction in suicidal ideation alone following treatment with sertraline (Lapierre 1991), viloxazine (Corona et al 1987) or fluvoxamine (Gonella et al 1990).

5.38 In contrast to the positive outcomes for treatment with anti-depressants reported above, a number of studies failed to find any reduction in suicidal behaviour or suicidal ideation following treatment with anti-depressants. These studies reported combined outcomes for the use of any anti-depressant (evaluated by three studies reported in five papers: Khan et al 2001 and 2006; Oquendo et al 1999 and 2002; Valuck et al 2004), outcomes for fluoxetine and venlafaxine in head-to-head comparison (Mitchell et al 2004) and escitalopram compared to placebo (Wagner et al 2006). A number of studies evaluating non-pharmaceutical interventions also failed to find reductions in suicidal behaviour or ideation. These studies evaluated matching or mismatching treatment to patterns of cognitive impairment (Miller et al 2005), interpersonal psychotherapy (Mufson et al 2004), brief training of GPs (Nutting et al 2005) and the provision of mental health services in rural areas (Rost et al 1998b).

5.39 In summary, substantive attention has been paid to interventions for depression in this literature and the overall quality of the studies available in this context is relatively high. This notwithstanding, there are again few consistent pointers to effective intervention. The chief focus of research in respect of interventions for depressed patients has been on pharmaceutical intervention. A wide range of individual drugs, primarily anti-depressants, have been evaluated in this context, with somewhat equivocal outcomes. Despite the comparatively large number of studies addressing treatment for depression, only two studies have reported positive outcomes for suicidal behaviour (paroxetine to prevent self-harm and SSRIs to prevent suicide and attempted suicide), the remaining studies have either failed to find any evidence of a reduction in suicidal behaviour or have focussed on suicidal ideation alone. The ambiguity in overall outcomes for anti-depressant treatment may well be dependent on distinctions between individual drug types, but given the current state of the evidence base, outcomes could as easily be dependent on differences in study methodology, differences in treatment context and/or differences in the response of particular individuals or groups to the anti-depressants evaluated. Future research focussed on anti-depressant treatment would benefit from more careful targeting.

5.40 A number of studies have provided evidence in favour of non-pharmaceutical intervention for people with depression. However, the diverse range of interventions considered and the limited number of studies overall again precludes any firm conclusions being reached regarding the effectiveness of any specific intervention. As in other contexts, there is some evidence for the effectiveness of maintaining ongoing contact and providing support to the depressed person, with more equivocal support for psychotherapeutic interventions. As is the case for pharmaceutical intervention, the positive outcomes reported are primarily for suicidal ideation. Future research focussed on treatment for depression should evaluate interventions against outcomes for suicidal behaviour as well as for suicidal ideation.

Personality disorder or borderline personality disorder ( BPD)

5.41 In contrast to outcomes for other 'priority' groups, the majority (65%) of studies evaluating interventions for suicidal behaviour and ideation in people with personality disorder or borderline personality disorder reported positive outcomes supported by statistical analysis. In contrast to the approach adopted with other mental health populations, the primary focus in this context was on non-pharmaceutical intervention. Reductions in attempted suicide are reported following partial hospitalization (Bateman & Fonagy1999), 'step down' care management following inpatient treatment (Chiesa & Fonagy 2003), psycho-analytically oriented residential treatment (Chiesa et al 2004), manual-assisted CBT (Evans et al 1999), and DBT (Linehan et al 1993 and 2006, Turner 2000).

5.42 A number of the above studies reported similarly promising outcomes for the same interventions in respect of a reduction both in self-harm and in suicidal ideation. Additional studies found reductions only in either self-harm or suicidal ideation for DBT (Bohus et al 2004, Low et al 2001) and CBT (Brown et al 2004). The more limited range of studies evaluating pharmaceutical interventions for people with BPD provide support for clozapine in reducing self-mutilation (Chengappa et al 1999), lithium, in comparison to carbamazepine, in reducing suicide and attempted suicide (Kleindienst & Greil 2000), nortriptyline in reducing suicidal ideation (Papakostas et al 2003) and both imipramine and fluoxetine in reducing suicidal ideation (Tollefson et al 1994).

5.43 Additional narrative accounts of intervention provide further support for the efficacy of DBT in reducing attempted suicide, self-harm and suicidal ideation in this population (Perseius et al 2003); clozapine in reducing self-mutilating behaviour (Ferreri et al 2004) and venlafaxine (Markovitz & Wagner 1995) and behaviour therapy (Bloxham et al 1993) in reducing self-harm. In contrast to the general run of evidence, one study failed to find any impact of DBT on attempted suicide (Verheul et al 2003) although the same study did find a significant reduction in self-harm. One other study also failed to find any statistically significant reduction in either attempted suicide or self-harm following treatment with transference-focussed psychotherapy (Clarkin et al 2001). In the case of the latter study, the authors noted that both behaviours did decrease in absolute terms in the treatment group and the very small sample size (only 17 cases were available for analysis at the endpoint) may well have been an issue here. Finally, one study, as previously reported, found significant increases in self-harm following intervention with mianserin (Hopko et al 2003).

5.44 In summary, clients diagnosed with borderline personality disorder represent a rare instance in which a sizeable number of studies, following distinct methodologies, have reported fairly consistent outcomes in favour of a single intervention ( DBT). It is currently unclear whether the overall more positive tone of studies reporting outcomes for people with personality disorder or borderline personality disorder is due to features of the disorder itself or perhaps to the greater focus on cognitive and behavioural interventions that is found in this context. Both possibilities are worth pursuing in future research. In respect of current practice however, there does seem to be some concrete evidence in favour of DBT and possibly also other cognitive and/or behavioural approaches in reducing suicidal ideation and suicidal behaviour in people with personality disorder. This notwithstanding, the specific parameters of the studies cited did vary and some replication of outcomes would still be of value. Available support for pharmaceutical intervention in this context is limited and further replication of the individual studies supporting particular drug types (in particular clozapine for self-mutilating behaviour and anti-depressants for suicidal ideation) would be required before such interventions could be recommended.

Schizophrenia or schizoaffective disorder

5.45 Counter to the pattern observed in the evaluation of treatment for personality disorder, the primary focus of the rather more limited number of studies evaluating interventions for people with schizophrenia or schizoaffective disorder was firmly on pharmaceutical intervention. Seven out of 10 studies addressed pharmaceutical intervention, with two further studies evaluating outcomes for ECT. The one remaining study (Cunningham-Owens et al 2001) evaluated an educational intervention, but this study reported a statistically significant increase in suicidal ideation following the intervention.

5.46 Overall, outcomes for schizophrenia matched those for other priority groups, with fewer than half of the available studies reporting positive outcomes supported by statistical analysis. The five studies which did provide such evidence reported reductions in both suicide and attempted suicide following the use of concomitant psychotropic medication (Glick et al 2004) and atypical anti-psychotics (Barak et al 2004) and reductions in attempted suicide following treatment with clozapine, either versus olanzapine (Meltzer et al 2003, Potkin et al 2003) or versus traditional anti-psychotics (Spivak et al 1999). Additional narrative support was provided for both the efficacy of clozapine in reducing suicide (Reid et al 1998) and its cost-effectiveness (Duggan et al 2003ps). In respect of the latter study, there are, however, some concerns regarding the assumptions made in modelling outcomes, for example the assumption that clozapine is cost neutral in comparison to the use of comparable pharmaceutical interventions, despite the established need to monitor people taking the drug for agranulocytosis. Of the two studies evaluating the use of ECT, one provided no evidence for a reduction in suicides (Tsuang et al 1979), the other provided a purely narrative report of reductions in self-harm following maintenance therapy with ECT for people with treatment-resistant schizophrenia (Dean 2000).

5.47 In summary, studies evaluating treatment for people with schizophrenia or schizoaffective disorder focussed almost exclusively on treatment with clozapine, or, to a lesser extent, other pharmaceutical therapies. Whilst outcomes for treatment with clozapine were universally favourable, none of the available studies chose to evaluate clozapine against either a placebo or against treatment as usual ( TAU) and only three of five studies provided statistical evidence of reductions in suicidal behaviour. Given the known side effects of clozapine, there is some virtue in taking a cautious approach to its use and further studies providing statistical evidence of its effectiveness in comparison to placebo or to non-pharmaceutical options for treatment would be of value. The limited range of studies evaluating other treatment options to reduce suicidal behaviour and suicidal ideation in people with schizophrenia restricts both current practice and directions for future research. It would clearly be of value to broaden the range of interventions evaluated but it is currently unclear which direction research should take. Exploratory pilot studies of alternative pharmaceutical and non-pharmaceutical treatment options would be of value in this context. As the evidence currently stands, treatment with clozapine appears promising, but, as stated, this option would also benefit from further evaluation against placebo or other non-pharmaceutical interventions.

Bipolar disorder

5.48 Only four studies specifically evaluated interventions for people with bipolar disorder. All of these studies evaluated treatment with lithium, either as a main treatment or as an adjunct. The two studies evaluating lithium alone reported statistically significant reductions in suicide and attempted suicide (Goodwin et al 2003) and in 'suicidal acts' (Tondo et al 1998). As previously discussed, some concerns are raised by the latter study and by an additional study (Oerlinghausen et al 1994) which, respectively, give evidence of possible increases in suicide following discontinuation of lithium treatment and following initial treatment with lithium. A study evaluating lithium (together with a limited range of other pharmaceutical interventions) as an adjunct to interpersonal and social rhythm therapy and primarily focussed on the latter treatment (Rucci et al 2002) reported no significant impact of the combined therapy on any measure of suicidal behaviour or suicidal ideation. Finally, one study (Baker et al 2004) comparing treatment with olanzapine plus adjunctive lithium with olanzapine plus adjunctive valproate failed to find any significant differences between the two adjunctive treatments but did find a statistically significant reduction in suicidal ideation in both conditions. These very limited outcomes provide clinicians with few choices in treating people with bipolar disorder for suicidal behaviour or suicidal ideation. It is important that further research is carried out if therapy for this diagnostic group is seen as a priority.

5.49 In summary, there are very few available studies evaluating treatment options to prevent suicidal behaviour and suicidal ideation in people with bipolar disorder. There is some support for treatment with lithium in potentially reducing suicidal behaviour, but significant caveats are raised by one high quality study which also reported initial increases in suicide and subsequent regression to base rates of suicide following discontinuation of long-term treatment. If people with bipolar disorder are seen as a priority group in respect of intervention for suicidal behaviour and suicidal ideation, further evidence is needed regarding outcomes for treatment with lithium and additional pilot research exploring other treatment options should urgently be carried out.

Other affective disorders

5.50 Finally, we were able to identify 10 studies which focussed on people with other affective disorders, or which combined outcomes for one or more different affective disorders. All 10 studies evaluated outcomes for pharmaceutical intervention and six studies again focussed on the impact of lithium. Of the four studies not focussed on lithium, three evaluated the impact of anti-depressant treatment by combining outcomes for all types of anti-depressants being prescribed for their participants, one evaluated the impact of treatment with fluoxetine compared to any other anti-depressant treatment and to no treatment at all. The latter study, evaluating fluoxetine, (Leon et al 1999) aimed to address outcomes for completed suicide, but was unable to do so since in the event only one completed suicide occurred, despite the comparatively large sample size of individuals perceived to be at risk (N=643 at endpoint) and long follow-up period (15 years). The study also evaluated fluoxetine in respect of outcomes for attempted suicide. Here the outcomes were equivocal, but certainly not in favour of fluoxetine. Whilst the numbers of attempted suicides in those continuing to take fluoxetine subsequent to the trial end-point reduced significantly, during the course of the trial the proportion of suicide attempts was significantly higher in the fluoxetine group than in either the group administered other anti-depressants or in the group receiving no treatment at all.

5.51 All three studies focussed on anti-depressant treatment as such (that is, regardless of the type of anti-depressant used) reported a statistically significant reduction either in population rates of suicide (two studies evaluated the same population at different periods in time, Isacsson et al 1997, Isacsson et al 2000) or in the incidence of suicide during naturalistic follow-up (Coryell et al 2001). In respect of lithium treatment, two of three linked studies by the same author identified a statistically significant reduction in suicide following treatment with lithium (Ahrens et al 1995a, 1995b), the third study failed to differentiate between deaths attributable to suicide and to other causes (Ahrens et al 1993). One additional independent study (Kessing et al 2005) also reported a statistically significant reduction in suicide following treatment with lithium. Limited additional support for a reduction in suicide following lithium treatment was provided by a single study (Thies-Flechtner et al 1996) providing only a narrative report of outcomes, whilst a study of long-term lithium treatment (Nilsson et al 1989) failed to find any positive outcomes for suicidal ideation.

5.52 In summary, studies addressing interventions for suicidal behaviour and suicidal ideation in people with affective disorders other than depression or bipolar disorder, or combining outcomes across different forms of affective disorder, have focussed exclusively on pharmaceutical intervention. The available research suggests, on balance, that outcomes for lithium therapy are promising. However, as previously, the diverse range of study designs, approaches to treatment evaluation (e.g. short term versus long term treatment) and tendency to combine outcomes for distinct demographic and clinical groups suggests that further research is needed in order to target lithium treatment appropriately. Similarly, outcomes for treatment with anti-depressants are positive where these combine across a wide range of distinct drug types, but the one available study focussed on a single anti-depressant (fluoextine) was not favourable. This also argues for further targeted studies focussed on specific drug types evaluated in the context of more clearly defined clinical groups. Further research exploring the use of non-pharmaceutical treatment options could also be of value in increasing the treatment options available.

Intervention for people who abuse substances

5.53 The available evidence for interventions to reduce suicidal behaviour or ideation in people who abuse substances represents a significant 'missed opportunity'. Epidemiological research (cf. Leitner & Barr 2003, Hawton et al 2005) suggests that a not insubstantial proportion of the people taking part in the available intervention studies are likely to have a history of substance abuse. It is clear that people who abuse substances represent a special case in clinical terms, both in respect of their risk profile and in respect of the range of interventions which are likely to be appropriate to their needs. Unfortunately, the issue of substance abuse was poorly addressed by the studies identified for inclusion in the review. Only one fifth of studies (21.6%) either reported whether or not participants had a substance abuse diagnosis or were using alcohol or illicit substances during the course of the study or explicitly excluded people who misused substances. Of the studies which explicitly included people with diagnosed substance abuse alongside other participants, not a single study carried out sub-group analyses to distinguish outcomes for substance users from outcomes for those not using substances.

5.54 Only three studies were identified which specifically focussed on interventions to prevent suicidal behaviour and suicidal ideation in people who misuse substances. None of these studies provided either narrative or statistical support for the effectiveness of the interventions evaluated. The interventions focussed on were fluoxetine for the treatment of depression presenting with co-morbid alcohol abuse (Cornelius et al 1993), an unspecified 'drug misuse treatment programme' (Magruder-Habib 1992) and an unspecified programme of 'aftercare' for alcoholic patients known to self-harm (Haw et al 2001). In the fluoxetine study, significant improvements in self- and other-report depression were noted, but no indication of any impact on suicidal behaviour or ideation was presented, despite a high reported incidence of suicidal ideation at baseline.

5.55 The unspecified 'drug treatment programme' reported no significant improvement in rates of attempted suicide or suicidal ideation as a result of the programme and, similarly, the aftercare programme for people diagnosed as suffering from alcoholism failed to find any improvements in self-harm as a result of the programme. In the absence of further individual data taken from the comparatively large number of studies known to include people with substance abuse, these outcomes provide virtually no information to help practitioners in making decisions regarding intervention with people who abuse substances. Given the clear importance of this issue in the context of suicidal behaviour and ideation, secondary research using available data or novel studies specifically focussed on this group are a clear priority for future research.

5.56 In summary, only three studies included in the review focussed on interventions for suicidal behaviour or suicidal ideation in people who misuse substances. None of these studies reported successful outcomes. Despite known associations between acute and chronic substance abuse and suicidal behaviour, only one fifth of studies identified whether or not participants had any current or recent history of substance abuse. Studies including participants with and without a history of substance misuse consistently failed to carry out sub-group analyses to distinguish outcomes for the two groups. The lack of evidence indicating appropriate treatment options for people who misuse substances is a critical gap in this literature. It is of particular importance that research is carried out in the short-term to inform clinical decision making in this context.

Intervention for people in contact with the Criminal Justice System

5.57 The available evidence in respect of effective interventions for people within the criminal justice system, who again are identified by the epidemiological evidence (cf. Wilson 2005) as a key group for preventive initiatives, is also sparse. Whilst it is possible that a number of the studies identified had participants who may have been in contact with the prison service, either during the course of the study or previously, we were able to identify only one experimental study which focussed specifically on interventions for suicidal behaviour and suicidal ideation in this population. This study (Condelli et al 1997) evaluated the novel approach of providing intermediate care (akin to psychiatric admission) within the prison setting and reported a statistically significant reduction in attempted suicide. In the absence of further evidence, this programme may be one which is worth pursuing both in future research and perhaps also in practice, given the weight of evidence supporting the likelihood of a high incidence of mental ill health in people within the criminal justice system.

5.58 Reviewing evidence drawn from our previous systematic review of interventions for other-directed violence, it seems likely that further information of relevance to intervention for suicidal behaviour and ideation in the prison population can be gleaned from the secondary outcomes and sub-analyses reported for studies primarily addressing other-directed violence. It may also be that other intervention literatures have included suicidal behaviour and suicidal ideation as subsidiary outcomes. Additional secondary analysis of these types of data may be a way forward in the short-term, but the paucity of studies specifically focussed on evaluating interventions for suicidal behaviour and ideation in the prison population is nevertheless disappointing.

5.59 In summary, only one of the studies included in the review specifically focussed on interventions for suicidal behaviour and ideation in people in contact with the Criminal Justice System. This represents a further significant gap in the literature, since people in this context are known to have a substantially increased risk of suicide. It is likely that additional evidence can be drawn from the literature on other-directed violence, or from other literatures including suicidal behaviour and suicidal ideation as secondary measures. Nevertheless, additional purpose-designed experimental research on interventions for suicidal behaviour and suicidal ideation within the Criminal Justice setting remains a clear priority. In the absence of further evidence, 'real world' evaluation of the approach taken in the one study identified in this review would be justified. This study evaluated an intermediate care service, similar to psychiatric admission, but located within the prison setting and reported a statistically significant reduction in attempted suicide.

What is known about intervention for other priority groups?

5.60 The Research Advisory Group identified a range of other priority groups for which we identified a similar paucity of information. Searches within the database containing all downloaded citations suggest that information relating to risk assessment is likely to be more common for these groups (notably for the unemployed, socio-economically deprived and for ethnic minorities) and that purely polemical discussions of both intervention and risk assessment are also likely to be relatively prevalent. However, concrete, experimental evaluation of specific interventions is scarce at best. We were unable to identify any intervention studies specifically addressing interventions with the following identified priority groups: asylum seekers, lesbian, gay, bisexual or transgender people, the recently bereaved (note here that studies addressing only postvention are not within the remit of the review), or socio-economically deprived, unemployed and homeless populations. Outcomes for the limited range of studies which addressed other priority populations specified by the Research Advisory Group are set out below. It is important to note here that within the studies included in the review, participants who are members of the above groups will undoubtedly exist, however study authors have not chosen to single out these groups for specific analysis. Secondary analysis of existing individual-level data may therefore provide additional evidence relating to these population groups in the short term if researchers are willing to volunteer their data for this purpose.

Interventions for ethnic minorities

5.61 Six of the studies identified for inclusion in the review specifically focussed on ethnic minority groups. Of these, one reported adverse outcomes, with a statistically significant increase in self-harm following treatment with mianserin (Hopko et al 2003). Three reported statistically significant positive outcomes. Of these, one study is the previously referred to video-focussed educational intervention aiming to modify family expectations regarding self-harm (Rotherham-Borus et al 1996). This study focussed on Latin-American families and reported a significant reduction in suicidal ideation. A second study by the same authors in the same population group provides narrative support for a reduction in attempted suicide and suicidal ideation. The second study reported on a broad based and culturally tailored community-wide intervention with Native Americans. This resulted in statistically significant reductions in attempted suicide and suicidal ideation (LaFromboise & Howard 1995) but, as mentioned previously, the components of the intervention are not outlined in sufficient detail and no attempt has been made to link specific components of the programme with the outcomes identified. The third study reported statistically significant reductions in attempted suicide and suicidal ideation for a community-wide public health oriented programme targeted at ethnic minorities (May et al 2005). Similar limitations apply to this study in terms of transferring the intervention to other community settings. Finally, one additional study (Zenere & Lazarus 1997) provided a narrative report of reductions in suicides, suicide attempts and suicidal ideation following a school-based initiative involving training of school staff and students to respond to suicidal crises. This study focussed primarily, although not wholly, on black ethnic minority groups .

Interventions for survivors of sexual abuse

5.62 The review failed to identify any studies which focussed specifically on intervention with survivors of sexual abuse to reduce suicidal behaviour or ideation. The Choose Life National Implementation Support Team flagged one very recent study, unpublished at the time this review was undertaken (Haslam, 2006) which addressed an intervention for male survivors of sexual abuse at low risk of suicide. However, the one report available for this study focuses primarily on process and implementation issues rather than providing an outcomes-based evaluation.

Interventions for rural populations

5.63 Four studies specifically evaluated interventions for rural populations. None of the studies provided statistical evidence of significant reductions in suicidal behaviour or suicidal ideation. Three of the studies were linked studies by the same author (Oyama et al 2004, 2006a, 2006b) focussing on similar community-based intervention programmes targeted at older people living in rural settings. Two of these studies provide narrative support for reductions in suicide. These programmes focussed, respectively, on psychoeducational interventions and on depression screening together with psychiatric or other health care and health education. The third study, which combined depression screening with group activity for the elderly, failed to demonstrate any reduction in suicide following the intervention. One final study (Rost et al 1998) evaluated the adequacy of provision of mental health services in rural areas but found no grounds for the assumption that inadequate facilities in rural areas were responsible for higher rates of attempted suicide.

5.64 In summary, the available evidence in respect of intervention with a broad range of population groups identified as a priority by Choose Life (asylum seekers, lesbian, gay, bisexual or transgender people, the recently bereaved, survivors of sexual abuse, socio-economically deprived, unemployed and homeless people) is extremely limited. No single intervention with any of these groups can currently be regarded as evidence based. The available evidence for intervention with other identified priority groups (ethnic minorities, rural populations) is also sparse. In respect of ethnic minority populations, there is some measure of support for a number of interventions (culturally tailored programmes, video-based educational and training initiatives, and educational public health programmes). However each intervention has been addressed by only one study and each in a different ethnic minority population. This provides little clear direction for targeted prevention strategies.

5.65 With regard to rural populations, there is some evidence that a range of support programmes for older people in rural areas may help to reduce suicide. However, the studies evaluating these outcomes failed to provide statistical evidence of a reduction in suicide and further evaluations would be necessary to confirm outcomes. It is clear that prevention strategies focussed on the priority groups identified for Choose Life are likely to be hampered by a lack of research evidence. In the short term, additional evidence may be obtained by an analysis of individual-level data from existing studies. In the longer term however, if prevention is to be targeted at the priority groups identified , the only solution will be to focus substantive additional resources on research addressing intervention in these groups.

Interventions

5.66 Outcomes for the very wide range of interventions which have been evaluated in the literature have been discussed from various perspectives throughout the report. Here we aim to provide a brief overview of outcomes for the interventions highlighted by the Research Advisory Group as of particular importance to the Scottish suicide prevention initiative. It should be noted that a number of these priority interventions overlap to some extent (for example gun control is a whole population initiative but is also a means of restricting access to lethal means in the individual case). We will discuss specific interventions under the heading which seems most appropriate given the primary focus identified by the study authors.

Whole population interventions

5.67 The vast majority of studies identified for inclusion in the review focussed on intervention with individuals. Aside from national initiatives targeting the restriction of access to means, which will be reported on later in this section of the report, only four studies evaluated interventions which could truly be referred to as 'whole population' initiatives. Three of these studies evaluated the introduction of suicide prevention centres. These in and of themselves could be regarded as service-based initiatives rather than whole population initiatives, but the focus of the studies was on national or state reductions in suicidal behaviour as a result of the decision to establish programmes for the introduction of suicide prevention centres. All three studies reported positive outcomes substantiated by statistical analysis, two studies reporting significant reductions in completed suicides (Leenaars & Lester 2004, Miller et al 1984) and one (Nordentoft et al 2005) a reduction in attempted suicides. The fourth study addressing a 'whole population' initiative (Metha et al 1998) evaluated all available legislative, public health and other state-level initiatives directed at youth suicide prevention in fifty US states. This exhaustive study of available whole population youth initiatives failed to find any evidence of a consequent reduction in completed suicide. The authors concluded that this was, at least in part, due to inadequate implementation of many of the initiatives.

Community-wide interventions aimed at the general population

5.68 Eight studies focussed specifically on community-wide initiatives. Two of these studies have been discussed above in the context of interventions for ethnic minority populations (LaFromboise & Howard 1995, May et al 2005). These studies report, respectively, reductions in both attempted suicide and suicidal ideation and suicidal ideation alone for community programmes targeted at ethnic minority groups. Three further studies evaluating programmes targeted at the rural elderly have also been discussed above, in the context of interventions for older people and for rural populations, two of these three studies (Oyama 2004, 2006a) provide some narrative support for the efficacy of the interventions evaluated. Both of the additional studies identified for inclusion and evaluating community-wide interventions focussed on programmes targeted at young people. Neither provided substantive support for the efficacy of these programmes. One study evaluating the 'Stop Youth' suicide campaign (Omar 2005) provided narrative support of a reduction in suicidal ideation only, the second study (Deykin et al 1986) failed to provide any evidence of a reduction in suicide as a consequence of a community-based educational initiative.

Reducing access to lethal means

5.69 Three studies specifically focussed on the impact of a restriction of access to means. All three reported positive outcomes. One of the studies focussed on national-level legislation relating to the control of firearms and evaluated outcomes on the basis of population statistics (Leenaars et al 2003). A second (Brent et al 1993) focussed more closely on the impact of access to firearms at the individual-level, using a community case-control approach. Both reported statistically significant reductions in rates of suicide. The third study (Landers 1981) focussed on state legislation regarding carbon monoxide emissions, but evaluated outcomes with reference to a single case study only. This study provides a physiological and clinical account of how deaths from suicide may be reduced by controlling carbon monoxide emissions but gives only limited insight into the likely impact of legislation on completed suicide.

5.70 Finally, 2 studies included an evaluation of the impact of restriction of access to means alongside a broader-based evaluation of a range of alternative options. One of these, identified as one of the 'highest quality' qualitative studies (Kuipers & Lancaster 2000), found that individualised restriction of access to 'preferred' means of self-harm was reported by people who had self-harmed to have been an important mechanism in helping them to stop their self-harming behaviour. The final study, also referred to in the context of whole population initiatives (Metha et al 1998), evaluated a very comprehensive range of national and state-level initiatives to prevent youth suicide. Amongst these initiatives, the researchers included a number of different approaches to the restriction of access to means, including control of firearms and control of access to drugs. The authors concluded that whilst such approaches appeared promising, they had in the main been poorly implemented and this resulted in disappointing outcomes in respect of their impact on suicide.

Media reporting

5.71 We were unable to find any study specifically addressing the impact of media reporting on suicidal behaviour or suicidal ideation. Searches of the full citation database initially downloaded indicates that several studies are available which refer to or evaluate this type of intervention, but none provide evidence of direct relevance to the outcomes considered here. This issue tends to be one which is extensively discussed but which is in fact rarely evaluated using concrete outcome measures.

Awareness raising / encouraging help-seeking

5.72 A number of interventions which are outlined here under other headings, in particular school-based programmes, could be regarded as focussed on awareness raising or encouraging help-seeking. However, four studies focussed more explicitly on the possible benefits of these approaches. Two of these studies evaluated very broad-based initiatives targeting military personnel. The interventions evaluated combined a package of training for higher level staff in recognising and providing support for suicidal behaviour and ideation, together with educational and awareness raising initiatives throughout the military population plus specific encouragement for individuals to seek help. One of these studies (Mcdaniel et al 1990) reported a statistically significant reduction in suicide attempts over the course of the programme and narrative support for a reduction in suicidal ideation. In respect of suicide attempts the outcomes should be treated with some caution however, since base rates of attempted suicide were in fact very low. The second study (Rozanov et al 2002) provided a purely narrative report of reductions in suicide.

5.73 Of the two other studies addressing awareness raising/help-seeking interventions, one focussed on empowerment-based parent education groups (Toumbourou & Gregg 2002) and reported no significant change in either self-harm or suicidal ideation. The other focussed on motivational visits by nurses to the homes of 'non-compliant' patients. This study (Vanheeringen et al 1995) reported a statistically significant reduction in attempted suicide. However, ethical concerns may be raised in respect of any widespread implementation of the intervention. Identifying people who have, in effect, refused the treatment offered and targeting these people for further contact may be seen both as undermining patient autonomy and as an invasion of privacy.

Mental health improvement

5.74 To a greater or lesser extent, virtually all of the interventions included in the review could be considered as interventions to improve mental health, if only by the very nature of the outcomes they focus on. However, no one study specifically addressed the issue of whether mental health improvement per se had any impact on outcomes. Studies understandably focussed instead on the more specific effect of particular mental health interventions on outcomes. The approach taken by two such studies does provide some indirect insight into whether or not mental health improvement as such associates with reductions in suicidal behaviour or ideation.

5.75 One of these studies (Etzsersdorfer 1993) reported unsuccessful outcomes for a single case study of fairly intensive inpatient treatment for suicidal behaviour, the second study (Suominen et al 1998) evaluated the impact of any treatment for depression on suicidal behaviour and reported no statistically significant differences in outcome between those who were treated and those who remained untreated. The question of whether interventions which succeed in improving mental ill health also result in reductions in suicidal behaviour and/or suicidal ideation remains an issue urgently requiring further elucidation. Evidence from the National Confidential Inquiry into Suicide and Homicide by People with Mental Illness (e.g. Appleby et al 1999) may be able to address this issue to some extent, but to date the papers resulting from this study have not provided evidence to this effect. Given the clear focus on mental ill health as a precursor to suicidal behaviour in both research and practice, it is perhaps surprising that this issue has not been addressed in any great depth in the literature to date.

School-based programmes

5.76 A number of initiatives targeting school-age children and/or their families have been considered in other contexts, similarly, studies focussed on psychotherapeutic and psychoeducational interventions which happen to take place in the school setting are considered elsewhere. Five studies identified for inclusion in the review evaluated programmes which were specially designed for the school setting. These focussed on quite distinct approaches to addressing the problem of youth suicide. Only one of the studies reported statistically significant reductions in suicidal behaviour or ideation. This study (Aseltine & DeMartino 2004) focussed on the SOS programme, which involved teaching young people to recognise signs of suicidal behaviour or ideation in themselves or in their peers. Significant reductions in attempted suicide over the course of the above programme were reported, although no change in suicidal ideation was identified.

5.77 Two other studies (Ross 1980, Zenere & Lazarus 1997) provided narrative support for reductions in self-harm. The first of these studies evaluated a training initiative for school personnel, the second study, which in addition gave narrative report of reductions in suicide and suicidal ideation, evaluated a programme focussed on crisis intervention and management. The final two studies evaluating interventions specifically designed for the school setting failed to report any impact on outcomes for either suicidal behaviour or ideation. The first of these (Randell et al 2001) evaluated coping and support training, with or without the addition of counselling provision, the second evaluated a purely educational initiative (Vieland et al 1991).

Training and peer-education

5.78 Training and peer-education programmes for families, individuals and other key non-health professionals such as school staff have already been discussed in other contexts. Here we focus on the eight studies which evaluated interventions focussed on the training of frontline health care professionals. One of these (Rotherham-Borus et al 2000) has been mentioned in a number of previous sections of the report. This study focussed on video-based training of emergency room staff. The authors provide narrative support for a reduction in self-harm and suicidal ideation as a consequence of staff training. Interestingly, this is the only study of training for health care professionals which focuses on any professional group other than staff in primary care general practice settings.

5.79 Three studies focused specifically on training GPs to recognise and treat depression or other mental ill health. None of these studies (Rutz & Walinder 1992, Owens et al 2004, Alexopoulos et al 2005) reported any change in outcomes for suicidal behaviour or ideation either with or without the support of statistical analysis. Three studies provided evaluations of more broadly based GP and nurse training initiatives, based on encouraging staff to follow a care management approach. One of these studies reported statistically significant reductions in suicidal ideation (Bruce et al 2004ps), one provided narrative support of a reduction in suicides (Rutz 2001), the third (Nutting et al 2005) failed to find any change in suicidal ideation as a consequence of the intervention ( STORM). Finally, one study (Morriss et al 2005) evaluated a brief educational intervention for a range of health professionals (but primarily GP practice staff) but failed to find any significant reductions in completed suicide.

Dialectical Behaviour Therapy ( DBT)

5.80 DBT was evaluated by eight studies and appears to be one of the more promising interventions identified. Only two of the eight studies failed to report any statistically significant outcomes relating to suicidal behaviour or ideation and of these, one study (Perseius et al 2003) was a relatively high quality qualitative study providing narrative report of a reduction in suicide attempts, self-harm and suicidal ideation. Whilst none of the studies reported outcomes for completed suicide, two reported statistically significant reductions in attempted suicide (Linehan et al 2006ps, Turner 2000), six reported significant reductions in self-harm (Bohus et al 2004, Linehan et al 1993, Linehan et al 2006, Low et al 2001, Turner 2000, Verheul et al 2003) and three reported significant reductions in suicidal ideation (Linehan et al 2006, Low et al 2001, Turner 2000). As previously noted, the majority of these outcomes were achieved in studies focussing on people with borderline personality disorder. It would be of interest to evaluate this intervention more widely outside of this population and also to evaluate whether the cognitive or behavioural components are primarily responsible for successful outcomes.

Cognitive Behaviour Therapy ( CBT)

5.81 A number of studies included in the review evaluated interventions based to some extent on a CBT model, or evaluated interventions which involved CBT as one element in a multi-modal approach. Here we consider outcomes only for those studies which specifically aimed to evaluate the efficacy of CBT alone. Eight studies provide outcomes on this basis. Of these, four reported statistically significant reductions in suicidal behaviour or ideation as a consequence of treatment with CBT, the remainder failed to provide any support, narrative or otherwise for the intervention. None of the studies reporting positive outcomes address reductions in completed suicide.

5.82 Two studies (Brown et al 2005, Salkovskis et al 1990) reported significant reductions in attempted suicide, one (Tyrer et al 2004) reported significant reductions in self-harm and one (Brown et al 2004) reported statistically significant reductions in suicidal ideation. An additional unpublished study, flagged to the review team by NIST (Gerber et al 2003) referred to reductions in suicidal ideation as a possible outcome of a study evaluating CBT, but provided no outcome data. The three studies which failed to identify positive outcomes for CBT compared brief manual assisted cognitive behavioural therapy with treatment as usual for the prevention of self-harm (Tyrer et al 2003); CBT with CBT plus fluoxetine and fluoxetine alone for attempted suicide and suicidal ideation (March et al 2004) and CBT for the prevention of attempted suicide and suicidal ideation in women who repeatedly attempted suicide (Hengeveld et al 1996).

Psychodynamic Interpersonal therapy

5.83 Two studies evaluated psychodynamic interpersonal therapy as a sole intervention rather than as part of a multi-modal intervention. The first of these studies provided evidence of a statistically significant reduction in self-harm and suicidal ideation (Guthrie et al 2001), the second found no significant reduction in either attempted suicide or suicidal ideation (Mufson et al 2004). A third study (Clarkin et al 2001) evaluating a transference focussed psychotherapy containing some elements of the interpersonal approach also failed to find significant reductions in suicidal behaviours or interventions. It is unclear if differences in protocol, differences in the therapeutic approach or differences between people account for the difference in outcomes between these studies.

Flupenthixol

5.84 None of the studies included in the review specifically addressed intervention with flupenthixol. Other related pharmaceutical approaches are evaluated and outcomes discussed within the report and a search of the full range of citations initially downloaded suggests that a number of studies have evaluated flupenthixol in the treatment of depression or other mental health problems, but none of these studies have specifically addressed the outcomes of interest here.

Crisis cards

5.85 One study, focussed on preventing attempted suicide (Cotgrove et al 1995), evaluated the efficacy of providing a green card/token for readmission to hospital. Although outcomes from this study look promising, no statistically significant differences were noted and the authors made no narrative claims regarding efficacy. The crisis card approach may, however, be worth pursuing in additional studies and in the context of other measures of suicidal behaviour and ideation. Although the difference in outcomes for people with and without a green card did not reach statistical significance, only three out of forty-seven people (6%) given a green card attempted suicide on a subsequent occasion compared to seven out of fifty-eight people (12%) not given a green card. Furthermore, a post-hoc analysis carried out by the authors demonstrated that people allocated to the green card group were at significantly greater risk of further suicide attempts than those allocated to the group not receiving a green card. It is possible that a combination of the comparatively small sample size (outcomes for one hundred and five people were available for analysis at the study end-point), the unintended bias in the distribution of risk between groups and the focus on a relatively rare behaviour (attempted suicide) accounted for the failure to identify a statistically significant difference in outcomes. Indirect support for crisis card initiatives is also provided by the small additional number of studies evaluating ongoing contact as outlined below.

Telephone and other contact

5.86 Four studies evaluated the efficacy of the simple intervention of staying in regular contact with a person known to be subject to suicidal behaviour or suicidal ideation. A fifth study, which has already been referred to in the context of the provision of services to rural populations (De et al 1995), evaluated the efficacy of telephone contact offering support to older people. Of these five studies, three reported statistically significant improvements in suicidal behaviour and only one study (Cedereke et al 2002a) failed to report positive outcomes for either suicidal behaviour or suicidal ideation. This latter study evaluated a very limited form of telephone-based contact (two phone calls at four month intervals). Two linked studies reporting positive outcomes also evaluated fairly minimal levels of contact (Motto1976, Motto & Bostrom 2001). These compared contact versus no contact options for patients refusing follow-up treatment. The first of these two studies provides only narrative support for a reduction in completed suicide following the intervention, but the second reported a statistically significant reduction in completed suicide in the contact group. One study (Carter et al 2005ps) evaluating very frequent contact post-discharge with people known to have engaged in suicidal behaviour reports statistically significant reductions in subsequent self-harm. Finally, the evaluation of telephone-based support for older people (De et al 1995) reported statistically significant reductions in completed suicide. It should be noted that the latter study offered additional support rather than simply contact alone. The simplicity and potential cost-effectiveness of maintaining contact, with or without additional provision of support, as an intervention, combined with the relatively promising outcomes outlined, suggest that this may well be an approach which is worth pursuing. Further 'real world' evaluations of this approach in clinical and other settings would be of particular value.

Service re-structuring and case management

5.87 A number of the interventions evaluated can be described as service-based interventions. However, the studies we focus on here are those which either involved a change in current service provisions, or which explicitly compared two or more options for service delivery. Eight studies evaluated interventions falling within these criteria. Two linked studies (Aoun 1999, Aoun & Johnson 2001) focussed on the introduction of intensive outreach services provided by a suicide intervention counsellor. Aoun (1999) reported statistically significant reductions in suicide attempts following the introduction of this form of service delivery, Aoun & Johnson (2001) provide narrative support for a reduction in attempted suicide and suicidal ideation based on further details from a consumer survey. Four studies compare the introduction of a service with 'treatment-as usual'. Three of these studies, two reporting on nurse-led case management (Clarke et al 2002, Congdon & Clarke 2005) and one on integrated treatment (Nordentoft et al 2002) failed to identify any reductions in suicidal behaviour or ideation.

5.88 Two additional studies by the same author identified statistically significant reductions in suicide attempts and self-harm in participants allocated to medium stay inpatient care plus subsequent 'step-down' planned care (Chiesa et al 2003ps) and in participants allocated to a phased step-down programme instead of a psychoanalytically oriented speciality treatment programme (Chiesa et al 2004). Finally, one study (Waterhouse & Platt 1990) failed to find any significant differences in outcomes between general hospital admission and discharge home. Although there are some promising outcomes here, the limited number of studies combined with the diverse range of service interventions evaluated provides no clear direction either for future research or for clinical practice. An appropriate and, in respect of the existing literature, novel way forward may be to use in-depth qualitative research with service users, carers and people involved in service delivery to identify aspects of service delivery which are seen as helpful or unhelpful and develop further pilot evaluations around themes identified in this way.

5.89 In summary, the number of studies addressing each individual approach to intervention is very limited. As the evidence currently stands, the interventions which have been highlighted by NIST as of particular interest to the Scottish suicide prevention strategy and which find the most consistent and substantive support in the literature are DBT (for people with personality disorder) and restriction of the access to means of suicide or self-harm. In the latter case, further exploration in contexts other than firearms control would be of value, as would studies exploring individual-level approaches to restricting the access to means. The minimalist intervention of simply maintaining ongoing contact with people known to be subject to suicidal behaviour or suicidal ideation also finds quite consistent support in the limited number of studies available. This approach could have additional merit in respect of its likely cost-effectiveness. Looking more broadly at national-level and service-based initiatives, which are identified as a priority for Choose Life, there is support for service provision based around specialist centres. In terms of service re-structuring at the local level, there is some evidence of positive impact, but existing studies are too small in number and focussed on too diverse a range of service initiatives to provide a clear direction either for current practice or for future research. Broader national initiatives such as school-based educational initiatives, public education and media campaigns and training initiatives for health care professionals have been under-evaluated in the literature and are lacking in consistent support where they have been evaluated.

Relevance to the Scottish Context

5.90 Part of the remit of the review was to assess the relevance of the available research evidence to the Scottish situation. We have addressed this issue as follows: firstly, by evaluating outcomes from any studies directly focussed on the Scottish population; secondly, by comparing demographic profiles for suicidal behaviour and ideation in Scotland with the profile of the available intervention studies; finally, by setting known outcomes against recent evidence addressing intervention priorities or constraints in the Scottish context.

Studies directly addressing the Scottish context

5.91 Of the 38 studies carried out in the UK, 8 related directly to the Scottish population. However, of these, four studies (Davidson et al 2004, Evans et al 1999, Tyrer et al 2003 and Tyrer et al 2004) reported on different aspects of the same multicentre trial (the POPMACT study). Only one of the five centres taking part in the study was sited in Scotland. Four additional and independent Scottish studies (Cunningham-Owens et al 2001, Eagles et al 2003, Gerber 2003 and Thrive Initiative 2006) have been carried out, but one of these (Gerber 2003) although referring to suicidal ideation as a potential outcome of the CBT intervention evaluated, in fact provided no data addressing this issue. It can be seen therefore that the direct evidence for intervention in the Scottish context is very limited.

5.92 Taking the evidence as it stands, the POPMACT study (a multi-centre RCT) failed overall to find any statistically significant outcomes favouring a brief form of manual-assisted cognitive behaviour therapy ( CBT) over treatment as usual in reducing self-harm. The study did report potential economic benefits of the treatment in comparison to TAU, but in the absence of convincing evidence of the intervention's effectiveness this information is of limited value. Of the three additional independent studies reporting relevant outcomes, one study (Cunningham-Owens et al 2001) found statistically significant evidence of an increase in suicidal ideation following a brief educational intervention for people with schizophrenia.

5.93 A second study (Eagles et al 2003) reported more promising outcomes, with evidence of a significant decrease in suicidal ideation following some if not all of the interventions addressed. This study surveyed people with serious mental health problems to identify which of the interventions they had experienced had served to reduce their suicidal ideation. On the basis of self-report, informal social networks and support by psychiatrists had proven substantially more helpful than contact with a GP. Outcomes from this study are interesting, but would need to be replicated using more objective measures of outcome and a larger sample size before any clear policy decisions could be taken in respect of service-based initiatives.

5.94 Finally, one study evaluated a counselling and support service focussed on male survivors of childhood sexual abuse at low risk of suicide (Haslam, 2006) and reported positive outcomes in relation to a reduction in suicidal ideation. However, this study was beset by pragmatic and ethical constraints and to date the evidence for a reduction in suicidal ideation is based solely on spontaneous reports of a reduction in suicidal ideation by 10 participants. The bulk of the data reported by this study in the one available report are focussed on issues relating to service process and service delivery rather than on the outcomes of interest here.

5.95 In summary, the direct evidence we were able to identify for intervention in the Scottish context provides few if any firm pointers towards initiatives which would be of particular value. Since there is little reason to assume that the Scottish context, except perhaps in terms of protocols for service delivery, is distinct from either the rest of the UK or from other countries, the interventions identified as promising in other populations may be equally applicable to the Scottish situation. In support of this assumption, we found few significant differences between outcomes in studies evaluating similar interventions in quite diverse countries. It is likely that both the triggers for suicidal behaviour and ideation and the interventions needed to resolve these behaviours are largely universal. However, if studies focussed explicitly on the Scottish population are seen as a priority, then it is clear that the current evidence base poorly serves this need.

Profile of suicidal behaviour in Scotland

5.96 In the background to the report we briefly outlined the known profile of suicidal behaviour and ideation in Scotland. Here we map what is known about this profile onto the available evidence base for intervention, to explore how well the existing evidence base fits the needs of the Scottish population. Given, in particular, the poor reporting of demographic and other characteristics in the studies included in the review and also the limited range of national statistics relating to suicide and self-harm available for Scotland ⁹, the picture we are able to paint is of necessity somewhat limited. Nevertheless, it provides at least a crude account of how informative the available evidence is likely to be for Scottish prevention and intervention initiatives.

Overview

5.97 The most pertinent and recent national statistics available to us for comparison were General Register Office for Scotland ( GROS) figures for deaths by suicide and events of undetermined intent in 2006; statistics for discharges from Scottish acute hospitals with a diagnosis of deliberate self-harm in 2004 collated by the Information Services Division ( ISD, Scotland's national organisation for health information and statistics) and Community Health Index ( ISD Scotland) figures for self-harm presentations to GPs and GP Practice Teams during 2004. The GP Practice self-harm figures represent a composite of presentations for self-harm and attempted suicide as the two are not separated out. No formal national figures specific to Scotland are available in respect of suicidal ideation, so in this context we have drawn on the one available UK-wide survey (Singleton op cit) which included a large general population random sample drawn from Scotland.

5.98 As noted above, the data available for comparison are limited. The national statistics used to assess the profile of suicide and attempted suicide/self-harm provide, respectively, summative figures for suicide set out by age categories and by gender and method of suicide and summative figures for self-harm/attempted suicide set out by age categories and broad method (poisoning versus other or unknown method) alone. The available survey providing an indicative profile for suicidal ideation gives summative data for a wider range of demographic characteristics, but few detailed statistics are given. Since we do not have access to individual participant data for the intervention studies included in the review, we are also dependent on the data provided by study authors to draw comparisons between the actual profile of suicidal behaviour and ideation in Scotland and the focus of the existing evidence base for intervention. Unfortunately, as outlined earlier, a particular problem in the literature is the poor reporting of basic participant characteristics.

5.99 It is important to recognise that the lack of information available, both with regards to the included studies and more generally, in itself gives a valuable insight into how readily, or otherwise, the existing evidence base can be used to inform national or local initiatives for prevention and intervention. It is clear, for example, that there is a mismatch between the general tendency in the intervention literature to recruit participants from a wide spectrum of demographic and other groups and the emphasis of national initiatives such as Choose Life on intervention tailored to the needs of priority groups. What works for whom is an issue rarely addressed in the literature and this, combined with generally poor reporting of participant characteristics, means that we have little evidence to inform targeted prevention and intervention strategies. In addition to focusing on the particular demographic characteristics of individuals, both national statistics and prevention and intervention initiatives also commonly focus on methods of self-harm and suicide. Again, studies evaluating interventions in contrast generally fail to draw distinctions between different methods, combining outcomes across participants whose distinct choice of methods may in fact indicate the presence of other important differences of relevance to effective prevention.

5.100 One other notable gap between the available evidence and the requirements of an effective prevention and intervention strategy is the lack of information regarding 'hidden' populations. It is apparent from self-report surveys (e.g. Hawton et al 2002) that only a comparatively small proportion of incidents of self-harm result in presentation to services. Formal counts of the incidence and also, potentially, of the distribution of self-harm within the population in Scotland and elsewhere are therefore likely to be inaccurate and this has quite significant implications for prevention. Yet, to date, there has been little research effort directed towards establishing the extent and profile of this unknown and hence hard to reach population. Future research intended to inform prevention and intervention initiatives could benefit from taking on board the approaches used in other fields which deal with hidden as well as explicit behaviours (e.g. the 'capture-recapture' methods used in identifying populations using illicit drugs but not presenting to services). These, and other mismatches between the nature of the existing evidence base and the profile and needs of the populations experiencing suicidal behaviour and suicidal ideation, support the need for a significant expansion of research effort and also closer liaison between policy makers, practitioners and the research community.

Suicide

5.101 The General Register Office for Scotland ( GROS) provides publicly accessible annual summary figures for the number of deaths caused by intentional self-harm and events of undetermined intent in Scotland. The classification of cause of death is based on information taken from the death certificate, together with any additional information provided subsequently by the certifying doctor and is coded as per the relevant International Classification of Diseases ( ICD-9/ ICD-10) codes. The figures do not relate to the actual year of death, but to deaths registered within the given year, although in the majority of cases these two will be the same. The rationale for including events of undetermined intent is to allow for the possibility of undercounting in the recognition of and/or reporting of a death as being due to self-harm. Tables 5.1-5.3 below set out the profile of suicide in Scotland in 2006 as represented in the GROS figures.

Table 5.1 Age and gender profile of deaths by suicide and deaths of undetermined intent in Scotland in 2006

Notes to Table
Column figures do not sum to the totals. These figures are as presented by GROS and presumably indicate that the age at death of two males and two females was not established.

5.102 There has been a consistent downward trend in suicide rates between 2001-2006 (the Scottish Public Health Observatory reports a 14% decrease for males and a 9% decrease for females). However, Scotland's suicide rate remains higher than rates in all other parts of the UK and the longer-term general trend (last 25 years) has been upwards (cf. Platt et al 2007). Both the absolute figures (Table 5.1) and rates per 100,000 (Table 5.2) suggest that, within the age categories set by GROS, the peak age range for completed suicide in Scotland is currently between 30-49 years of age. The Scottish Public Health Observatory in addition cites suicide as a leading cause of death in people aged under 35 years.

5.103 It is worth bearing in mind that whilst the absolute number of deaths at both ends of the age spectrum (19 or younger and 60 or older) is substantially smaller than the number of deaths in the mid-range age categories, taken together the oldest and youngest age groups nevertheless account for one-fifth of all completed suicides. Current gender differences in suicide are, however, more marked than age differences, with the rate for males in 2006 almost four times that for females. Males show higher rates of suicide than females across all age categories, although this is most notable in the youngest (10-19) and oldest (80-89) age categories. One final demographic difference of note is the clear association of suicide with economic deprivation (reported in Platt et al op cit). The most deprived areas in Scotland have a risk of suicide double that of the Scottish average. Geographic variation in suicide rates is also evident across health board and local authority areas.

Table 5.2 Deaths by suicide and deaths of undetermined intent in Scotland in 2006 combined rates per 100,000 by age and gender

5.104 The methods used to commit suicide have the potential to be highly informative in respect of mechanisms for intervention and prevention. The profile for Scotland, in terms of the rank ordering of methods used, has remained broadly similar over time to the national profile for the UK as a whole (Platt et al op cit). Comparing the most recent national figures available for Scotland (Table 5.3) with those for England and Wales (Table 5.4) however, there are some absolute differences in the distribution of the most common methods currently used. In Scotland deaths by both drowning and jumping from a high place are roughly twice that noted in England and Wales, albeit still a relatively small absolute proportion of total deaths.

5.105 The proportion of deaths due to poisoning is also slightly higher in Scotland than in England and Wales. In contrast, deaths by hanging, by use of firearms and by 'other and unspecified' means are lower in Scotland than in England and Wales. These differences, although minor in absolute terms, are indicative of the need to tailor the Scottish strategy to the Scottish situation. The depth of information required to achieve this, however, is currently lacking. A targeted prevention strategy requires detailed information both about people and methods of self-harm beginning far earlier in the chain of events leading to an eventual death. Routine collection of cross-service data with attention to the methods used in self-harm would have the potential to significantly improve the effectiveness of future prevention initiatives.

Table 5.3 Specific causes of deaths by suicide and deaths of undetermined intent in Scotland in 2006

5.106 The broad profile of methods used by males and females in Scotland compared to England and Wales is also, in terms of rank order, quite similar. Again, however, there are slight, but, in terms of prevention, potentially important differences. Overall, a higher proportion of males commit suicide by either hanging or use of firearms than females and a higher proportion of females commit suicide by self-poisoning. In Scotland, however, death by poisoning accounts for a higher proportion of deaths overall (34.6% versus 27%) and also shows a greater disparity between men and women than is found in England and Wales (a difference in proportions of 21.7% between males and females in Scotland compared to 17.5% in England and Wales). In contrast to the pattern for poisoning, the proportion of females committing suicide by either drowning or jumping from high places is slightly closer to the proportion of males doing in so in Scotland than is the case for England and Wales.

Table 5.4 Specific causes of deaths by suicide and deaths of undetermined intent in England and Wales in 2005

5.107 Roughly one third of all the studies included in the review (33%) evaluated outcomes for completed suicide. In terms of the age distribution of studies, where this was adequately reported, the focus of the intervention literature broadly matched the age distribution noted for suicide in Scotland. Around half of the studies giving sufficient information to allow a comparison to be drawn (N=27) focussed on the 30-49 age group also identified in GROS figures as accounting for around half of the completed suicides identified in 2006, with 22% of studies focussed on people aged 29 or younger, who accounted for around 20% of deaths in the GROS figures.

5.108 In drawing comparisons purely between the focus of the available informative literature and current profiles for the age distribution of completed suicide, the most poorly served population are those aged 60 and over, who accounted for 15% of deaths in Scotland in 2006. Only 7% of studies providing adequate information regarding the age of their participants focussed on this age group in evaluating interventions. Although the gender of participants was better reported than their age (61 studies addressing completed suicide gave details of their participants' gender), all but four studies (three focussed solely on males, one on females) reported outcomes for mixed participant groups. There is therefore very limited available information to inform gender specific interventions for suicide. Attempts to target intervention initiatives at even the quite basic level of focussing on distinct methods of self harm for males and females are currently severely hampered by the lack of specific evidence. The intervention studies as a whole provide too little information regarding distinct methods of suicide even to draw meaningful comparisons between the methods focussed on in the literature and the methods of greatest prominence in recorded deaths in Scotland. This