The Scottish Government

« Previous | Contents | Next »

Listen

CHAPTER SIX: RECOMMENDATIONS FOR FUTURE RESEARCH & PRACTICE

Research

Current state of the evidence base

6.1 There has been a rapid expansion in published research in the field of suicide and self-harm over recent years, with over half of the available intervention studies (54%) published between 2000 and 2006. However, research in this area to date has adopted a 'scattergun' approach and if useful insights for policy and practice are to be taken from the research evidence, there is an urgent need for a more co-ordinated and focussed research effort. Precisely where this effort should be directed remains an issue for further consideration, since there are currently few unequivocal pointers towards interventions which are likely to be consistently effective. In part, this is due to the lack of specific focus in the literature to date. The literature has evaluated an extremely broad range of interventions, but each intervention has been addressed only by a very small number of studies (we found a total of 200 studies evaluating 150 distinct interventions). In addition, studies have tended to combine outcomes for participants drawn from a range of diverse demographic groups and similarly have tended to combine outcomes for people using quite distinct methods of self-harm and suicide. Targeted intervention based on the current evidence is therefore not a particularly viable option.

6.2 In contrast to the lack of information available for clearly defined populations, the weight of evidence suggests that the literature has clear biases in focus. Nearly half of the available studies (46%) evaluated interventions for psychiatric populations, focussing in particular on people with depression or, to a lesser extent, borderline personality disorder. Whilst the attention given to these populations is understandable, it is disproportionate in respect of the actual distribution of suicidal behaviour in the population. As a consequence of this 'bias' in the literature, we currently have little insight into interventions which may be effective for the general population or for specific populations other than people with depression or borderline personality disorder.

6.3 In spite of the focus on psychiatric populations, the majority of studies (56%) evaluated outcomes for people living in the community and we also have little evidence available to address intervention in institutional settings, in particular, A&E settings, outpatient units and residential facilities such as nursing homes and prisons or secure units. Although there has been a recent shift of emphasis away from the evaluation of pharmaceutical interventions, with an increasing emphasis on psychotherapeutic and service delivery initiatives, the mode of intervention for which there is currently the greatest bulk of evidence is pharmaceutical intervention. Unfortunately, for the most part, the focus of the literature on pharmaceutical intervention has not been rewarded by substantive evidence of effectiveness for the majority of individual drug types evaluated. Finally, with regard to the spectrum of suicidal behaviour which has been explored in the literature, the main focus (47% of studies) appears to have been on suicidal ideation. Attempted suicide (37% of studies) and completed suicide (33% of studies) have received slightly less attention and interventions for self-harm (22% of studies) are substantially under-evaluated given the relative prevalence of this form of behaviour.

Specific outcomes and issues

6.4 Looking in greater detail at the populations and particular forms of suicidal behaviour and ideation which the literature has chosen to address, it is apparent that certain areas of need are poorly served. Interventions addressing self-harm and, in particular, self-cutting are under-represented. This appears to be a general feature of this literature as a whole rather than an issue which is specific to the evaluation of interventions. From a population perspective, both ends of the age continuum (under 15 and, notably, over 65) are under-represented in comparison to the incidence of suicidal behaviour and ideation noted for these populations. In this context also it is of considerable importance that research studies define their participant populations more clearly in future.

6.5 The tendency of studies to recruit participants from a wide variety of demographic groups and to combine across a number of distinct modes of suicidal behaviour is exacerbated by poor reporting of population characteristics. This includes poor reporting of basic demographic characteristics and characteristics of relevance to both the mode of suicidal behaviour or ideation and the outcome measures used. The evidence base would benefit substantially from improvements in this aspect of study design and reporting. In addition, researchers recruiting people experiencing mental health problems should be encouraged to focus on single diagnostic groups rather than recruiting from a range of sub-populations. The potential benefits of targeted intervention are often obscured by a tendency to recruit study participants from a broad range of very diverse sub-groups without either the sample size necessary to carry out sub-group analyses or any attempt to control for inherent variation.

6.6 As noted above, pharmaceutical interventions are the single most prominent focus of the literature to date, with 30% of the identified studies evaluating individual drug treatments or comparing different drugs in head-to head trials. Despite their relative popularity as a research focus and, arguably, also in clinical practice, pharmaceutical interventions did not outperform non-pharmaceutical interventions overall (19.7% of pharmaceutical studies reported successful outcomes compared to 25.2% of studies evaluating non-pharmaceutical interventions). The main focus of attention in respect of pharmaceutical intervention to date has been on treatment with either lithium (in the context of affective disorder, including bipolar disorder) or anti-depressants (both for the treatment of depression and more widely). Whilst there is evidence that both lithium and anti-depressants can have a positive impact on outcomes for suicidal behaviour and ideation, studies reporting successful outcomes need to be weighed against a smaller number of well-conducted studies reporting increases in completed and attempted suicide associated with these therapies. Comparatively few pharmaceutical studies have evaluated the impact of treatment on self-harm and those which have report outcomes which compare unfavourably with outcomes reported for non-pharmaceutical interventions. The failure of the majority of pharmaceutical studies to compare drug treatments with either placebo or non-pharmaceutical treatment as usual' options also serves to undermine confidence in the use of these therapies.

6.7 With regard to the highest quality studies available, outcomes supporting the effectiveness of pharmaceutical intervention rely on scale-based evaluations of outcomes for suicidal ideation. Outcomes for suicidal behaviour are largely equivocal. Future evaluation of pharmaceutical intervention would benefit from following the methods adopted by the more successful trials in this area. These trials have tended to adopt a specific focus, with a limited number of well-defined outcome measures and a single, clearly defined, client population (e.g. specified doses of lithium in long- versus short-term treatment of bipolar disorder). Additional improvements to the utility of research outcomes could be made if studies evaluating pharmaceutical interventions based the choice of drug to be evaluated on clearly articulated theoretical principles regarding the mechanism of action with specific regard to suicidal behaviour and used behavioural outcome measures rather than purely scale-based outcomes.

6.8 The range and diversity of non-pharmaceutical interventions is such that few direct comparisons between modes of intervention can be made within this broader grouping. However, the most prominent focus in the literature to date has been on either psychological/psycho-social interventions (16% of all intervention studies) or service delivery initiatives (10% of all intervention studies). Of these two broad categories, service delivery initiatives were the least likely overall to show positive outcomes when evaluated (14.3% of service delivery initiatives were reported as having positive outcomes, compared to 18.2% of psychological/psycho-social initiatives and 30.6% of 'other' non-pharmaceutical initiatives). In respect of the highest quality studies available, outcomes favour DBT within the category of psychological interventions and either the restriction of access to means or contact-based initiatives in respect of the eclectic category of 'other' interventions.

6.9 The only initiative broadly relating to service delivery and supported by the outcomes of higher quality studies was the introduction of specialist centres. It is important to recognize that the number of studies addressed in this context is small in absolute terms. Nevertheless, there is some consistency in the nature of the interventions which have resulted in positive outcomes. The successful service delivery initiatives, for example, are in effect also either 'contact-based' initiatives (intensive outreach support; introduction of telephone support) or 'specialist centre' service provision (specifically, hospitalization in a specialist crisis centre). To date, research focussed on non-pharmaceutical intervention has suffered both from an over-complexity in the design and/or implementation of interventions and equally from a consistent failure to identify the discrete components within a complex intervention and to evaluate which component(s) are responsible for achieving the outcomes observed.

6.10 Since the evidence to date suggests that relatively simple interventions such as providing a person with ongoing contact and support may achieve significant improvements in suicidal behaviour and ideation, future research would benefit from going 'back to basics' and exploring in greater depth this type of minimalist approach. Where more complex and/or intensive approaches have been shown to work ( DBT; restriction of access to means; specialist centres or specialist care) future research could similarly benefit from 'unpacking' both the concepts behind the interventions and the components involved in successful intervention.

6.11 In the case of DBT, 'unpacking' the intervention would, for example, establish the relative contribution to outcomes of the behavioural and cognitive components of therapy. Similarly, further comparative research could establish whether any form of intensive one-to-one support could achieve the same outcomes or whether the therapy as constituted is required to achieve positive outcomes. In respect of the apparently simple intervention of restricting a person's access to the means of suicide or self-harm, it is equally important to establish which 'means' are amenable to restriction, at what level (individual, community, population) and in which contexts. In the case of specialist centres, it would be of value to establish which features of this form of service structure produce improved outcomes and whether these key characteristics could be emulated using less resource-intensive mechanisms for service delivery, for example by better co-ordination of existing care pathways.

6.12 A number of the more specifically methodological points raised in respect of pharmaceutical studies also apply equally to non-pharmaceutical studies. In particular, the need to use a smaller number of more tightly defined outcome measures and to develop in greater depth the theoretical underpinning of the intervention evaluated. In contrast to pharmaceutical studies, studies evaluating non-pharmaceutical interventions also seem to have shied away from the evaluation of outcomes in respect of completed suicide. This is particularly true for studies focussed on psychological interventions. If such therapies are to act as primary rather than adjunctive interventions, this gap in the research evidence needs to be addressed.

Issues of particular relevance to the Scottish context

6.13 Currently, there is very limited evidence specific to the Scottish population available. Only 5 independent studies to date have evaluated interventions for suicidal behaviour and suicidal ideation in exclusively Scottish populations. Compared to available information suggesting the likely profile of suicidal behaviour and ideation in Scotland, the literature in general 'over-emphasises' suicidal ideation in young people (aged 16-24) and suicidal behaviour in slightly older groups (20-44) at the cost of failing to explore interventions for the youngest (14 and under) and oldest (aged 65 and older) age groups. The general failing to evaluate interventions for clearly defined population groups and to explore interventions specific to particular demographic groups affects Scotland as it does intervention in all other locations. There is, for example, very little and in some cases no evidence for effective intervention in population groups identified as a priority by Choose Life. For example, rural populations, ethnic minorities, asylum seekers, lesbian, gay, bisexual or transgender people, the recently bereaved, survivors of sexual abuse, socio-economically deprived, unemployed and homeless people.

6.14 There has also been little attention in the literature to date on the type of interventions which have been highlighted as of importance to Choose Life. In particular, few studies have evaluated whole population or whole community initiatives and, where these have been evaluated, there is little evidence of their effectiveness. Since there is little reason to assume that the effectiveness of individual interventions varies significantly between locations, intervention in Scotland can, in the short term, draw on the existing evidence base for other countries. In doing so, a pragmatic approach would be first to evaluate in the Scottish context those interventions which are comparatively simple to implement and which can easily be withdrawn or reversed if they fail to transfer successfully to the Scottish context. In particular, 'minimalist' interventions such as the maintenance of ongoing contact with people known to be experiencing suicidal behaviour or suicidal ideation, provision of informal social support, including for example the provision of telephone helplines and short-term targeted interventions with individuals, such as DBT for people with borderline personality disorder or treatment with sertraline or fluvoxamine for people with depression.

Focus of Future Research

6.15 In order to serve the needs of existing and future prevention and intervention initiatives, research needs to address the issues which have been outlined above and also to explore in greater depth and, ideally, in a 'real world' context interventions which the existing literature suggests show some promise of effectiveness. With regard to addressing gaps in the existing evidence base, additional studies are needed to explore outcomes for:

• clearly defined populations, in particular specific demographic groups and psychiatric populations other than people with depression or borderline personality disorder
• populations identified as a priority for national initiatives
• populations which have been under-researched compared to their known prevalence of suicidal behaviour, in particular children (aged 14 and under), older people (aged 65 and older), people who misuse substances and people who do not have a mental health diagnosis
• settings other than the community, in particular A&E, oupatient units and institutional settings such as prisons, secure units and in-patient open wards
• suicidal behaviour differentiated by method, in particular interventions for self-harm, notably self-cutting and self-harm involving the use of multiple methods
• modes of intervention which have been under-evaluated to date, for example, whole population interventions such as educational initiatives and individual interventions such as behaviour therapy.

6.16 Drawing on the existing evidence base and, in particular, on the limited number of high quality studies available, the following interventions are likely to be worth pursuing in future evaluations:

For the prevention of suicide:

• restriction of access to means (this approach needs further evaluation in contexts other than firearms control)
• maintenance of ongoing contact with the suicidal person
• service delivery via specialist centres with appropriately trained staff. ¹⁵

For the prevention of attempted suicide:

• restriction of access to means (with further evaluation as above)
• informal social support and support in developing social networks
• treatment with lithium for people with bipolar disorder (this approach needs further cautious evaluation given reports of possible increases in suicide and suicidal behaviour associated with it, research focussed on better targeting of lithium treatment could be beneficial)
• dialectical behaviour therapy (this approach needs further evaluation in people who have not been diagnosed with borderline personality disorder).

For the prevention of self-harm:

• dialectical behaviour therapy (with further evaluation as above)
• maintenance of ongoing contact.

For the prevention of suicidal ideation:

• treatment with fluvoxamine for people with depression
• treatment with sertraline for people with depression
• telephone support for people experiencing a suicidal crisis (further evaluation of the long-term efficacy of this intervention is needed).

6.17 It is important to note that the above recommendations regarding particular interventions which show some promise of effectiveness are based on a very restricted evidence base. Currently no one intervention finds extensive and concrete support in the literature. Implementation of any of the above interventions either in the context of individual clinical practice or in the broader context of national initiatives must consequently be regarded in the light of a 'real world' evaluation rather than a truly evidence-based approach to prevention and intervention. Implementation should therefore involve appropriate evaluation of outcomes. Whilst it is crucial that the intervention literature becomes more focussed, there is also a need to expand the treatment options available for people with and without a diagnosed mental health problem. Well conducted pilot evaluations of novel or under-researched interventions should therefore also be seen as a priority for future research.

Methodological considerations

6.18 In order to improve the relevance of the existing evidence base to future prevention and intervention initiatives, it is also important to improve the methodological quality of future studies. Whilst this literature compares favourably with other public health literatures in this regard, a number of issues nevertheless need to be resolved. Qualitative research in this area is sparse and is also, in the main, poorly carried out, with few attempts to follow protocols for well-established qualitative methodologies. Future qualitative research should improve on current methods and would, ideally, 'piggy-back' with larger scale quantitative studies to allow interventions to be addressed from distinct perspectives. This would offer both objective and subjective insight into what works and how and, indeed, how acceptable individual interventions are to people experiencing suicidal behaviour or suicidal ideation, their carers and service providers.

6.19 Particular problems for the quantitative research base, which should be addressed in future studies, are comparatively high drop-out rates (notably in studies evaluating interventions for suicide), a failure to blind investigators to assignment and, where possible, to blind participants to interventions, adequate controls on the fidelity of implementation of an intervention and, in particular with regard to pharmaceutical studies, a failure to evaluate outcomes against placebo, or against non-pharmaceutical treatment as usual and a failure to provide a true 'washout' period in order to evaluate interventions in isolation from the impact of ongoing treatment. Whilst in respect of the latter issues it is clear that ethical concerns may be raised, it is fair to reason that providing a client with an ineffective intervention which may have adverse outcomes poses at least as great an ethical dilemma.

6.20 There is no indication from the outcomes of available studies that complex, multi-component interventions are needed to prevent and treat suicidal behaviour and ideation. Indeed, as outlined above, the vast majority of the more successful interventions evaluated to date are conceptually quite simple. This notwithstanding, between one quarter (24%) and one half (49%) of the available studies evaluated interventions for suicidal behaviour and suicidal ideation which would fall within the Medical Research Council's ( MRC) definition of a 'complex intervention' (Campbell 2000) ¹⁶. Not one of these studies, whether conducted before or after the publication of the MRC's 'Framework for the development and evaluation of RCTs for complex interventions to improve health', followed the recommended pathway for development and evaluation of a complex intervention.

6.21 Putting aside the potentially controversial issues of whether the MRC's admittedly rather broad-ranging definition of 'complex' is viable in the current context ¹⁷ and whether or not RCTs are always the best approach to evaluation, the core tenets of the MRC guidelines are well established and failure to adhere to these raises concerns regarding the robustness of the available evidence base. To paraphrase the guidelines, studies should ideally follow the following stages:

Stage 1 (Theory) Identify or develop an adequate theoretical underpinning to support the choice of intervention and identify major confounders and strategic design issues.

Stage 2 (Modelling) Identify the components of the intervention and the underlying mechanisms by which they will influence outcomes directly or via identifiable interactions.

Stage 3 (Exploratory Trial) Identify (and test) a replicable intervention and a feasible protocol for evaluating the intervention against an appropriate alternative.

Stage 4 (Definitive ' RCT') Compare the intervention with one or more appropriate alternatives using a theoretically defensible protocol that is reproducible and adequately controlled in a study with appropriate statistical power.

Stage 5 (Implementation) Determine whether (positive) outcomes form the intervention can be replicated outside the controlled research setting over the longer term.

6.22 The stages in evaluation most commonly overlooked by the studies evaluating complex (and indeed also 'simple' interventions) for the prevention of suicidal behaviour and suicidal ideation were the preliminary and final stages (theory/modelling and implementation). This is not to say that the literature as such is lacking in theoretical analysis or that clinical practice is entirely lacking in the implementation of novel approaches. The problem seems rather to be that the three components (theory/modelling; experimental evaluation; implementation) rarely follow the well-established sequence recommended by the MRC. Interventions without clear theoretical underpinnings are evaluated, with the result that even if they work it is unclear why; interventions may reach the stage of a pilot evaluation, with promising results, but only rarely then progress to a full-scale evaluation; interventions which have successfully progressed through the relevant theory/modelling and full-scale evaluation phases are then not implemented or further evaluated over the long term in 'real world' settings and finally, interventions are implemented in clinical practice without the benefit of either detailed theoretical development or evaluation.

6.23 The above widespread failure to follow an apparently straightforward and well established pathway to developing and implementing effective interventions is not uncommon in respect of the health care disciplines. 'Evidence-based' health care is a surprisingly recent concept (cf. Cochrane 1972) and neither the funding nor the broader administrative control of research in the UK and internationally (cf. Gellert et al 1993; Geuna et al 1999, Holdcroft 2006) are well-aligned with the quite extensive demands which need to be met if health care is to be truly evidence-based. If anything, the research literature relating to suicidal behaviour and ideation, despite its failings, is comparatively sophisticated in its approach. Nevertheless, both the 'scattergun' approach to evaluating interventions and the general failure to take interventions through all essential stages of development suggest an urgent need for a well thought out, focussed and adequately funded national programme of intervention research. Since adequate funding is likely to be the key to the successful development and roll-out of such a programme, it is critical to the future prevention of suicidal behaviour and ideation that issues of resource allocation, encompassing all aspects of the pathway from initial theoretical research to final long- term implementation are addressed in the near future.

Practice

General issues

6.24 The concrete messages which can be drawn from the available evidence to inform practice are, sadly, rather more limited than the very clear messages which the evidence provides for researchers. Interventions for suicidal behaviour and ideation remain in their infancy at the current date. This does, however, suggest two broad points which should be taken on board. Firstly, intervention and prevention should begin with the least 'invasive' and most readily reversible options available to the practitioner. In the case of individual clients, this suggests approaches which are the least likely to result in adverse side effects and are also the least disruptive or disturbing for the client. Since there is little conclusive evidence for any given intervention, there are few grounds for escalating treatment without cause. That is, comparatively simple interventions which are also found to be acceptable to individual clients should in the first instance be favoured over complex interventions or interventions which the client is uncomfortable with. Secondly, both individual level intervention practice and intervention and prevention in the wider public health sphere should be more closely tied to ongoing research. A closer liaison between practitioners and the research community would be of substantial value and the key to this relationship is likely to be access to data.

6.25 Consistent collection of accurate day-to-day clinical information, including individual demographic and other relevant client details and details of the implementation and outcomes of any interventions used with a particular client or client group will serve to extensively supplement the available evidence base. It is unlikely that the number of funded research studies will be sufficiently great to increase the available evidence at the rate which is required if we are to improve outcomes in the short term. Clinical data have the potential to plug this gap, but, historically, clinicians and researchers have rarely worked together on an ongoing basis and the collection of routine data has been sketchy and often inaccurate. It is important that these issues are addressed. Initial pilot schemes evaluating the cost implications and most effective mechanisms for improving routine data collection in clinical settings would be of value. It is equally important that pertinent population-level data are recorded and collated in a way which can usefully inform prevention efforts across the full spectrum of suicidal behaviours. Currently, reliably collected information is largely restricted to national summary data for the prevalence of suicide.

6.26 There is a particular need for consistently collected and collated figures regarding the prevalence and profile of non-fatal self-harm and equally there is a need to establish the size and characteristics of the 'hidden' population of people engaging in self-harm but not presenting to services. Whilst there are increasing concerns regarding access to individual, non-summative, information for research purposes, the loss of such information represents a very significant obstacle to the development of effective prevention and intervention. Provision should be made for national level and local service level 'pseudonymised' ¹⁸ data to be made more readily available to researchers in this field.

Specific recommendations

Mode and type of intervention

6.27 One of the most prominent modes of intervention currently used for suicidal behaviour and ideation is pharmaceutical intervention. Given the ubiquity of this approach to treatment, it is important that clinicians recognise that the evidence base for this approach is in fact rather equivocal. It is clear that positive outcomes have been identified for pharmaceutical intervention within clearly defined clinical groups, for example lithium treatment in the context of affective disorders including bipolar disorder and fluvoxamine and sertraline in the treatment of depression. However, research outcomes overall are disappointing and individual studies, including studies evaluating lithium, have flagged concerns regarding possible adverse impacts associated with pharmacotherapy ¹⁹.

6.28 This situation is not purely the result of a lack of research evidence. A substantial number of studies have, for example, evaluated the use of anti-depressants (specifically as a means of reducing suicidal behaviour and ideation, as opposed to treating for depression) and yet there is no clear evidence that these drugs are consistently effective in preventing either suicidal behaviour or suicidal ideation. Even within the context of a single type of anti-depressant, studies can be found which report increases, decreases and no change in the same forms of suicidal behaviour. Such inconsistent outcomes could well be due to the need for more sophisticated theorising regarding the mode of action of particular drug types and, in line with this, better targeting of drug types to specific populations. Nevertheless, as the evidence currently stands no single pharmaceutical intervention can be recommended without caveat on the basis of the existing evidence. Clinicians wanting to use pharmaceutical therapies would currently be justified in using lithium to prevent attempted suicide in people with bipolar disorder (but should exercise caution given reports of adverse outcomes) and either fluvoxamine or sertraline to prevent suicidal ideation in people with depression, but beyond these rather limited options the available evidence becomes decidedly equivocal.

6.29 Non-pharmaceutical interventions have fared slightly better in the literature to date, but it would still be cavalier to recommend any specific intervention as being truly evidence-based. Bearing this in mind, there are a number of broad approaches to non-pharmaceutical intervention which have shown consistently positive outcomes in the small number of studies available. These are the restriction of access to means (currently only well-evaluated in respect of firearms control), the maintenance of ongoing contact with a person known to be subject to suicidal behaviour or to suicidal ideation, the provision of specialist services (either via specialist centres or through specialist follow-up of individual clients) and treatment with intensive cognitive/behavioural therapies, in particular dialectical behaviour therapy ( DBT). It should be noted that whilst the first three of these approaches are likely to be generally applicable, the last approach ( DBT) has been evaluated almost exclusively in the treatment of people with borderline personality disorder and further research would therefore be required to justify its use as a treatment for other populations.

6.30 It is also worth noting that, with the partial exception of the restriction of access to means ²⁰, none of these promising approaches are currently routinely used either in Scotland or in the UK as a whole. In the case of ongoing contact and specialist service provision, structural or cultural issues may need to be addressed to ensure that such approaches become more widespread. In the case of DBT, the primary limiting factor is likely to be resource allocation, as there are already waiting lists for the therapy, but currently too few trained therapists to satisfy existing demand.

6.31 The range of evidence-based non-pharmaceutical options for clinicians working with clients at the individual level may seem limited, but the options for practitioners aiming to address prevention at the level of whole communities or the general population are even more restricted. Three of the four broad approaches to intervention outlined above (restriction of the access to means, maintenance of contact and specialist service provision) can be applied at the level of entire populations. For example via legislative initiatives, national telephone-based services, re-structuring of existing formats for NHS service provision etc. However, effective interventions specifically designed as national initiatives are in very short supply. The forms of national initiative identified as a priority by Choose Life, for example, (general population educational campaigns and, to a lesser extent, curriculum-based initiatives in schools) have barely been considered in the literature to date and, where evaluations have taken place, these have rarely supported the effectiveness of such approaches.

6.32 There is an urgent need to explore in greater depth suitable approaches to national-level prevention and intervention. However, both the development of such approaches and their evaluation would need to be based on more detailed information regarding the prevalence and profile of suicidal behaviour and suicidal ideation than is currently available. Despite the paucity of existing evidence of effectiveness overall, a good start-point in developing suitable initiatives may be to consider what works at the individual level.

Population

6.33 Practitioners working with people who have identified mental health problems have a greater range of evidence available to them than practitioners working with other populations. This having been said, the treatment options which have been evaluated for people with mental health problems are primarily pharmaceutical and, outside of the very limited number of options referred to above, there is only equivocal evidence regarding their effectiveness. Practitioners working with people who have bipolar disorder, depression or borderline personality disorder have some relatively well supported mechanisms for intervention, respectively, lithium, sertraline or fluvoxamine and DBT, although again there are some significant caveats around the first of these options. Despite the prominence of schizophrenia/schizo-affective disorder in studies focussed on populations with mental health problems, the one intervention which has been fairly widely evaluated in this context (clozapine) receives insufficient support from the existing literature to justify its widespread use, notably given the known side-effect of agranulocytosis. The available evidence to inform practitioners working with clients who have other mental health problems currently precludes evidence based-practice defined in the strictest sense.

6.34 Outside the mental health context, the most extensive evidence available for practitioners working with particular groups is the evidence relating to interventions for 'high risk' groups, defined by their previous experience of engaging in self-harm. In this context the focus of the literature appears to be age-dependent, with pharmacological intervention evaluated for older age groups (40-59) and psychological or outreach interventions evaluated for younger age groups (12-30). Neither set of outcomes receives sufficiently strong support to guide evidence-based practice. Since, currently, there is also no evidence to suggest that interventions used successfully with other populations will not work with 'high risk' populations, clinicians would be justified in using the limited range of interventions identified in other contexts with similar 'high risk' clients. Treatment with lithium is an exception here, given concerns outlined previously.

6.35 With regard to interventions for particular demographic groups, the available evidence is even more restricted. Practitioners working with particular age groups have the greatest weight of evidence to support their practice, but even here the evidence for some key groups (specifically children aged 14 and under and people aged 65 and older) is sparse to say the least. The majority of interventions have been evaluated with people aged between 20-44 and hence recommendations made in other contexts apply largely to this group. Practitioners aiming to provide interventions for children (aged up to 15) and young people (aged 16-24) are limited to lower quality evidence suggesting that a surprisingly broad range of psychotherapeutic approaches may be effective. Practitioners aiming to provide interventions for people aged 65 and older are limited to evidence, based largely on the narrative report of study authors, that support-based interventions (for older people in rural environments at least) may be effective.

6.36 Practitioners addressing intervention with other demographic groups, including practitioners attempting to target interventions at either males or females, have no clear evidence to guide their treatment options. There is a particular dearth of evidence in respect of populations which have been highlighted by Choose Life as priority groups, including groups known to have a high prevalence of self-harming behaviour, for example, people who misuse substances and people within the Criminal Justice System. Essentially, targeted intervention cannot currently be usefully informed by the existing evidence base in the context of most demographic groups.

Setting

6.37 Practitioners working with people living in the community are comparatively well served by the existing literature. Over half of all studies (53%) evaluated outcomes for people living in the community and in the case of interventions for suicide these studies were also more likely overall to report positive outcomes. The range of interventions reported as having positive outcomes in community settings includes the full range of interventions outlined previously in other contexts. In respect of outcomes reported by the highest quality studies, practitioners intervening to prevent suicidal behaviour in community settings have reliable, if limited, evidence to support the maintenance of ongoing contact and specialist follow-up and, again with caveats as outlined previously, treatment with lithium for people with bipolar disorder and treatment with clozapine ( in preference to olanzapine) for people with schizophrenia. With regard to the latter recommendation, it should be noted that the general run of studies in other settings do not provide consistent support for the use of clozapine as such. The evidence cited here therefore relates specifically to situations in which there is, for other reasons, a choice between treatment with olanzapine and clozapine in the community setting and a client is known to be experiencing suicidal behaviour or suicidal ideation.

6.38 In the case of practitioners working to prevent suicidal ideation in people living in the community, the two higher quality studies available provide some support for the generic use of telephone counselling and for treatment with moclobemide specifically for people with major depression. One clear point of contact for people living in the community and experiencing suicidal behaviour or suicidal ideation is the GP surgery. Unfortunately, although there have been a number of studies, including high quality studies, addressing intervention based on service provision by GP practices, there is currently no substantive evidence to support the interventions evaluated. The majority of studies in this context to date have evaluated some form of training programme for GPs. It may be that future research needs to evaluate other options, for example provision by GP practices of interventions which have some evidence of effectiveness in other contexts.

6.39 The tendency of the literature to focus on intervention in the community may be justified in terms of the absolute prevalence of suicidal behaviour occurring in this setting. However, it leaves practitioners working in settings where there is a high relative prevalence of suicidal behaviour and suicidal ideation (for example prisons, secure units) with very sparse evidence to inform their practice. Similarly, it leaves a dearth of information relevant to settings which form the point of first contact for many people who engage in suicidal behaviour, in particular A&E. Considering the limited evidence which is available, practitioners would be justified in introducing ongoing contact and specialist follow-up care in the A&E setting and there is some evidence that training and educational videos for staff and family members may have the potential to improve outcomes for young people attending A&E.

6.40 In the secure in-patient setting, individual evaluations suggest that DBT, anti-depressant therapy and behaviour therapy may be beneficial in reducing suicidal behaviour. Options derived from the current literature for practitioners working with people within the Criminal Justice System ( CJS) are very limited indeed. There is support from one study for the provision of intermediate care, similar to psychiatric admission, in the prison setting, but beyond this option any interventions are largely speculative. It is likely that further options have in fact been explored as secondary outcomes in research which is primarily focussed on, for example, other-directed violent behaviour in people within the CJS. However, except where authors have flagged secondary outcomes in the titles or keywords of their research reports, this literature will have fallen outside the remit of the current review.

6.41 Considering the evidence which is available to inform practitioners working in other settings, there is some support for the use of pharmaceutical intervention (as previously, lithium for people with bipolar disorder, again with caveats regarding possible adverse outcomes and either fluvoxamine or sertraline for people with depression) with people in open in-patient psychiatric wards. Practitioners aiming to provide people in in-patient settings with non-pharmaceutical interventions are particularly poorly served by the existing evidence. Evidence-based options for intervention in the out-patient setting are equally limited, although a more diverse range of interventions have been evaluated. There is some indication, as in the A&E setting, that educational videos for staff and family members may help to reduce suicidal ideation and there is also evidence to support treatment with anti-depressants in this context. There are no firm pointers towards interventions to reduce suicidal behaviour in the outpatient setting.

6.42 Intervention in the context of residential units set in the community, for example nursing homes, or during general hospital admission, has not been specifically addressed in the literature to date. A number of studies have addressed intervention in schools, primarily using complex multi-component initiatives based around educational or psycho-educational programmes. The limited number of positive outcomes reported for these studies suggest that practitioners working within schools may be justified in taking a simpler approach to intervention. In particular, basing interventions around the provision of crisis support and/or training school students to recognise the signs of suicidal behaviour and suicidal ideation in themselves and others.

6.43 Individual studies have addressed a range of other settings, including, for example, military bases and palliative care units, but additional research would be required in these settings to provide reliable evidence to inform practice. Whilst there is currently no reason to believe that an intervention delivered in one setting will not be as effective if delivered in another, the lack of information specific to particular contexts leaves practitioners with the challenge of exporting interventions from one context to another without certain knowledge that this is appropriate.

Form of behaviour

6.44 Defining suicidal behaviour and suicidal ideation is not straightforward and few studies included in the review provided clear and precise descriptions of the particular forms of suicidal behaviour and/or ideation engaged in by their participants. The broad descriptors we have used ('suicide', 'attempted suicide', 'self-harm', 'suicidal ideation') track the labels commonly used by study authors. The absence of more detailed descriptions in the intervention literature and the general tendency to combine outcomes for people engaging in potentially quite distinct behaviours (e.g. self-poisoning versus self-cutting) is likely to be frustrating for practitioners. Practitioners need information which is specific to their client group. The reality, however, is that interventions have largely been evaluated against generic outcomes with participants drawn from a wide range of population groups.

6.45 Taking at face value the broad categories used both in this report and within the literature as a whole, practitioners working with people experiencing suicidal ideation (addressed by 47% of studies) have the most extensive evidence base to draw on in taking forward their practice. The range of interventions supported by this evidence base is nevertheless limited. Specifically, there is reasonable evidence to suggest that treating people with depression with either fluvoxamine or sertraline may be effective in reducing suicidal ideation. There is also some limited evidence that providing telephone support may resolve suicidal ideation in the context of a short crises-driven telephone call, although currently there is no further evidence to determine whether positive outcomes persist over the longer term.

6.46 Practitioners working with clients at risk of attempted suicide are possibly best served overall, with access to a relatively broad evidence base (37% of studies addressed attempted suicide), but also to a number of interventions which are supported by comparatively strong evidence. Specific interventions finding favour in this context are the restriction of access to means (both at the population and, although on the basis of less substantive evidence, at the individual level), the provision of informal social support and/or help in developing supportive social networks, DBT for people with borderline personality disorder and treatment with lithium for people with bipolar disorder. Again, the latter recommendation comes with the caveat that adverse outcomes have also been reported for treatment with lithium.

6.47 There is a contrast in the above recommendations between interventions which are as yet supported only in specific diagnostic groups and interventions which are generally applicable. The more generally applicable interventions (restriction of access to means and support-based initiatives) have been shown to have positive outcomes also in respect of suicide prevention. Around one third of studies (33%) evaluated outcomes for suicide, with the available evidence favouring the restriction of access to means and 'support', in the sense of maintaining ongoing contact, with the suicidal person as effective approaches to intervention. The one other intervention finding fairly robust support in suicide prevention is service delivery via specialist centres. This is not an intervention which is generally within the purview of individual practitioners. However, re-structuring of care pathways or the setting up of specialist teams may be able to achieve a similar profile for service provision at the local level.

6.48 Practitioners working with people who self-harm are the least well served by the available evidence. The lack of focus on self-harm in comparison to completed suicide or behaviour seen as attempted suicide is a significant problem given the known prevalence of self-harming behaviour. This is exacerbated by the fact that few studies differentiate between distinct methods of self-harm, with the result that practitioners are again left with little specific evidence to inform their practice. Fewer than one quarter of studies (22%) evaluated outcomes for self-harm and the only intervention finding consistent support in the literature currently is DBT for people with borderline personality disorder. Although based on weaker evidence overall, outcomes from one high quality study suggest that it would be justifiable also for practitioners to use maintenance of ongoing contact as an intervention for people who self-harm.

6.49 The lack of evidence specific to particular forms of self-harm highlights a key issue for this literature as a whole. The relationship between different behaviours within the spectrum of suicidal behaviour and suicidal ideation is under-explored. Pathways both into and out of particular behaviours may not be the same and if intervention is to be effective, it is essential that the mechanisms behind individual forms of behaviour are well understood. The reality of clinical practice is that clients present with individual and quite specific problems and future research needs to be sufficiently well-focussed to allow targeted intervention based on finer-grained distinctions between different forms of suicidal behaviour and different population groups.

Limitations of the Review

6.50 The review had a very wide remit and needed to meet a tight timescale and budget. To achieve these goals certain pragmatic decisions were taken. The final searches were restricted to searches on the titles and specified key words of the research material identified. All searches depended on electronic access to data sources and we were unable to carry out hand-searching. We were also, largely, unable pro-actively to contact other researchers and practitioners in the field to canvass for additional material. Equally, we were unable to search through the bibliographies of all retrieved studies, searching was instead restricted to the bibliographies of articles specifically included in the review and, for the most part, was carried out by one reviewer alone. Whilst final decisions regarding the material identified as meeting our inclusion and exclusion criteria were based on at least two reviewers reading each paper identified, data extraction for this review was intensive given the number of articles included and only around 32% of the material was extracted by two reviewers. Finally, since our remit here was to provide a broad overview of the available evidence, we were unable to delve more deeply into individual studies or to canvass study authors for additional information or for individual level data.

Carer's Perspective

6.51 One of the key gaps in the literature identified by the review was the lack of studies addressing the lived experience of pathways into and out of suicidal behaviour and suicidal ideation. The lack of emphasis on the relevance of the individual's experience is highlighted by the fact that only one of the 200 studies included in the review actually focussed on asking people what had worked for them. The literature has similarly failed to take advantage of insights which may be gained through the experience and knowledge of people caring for friends or relatives experiencing suicidal behaviour or ideation. In conducting this review, we were fortunate to have working with us two people ( KM and AF) ²¹ with direct experience of the impact of suicidal behaviour. These advisors have also had direct experience of how interventions for suicidal behaviour and suicidal ideation feel from the service user's and carer's perspective. We provide below the response of KM to the review outcomes, giving a verbatim account of how the current state of the literature and consequent evidence base is perceived by someone with personal experience of 'real world' outcomes.

Comments on review outcomes ( KM)

6.52 At the start of the research, I expected there to be a large body of evidence and research which the team would need to look through. Given recent health policies acknowledging the large numbers of suicides and the need to decrease the number of deaths, I assumed that there would be well-researched evidence of interventions that work. I thought there would be evidence about a whole range of interventions, as well as consideration of specific groups. I thought there would have been studies targeting specific ages, cultures and diagnoses. Having been involved with carers' issues, I suspected that there may be less information about interventions involving the whole family. I commend the academic researchers for gathering the information based on 26085 original citations. As a lay person, I was surprised that within these, there were only 37 systematic reviews and 200 primary empirical studies which were of relevance to the study. I assumed then that these would be of a high standard and show some clear recommendations for specific interventions. It would seem commonsense that restriction of access to means and maintaining contact are likely to have some success in preventing suicides. This is what families do instinctively. I would have liked to have seen evidence of something more that could be done, especially for young people.

6.53 I was very surprised, though relieved at their honesty, that 2 of the systematic reviews described pharmaceutical interventions that actually made things worse. I was disappointed that there were such equivocal results from the reviews addressing psychosocial intervention and that there didn't seem to be any interventions consistently supported by the literature. I was disappointed that there weren't more studies about particular groups of people, considering specific ages, cultures or diagnoses. I had hoped that there would be something to learn from other cultures.

6.54 I thought that there would be more evidence about interventions that would help known vulnerable groups, such as those with a dual diagnosis. And I assumed that children and young people would be a high priority for research. As a carer, I wanted to see studies that included more details about the whole family. Suicide affects the whole family and wider network. I hoped there would be evidence about how family and friends can intervene. I was surprised that more studies didn't include details about the family background and carers issues. I don't know how the studies ruled out the differences in support between families.

6.55 My conclusions and recommendations would emphasise the need for research which involves the whole family (in the widest sense to include a network of supportive friends). Are there 'supportive families' who have developed their own forms of intervention which work? Families are on the front-line of suicide intervention and most instinctively protect their loved one. It surprises me that no-one seems to have asked them. I also wonder about the impact that suicide attempts and ideation have on the family's mental health, and then back in a downwards spiral of a loop to the suicidal individual. Is there any intervention to reverse this by improving the emotional health of the family?

6.56 I would also like to see a lot more evidence about what can help for young people and for the elderly. Does the same intervention work across all age groups? Personally, I feel that if our family had had excellent support after the first crisis, then maybe our story would have ended differently. I would like to see clear guidelines for intervention following a first attempt, before behavioural habits have been developed. Any work which goes towards describing an ideal service which could prevent deaths is of interest internationally. Three aspects instinctively seem to offer potential: specialist crisis centres, on-going contact using peer support and informal networks, and training of health care workers and the wider community, including families. These would seem to provide a service which should help - I would like to see some evidence of their success.

6.57 I am impressed that the Scottish Executive have been so forward thinking to pursue this piece of work and thank you for the opportunity to be involved.

« Previous | Contents | Next »