The Scottish Government

« Previous | Contents | Next »

Listen

CHAPTER THREE: OVERVIEW OF RETRIEVED MATERIAL

Profile of retrieved and rejected material

3.1 The 646 papers retrieved in full-text format and screened by two reviewers provide a backdrop against which to set the core material included in the review. A number of papers were rejected simply because, although referring to intervention in their abstracts, they addressed only risk assessment. These aside, the rejected papers provide some insight into the way in which intervention has been addressed in the literature. Taken together with the minority of studies which were eventually included and seen as a snapshot of available research, it becomes apparent that, as with comparable public health literatures (such as other-directed violence cf. Leitner et al 2006), much of the available research material focuses on issues which are, in fact, peripheral to prevention or intervention as such. Table 3.1 below summarises the reasons for excluding material following full-text retrieval.

Table 3.1 Reasons for Study Exclusion (% of rejected studies)

3.2 The reasons for rejection included in the comparatively large 'other' category in Table 3.1 are quite diverse, but primarily, as suggested above, relate to studies which focussed on issues relevant but peripheral to intervention, with no attempt to evaluate outcomes of direct relevance to self-harm. In addition to purely polemical material and to reviews of the primary literature which failed to meet the criteria for a systematic approach, this category included a number of studies with the potential and stated intention to provide material of direct relevance to intervention. For example, a study of the frequency with which GPs questioned their patients regarding the presence of firearms in their house; a study focussed on how best to teach medical students about self-harm; and a study interviewing families bereaved by suicide about their experience of medical care for the family member who had completed suicide. These studies were excluded because, although they addressed issues of direct relevance to intervention, they failed to provide any data relevant to the evaluation of the intervention discussed.

3.3 The proportion of studies rejected for describing an intervention without presenting data is comparatively high, but not unusually so for the mental health literature. This may be a reflection of a lack of funding to complete the research or possibly of the difficulty of carrying studies through to completion, given actual or perceived ethical and pragmatic constraints. Studies which are categorised as purely epidemiological in the above table provided only simple 'headcounts' or rates rather than data which could be used to directly address outcome.

Profile of systematic reviews and protocols retrieved

3.4 The aim of the review was to provide an account of the existing primary research evidence in relation to interventions for suicidal behaviour and suicidal ideation. The intention was to provide a novel synthesis of the evidence, giving a broader overview of the available information than that provided by existing reviews. However, it would be cavalier to wholly ignore the outcomes of the many systematic reviews and meta-analyses carried out on aspects of the evidence base to date. Annex E provides an overview of the systematic reviews retrieved by our search strategy and a more detailed account of the outcomes of these reviews. The main issues and outcomes identified are summarised below:

• The bulk of previous systematic reviews (49%) have focussed exclusively on pharmaceutical interventions
• Around one third of the available reviews include only material derived from Randomised Controlled Trials ( RCTs)
• None of the systematic reviews identified focussed specifically on outcomes from qualitative research
• Nearly half of the reviews (46%) focussed solely on populations with a diagnosed mental health problem
• A high proportion (43%) of the reviews identified positive outcomes for the intervention evaluated
• Despite the positive tone of many of the reviews, the evidence base cited for individual interventions was consistently weak and reviews of the same intervention often reported contradictory findings ⁶

3.5 Taking the outcomes of the existing reviews at face value, reductions in suicidal behaviour or ideation are reported for the following interventions:

• Lithium for bipolar, affective and mood disorders
• Alprazolam for depression
• Fluvoxamine for depression
• Paroxetine for depression
• Fluoxetine for depression and mood disorders
• Clozapine for schizophrenia or schizoaffective disorders
• Cognitive Behaviour Therapy in self-harming populations
• Dialectical Behaviour Therapy (no specific population)
• Green Card Initiatives in self-harming populations presenting to A&E
• Physician education in recognising and treating depression
• Restriction of access to means (primarily evaluated in the context of firearms control)

3.6 Adverse outcomes (an increase in suicidal behaviour or ideation) are reported for the following interventions:

• Levetiracetam for epilepsy
• Naltrexone for opioid dependence

3.7 Equivocal outcomes (evidence of both increases and decreases in suicidal behaviour or ideation or calls for further evidence) were reported for the following interventions:

• Lithium for mood disorders
• Lithium as an adjunct to olanzapine
• Fluoxetine as an adjunct to olanzapine
• SSRIs for depression
• SSRIs used to reduce self-harm in the general population
• Pharmacological treatment of any kind for borderline personality disorder
• Psychosocial programmes for depression
• Psychosocial programmes for low risk groups
• Curriculum based educational initiatives
• Suicide prevention centres
• Electro-Convulsive therapy
• No-suicide contracts
• Contact with clinicians or with liaison psychiatry
• Prevention programmes based in general hospital or A&E settings

3.8 The above profile of outcomes from previous reviews can be compared to the outcomes from this review, which are based, for the most part, on a broader overview of the literature as a whole and on a more recent evidence base. The main differences between this and previous reviews lie in the greater caution expressed in this review regarding the quality and reliability of the evidence base, in particular in respect of pharmaceutical interventions and in the addition of more recent evidence in this review regarding a range of interventions not currently in general usage including psycho-therapeutic and service-based initiatives. One chief advantage this review holds over previous reviews is the ability to directly compare and contrast outcomes for different modes of intervention and different modes of behaviour across different populations and settings.

Profile of primary studies included

3.9 A further key advantage of carrying out a review with very broad inclusion criteria is that the review provides a comprehensive overview of the nature of the existing primary literature. This overview is of benefit both in informing the development of future research programmes and in evaluating the extent to which current practice can be regarded as truly 'evidence based'. In Annex F, we provide a detailed overview of the nature and focus of the primary studies retrieved. A summary of key points is given below:

• The intervention literature is rapidly expanding with the majority of available studies (54%) published between 2000 and 2006
• The focus of the research literature shows a shift of emphasis away from pharmaceutical interventions and towards psycho-therapeutic and service-delivery initiatives
• The research literature has adopted a 'scattergun' approach (a total of 200 studies have evaluated 150 separate interventions). The evidence base for any single form of intervention is therefore very limited
• Despite an international focus (21 countries provided relevant evidence), the bulk of current research evidence derives from the US and Canada
• In contrast to other public health literatures, the UK as a whole has provided a substantive contribution to the evidence base (19% of available studies, N=38)
• Very limited evidence specific to the Scottish context is available (only 5 independent studies have evaluated interventions in the Scottish population).
• A high proportion of studies (46%) have focussed exclusively on interventions with psychiatric populations
• There is very limited evidence relating to intervention within the general population and currently we are lacking even accurate estimates of the prevalence of suicidal behaviour and ideation in either the Scottish or UK-wide general population (cf. Hawton et al 2002)
• In respect of mental health problems, the research literature shows a strong focus on depression (38% of studies) and on borderline personality disorder (24% of studies)
• Outside the mental health context, the main focus (30% of studies) is on people who have previously presented with self-harming behaviour
• Evidence specific to particular demographic groups is lacking, as studies commonly fail to report relevant details of their participants and also tend to combine outcomes for participants from different gender, ethnic or socio-economic groups
• Studies also commonly fail to identify whether participants have previously engaged in suicidal behaviour or have reported suicidal ideation
• Very few studies have focussed specifically on participants with diagnosed substance misuse and the majority of studies (73%) have failed to identify whether participants are currently or have previously engaged in substance misuse
• Just over half of all studies (53%) evaluated outcomes with participants living in the community
• The literature is lacking in high quality qualitative studies capable of providing information regarding the 'lived experience' both of pathways to suicidal behaviour and of intervention for suicidal behaviour

Scoping review of study outcomes

3.10 In line with the goal of providing a broad overview of the literature as a whole, we carried out a scoping review to identify trends in the literature. This evaluated outcomes for the full range of studies available, irrespective of the quality of individual study design, in an attempt to identify possible 'promising' interventions requiring further evaluation and to assess the impact on outcomes of key features of the available studies such as populations, settings and mode of intervention. The scoping review also attempted to address the likely cost-effectiveness of interventions, but in the event we were only able to identify three studies focussing on this issue. In considering the outcomes of the Scoping Review and also of the evaluation of the highest quality evidence set out in subsequent sections of the report, it is important to recognise that, where research is lacking, this cannot be taken as an indication that the intervention does not work, it is simply that there is, as yet, insufficient evidence to determine whether the intervention works or not. This is, unfortunately, the case for very many interventions which have been tried for suicidal behaviour and ideation. A detailed account of the findings of the scoping review is given in Annex G, here the main outcomes are summarised for each of the behaviours evaluated:

Suicide

• Around one third of studies (33%) evaluated intervention outcomes on the basis of their impact on completed suicide
• One third of these studies (33%) provided statistical evidence supporting the conclusion that completed suicide had been significantly reduced
• Studies with follow-up in the community and using official statistics as their main outcome measure were more likely to report positive outcomes
• Unequivocal support for the impact of specific interventions on completed suicide is nevertheless lacking due to the diverse focus of studies reporting successful outcomes
• Interventions which are consistently supported by a number of studies including some higher quality studies are the restriction of access to means and the maintenance of ongoing contact with the suicidal person; a promising approach to service delivery identified by a small number of higher quality studies is provision via specialist centres with highly trained personnel

Attempted suicide

• Over one third of studies (37%) evaluated intervention outcomes on the basis of their impact on attempted suicide
• Just under half of these studies (44%) provided statistical evidence supporting the conclusion that attempted suicide had been significantly reduced
• Higher quality studies and studies conducted on populations outside the US and Canada were more likely to report positive outcomes
• No specific intervention is consistently supported by more than a very small number of studies within the literature, promising interventions are treatment with lithium for bipolar disorder ( with the caveat that one study has identified increases in suicide and an additional study has identified increases in risk following discontinuation of treatment), restriction of access to means and the setting up of informal social support networks

Self-harm

• Only around one fifth of studies (22%) evaluated intervention outcomes on the basis of their impact on self-harm
• Just over one third of these studies (34%) provided statistical evidence supporting the conclusion that self harm had been significantly reduced
• Higher quality studies, in particular randomised controlled trials, were more likely to report positive outcomes, whilst studies focussed on older people (aged >65) and children or young adults (aged <25) were less likely to do so
• The only specific intervention finding consistent support from a number of studies within this literature is Dialectical Behaviour Therapy ( DBT) in the treatment of borderline personality disorder. Cognitive Behaviour Therapy ( CBT) and the maintenance of ongoing contact also find some support as promising approaches, although the evidence for these options is more limited.

Suicidal ideation

• Close to half of the studies (47%) evaluated intervention outcomes on the basis of their impact on suicidal ideation
• Just under half of these studies (43%) provided statistical evidence supporting the conclusion that suicidal ideation had been significantly reduced
• The diversity of interventions, methodologies and populations studied is too great to allow for comparative analysis of outcomes
• No specific intervention is consistently supported by the literature: promising approaches are treatment of depression with sertraline or fluvoxamine and non-directive telephone-based support

3.11 A key message to take away from the above summary of the scoping review is that the 'scatter-gun' approach adopted by the research literature has hampered efforts to identify successful interventions. There are numerous reports of interventions which show promise, but very few interventions have been evaluated by more than one or two studies. Where an intervention has been evaluated by a reasonable number of studies, outcomes tend to be more equivocal, suggesting that the impact of specific interventions on suicidal behaviour or ideation may also be context dependent (an intervention of benefit to one population or within one setting may not prove effective in another). Finally, it is clear that both the methodological approach taken and the quality and rigour of study design also have an impact on outcomes. This argues for caution in the interpretation of outcomes from poorer quality studies.

3.12 To explore further the impact of different features of study design and focus on the outcomes reported, we carried out a multivariate regression analysis. Details of this analysis are set out in Annex G. The analysis suggests that the likelihood that a study will report positive outcomes is most strongly influenced by whether the study design is quantitative or qualitative (qualitative studies are significantly less likely to report that the intervention was successful); whether or not the study focuses on people with mental health problems (with a greater likelihood that successful outcomes will be reported for psychiatric populations) and on the primary focus of the study (with studies evaluating interventions for self-harm being significantly less likely to report positive outcomes than studies focussing on other forms of suicidal behaviour or ideation). These outcomes emphasise the need for greater attention to detail in both research and practice - suicidal behaviour is not a unitary phenomenon and the outcome of intervention is likely to be highly context specific.

« Previous | Contents | Next »