The challenge of suicide prevention

It has been estimated that globally there are over 800 000 deaths by suicide each year (approximately 1 every 40 seconds), ¹ a figure that is likely to be a considerable underestimate of the real size of the problem. Suicide is a highly complex issue, with a range of causes at both the individual and contextual levels. Examples of individual factors are genetic influences, mental disorders and psychological vulnerabilities. Contextual contributions include, for example, family influences, abuse, socioeconomic downturn, exposure to suicidal behaviour by others, access to means for a suicidal act and lack of availability of help at times of crisis. Careful investigation of the processes that lead to specific deaths by suicide usually identify multiple likely contributory factors. Suicide is rarely because of a single cause.

As suicide can be caused by a complexity of factors, if suicide prevention policies are to be effective they must include a range of initiatives. ² These include, first, what are termed ‘universal’ interventions, aimed at reducing risk at the population level. Examples include reduction of access to means of suicide through initiatives such as increasing the safety of sites that are often used for suicide (such as certain bridges, cliffs, railways, multi-storey car parks) by installing barriers and other structures that reduce access; limiting amounts of dangerous medicines and withdrawal of more toxic drugs and other substances (for example pesticides) that have frequently been used for fatal self-poisoning; public education and prevention programmes in schools; and promoting responsible media reporting and portrayal of suicide.

Second, there are ‘selective’ interventions focused on subgroups of the population who may have heightened risk but are not necessarily having suicidal thoughts or engaging in self-harm, such as individuals with mental illness, substance misuse, financial problems and chronic pain. Examples of such interventions include psychological therapy for people with depression, specific programmes for people who have experienced sexual and physical abuse, improvements to the safety of psychiatric in-patient units, and training opportunities for clinicians, police, prison staff and others likely to encounter people at risk in their daily work.

Third, there are ‘indicated’ interventions for individuals who are showing more evident or immediate signs of suicide risk, such as those who have self-harmed, expressed suicidal thoughts or recently been discharged from psychiatric hospital care. Possible interventions include psychological therapy, frequent appointments with clinical services, ensuring early follow-up of patients leaving psychiatric hospitals and use of medication thought to have anti-suicidal effects (for example lithium, ketamine).

The challenge of evaluating suicide prevention initiatives

As a result of the range of types of potential suicide prevention initiatives, evaluation of their effectiveness requires different methodologies. In this issue of the BJPsych, Riblet and colleagues ^{Reference Riblet, Shiner, Young-Xu and Watts3} have focused on one approach to evaluation, namely randomised controlled trials (RCTs). They have conducted a systematic review and meta-analysis of evaluations using this design, using suicide as the sole outcome measure. They have concluded that just one initiative, the World Health Organization brief intervention and contact intervention for people who have self-harmed that was conducted in low- and middle-income countries, ‘is a promising suicide prevention strategy’. This is based on meta-analysis of just three trials.

Policy-makers and clinicians looking for evidence on which to base development of national or local suicide prevention programmes are likely to be disappointed with this conclusion as it does not provide guidance on which other approaches to suicide prevention might be effective. RCTs are undoubtedly valuable for evaluating certain types of approaches, especially some of those in the ‘indicated’ intervention category, but it is not always feasible or appropriate to use them to establish whether many of the other possible approaches are effective. This is a particular problem in the field of suicide prevention, especially for ‘universal’ interventions. For example, erection of a barrier to prevent suicide by jumping from a bridge known to be used frequently for this purpose is clearly going to be introduced for the whole community; it is not feasible to randomise half the community to have exposure to the barrier and half not. However, a quasi-experimental approach could be used, involving a before-and-after design, perhaps augmented with a parallel comparison with a similar bridge in another area that also has a relatively high incidence of suicide by jumping but where a barrier has not been erected. Similar problems are encountered in evaluation of other ‘universal’ approaches where, for example, availability of dangerous medication known to be used in fatal self-poisonings is restricted or terminated across the population. Here, just a before-and-after approach to evaluation may be feasible. An example from the UK was the withdrawal of co-proxamol (dextropropoxyphene and paracetamol combination analgesic) in the UK in the mid-2000s. This was shown to not only reduce the toll of suicidal and accidental deaths involving this drug but also to not lead to an increase in deaths involving ingestion of other (less toxic) analgesics, at least during the first 3 years following withdrawal of the drug. ^{Reference Hawton, Bergen, Simkin, Wells, Kapur and Gunnell4}

Another important issue is that although suicide is a relatively common cause of death, indeed the most common cause of death in men aged under 50 years in England in 2015, ⁵ the population rate of suicide means that huge and unrealistic numbers of participants are required in RCTs to evaluate whether a specific intervention has an impact on suicides compared with a control intervention. Therefore, other related outcomes need to be assessed. The most obvious is non-fatal self-harm, which occurs very many times as frequently as suicide. Previous self-harm is recognised as a major risk factor for suicide, so it makes good sense that reducing self-harm would ultimately reduce suicide. In any case, self-harm is worthy of attention in its own right, as indicated by its acknowledgement in the recently refreshed National Suicide Prevention Strategy for England. ⁵ A recent review and meta-analysis of 18 trials of cognitive–behavioural-based psychotherapy has shown a significantly greater impact on the proportion of trial participants repeating self-harm compared with treatment as usual, and on the basis of three trials, a reduction in frequency of self-harm associated with receiving dialectical behaviour therapy. ^{Reference Hawton, Witt, Salisbury, Arensman, Gunnell and Hazell6} There were insufficient numbers of suicides in these trials to meaningfully assess the impact on this outcome.

What is the current evidence for effectiveness of measures to prevent suicidal behaviour?

So, what is the current position regarding which suicide prevention initiatives are effective? In the seminal review by Mann and colleagues of effectiveness of interventions the main conclusion was that restriction of access to means of suicide and education of primary care physicians are effective. ^{Reference Mann, Apter, Bertolote, Beautrais, Currier and Haas7} A similar recent review by Zalsman and colleagues has provided a more contemporary overview of what may work. ^{Reference Zalsman, Hawton, Wasserman, van Heeringen, Arensman and Sarchiapone8} Evidence regarding effectiveness of restricting access to methods of suicide appears to have strengthened, particularly measures to prevent suicide at sites popular for suicide. ^{Reference Pirkis, Too, Spittal, Krysinska, Robinson and Cheung9} In contrast, evidence for beneficial effects on suicidal behaviour of physician education regarding detection and management of depression appears weaker than it did a decade previously. School-based awareness programmes are beginning to show positive benefits with regard to prevention of self-harm (and suicidal ideation). It remains unclear whether antidepressants reduce suicide risk, except in older adults. Lithium is probably the drug with most evidence of anti-suicidal effects, although Riblet and colleagues in their review concluded that the evidence was less strong than it had been, but largely because of inclusion of a recent trial of poor quality. ^{Reference Riblet, Shiner, Young-Xu and Watts3} Further work based on the National Confidential Inquiry into Suicide and Homicide by People with Mental Illness that was not included in the review by Zalsman and colleagues ^{Reference Zalsman, Hawton, Wasserman, van Heeringen, Arensman and Sarchiapone8} indicates that certain aspects of care by mental health services may reduce risk of suicide, namely removal of ligature points from hospital wards, specialised community teams such as outreach and crisis resolution/home treatment teams, services for patients with dual diagnoses, implementation of National Institute for Health and Care Excellence guidance for depression and self-harm and multidisciplinary review after suicide, with family input. ¹⁰