Predicting suicide following self-harm: risk analysis provides little help


The impact of suicide is devastating for all those affected and for society as a whole. Suicide is the result of self-inflicted injuries, and suicides are therefore preventable. It is important to understand the factors that are associated with suicide so that we can better understand how to prevent deaths.

The WHO reports that on a global level there are approximately one million deaths by suicide annually. There were 6,581 recorded cases of suicide in 2014 in the UK and Republic of Ireland (Samaritans Suicide Statistics Report 2016).

Evidence indicates that a history of self-harm is an important risk factor for future suicide. It elevates the risk of suicide 50 to 100 fold within the year following self-harm (Chan et al, 2016). Suicidal behaviour is described as a means of addressing unbearable psychological pain rather than a definite decision to die.

People who self-harm do so for a variety of reasons, almost always as an expression of emotional pain, and with a variety of intentions; not always with suicidal intent. Sometimes it is even intended as an act of self-preservation. The behaviour is therefore often associated with ambivalence and self harm may be viewed as lying on a continuum of suicidal behaviour, with death by suicide at the extreme end of the continuum.


How can we predict when self-harm will escalate to suicidal feelings?

In support of this, a previous large-scale study of Scottish adolescents found that of those who had engaged in self-harm, four out of ten did so with the intent to take their own life (O’Connor et al, 2009). The incidence of self-harm in the UK has continued to increase over the past 20 years and is among some of the highest in Europe for young people (RCP CR158). There are approximately 200,000 hospital attendances annually in England and Wales as a result of self-harm (Chan et al, 2016).

Service providers therefore have a dilemma with regards to how to manage this behaviour and provide appropriate treatment. The application of blanket interventions, such as automatic admission to all those who self-harm, will be overly restrictive for individuals who require a safety plan. Yet the absence of psychiatric care may be insufficiently supportive for those patients with a suicide plan that will be enacted in the near future. The challenge lies in the identification and stratification of patients who would benefit from more tailored interventions to reduce the likelihood of suicide.

A recent BMJ review of suicide risk assessment and intervention in people with mental illness illustrates that risk assessment tools which place particular emphasis on demographic factors are unable to predict suicide risk accurately and should not be relied upon (Bolton et al, 2015).

The current paper, a systematic review by Chan et al (2016) provides a more comprehensive assessment of the literature on risk assessment and includes scales that involve different types of variables.

What self harm risk factors can be identified to help provide better care?

What self harm risk factors can be identified to help provide better care?


The stated aim was to examine both individual risk factors and combinations of risk factors (evaluated by risk scales).

The study distinguishes these two and constructs a ‘two arm’ model. It describes itself as the first systematic review and meta-analysis of:

  • Prospective studies examining the factors associated with suicide following self-harm, and
  • Risk assessment scales predicting suicide in people who have self-harmed or were under specialist mental healthcare.

Four databases were searched for English-language prospective cohort studies of populations who had self-harmed, from inception of the database to 2014.

For the Risk Factors review the authors included:

  • Studies of people who presented to hospital following self-harm,
  • All types of self-harm (irrespective of the motive),
  • Participants who had experienced at least one episode of self-harm (recruited in the hospital setting).

They were followed up for variable time periods. Suicide was the main outcome measure, most commonly determined from national registers. Studies were included if they reported an effect estimate (odds ratio, risk ratio or hazards ratio with their 95% confidence interval) for the association between the examined risk factor and suicide following self-harm. Each of the risk factors was listed with the effect estimate from each study for that risk factor, grouped together and meta-analysed.

Twelve prospective cohort studies were included in the meta-analysis for risk factors associated with suicide following self-harm. Seven risk factors were evaluated:

  • Gender
  • Previous episodes of self-harm
  • History of psychiatric contact
  • Suicidal intent
  • Alcohol misuse
  • Poor physical health
  • Economic status

Exclusion criteria included:

  • Retrospective study design, if the main outcome was not relevant or extractable and if the population did not meet criteria
  • Those with a specific mental health disorder
  • Older adults and adolescents in school
  • People with intellectual disabilities
  • Those with suicidal ideation only who did not meet the population criteria.

For the ‘Risk Scale’ review, participants had self-harmed or were under mental healthcare and had been assessed using a risk assessment scale. For this group, the numbers of deaths by suicide were calculated to determine the predictive validity of each risk scale. Risk assessment scales were only included if they had been validated by one previous study. They examined the sensitivity, specificity, positive predictive value and negative predictive value of each risk scale.

Seven prospective cohort studies were included in the review of risk scales, which included three scales: the Beck Hopelessness Scale (BHS), the Suicide Intent Scale (SIS), and the Scale for Suicide Ideation (SSI).

Meta-analysis was only performed on the BHS and the SIS due to lack of data for the SSI. Exclusion criteria for the risk scales included when relevant material was not provided (such as no data on completed suicide).


Of the two risk scales evaluated in the meta-analysis (BHS and SIS), both had a low positive predictive value and significant numbers of false positives. There was a lack of evidence to support the superiority of either risk scale. Therefore the authors concluded that:

these scales should not be used alone in clinical practice to assess the future risk of suicide.

Of the seven risk factors evaluated, four had robust evidence to support their association with suicide following the index episode of self-harm:

  • Previous episodes of self-harm (prior to the index episode)
  • Male gender
  • Suicidal intent
  • Poor physical health.

There was insufficient evidence to make a firm conclusion on the association between the remaining three risk factors with suicide following an index episode of self-harm. These three inconclusive risk factors were:

  • History of psychiatric contact
  • Alcohol misuse
  • Economic status.
Previous self-harm, being male, having suicidal intent and being in poor physical health were all associated with suicide following self-harm.

Previous self-harm, being male, having suicidal intent and being in poor physical health were all associated with suicide following self-harm.

Strengths and limitations

This study is unique in being the first review and meta-analysis of risk assessment scales in people who have self-harmed or were under mental healthcare (Mulder et al 2016). There were a relatively small number of studies suitable for inclusion, particularly in the risk-scale review. This limited evidence base is in itself disappointing and indicates a need for further research in this area. Furthermore, the four risk factors identified are common in the clinical self-harm population and are of limited clinical value in predicting suicide.

It would potentially be useful to have a meta-analysis that included other risk factors not included in the current study (e.g. isolation, recent bereavement, migrant status, sexual orientation, or a mental health disorder).

However, reliance upon demographic risk factor identification is of limited value to both clinicians and patients. Prediction studies offer minimal clinical value in determining the risk of individual patients, as even the risk factors associated with the highest odds ratios and significant statistical associations may not be clinically useful when assessing an individual patient (Christensen et al, 2016).

NICE guidelines for long-term management of self-harm state

Do not use risk assessment tools and scales to predict future suicide or repetition of self-harm.

However understanding which factors differentiate those who will have thoughts of suicide, and those who will act upon those thoughts and attempt suicide is a key goal of current research programmes (Klonsky et al, 2014; O’Connor et al, 2014). In particular, alcohol and drug misuse was found to be only marginally significant, and yet we know that it is implicated in a high proportion of suicide deaths (O’Neill et al, 2016). Further research, with more detailed delineation of alcohol and the use of different categories of drugs may facilitate clearer conclusions.

It bears repetition that the two risk scales evaluated in the meta-analysis (BHS and SIS) had a low positive predictive value and high proportions of false positives. Risk scales claim to reduce uncertainty and inform intervention strategies. However this paper is a reminder that:

the idea of risk assessment as risk prediction is a fallacy.

As the Royal College of Psychiatrists reminds us,

Risk assessment per se has a very limited, and short-term, predictive power of a person’s future risk’ (Royal College of Psychiatrists, 2010).

So managing uncertainty remains a core aspect of clinical practice when dealing with suicide risk (Dixon et al, 2007). Yet, in a culture of blame is it inappropriate to suggest to a busy clinician that all they can rely upon is professional judgement?


Risk analysis provides little help to busy clinicians supporting people at risk of suicide.

A suicide mitigation approach may be more useful in this context (Cole-King, 2011). A mitigation approach needs to be adopted within a context of multi-disciplinary practice, at the assessment as well as the management and therapeutic intervention levels. A suicide mitigation approach recognises that self-harm increases the likelihood of future suicide which is why every episode of self-harm needs to be taken seriously. Self harm behaviour, when previously used as a method of managing pain, can, in the context of unbearable distress, escalate and result in death. Early identification and intervention can minimise distress and reduce the likelihood of such a coping mechanism becoming established and entrenched (Cole-King and O’Neill, 2016).

Whilst the use of risk indices to dictate treatment and referral following self-harm is strongly discouraged, the clinician should nonetheless be familiar with established risk factors and risk groups for suicide at a population level. However it is important that clinicians do not rely wholly on this knowledge when assessing specific individuals. A person may be at high risk of suicide even though not a member of a high-risk group and conversely, not all members of high-risk groups are at equal risk of suicide. The presence of ‘red flag’ warning signs suggests that someone may be particularly at risk of suicide (Cole-King, 2013). However risk factors and ‘red flag’ warning signs should not be used to predict, or rule out an individual suicide (attempt) (Cole-King and O’Neill, 2016).

Suicide mitigation involves encouraging help-seeking behaviour, removing or restricting access to means, and ensuring an appropriate and early response to suicidal behaviour (Zalsman et al 2016). Increasing hopefulness, emotional resilience, and helping someone to identify their reasons for living, have all been proven to lead to a reduction in suicide rates.

Every person considering self-harm or suicide needs to know they can be supported through tough times.

People at risk of suicide or after self-harm should be helped to develop a Safety Plan with details of what they can do to help themselves and who they can contact for support. Information about how to make a Safety Plan can be found in the Feeling Overwhelmed and Helping You Stay Safe resource. This is an Interactive online resource for anyone struggling to offer hope, compassion and practical ideas and suggestions on how to find a way forward.

Every person considering self-harm or suicide needs to know they can be supported through tough times.

Every person considering self-harm or suicide needs to know they can be supported through tough times.

If you need help

If you need help and support now and you live in the UK or the Republic of Ireland, please call the Samaritans on 116 123.

If you live elsewhere, we recommend finding a local Crisis Centre on the IASP website.


Primary paper

Chan MKY, Bhatti H, Meader N, Stockton S, Evans J, O’Connor RC, Kapur N, Kendall T. (2016) Predicting suicide following self-harm: systematic review of risk factors and risk scales. The British Journal of Psychiatry Oct 2016, 209 (4) 277-283; DOI: 10.1192/bjp.bp.115.170050 [Abstract]

Other references

Bolton JM, Gunnell D & Turecki G, (2015) Suicide risk assessment and intervention in people with mental illness. BMJ 2015; 351 doi: [article]

Suicide risk assessment and intervention: need for a new approach?

Christensen H, Cuijpers P, Reynolds CF. (2016) Changing the Direction of Suicide Prevention Research: A Necessity for True Population Impact. JAMA Psychiatry. 2016 May 1;73(5):435-6. doi: 10.1001/jamapsychiatry.2016.0001.

Cole-King A, Lepping P (2010) Suicide mitigation: Time for a more realistic approach. Br J Gen Pract. 2010 Jan 1; 60(570): e1–e3. doi: 10.3399/bjgp10X482022

Cole-King A, Green G, Gask L, Hines K, Platt S. (2013) Suicide mitigation: a compassionate approach to suicide prevention. Advances in Psychiatric Treatment Jul 2013, 19 (4) 276-283; DOI: 10.1192/apt.bp.110.008763

Cole-King A, O’Neill S. 2016 Suicide prevention for physicians: identification, intervention and mitigation of risk in mental Health In Primary Care Mental Health Gask L, Peveler R, Royal College of Psychiatrists Second Edition (in submission)

O’Neill S, Corry C, McFeeters D, Murphy S, Bunting B. Suicide in Northern Ireland. Crisis. 2016 Jan;37(1):13-20. doi: 10.1027/0227-5910/a000360. Epub 2015 Dec 23. [Pubmed abstract]

Samaritans Suicide Statistics Report 2016

WHO Mental Health Suicide data

Royal College of Psychiatrists (2010) Self-harm, suicide and risk: helping people who self-harm (PDF). Final report of a working group, College Report CR158, June 2010.

The Futility of Risk Prediction in Psychiatry

R.L. Logan, P.J Scott (1996) Uncertainty in clinical practice: implications for quality and costs of health care. The Lancet, Volume 347, No. 9001, p595–598. [Lancet abstract]

Mandy Dixon & Femi Oyebode (2007) Uncertainty and risk assessment. Advances in Psychiatric Treatment (2007), vol. 13, 70–78 doi: 10.1192/apt.bp.105.002022 [PDF]

Self-harm, suicide and risk: a summary. Royal College of Psychiatrists, Position Statement June 2010 [PDF]

Klonsky, E. D. and May, A. M. (2014), Differentiating Suicide Attempters from Suicide Ideators: A Critical Frontier for Suicidology Research. Suicide Life Threat Behav, 44: 1–5. doi:10.1111/sltb.12068 [Pubmed abstract]

O’Connor, R. C., & Nock, M. K. (2014). The psychology of suicidal behavior. Lancet Psychiatry, 1, 73–85. doi:10.1016/S22150366(14)70222-6 [Pubmed abstract]

O’Connor, R.C., Rasmussen, S., Miles, J., & Hawton, K.  (2009). Self-harm in adolescents: self-report survey in schools in Scotland. British Journal of Psychiatry, 194, 68-72 [article]

Zalsman G, Hawton K, Wasserman D, van Heeringen K, Arensman K, Sarchiapone M, Carli V, Höschl C, Barzilay R, Balazs J, Purebl G, Kahn JP, Sáiz PA, Bursztein Lipsicas C, Bobes J, Cozman D, Hegel U, Zohar J. (2016) Suicide prevention strategies revisited: 10-year systematic review, The Lancet Psychiatry, Available online 8 June 2016, ISSN 2215-0366. [abstract]

Photo credits

Share on Facebook Tweet this on Twitter Share on LinkedIn Share on Google+
Mark as read
Create a personal elf note about this blog
Profile photo of Camille Hart

Camille Hart

Camille Hart is a Specialist Trainee in General Adult Psychiatry and Medical Psychotherapy.

More posts

Follow me here –

Profile photo of Alys Cole-King

Alys Cole-King

Dr Alys Cole-King works nationally with Royal Colleges, voluntary bodies, academics, patients, and carers. Alys is a reviewer for journals and has published several articles, podcasts, webinars and book chapters in the field of compassionate care, suicide and self-harm mitigation and prevention. Alys has led international campaigns via social media and works with the media (film, radio and newspapers). Alys is the co-founder and Clinical Director of Connecting with People, a social enterprise and has received funding for developing and delivering MH awareness, emotional resilience and resourcefulness, suicide and self-harm awareness and response training across different sectors.

More posts - Website

Follow me here –

Profile photo of Siobhan O'Neill

Siobhan O'Neill

Siobhan is a Professor of Mental Health Sciences at Ulster University. Her current research programmes focus on trauma and suicidal behaviour in Northern Ireland (NI) and novel interventions for mental health and suicidal behaviour. Siobhan has expertise in qualitative and quantitative (epidemiology and survey) research methods. She coordinated the largest ever study of mental health in Northern Ireland, which revealed the high proportions of the NI population who had unmet mental health needs and the extent of mental health disorders and suicidal behaviour associated with the NI conflict. Siobhan is also a coordinator of the NI suicide study, a study of the characteristics of completed suicides and undetermined deaths.

More posts - Website

Follow me here –