NCBI Bookshelf. A service of the National Library of Medicine, National Institutes of Health.

National Collaborating Centre for Mental Health (UK). Self-Harm: Longer-Term Management. Leicester (UK): British Psychological Society; 2012. (NICE Clinical Guidelines, No. 133.)

Cover of Self-Harm: Longer-Term Management

Self-Harm: Longer-Term Management.

Show details



Many people who self-harm take psychotropic medication (Murphy et al., 2007), often as treatment for associated conditions such as depression. However evidence for the efficacy of pharmacological interventions to reduce self-harm is lacking. Some research suggests that lithium and clozapine may have specific anti-suicidal properties (Cipriani et al., 2005; Meltzer et al., 2003). Other studies have reported that prescription of certain classes of antidepressants (for example, selective serotonin reuptake inhibitors [SSRIs]) may be associated with an increase in suicidal behaviour particularly in young people (Barbui et al., 2009; Fergusson et al., 2005). Those who self-harm are at increased risk of future episodes, including overdoses of medication. There are large differences in the toxicity of medication prescribed to people who self-harm (Hawton et al., 2010).

Other NICE guidelines discuss the pharmacological treatment of conditions that may be associated with self-harm (for example, NICE, 2009a, 2009b and 2009e). The aim in the current chapter is to review the RCT evidence specifically for pharmacological treatment of self-harm. Because of variation in the toxicity of medication, a discussion of studies that could help to inform safer prescribing practices is also included.


8.2.1. Studies considered20

An existing systematic review was identified (Hawton et al., 2011) and the authors made their data available to the NCCMH team. The review for this guideline includes five studies (BATTAGLIA1999 [Battaglia et al., 1999]; HALLAHAN2007 [Hallahan et al., 2007], HIRSCH1982 [Hirsch et al., 1982]; MONTGOMERY1979 [Montgomery et al., 1979]; LAUTERBACH2008 [Lauterbach et al., 2008]). The GDG decided to exclude two studies (MONTGOMERY1983 [Montgomery et al., 1983], VERKES1998 [Verkes et al., 1998]) from the meta-analysis because they looked at people who had a diagnosis of personality disorder. Pharmacological treatment options in the treatment of personality disorder are partly covered in the NICE Borderline Personality Disorder guideline (NICE, 2009e).

Additional systematic searches were undertaken to update the review. The last search was dated in January 2011. No additional studies that met inclusion criteria were found.

Further information about both included and excluded studies can be found in Appendix 15f.

8.2.2. Clinical evidence for antidepressants versus placebo

One study compared antidepressants with placebo (HIRSCH1982). This was reviewed in the NICE guideline Self-harm: the Short-term Physical and Psychological Management and Secondary Prevention of Self-harm in Primary and Secondary Care (NICE, 2004a) and there were no new studies identified for this comparison.

Evidence from each important outcome and overall quality of evidence are presented in Table 56. The full evidence profiles and associated forest plots can be found in Appendix 17b and Appendix 16b, respectively.

Table 56. Summary study characteristics of trials comparing antidepressants versus placebo.

Table 56

Summary study characteristics of trials comparing antidepressants versus placebo.

There was insufficient evidence to differentiate clinical effectiveness between treatment and placebo on the reduction of repetition during the first 6 weeks of treatment (N = 114; RR 1.6, 95% CI, 0.63 to 4.04). There was insufficient evidence to determine the effect on death by suicide, or the acceptability of treatment.

Borderline personality disorder

Two studies (MONTGOMERY1983, VERKES1998) compared antidepressants versus placebo among people with personality disorders. All participants in MONTGOMERY1983 and 92% of participants in VERKES1998 had a diagnosis of personality disorder. Thus, the GDG decided not to include these two studies in the meta-analysis. In MONTGOMERY1983 participants took 30 mg of mianserin for 6 months. Participants in VERKES1998 took 40 mg paroxetine for 12 months and both groups received psychotherapy. However, there was insufficient evidence to determine differences between groups.

8.2.3. Clinical evidence for antipsychotic medication versus placebo or low-dose antipsychotic medication

Two studies included antipsychotics as one of their treatment arms (BATTAGLIA1999, MONTGOMERY1979). A narrative synthesis for each study was included in the NICE guideline on the short-term management of self-harm (NICE, 2004a) and no new studies had been identified since then. In MONTGOMERY1979, flupenthixol depot (20 mg) or placebo was administered every 4 weeks for 6 months. In BATTAGLIA1999, 12.5mg of fluphenazine or 1.5 mg fluphenazine was administered once a month for 6 months. Study characteristics can be found in Table 57.

Table 57. Summary study characteristics of trials comparing antipsychotics versus other comparators.

Table 57

Summary study characteristics of trials comparing antipsychotics versus other comparators.

There was limited evidence (MONTGOMERY1979) suggesting that there was a statistically significant clinical difference between flupenthixol and placebo on reducing repetition of self-harm (N = 37; RR 0.29, 95% CI, 0.1 to 0.81). Despite the observed effect, a wide variability in the CI was observed due to the small sample size. No statisticaly significant difference was found between groups regarding treatment complaince (N = 37; RR 0.92, 95% CI, 0.67 to 1.26). There were a total of seven dropouts, two of which were due to Parkinsonian side effects and the remaining five were not specified. As a result, it was not possible to make a recommendation based on this single trial.

There was insufficient evidence (BATTAGLIA1999) to differenitate clinical effectiveness between 12.5 mg and ultra low-dose 1.5 mg fluphenazine on reducing repetitions during 6 months after trial entry (N = 53; RR 1.28, 95% CI, 0.65 to 2.52). There were no suicide deaths reported in either trial arm. It was also unclear how the different dosage reduced the likelihood of leaving treatment early (N = 58; RR 1.12, 95% CI, 0.71 to 1.76).

8.2.4. Clinical evidence for other pharmacological interventions versus placebo

Two studies (HALLAHAN2007, LAUTERBACH2008) included medication other than antidepressants or antipsychotics. In LAUTERBACH2008, lithium was administered using a fixed schedule of dose augmentation increased by 200 mg per week for 3 to 4 weeks. At 1 year, the doses were reduced by half and discontinued at the 13th month. The majority of the participants had a diagnosis of depression. In HALLAHAN2007, participants were randomised to receive omega-3 fatty acid supplement (n = 22) of 1.2 g eicosapentaenoic acid and 0.9 g decosahexaenoic acid, or placebo for 12 weeks. Four identical capsules were given each morning to each group containing either an active ingredient or placebo. The active capsules contained a total dose equal to 2,128 mg per day of eicosapentaenoic acid plus decosahexaenoic acid. Patients continued to receive psychiatric care. The majority of participants had a diagnosis of personality disorder. Study characteristics can be found in Table 58.

Table 58. Summary study characteristics of trials comparing other medications versus placebo.

Table 58

Summary study characteristics of trials comparing other medications versus placebo.

There were no statistically significant clinical differences between lithum and placebo on any reported outcomes (LAUTERBACH2008) and no differences in reducing repetition of self-harm at 12 months from trial entry (N = 167; RR 0.99, 95% CI, 0.36 to 2.69). There were also no differences in terms of depression scores (measured by Hamilton Depression Rating Scale) at 3 months (SMD −0.05, 95% CI, −0.42 to 0.33), 6 months (SMD 0.07, 95% CI, −0.36 to 0.50) and 12 months (SMD −0.05, 95% CI, −0.54 to 0.44), and no differences in BHS score at 3 months (SMD −0.1, 95% CI, −0.49 to 0.3), 6 months (SMD −0.12, 95% CI, −0.67 to 0.42) and 12 months (SMD −0.03, 95% CI, −0.58 to 0.52). Three suicides occurred in the placebo arm and none in the treatment arm, but it should be noted that these suicides occurred within the context of a depressive spectrum disorder. Several limitations should be noted: there were more people who had personality disorders (p < 0.05) and multiple suicide attempts (p < 0.001) in the lithium group, and participants in the placebo group had higher baseline suicide ideation scores (p < 0.05). Furthermore, the high proportion of participants lost to follow-up (approximately 60%) might indicate the results were overestimated.

There was no statistically significant difference between omega-3 fatty acid supplement and placebo group during 12 weeks' treatment when looking at repetition as an outcome (N = 49; RR 1.23, 95% CI, 0.51 to 2.97) (HALLAHAN2007).

There was limited evidence to show a small reduction in depression scores (N = 49; SMD = −0.3, 95% CI, −0.87 to 0.26). Fewer participants in the treatment group reported suicide ideation at 12 weeks after trial entry compared with placebo (N = 49; RR 0.47, 95% CI, 0.24 to 0.9); there were slightly more participants (but not statistically significant) completing treatment (N = 49; RR 1.17, 95% CI, 0.88 to 1.54). A few limitations should be noted: at baseline, there were more married participants in the treatment group, and the depression scores in the treatment group were higher than the placebo group. These limitations might inflate the effects.

8.2.5. Clinical evidence summary

The evidence base for the pharmacological treatment of self-harm remains very limited since the publication of the NICE guideline Self-Harm: the Short-term Physical and Psychological Management and Secondary Prevention of Self-harm in Primary and Secondary Care (NICE, 2004a). With regard to the effects of antidepressants and antipsychotics on the reduction of self-harm behaviour, no new trials were identified. The clinical efficacy of these medications remains uncertain. The variations in the treatment lengths, follow-up period, and participants' psychiatric diagnosis in these trials made it more difficult to make any conclusions about the clinical effects of these medications.

There were two new trials looking at effects of lithium and omega-3 fatty acid supplement versus placebo. There was no evidence of reduction in repetitions in either trial. There might be a small improvement in a few symptom measures; however, these trials were too small to detect a statistically significant effect. There were baseline differences between groups in the two studies that might overestimate the clinical effects for some measures. Moreover, the population of these two studies had a high prevalence of psychiatric disorders (depressive disorder, alcohol misuse and personality disorder), which might limit generalisability of the findings.

8.2.6. Health economic evidence

No evidence on the cost effectiveness of pharmacological interventions for the management of self-harm was identified by the systematic search of the economic literature. Details on the methods used for the systematic search of the economic literature are described in Section 3.6.1.

According to the guideline systematic review of clinical evidence, the clinical efficacy of pharmacological interventions for the treatment of self-harm is uncertain; therefore, no economic modelling was undertaken in this area.


The issue of safer prescribing is not limited to treatment with psychotropic medications and is relevant to all prescribing to those with a known history of self-harm or who are at risk of self-harm. Prescribed medications may also be used to effect self-harm either as an end in itself, or as a consequence of use for other aims, for example the manipulation of, or neglect of insulin regimes to influence weight. The risks associated with the prescription of other potentially dangerous drugs such as warfarin should be assessed with reference to the health consequences of not prescribing it and considering alternatives.

There have been wider public health measures to limit the volumes of potentially hazardous drugs in the population. Consideration of some of these measures are beyond the remit of this guideline, but the recent limitation on prescription of co-proxamol has already been shown to result in fewer deaths from poisoning with no increase in those due to other analgesics (Hawton et al., 2009).

Generally, prescription of potentially toxic psychotropic medications such as lithium is undertaken in secondary care with close attention to the risks of overdose. The majority of antidepressants are prescribed in primary care. Commonly, SSRIs are regarded as being of low toxicity and knowledge of the variation within this group as well as differences between SSRIs, selective noradrenaline reuptake inhibitors and other antidepressants may not be widely appreciated.

8.3.1. Studies considered

A comprehensive search was conducted that resulted in 6,183 references being generated. Sifting was conducted by three members of the technical team based on the titles and abstracts of the references. Full texts of the studies of potential relevance were retrieved. Studies were excluded on the basis of the outcomes reported. Studies in which the fatal toxicity index or case fatality index were reported were included. A total of 18 studies met these criteria, of which the GDG decided to include only the most recent. Due to the changes in regulatory policy and development of newer drugs over years, only studies published in the last 5 years were reviewed. AFSHARI2005 (Afshari et al., 2005), HAUKKA2009 (Haukka et al., 2009) and HAWTON2010 (Hawton et al., 2010) were included. Due to the small number of studies, these have been narratively synthesised.

Toxicity is the primary outcome examined in the review. Toxicity can be measured by the fatal toxicity index or case fatality index. Fatal toxicity index is calculated by the number of deaths divided by the number of prescriptions of a particular drug. However, the interpretation of toxicity using the fatal toxicity index can be confounded by differential prescribing of drugs to particular groups of people (for example, people at highest risk of self-harm being preferentially prescribed particular medications). This is referred to as confounding by indication. Case fatality index is calculated by dividing the number of suicides by the number of fatal and non-fatal poisonings of a particular drug. Case fatality index might be a more reliable indicator of toxicity because it partly accounts for this ‘confounding by indication’.

Further information about both included and excluded studies can be found in Appendix 15f.

8.3.2. Narrative synthesis


This study was conducted in Scotland and aimed to look at the relative toxicity of co-proxamol in overdose compared with co-codamol and co-dydramol. Prescription data, number of overdoses and deaths relating to these popular paracetamol–opioid compound analgesics were collected. Co-proxamol was ten times more toxic than co-codamol or co-dydramol in terms of its fatal toxicity index, even after the differences in prescription data were accounted for.


This study, conducted in Finland, aimed to look at a national cohort of antidepressant users and how antidepressants related to the risk of suicide from 1999 to 2003. Data included in the study were the participants' years of usage and the number of suicides relating to that drug. Data were reported both by drug class and by individual drug. It was possible to calculate the fatal toxicity index for drug classes or individual drugs based on the data reported in the study. This showed that tricyclic antidepressants (TCAs) were more toxic than SSRIs. When individual drugs were considered, mirtazapine was the most toxic followed by venlafaxine and then moclobemide. It was unclear whether confounding by indication was accounted for. People at higher risk of self-harm might be prescribed a certain drug, which might not be accounted for in the calculation of the fatal toxicity index.


This was an observational study of prescriptions and suicide by self-poisoning in the UK that aimed to provide updated toxicity data of antidepressants, to aid clinicians' decision-making about prescriptions. Data included the death rate by suicide, prescriptions rate and self-poisoning rate for each individual antidepressant for people aged over 10 years from 2000 to 2006. The fatal toxicity index and case fatality index were then calculated. Data can be found in Appendix 16c. The study reported a very high correlation between the rankings of the results from the fatal toxicity and case fatality index (which may be a more reliable indicator of potential toxicity because case fatality index accounted for confounding by indication). The findings showed that TCAs as a drug class were more toxic than SSRIs. Dosulepin and doxepin were the most toxic antidepressants in terms of the fatal toxicity index and case fatality index.

In addition, the study reported venlafaxine was less toxic in overdose than other TCAs. It was, however, still more toxic than SSRIs. Although SSRIs generally had lower toxicity, not all drugs within the class were the same. There was a greater than three-fold variation in case fatality rates between individual SSRIs.

8.3.3. Clinical evidence summary

Three recent studies reviewed the toxicity of different drugs that were commonly used for self-poisoning (analgesics and antidepressants). These studies included different individual drugs in their comparison, hence it was not possible to synthesise the toxicity data across these studies. Nevertheless, a common finding could be concluded for antidepressants. TCAs as a drug class were more toxic than SSRIs.


There was insufficient evidence to determine whether the provision of pharmacological treatment would reduce the likelihood of repetition of self-harm. No new trials looking at antidepressants or antipsychotics had been identified. Hence, no recommendations could be made.

It is recommended that healthcare professionals provide pharmacological interventions for any associated conditions as described in the relevant NICE guidelines. When prescribing drugs, toxicity of prescribed drugs in overdose should be taken into consideration. There was recent evidence suggesting TCAs as a drug class were more toxic than SSRIs. When clinicians are considering antidepressants, SSRIs might be preferred in those at risk of suicidal behaviour. In particular, the more toxic TCAs such as dosulepin should be avoided.


8.5.1. Clinical practice recommendations

Do not offer drug treatment as a specific intervention to reduce self-harm.

Provide psychological, pharmacological and psychosocial interventions for any associated conditions, for example those described in the following published NICE guidance:

When prescribing drugs for associated mental health conditions to people who self-harm, take into account the toxicity of the prescribed drugs in overdose. For example, when considering antidepressants, selective serotonin reuptake inhibitors (SSRIs) may be preferred because they are less toxic than other classes of antidepressants. In particular, do not use tricyclic antidepressants, such as dosulepin, because they are more toxic.



Here and elsewhere in the guideline, each study considered for review is referred to by a study ID in capital letters (primary author and date of study publication, except where a study is in press or only submitted for publication, then a date is not used).


This recommendation also appears in Section 7.3 where the psychosocial data is presented.

Copyright © 2012, The British Psychological Society & The Royal College of Psychiatrists.

All rights reserved. No part of this book may be reprinted or reproduced or utilised in any form or by any electronic, mechanical, or other means, now known or hereafter invented, including photocopying and recording, or in any information storage or retrieval system, without permission in writing from the publishers. Enquiries in this regard should be directed to the British Psychological Society.

Bookshelf ID: NBK126780


Recent Activity

Your browsing activity is empty.

Activity recording is turned off.

Turn recording back on

See more...