Psychosocial and Pharmacological Treatments For Deliberate Self Harm (Review)

Psychosocial and pharmacological treatments for deliberate
self harm (Review)
Hawton KKE, Townsend E, Arensman E, Gunnell D, Hazell P, House A, van Heeringen K
This is a reprint of a Cochrane review, prepared and maintained by The Cochrane Collaboration and published in The Cochrane Library
2009, Issue 1
http://www.thecochranelibrary.com
Psychosocial and pharmacological treatments for deliberate self harm (Review)

Copyright © 2009 The Cochrane Collaboration. Published by John Wiley & Sons, Ltd.
TABLE OF CONTENTS
HEADER . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
ABSTRACT . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
PLAIN LANGUAGE SUMMARY . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
BACKGROUND . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
OBJECTIVES . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
METHODS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
RESULTS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
DISCUSSION . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
AUTHORS’ CONCLUSIONS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
ACKNOWLEDGEMENTS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
REFERENCES . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
CHARACTERISTICS OF STUDIES . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
DATA AND ANALYSES . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33
Analysis 1.1. Comparison 1 Problem solving therapy vs Standard aftercare, Outcome 1 Repetition. . . . . . . 35
Analysis 2.1. Comparison 2 Intensive intervention plus outreach vs. Standard aftercare, Outcome 1 Repetition. . . 36
Analysis 3.1. Comparison 3 Emegency card vs. Standard aftercare, Outcome 1 Repetition. . . . . . . . . . 36
Analysis 4.1. Comparison 4 Dialectical behavior therapy vs. Standard aftercare, Outcome 1 Repetition. . . . . . 37
Analysis 5.1. Comparison 5 Inpatient behavior therapy vs Inpatient insight-orientated therapy, Outcome 1 Repetition. 37
Analysis 6.1. Comparison 6 Same therapist vs. Different therapist, Outcome 1 Repetition. . . . . . . . . . 38
Analysis 7.1. Comparison 7 General hospital admission vs. Discharge, Outcome 1 Repetition. . . . . . . . . 38
Analysis 8.1. Comparison 8 Flupenthixol vs. Placebo, Outcome 1 Repetition. . . . . . . . . . . . . . . 39
Analysis 9.1. Comparison 9 Antidepressants vs. Placebo, Outcome 1 Repetition. . . . . . . . . . . . . . 39
Analysis 10.1. Comparison 10 Long-term therapy vs. Short-term therapy, Outcome 1 Repetition. . . . . . . . 40
Analysis 11.1. Comparison 11 Homebased family therapy vs. Standard aftercare, Outcome 1 Repetition. . . . . 40
WHAT’S NEW . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40
HISTORY . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41
CONTRIBUTIONS OF AUTHORS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41
DECLARATIONS OF INTEREST . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41
SOURCES OF SUPPORT . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41
NOTES . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41
INDEX TERMS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41
Psychosocial and pharmacological treatments for deliberate self harm (Review) i

[Intervention Review]
Psychosocial and pharmacological treatments for deliberate

self harm
Keith KE Hawton1 , Ellen Townsend2 , Ella Arensman3 , David Gunnell4 , Philip Hazell5 , Allan House6 , K van Heeringen7
1
University Department of Psychiatry, Warneford Hospital, Oxford, UK. 2 Risk Analysis, Social Processes and Health Group, School
of Psychology, University of Nottingham, Nottingham, UK. 3 Vrije University, HV Amsterdam, Netherlands. 4 Department of Social
Medicine, University of Bristol, Bristol, UK. 5 Thomas Walker Hospital (Rivendell) Child, Adolescent and Family Mental Health
Services, Concord West, Australia. 6 Leeds Institute of Health Sciences, University of Leeds, Leeds, UK. 7 University Department of
Psychiatry, University Hospital, Gent, Belgium
Contact address: Keith KE Hawton, University Department of Psychiatry, Warneford Hospital, Oxford, OX3 7JX, UK.
keith.hawton@psych.ox.ac.uk.
Editorial group: Cochrane Depression, Anxiety and Neurosis Group.

Publication status and date: Edited (no change to conclusions), published in Issue 1, 2009.
Review content assessed as up-to-date: 29 July 1999.
Citation: Hawton KKE, Townsend E, Arensman E, Gunnell D, Hazell P, House A, van Heeringen K. Psychosocial and pharma-
cological treatments for deliberate self harm. Cochrane Database of Systematic Reviews 1999, Issue 4. Art. No.: CD001764. DOI:
10.1002/14651858.CD001764.
ABSTRACT
Background
Deliberate self-harm is a major health problem associated with considerable risk of subsequent self-harm, including completed suicide.
Objectives
To identify and synthesise the findings from all randomised controlled trials that have examined the effectiveness of treatments of
patients who have deliberately harmed themselves.
Search strategy
Electronic databases screened: MEDLINE (from 1966-February 1999); PsycLit (from 1974-March 1999); Embase (from 1980-January
1999); The Cochrane Controlled Trials Register (CCTR) No.1 1999. Ten journals in the field of psychiatry and psychology were hand
searched for the first version of this review. We have updated the hand search of three specialist journals in the field of suicidal research
until the end of 1998. Reference lists of papers were checked and trialists contacted.
Selection criteria
All RCTs of psychosocial and/or psychopharmacological treatment versus standard or less intensive types of aftercare for patients who
shortly before entering a study engaged in any type of deliberately initiated self-poisoning or self-injury, both of which are generally
subsumed under the term deliberate self-harm.
Data collection and analysis
Data were extracted from the original reports independently by two reviewers. Studies were categorized according to type of treatment.
The outcome measure used to assess the efficacy of treatment interventions for deliberate self-harm was the rate of repeated suicidal
behaviour. We have been unable to examine other outcome measures as originally planned (e.g. compliance with treatment, depression,
hopelessness, suicidal ideation/thoughts, change in problems/problem resolution).
Psychosocial and pharmacological treatments for deliberate self harm (Review) 1
Main results
A total of 23 trials were identified in which repetition of deliberate self-harm was reported as an outcome variable. The trials were
classified into 11 categories. The summary odds ratio indicated a trend towards reduced repetition of deliberate self-harm for problem-
solving therapy compared with standard aftercare (0.70; 0.45 to 1.11) and for provision of an emergency contact card in addition to
standard care compared with standard aftercare alone (0.45; 0.19 to 1.07). The summary odds ratio for trials of intensive aftercare
plus outreach compared with standard aftercare was 0.83 (0.61 to 1.14), and for antidepressant treatment compared with placebo was
0.83 (0.47 to 1.48). The remainder of the comparisons were in single small trials. Significantly reduced rates of further self-harm were
observed for depot flupenthixol vs. placebo in multiple repeaters (0.09; 0.02 to 0.50), and for dialectical behaviour therapy vs. standard
aftercare (0.24; 0.06 to 0.93).
Authors’ conclusions
There still remains considerable uncertainty about which forms of psychosocial and physical treatments of self-harm patients are most
effective, inclusion of insufficient numbers of patients in trials being the main limiting factor. There is a need for larger trials of
treatments associated with trends towards reduced rates of repetition of deliberate self-harm. The results of small single trials which
have been associated with statistically significant reductions in repetition must be interpreted with caution and it is desirable that such
trials are also replicated.
PLAIN LANGUAGE SUMMARY
Psychosocial and pharmacological treatments for deliberate self harm
Deliberate self-harm is a major health problem associated with considerable risk of subsequent self-harm, including completed suicide.
This systematic review evaluated the effectiveness of various treatments for deliberate self-harm patients in terms of prevention of
further suicidal behaviour. From the results of 23 randomized controlled trials the reviewers concluded that more evidence is required
to indicate what the most effective care is for this large patient population. Promising results were found for problem-solving therapy,
provision of a card to allow emergency contact with services, depot flupenthixol for recurrent repeaters of self-harm and long-term
psychological therapy for female patients with borderline personality disorder and recurrent self-harm. However, insufficient numbers
of patients in nearly all trials limit the conclusions that can be reached. More evidence is required to determine the most effective
treatment for deliberate self-harm patients and larger trials are badly needed.
BACKGROUND
self-harm’ (WHO 1982; WHO 1986). The United Nations has
In many countries, suicidal behaviour has been identified as a also highlighted the importance of suicide prevention (UN 1996).
major public health problem, both with regard to mortality and
treatment of patients who have deliberately harmed themselves. In Europe, most attention to suicide prevention strategies has oc-
In 1984, all member states of the European Region of the World curred in Scandinavia, with Finland having a very sophisticated
Health Organization adopted a common European health policy programme based on a national psychological autopsy study, and
including 38 targets for attaining health for all. In one target the Sweden, Norway and Denmark having their own programmes.
public health importance of suicide and deliberate self-harm is
In Belgium and The Netherlands, suicide prevention is currently
recognized as a major cause of mortality and morbidity. The tar-
not included in public health policy. However, in Belgium efforts
get states that ’by the year 2000 there should be a sustained and
are being made by the Minister of Health to develop a suicide
continuing reduction in the prevalence of mental disorders, and
prevention programme.
improvement in the quality of life of all people with such disor-
ders, and a reversal of the rising trends in suicide and deliberate The Australian Government through its Commonwealth Depart-
ment of Human Services and Health has made a commitment to groupings of treatments which do not inform clinical practice (
reducing suicide and suicidal behaviour, with a particular focus on van der Sande 1997).
youth suicide. No specific targets have been set, but considerable
In this review we have focused on deliberately initiated acts of
funds have been allocated for programmes aimed at the reduction
self-harm with non-fatal outcome, including both self-poisoning
of suicide. The New Zealand Government has recently commis-
and self-injury. Different terms have been proposed to describe
sioned a report to assist its suicide prevention policy.
the same type of behaviour, such as ’attempted suicide’, ’parasui-
In the USA, attention has been paid to prevention of youth suicide, cide’, and ’deliberate self-harm’. No consensus has yet been reached
and a more general discussion of suicide prevention has taken place about one common term. In this review we will use the term ’de-
in Canada. liberate self-harm’ (DSH) as this term is more accurate in terms of
the range of motives that lead to self-destructive behavior. In the
In the UK the former Government established two suicide targets review we will not include studies concerning acts of self-injury as
for the year 2000. The first target was a reduction in the overall a symptom of mental retardation.
suicide rate by 15%, and the second a reduction in the suicide rate
of people with severe mental illness of 33% (DOH 1992). The
present Government has included a target of a reduction in the
OBJECTIVES
rate of death by suicide and undetermined injury of 17% by the
year 2010 (DOH 1998). There has until now been an absence 1) To identify all RCTs of treatments following DSH, and to con-
of assimilation of knowledge about the effectiveness of preventive duct a meta-analysis (where possible) to compare the effects of spe-
measures or of specific treatments of those at risk (Gunnell 1994). cific treatments (e.g., cognitive behavioural therapy or psychophar-
macological treatment) and standard types of aftercare (e.g., emer-
While it is difficult to examine the effects of treatments on rates of gency room service, psychiatric assessment) or control treatments
completed suicide, intervention following non-fatal suicidal be- (e.g. placebo) for DSH.
haviour is more amenable to evaluation. This is directly relevant to
suicide prevention, because the risk of suicide following deliberate 2) To test the hypothesis that specific treatments are more effective
self-harm (DSH) is considerable. Thus, at least 1% of patients for DSH than standard or other control types of aftercare.
referred to general hospitals in the United Kingdom for DSH die 3) To test the hypothesis that trials which include only patients
by suicide within a year of an episode of DSH, and 3-5% within who have a history of previous DSH result in greater differences
5-10 years (Hawton 1988). Rates of suicide following DSH are in outcome between treatment conditions than trials that include
considerably higher in some other countries where the DSH pop- a mixture of ’repeaters’ and ’first timers’.
ulation has an older age profile and includes more patients with
major psychiatric disorders (e.g. Rygnestad 1997). Looked at the In our previous version of this review we stated that we hoped to
other way around, 40-50% of people who die by suicide have pre- examine other issues. First, that female DSH patients comply with
vious episodes of DSH (Ovenstone 1974). In studies of risk in and also benefit more from treatment (both specific treatments
different psychiatric patient populations a history of DSH is the and standard types of aftercare) than males. Unfortunately, our
best predictor of eventual suicide, with those who have repeated efforts to obtain extra data from authors pertaining to this issue
episodes of self-harm being at the greatest risk. DSH is more com- have not been successful. Second, that there would be a difference
mon among females; in most studies, two thirds of patients who in outcome between patients carrying out a first act of DSH and
have self-harmed are females. However, since the mid 1980s, DSH those who are repeaters. We also wished to examine the effects
rates in some European countries have increased among young of treatment for episodic self-harm behaviour (e.g. habitual self-
males (Hawton 1992; Kerkhof 1994). cutting) but no such studies appear to have been published.
At the present time in the UK, partly as a result of the former Gov-
ernment’s suicide targets, there is considerable focus on improving
METHODS
the standards of general hospital services for suicide attempters.
The Royal College of Psychiatrists has published consensus guide-
lines on standards for such services which mainly address assess-
ment procedures (RCP 1994). There is a need for assimilation Criteria for considering studies for this review
of evidence both with regard to these procedures and especially,
subsequent treatment. Descriptive reviews of treatment outcomes
in DSH patients have been published previously but have not in- Types of studies
cluded systematic screening of the literature, quality ratings and We sought to identify all randomized controlled trials of specific
meta-analysis (Hirsch 1982; Dew 1987; Moller, 1992; Hawton psychosocial and physical treatments versus any control in the
1989a; Hawton 1997a), or have also been based on heterogeneous treatment of DSH.

Types of participants Journal of Psychiatric Research, 1961-1972; 1985-1996
Participants were males and females of all ages, who shortly before Social Psychiatry, 1966-1987 and Social Psychiatry & Psychiatric
entering the study had all engaged in any type of deliberately Epidemiology, 1988-1996
initiated self-poisoning or self-harm. We did not include trials in * Search updated for current version of review.
which the participants were suicide ideators (without self-harm) (iii) Checking reference lists:
nor those of people with depression in which DSH was an outcome The reference lists of all relevant papers known to the investigators
variable. of treatment of self-harm patients were checked.
We did not include studies concerning treatment in the mentally (iv) Personal communication:
handicapped where DSH was repetitive behaviour (e.g. head bang- The authors of trials and other experts in the field of suicidal
ing), because the intention is probably rather different to that in- behaviour were consulted to find out if they knew of any published
volved in DSH in other populations. or unpublished RCTs of treatment of self-harm patients.
Types of interventions Data collection and analysis

All psychosocial and/or psychopharmacological treatment versus
SELECTION OF TRIALS
standard or less intensive types of aftercare.
Electronic databases:
One reviewer screened the abstracts of all publications which were
Types of outcome measures obtained by the search strategy. A distinction was made between:
1) eligible studies, in which any psychological and/or psychophar-
The main outcome measure used to assess the efficacy of treatment
macological treatment was compared with a standard type of af-
interventions for DSH was the rate of repeated self-harm (fatal
tercare.
and non-fatal) within a follow-up period of up to 2 years.
2) general treatment studies, without any control treatment.
We had hoped to examine other outcome variables e.g. compli-
The original articles of the eligible studies were screened in order
ance with treatment, depression, hopelessness, suicidal ideation/
to determine the status (RCT/CCT) of the study, and whether
thoughts, change in problems/problem resolution. Unfortunately,
they were relevant for the purpose of the review.
our attempts to obtain data from trialists to make these analyses
DATA EXTRACTION
possible have not been successful.
We extracted data from each eligible trial concerning the charac-
teristics of patients, the details of the interventions used and out-
come measures used to evaluate the efficacy of the treatments stud-
Search methods for identification of studies ied. This was carried out by two reviewers independently of each
The search strategy included the following sources: other. Where disagreements occurred these were resolved through
i) Electronic databases: consensus discussions with a third member of the group of review-
MEDLINE (from 1966-February 1999); PsycLit (from 1974- ers.
March 1999); Embase (from 1980-January 1999); The Cochrane QUALITY ASSESSMENT
Controlled Trials Register (CCTR) No.1 1999. The quality of the papers was rated by two independent reviewers
The search term used was; blind to their authorship according to the recommended Cochrane
Suicid* criteria for quality assessment (Sackett 1997). This rating system
(ii) Full-text searching/hand searching: is based on the finding that the quality of concealment of allo-
For the previous version of our review we hand-searched ten spe- cation can affect the results of trials (Schulz 1995). Studies were
cialist journals in the fields of psychology and psychiatry includ- assigned a quality rating from C (poorest quality) to A (best qual-
ing all the English language suicidology journals. We have now ity). Thus, trials rated as inadequately concealed (e.g. via alterna-
updated the search for these specialist journals until the end of tion or reference to an open random number table) were given a
1998. The journals searched were as follows: rating of C. Trials that did not give adequate details about how
Archives of Suicide Research, 1995-1998* the randomization procedure was carried out were given a rating
Crisis, 1980-1998* of B. Trials that were deemed to have taken adequate measures
Suicide and Life-Threatening Behavior, 1971-1998* to conceal allocation (e.g. serially numbered, opaque, sealed en-
Der Nervenarzt, 1950-1979 velopes; numbered or coded bottles or containers) were rated as A
Journal of Adolescence, 1978-1996 quality. We contacted authors of trials for more information where
Journal of Affective Disorders, 1994-1996 the concealment of allocation was not clearly reported (i.e. where
Journal of the American Academy of Child and Adolescent Psy- trials were initially in category B). Where raters disagreed the final
chiatry,1978-1996 rating was made by consensus, including the opinion of a third
Journal of Clinical Psychiatry, 1978-1996 member of the group of reviewers.

GROUPING OF THE STUDIES The assessment of quality of concealment of allocation resulted
Studies which shared similar treatment strategies were grouped by in 16 trials (70%) being given rated as ’A’ quality (the highest
consensus of the reviewers, blind to the outcome data. The first rating - for adequate concealment), 3 trials (13%) being given a
category (PROBLEM SOLVING THERAPY VS. STANDARD rating of B (for unclear concealment), and 4 trials (17%) being
AFTERCARE) included studies in which participants in the ex- given a rating of C (inadequate concealment). Quality scores for
perimental group were offered some form of problem-solving ther- individual trials were as follows:
apy which was compared with standard aftercare. The second Chowdhury et al 1973 C
group (INTENSIVE INTERVENTION PLUS OUTREACH Welu 1977 C
VS. STANDARD AFTERCARE) included studies in which the Gibbons 1978 A
patients in the experimental group had greater access to thera- Montgomery et al 1979 A
pists than in standard care, and where efforts were made to keep Hawton 1981 A
contact with patients through some form of outreach (e.g. home- Liberman & Eckman 1981 B
based treatment either as standard or for those patients who de- Hirsch 1982 A
faulted on appointments at a clinic). The third group (EMER- Montgomery et al 1983 A
GENCY CARD VS. STANDARD AFTERCARE) included stud- Hawton 1987 A
ies in which patients in the experimental group, in addition to Torhorst 1987 B
being offered standard aftercare, were given an emergency contact Torhorst 1988 B
card with which they either had 24-hour access to emergency ad- Salkovskis et al 1990 A
vice from a psychiatrist (Morgan 1993), or could admit themselves Waterhouse & Platt 1990 A
to hospital (Cotgrove 1995). In only one other group (ANTIDE- Linehan 1991 A
PRESSANT MEDICATION VS. PLACEBO) was there more Allard et al 1992 A
than one trial. The remainder of the studies are reported singly. Morgan 1993 A
McLeavey 1994 C
Cotgrove 1995 C
van Heeringen et al 1995 A
RESULTS van der Sande et al 1997 A
Evans et al 1999 A
Harrington et al 1998 A
Verkes 1998 A
Description of studies
See: Characteristics of included studies; Characteristics of excluded
studies. Effects of interventions
Twenty three studies were identified through the combined search
strategies as eligible for inclusion in the study. All reports were pub- PROBLEM SOLVING THERAPY VS. STANDARD AFTER-
lished. However, two studies (Hirsch 1982; Montgomery 1979) CARE: All five studies (Gibbons 1978; Hawton 1987; Salkovskis
have not been published in full. A detailed report of one of the 1990; McLeavey 1994; Evans 1999) reported reduced repetition
studies was obtained from conference proceedings (Montgomery of DSH in patients in the experimental groups. However, the sum-
1979) and the full report of the other trial was provided by the au- mary odds ratio of 0.70 (95% confidence interval 0.45 to 1.11) was
thor in the form of an unpublished manuscript (Hirsch 1982). Six not statistically significant. Excluding the one trial which did not
studies were reported in more than one publication (Montgomery have the highest quality of concealment of allocation (McLeavey
1979; Hirsch 1982; Torhorst 1987; Linehan 1991; Harrington 1994; which also compared two types of problem-solving therapy,
1998; Evans 1999). Some of these reports included outcome data the control type being standard care at the time of the study) made
for different variables than those described in the initial reports. little difference to the summary odds ratio (0.74; 0.46 to 1.20). We
One further randomised controlled trial of DSH patients was iden- examined separately the two trials in this category which included
tified but this did not include repetition of deliberate self-harm as only repeaters (Salkovskis 1990; Evans 1999). The summary odds
an outcome variable (Patsiokas 1995). We tried, unsuccessfully, to ratio for this analysis was 0.43 (0.13 to 1.39). The summary odds
obtain this information from the authors of this trial. ratio for the trials which included both repeaters and participants
The included trials comprised 3014 randomised participants of without previous DSH was 0.77 (0.47 to 1.27).
whom 2832 had outcome data regarding repetition of DSH. INTENSIVE INTERVENTION PLUS OUTREACH VS.
STANDARD AFTERCARE: There was no consistent direction
of effect amongst studies in this group (Chowdhury 1973; Welu
1977; Hawton 1981; Allard 1992; van Heeringen 1995; Van der
Risk of bias in included studies Sande 1997); the summary odds ratio for this comparison was

0.84 (0.62 to 1.15). Inclusion of only the trials with the highest 0.83 (0.47 to 1.48). For one of the studies (Draper 1982; Hirsch
quality of concealment of allocation did not markedly alter the 1982), the results for the patients given mianserin were combined
summary odds ratio (0.86; 0.60 to 1.23). with the results for those who had received nomifensine to form
EMERGENCY CARD VS. STANDARD AFTERCARE: In both the experimental group of ’patients receiving anti-depressant med-
studies in this comparison (Morgan 1993; Cotgrove 1995) there ication’. (It should be noted that nomifensine is no longer avail-
was a trend towards less repetition of self-harm in the experimental able). In the study of paroxetine vs. placebo (Verkes 1998), the
group but the summary odds ratio of 0.45 (0.19 to 1.07) was not authors conducted subgroup analyses of patients with less than
significant. The odds ratio was similar when only the results of the five previous acts of DSH (’minor repeaters’) and those with five or
trial with the highest quality rating (Morgan 1993) were analysed more (’major repeaters’). On the basis of Kaplan-Meier Curves for
(0.43; 0.15 to 1.27). probability of another suicide attempt the authors reported a sig-
DIALECTICAL BEHAVIOUR THERAPY VS. STANDARD nificant reduction in repetition in patients in the minor repeaters
AFTERCARE: In this study (Linehan 1991) there was a signifi- group who received paroxetine compared with those who received
cantly lower rate of repetition of self-harm during the follow-up placebo, but not in the major repeaters group.
period in patients who received dialectial behaviour therapy (0.24; LONG-TERM THERAPY VS. SHORT-TERM THERAPY:
0.06 to 0.93). This comparison was restricted to a sub-group of There was no indication from this study (Torhorst 1988) that,
randomly assigned patients which was smaller than that which for patients with a prior history of self-harm, long-term therapy
entered the original trial (Linehan 1991). was more effective in terms of preventing repetition of DSH than
INPATIENT BEHAVIOUR THERAPY VS. INPATIENT IN- short-term therapy; odds ratio 1.0 (0.35 to 2.86).
SIGHT-ORIENTATED THERAPY: The small sample size of HOMEBASED FAMILY THERAPY VS. STANDARD AFTER-
the single study in this comparison (Liberman 1981) precludes CARE: The odds ratio from the one study in this category (
meaningful conclusions from the odds ratio analysis (0.60; 0.08 Harrington 1998) did not demonstrate a beneficial effect of fam-
to 4.45). ily therapy carried out in the patient’s home (1.02; 0.41, 2.51).
SAME THERAPIST VS. DIFFERENT THERAPIST: The repe- The authors of this trial reported a subgroup analysis of patients
tition rate in the group of patients who saw the same person for af- who were not depressed at entry to the trial: fewer of those who
tercare who assessed them in hospital following their initial episode received home-based family therapy reported suicidal ideation at
of DSH was significantly higher than that of patients who had both two months and six months follow-up compared to those
a change of clinician; odds ratio 3.70 (1.13 to 12.09). However, who received standard aftercare.
the authors (Torhorst 1987) reported that in spite of randomiza-
tion there were several imbalances between the experimental and
control groups resulting in a greater prevalence of risk factors for
repetition in the experimental group. It is of note that continuity DISCUSSION
of therapist resulted in 48/68 (71%) of patients attending at least
The results of this updated systematic review indicate that there
one outpatient treatment session compared with 34/73 (47%) of
continues to be insufficient evidence on which to make firm rec-
patients in the control group; odds ratio 2.75 (1.37 to 5.52).
ommendations about the most effective forms of treatment for
GENERAL HOSPITAL ADMISSION VS. DISCHARGE: The
patients who have recently engaged in DSH. This is a serious situ-
odds ratio from the one study in this category (Waterhouse 1990)
ation given the size of the problem of DSH throughout the world
did not indicate a beneficial effect of general hospital admission
(Morgan 1993; Davis 1991; Schmidtke 1996; Hawton 1997b),
following DSH (0.75; 0.16 to 3.60). It is important to note, how-
and its importance for suicide prevention (Gunnell 1994).
ever, that only 15% of patients referred were eligible for inclu-
sion in the study as only those attempters at low risk and without The main problem with nearly all trials in this review is that they
immediate medical or psychiatric needs could be considered for included far too few subjects to have the statistical power to detect
discharge without treatment. The follow-up period was relatively clinically meaningful differences in rates of repetition of DSH
short. between experimental and control treatments, if such differences
FLUPENTHIXOL VS. PLACEBO: In this single trial ( existed. Rarely was there evidence that prior power analyses had
Montgomery 1979) there was a significant reduction in repetition been conducted.
of DSH in patients receiving flupenthixol (0.09; 0.02 to 0.50).
In nearly all trials the subjects were recruited after general hospital
The trial was relatively small and all the patients were repeaters.
attendance because of DSH. Some trials just included self-poison-
ANTIDEPRESSANTS VS. PLACEBO: The pooled odds ratio
ing patients, who constitute the large majority of DSH patients
for the three studies in this category (Draper 1982; Hirsch 1982;
(Hawton 1997b), others both self-poisoning and self injury pa-
Montgomery 1983; Verkes 1998) indicates no apparent benefit
tients, whilst some did not specify the method of self-harm. Most
regarding repetition of DSH for patients treated with mianserin,
of the studies focused on patients who could be treated as outpa-
nomifensine or paroxetine compared with placebo medication;
tients. Patients who, for example, required psychiatric hospital in-

patient care because of severe mental illness and, or serious suicide were also trends favouring provision of an emergency access card
risk, were excluded, but these comprise the minority of DSH pa- in addition to standard aftercare but again a larger trial is required,
tients presenting to general hospitals (Hawton 1997b). The studies including specific attention to what role the card might play since
examined are, therefore, of relevance to a large proportion of DSH only a small minority of patients actually used the facility provided
patients who will be treated in the community. Most patients in the by possession of the card (Morgan 1993; Cotgrove 1995).
studies had a history of previous episodes of self-harm, and in nine
trials the whole sample consisted of repeaters (Chowdhury 1973; For female patients with borderline personality disorder who have
Montgomery 1979; Liberman 1981; Montgomery 1983; Torhorst a history of multiple episodes of DSH there have been promis-
1988; Salkovskis 1990; Linehan 1991; Verkes 1998; Evans 1999). ing results from a single study of dialectical behaviour therapy (
Only one study included only patients with no previous history Linehan 1991), which is similar to cognitive behavioural therapy
of self-harm (Morgan 1993). (Linehan 1993). A larger replication study is required because of
the small size of this study and the fact that follow-up informa-
The classification of the trials into groups presented some difficul- tion on repeated self-harm was only obtained in a subgroup of
ties, particularly the group of trials of intensive intervention plus patients. The intervention is very intensive, consisting of weekly
outreach. This group contained the most heterogeneous range of group and individual therapy for one year plus 24-hour access to
trials. the therapist. Development and evaluation of a shorter form of
In view of the considerable problem of DSH in adolescents in this treatment more suited to general psychiatric service provision
many countries (Hawton 1992; Schmidtke 1996) it is surprising and investigation of its efficacy in male patients are needed.
that only two trials focused specifically on this specific clinical The positive result of depot neuroleptic medication in a single
population (Cotgrove 1995; Harrington 1998). The trial of home- small study of patients with repeated self-harm (Montgomery
based family therapy versus standard aftercare for adolescents by 1979) suggests that this treatment should be subjected to further
Harrington et al (Harrington 1998) is interesting in that non- evaluation in a larger study, although reluctance of patients to ac-
depressed patients who were allocated to receive family therapy cept depot medication, side effects and other practical and ethical
showed a greater reduction in suicidal ideation at follow-up than implications, may limit its applicability.
those who were allocated to standard aftercare. This interesting
finding, which was based on post-hoc subgroup analyses requires There was little indication that intensive intervention plus out-
verification. reach was effective. However, in one relatively large study in this
group, which evaluated community follow-up of patients who did
The comparison intervention for most of the studies of psychoso- not attend outpatient appointments (van Heeringen 1995), there
cial intervention was standard care. This will probably vary from was a statistically significant increase in outpatient attendance from
centre to centre and details of this care were usually not provided. 42.5% before the home visit (39.8 % in the control group) to
Variability in standard aftercare of DSH patients in different coun- 51.2% after the visit (odds ratio 1.58; 1.15 to 2.33), and a near
tries and regions may influence the relative effectiveness of experi- significant difference in repetition of DSH of 10.7% compared
mental interventions in particular settings. Future studies in which with 17.4% (0.57; 0.32 to 1.02). Home treatment was also found
standard care is included should define precisely the nature of the to substantially increase the rate of take-up of treatment in another
treatment patients receive. trial in this review (Hawton 1981). Assertive outreach for poorly
The dependent variable studied so far in this review, namely rep- compliant patients may, therefore, be a necessary component in
etition of self-harm, was not consistently defined and measured maximising the delivery of any treatment which is shown to be
in a standard way across all studies. In most studies repetition effective.
was based on hospital referral for further DSH, whereas in some
There was no evidence that antidepressants were effective in pre-
studies interviews with patients and other informants also identi-
venting repetition of self-harm in DSH patients in general. How-
fied episodes of self-harm which did not result in hospital refer-
ever, it must be noted that one of the drugs investigated is no
ral. Furthermore, it is possible that different treatment conditions
longer available (nomifensine) and the other (mianserin) is now
may be associated with differences in the extent to which subjects
little used. This review does not give any indication of whether
who repeat actually present to hospital, because of the effects of
other antidepressants could be of benefit in preventing further
treatment on willingness to seek hospital help. Such an effect, if
episodes of self-harm. However, it is interesting that in a recent
marked, could seriously affect the apparent result of a trial. This is
trial of paroxetine (Verkes 1998), subgroup analyses indicated a
an important potential bias in the trials which use routine service
marked reduction in patients who received the active drug and
data (as opposed to interviews) to ascertain repetition.
who had a history of one to four acts of DSH compared to similar
Promising results were found for problem solving therapy, which patients who received placebo, but there was no indication of a
is a brief and reasonably easily taught form of treatment (Hawton similar benefit in those who had a history of five or more previous
1989a). A larger trial of this treatment approach is required. There acts of DSH. Verification of this result is required as it was based

on post-hoc subgroup analyses. This finding raises the question of clinically meaningful differences in rates of repetition of DSH
whether among patients with a history of DSH the type of treat- between experimental and control treatments, if such differences
ment that may be effective will vary according to the number of existed. The number needed is a function of both the expected
previous episodes. rate of repetition (i.e. that in the control group) and the size of
the difference. If the predicted rate were 10% in the experimental
Repetition of DSH has been the sole outcome variable investigated
group versus 15% in the control, with, a total of 687 subjects
in this review. We intended to investigate whether there is evidence
would be required in each treatment group (80% power and a
of benefits with regard to other outcomes (e.g. depression, problem
5% significance level), while if the rates were 20% and 30%, 293
resolution). However, the data for these factors, where reported,
subjects would be required in each group (Pocock, 1983). Even
were often inadequate for meta-analysis and we have been unable
when the results from similar trials are synthesized using meta-
to obtain further information from authors. The groups of patients
analytical techniques there are insufficient numbers of patients to
were often heterogeneous in term of gender, age and presenting
detect such differences. The only statistically significant findings
problems. Further work in this area should examine the efficacy
have come from smaller studies, which may reflect publication
of interventions according to such factors.
bias (Light and Pillemer, 1984).
We are keen to be informed of any current or planned trials in the
treatment of DSH patients.
ACKNOWLEDGEMENTS
AUTHORS’ CONCLUSIONS Sandy Bremner, Eleanor Feldman, Robert Goldney, David Owens,
Isaac Sakinofsky, and Lil Traskman-Bendz contributed to the first
Implications for practice stage of this review.
At present, evidence is lacking to indicate the most effective forms This study was carried out under the auspices of the Depression,
of treatment for DSH patients. This is a serious situation given Anxiety and Neurosis Review Group of the Cochrane Collabo-
the size of the DSH population and the risks of subsequent self- ration and was funded by the NHS Executive Anglia and Ox-
harm, including suicide. ford Research and Development Committee. We thank Rachel
Churchill and Drs. Douglas Altman, Henry McQuay, John Ged-
Implications for research des and Clive Adams for advice. We also thank the several authors
who supplied us with unpublished information regarding their
As noted in the discussion, there is a need for large trials of the
trials.
interventions shown in small trials to be of possible benefit. The
main problem with nearly all trials in this review is that they An earlier version of this review was published in the British Med-
included far too few subjects to have the statistical power to detect ical Journal (15 Aug. 1998, No. 7156 pp. 441-6).
REFERENCES
References to studies included in this review Draper 1982 {unpublished data only}
Draper R, Hirsch S. Treatment of parasuicide patients with
mianserin, nomifensine and placebo, a double blind placebo
Allard 1992 {published data only} controlled trial. Unpublished 1982.
Allard R, Marshall M, Plante MC. Intensive follow-up does not
Evans 1999 {published data only}
decrease the risk of repeat suicide attempts. Suicide & Life
Evans K, Tyrer P, Catalan J, Schmidt U, Davidson K, Dent J, et
Threatening Behaviour 1992;22(3):303–14.
al.Manual-assisted cognitive behaviour therapy (MACT): a
randomized controlled trial of a brief intervention with a
Chowdhury 1973 {published data only}
bibliotherapy in the treatment of recurrent deliberate self-harm.
Chowdhury N, Hicks RC, Kreitman N. Evaluation of an after-care
Psychological Medicine 1999;29(1):19–25.
service for parasuicide (attempted suicide) patients. Social
MacLeod AK, Tata P, Evans K, Tyrer P, Schmidt U, Davidson K, et
Psychiatry 1973;8(2):67–81.
al.Recovery of positive future thinking within a high-risk
parasuicide group: Results from a pilot randomized controlled trial.
Cotgrove 1995 {published data only} British Journal of Clinical Psychology 1998;37(4):371–9.
Cotgrove AJ, Zirinsky L, Black D, Weston D. Secondary
prevention of attempted suicide in adolescence. Journal of Gibbons 1978 {published data only}
Adolescence 1995;18(5):569–77. Gibbons JS, Butler J, Urwin P, et al.Evaluation of a social work
service for self-poisoning patients. British Journal of Psychiatry Montgomery 1982 {published data only}
1978;133:111–8. Montgomery SA. The psychopharmacology of borderline
personality disorder. Acta Psychiatrica Belgique 1987;87:260–6.
Harrington 1998 {published data only}
Montgomery SA, Montgomery D. Pharmacological prevention of
Byford S, Harrington R, Torgerson D, Kerfoot M, Dyer E,
suicidal behaviour. Journal of Affective Disorders 1982;4:291–8.
Harrington V, et al.Cost effectiveness analysis of a home-based
social work intervention for children and adolescents who have Montgomery 1983 {published data only}
deliberately poisoned themselves: Results of a randomized trial. Montgomery SA, Roy D, Montgomery DB. The prevention of
British Journal of Psychiatry 1999;174:56–62. recurrent suicidal acts. British Journal of Clinical Pharmacology
Harrington R, Kerfoot M, Dyer E, McNiven F, Gill J, Harrington 1983;15:183–8.
V, et al.Randomized trial of a home-based family intervention for
Morgan 1993 {published data only}
children who have deliberately poisoned themselves. Journal of the
Morgan HG, Jones EM, Owen JH. Secondary prevention of non-
American Academy of Child & Adolescent Psychiatry 1998;37:512–8.
fatal deliberate self-harm. The green card study. British Journal of
Hawton 1981 {published data only} Psychiatry 1993;163:111–2.
Hawton K, Bancroft J, Catalan J, et al.Domiciliary and out-patient
Salkovskis 1990 {published data only}
treatment of self-poisoning patients by medical and non-medical
Salkovskis P, Atha C, Storer D. Cognitive-behavioural problem
staff. Psychological Medicine 1981;11:169–77.
solving in the treatment of patients who repeatedly attempt suicide.
Hawton 1987 {published data only} A controlled trial. British Journal of Psychiatry 1990;157:871–6.
Hawton K, McKeown S, Day A, et al.Evaluation of out-patient Torhorst 1987 {published data only}
counselling compared with general practitioner care following Moller HJ. Attempted suicide: efficacy of different aftercare
overdoses. Psychological Medicine 1987;17:751–61. strategies. International Clinical Psychopharmacology 1992;6:1–12.
Hirsch 1982 {published and unpublished data} Moller HJ. Efficacy of different strategies of aftercare for patients
Draper R, Hirsch S. Treatment of parasuicide patients with who have attempted suicide. Journal of the Royal Society of Medicine
mianserin, nomifensine and placebo, a double blind placebo 1989;82:643–7..
controlled trial. Unpublished 1982. Torhorst A, Moller HJ, Buerk F, Waechtler C, Lauter H.
Hirsch SR, Walsh C, Draper R. Parasuicide: A review of treatment Outpatient aftercare following attempted suicide. Initial results of
interventions. Journal of Affective Disorders 1982;4:299–311. an experimental study. Suizidprophylaxe 1984;4:254–73.
Torhorst A, Moller HJ, Burk F, et al.The psychiatric management
Liberman 1981 {published data only} of parasuicide patients: A controlled clinical study comparing
Liberman RP, Eckman T. Behavior therapy vs insight-oriented different strategies of outpatient treatment. Crisis 1987;8:53–61.
therapy for repeated suicide attempters. Archives of General
Psychiatry 1981;38:1126–30. Torhorst 1988 {published data only}
Torhorst A, Moller HJ, Schmid-Bode KW, et al.Comparing a 3-
Linehan 1991 {published and unpublished data} month and a 12-month-outpatient aftercare program for
Linehan MM, Armstrong HE, Suarez A, et al.Cognitive-behavioral parasuicide repeaters. In: Moller HJ, Schmidtke A, Welz R editor
treatment of chronically parasuicidal borderline patients. Archives of (s). Current Issues of Suicidology. Berlin: Springer-Verlag, 1988:
General Psychiatry 1991;48:1060–4. 419–24.
Linehan MM, Heard HL, Armstrong HE. Naturalistic follow-up of
Van der Sande 1997 {published data only}
a behavioral treatment for chronically parasuicidal borderline
van der Sande R, van Rooijen E, Buskens E, et al.Intensive in-
patients. Archives of General Psychiatry 1993;50:971–4.
patient and community intervention versus routine care after
Linehan MM, Tutek DA, Heard HL, Armstrong HE. Interpersonal
attempted suicide: A randomised controlled intervention. British
outcome of cognitive behavioural treatment for chronically suicidal
Journal of Psychiatry 1997;171:35–41.
borderline patients. American Journal of Psychiatry 1994;151:
1771–6. van Heeringen 1995 {published data only}
Shearin EN, Linehan MM. Dialectical behaviour therapy for van Heeringen C, Jannes S, Buylaert W, et al.The management of
borderline personality disorder: theoretical and empirical non-compliance with referral to out-patient after-care among
foundations. Acta Psychiatrica Scandinavica 1994;89:61–8. attempted suicide patients: a controlled intervention study.
Psychological Medicine 1995;25:963–70.
McLeavey 1994 {published data only}
McLeavey BC, Daly RJ, Ludgate JW, et al.Interpersonal problem- Verkes 1998 {published data only}
solving skills training in the treatment of self-poisoning patients. Verkes RJ, Van der Mast RC, Hengeveld MW, Tuyl JP,
Suicide & Life Threatening Behavior 1994;24:382–94. Zwinderman AH, Van Kempen GM. Reduction by paroxetine of
suicidal behavior in patients with repeated suicide attempts but not
Montgomery 1979 {published and unpublished data}
major depresssion. American Journal of Psychiatry 1998;55:543–47.
Montgomery SA, Montgomery DB, Jayanthi-Rani S, et
al.Maintenance therapy in repeat suicidal behaviour: A placebo Waterhouse 1990 {published data only}
controlled trial. Proceedings of the 10th International Congress for Waterhouse J, Platt S. General hospital admission in the
Suicide Prevention & Crisis Intervention. Ottawa, Canada, 1979: management of parasuicide: A randomised controlled trial. British
227–9. Journal of Psychiatry 1990;156:236–42.
Welu 1977 {published data only} Additional references
Welu TC. A follow-up program for suicide attempters: Evaluation
of effectiveness. Suicide & Life Threatening Behavior 1977;7:17–30. Buglass 1974
Buglass D, Horton J. A scale for predicting subsequent suicidal
References to studies excluded from this review behaviour. British Journal of Psychiatry 1974;124:573–8.
D’Zurilla 1971
Bateson 1989 {published data only}
D’Zurilla TJ, Godfried MR. Problem solving and behavior
Bateson M, Oliver JP, Goldberg DP. A comparative study of the
modification. Journal of Abnormal Psychology 1971;78:107–26.
management of cases of deliberate self-harm in a district general
hospital. British Journal of Social Work 1989;19:461–77. Davis 1991
Davis AT, Kosky RJ. Attempted suicide in Adelaide and Perth:
Catalan 1980 {published data only}
Changing rates for males and females, 1971-1987. Medical Journal
Catalan J, Marsack P, Hawton K, Whitwell D, Fagg J, Bancroft JH.
of Australia 1991;154:666–70.
Comparison of doctors and nurses in the assessment of deliberate
self-poisoning patients. Psychological Medicine 1980;10:483–91. Dew 1987
Dew MA, Bromet JE, Brent D, et al.A quantitative literature review
Crawford 1998 {published data only}
of the effectiveness of suicide prevention centers. Journal of
Crawford MJ, Turnbull G, Wessely S. Deliberate self-harm
Consulting and Clinical Psychology 1987;55:229–44.
assessment by accident and emergency staff - an intervention study.
Journal of Accident and Emergency Medicine 1998;15:18–22. DOH 1992
Department of Health. The Health of the Nation: A Strategy for
Deykin 1986 {published data only}
Health in England. London: HMSO, 1992.
Deykin EY, Chung-Chen Hsieh PH, Joshi N, et al.Adolescent
suicidal and self-destructive behavior: Results of an intervention DOH 1998
study. Journal of Adolescent Health Care 1986;7:88–95. Department of Health. Our healthier nation: A contract for health.
London: HMSO, 1998.
Donaldson 1997 {published data only}
Donaldson D, Spirito A, Arrigan M, et al.Structured disposition Gunnell 1994
planning for adolescent suicide attempters in a general hospital: Gunnell D, Frankel S. Prevention of suicide: aspirations and
Preliminary findings on short-term outcome. Archives of Suicide evidence. BMJ 1994;308:1227–33.
Research 1997;3:271–82. Hawton 1982
Gardner 1977 {published data only} Hawton K, Catalan J. Attempted suicide: A practical guide to its
Gardner R, Hanka R, Evison B, Mountford PM, O’Brien VC, nature and management. Oxford: Oxford University Press, 1982.
Roberts SJ. Consultant liaison scheme for self-poisoned patients in Hawton 1988
a general hospital. BMJ 1978;2:1392–94. Hawton K, Fagg J. Suicide, and other causes of death, following
Gardner R, Hanka R, O’Brien VC, Page AG, Rees R. Psychological attempted sucide. British Journal of Psychiatry 1988;152:359–66.
and social evaluation in cases of deliberate self-poisoning admitted Hawton 1989a
to a general hospital. BMJ 1977;2:1567–70. Hawton K, Kirk J. Problem-solving. In: Hawton K, Salkovskis P,
Patsiokas 1995 {published data only} Kirk J, Clark DM editor(s). Cognitive behaviour therapy for
Patsiokas AT, Clum GA. Effects of psychotherapeutic strategies in psychiatric problems: A practical guide. Oxford: Oxford University
the treatment of suicide attempters. Psychotherapy 1985;22:281–90. Press, 1989:406–26.
Rotherham-Borus 1996 {published data only} Hawton 1989b
Rotherham-Borus MJ, Piacentini J, Van Rossem R, et al.Enhancing Hawton K. Controlled studies of psychosocial interventions
treatment adherence with a specialized emergency room program following attempted suicide. In: Kreitman N, Platt SD editor(s).
for adolescent suicide attempters. Journal of the American Acadamey Current Research on Suicide & Parasuicide. 2nd Edition. Edinburgh:
of Child & Adolescent Psychiatry 1996;35:654–63. Edinburgh University Press, 1989.
Termansen 1975 {published data only} Hawton 1992

Termansen P E, Bywater C. S. A. F. E. R.: A follow-up service for Hawton K, Fagg J. Deliberate self-poisoning and self-injury in
attempted suicide in Vancouver. Canadian Psychiatric Association adolescents: a study of characteristics and trends in Oxford. British
Journal 1975;20:29–34. Journal of Psychiatry 1992;161:816–23.
Wullimier 1979 {published data only} Hawton 1997a

Wullimier F, Bovet J, Meylan D. Comparative study of two Hawton K, Fagg J, Simkin S, et al.Trends in deliberate self-harm in
intervention modes on suicide attempters hospitalized in the Oxford, 1985-1995: Implications for clinical services in the
general hospital. Proceedings of the 9th International Conference prevention of suicide. British Journal of Psychiatry 1997;171:
on Suicide Prevention. 1977. 556–60.
Wullimier F, Bovet J, Meylan D. The future of suicidal patients Hawton 1997b
admitted to a general hospital. Comparative study of two methods Hawton K. Attempted suicide. In: Clarke DM, Fairburn CG
of preventing recurrence and suicides. (Title translated). Sozial-und editor(s). Science and practice of cognitive behaviour therapy. Oxford:
Praventivmedizin. 1979; Vol. 24:73–88. Oxford University Press, 1997:285–312.
Kerfoot 1986. suicide attempters during the periods 1989-1992. Results of the
Kerfoot M. Family therapy and psychotherapy following suicidal WHO/EURO Multicentre Study on Parasuicide. Acta Pschiatrica
behaviour by young adolescents. In: Diekstra R, Hawton K editor Scandinavica 1996;93:327–38.
(s). Suicide in Adolescence.. Dordrecht: Martinus Nijhoff, 1986: Schulz 1995
95–111. Schulz KF, Chalmers I, Hayes RJ.et al. Empirical evidence of bias:
Kerfoot 1995 Dimensions of methodological quality associated with estimates of
Kerfoot M, Harrington RC, Dyer E. Brief home-based intervention treatment effects in controlled trials. JAMA 1995;273:408–12.
with young suicide attempters and their families. Journal of
UN 1996
Adolescence 1995;18:557–68.
United Nations. Prevention of suicide guidelines for the formulation
Kerkhof 1994 and implementation of national strategies. New York, NY: United
Kerkhof A, Schmidtke A, Bille-Brahe U, De Leo D, Lonnqvist J. Nations, 1996.
Attempted suicide in Europe. Findings from the multicentre study on
van der Sande 1997
parasuicide by the WHO regional office for Europe. Leiden: DSWO
van der Sande R, Buskens E, Allart E, et al.Psychosocial intervention
Press, 1994.
following suicide attempt: a systematic review of treatment
Light 1984 interventions. Acta Pschiatrica Scandinavica 1997;96:43–50.
Light RJ, Pillemer DB. Summing up: The science of reviewing
Wexler 1978
research. Cambridge, MA: Harvard University Press, 1984.
Wexler L, Weissman MM, Kasl. SV. Suicide attempts 1970-75:
Linehan 1993 updating a United States study and comparisons with international
Linehan M. Cognitive behavioral treatment of borderline personality trends. British Journal of Psychiatry 1978;132:180–5.
disorder. New York, NY: Guilford, 1993.
WHO 1982
Ovenstone 1974 World Health Organization. Changing patterns in suicidal
Ovenstone IM, Kreitman N. Two syndromes of suicide. British behaviour. European reports and studies. Copenhagen: Regional
Journal of Psychiatry 1974;124:336–45. Office for Europe, 1982.
Pocock 1983
WHO 1986
Pocock SJ. Clinical trials: A practical approach. Chichester: John
World Health Organization. Summary report. Working Group on
Wiley & Sons, 1983.
Preventive Practices in Suicide and Attempted Suicide. York, 22-26
RCP 1994 Sept. Vol. Copenhagen, Copenhagen: Regional Office for Europe,
Royal College of Psychiatrists. Consensus guidelines on the general 1986.
hospital management of deliberate self-harm patients. London: Royal
College of Psychiatrists, 1994. References to other published versions of this review
Rygnestad 1997
Hawton 1998a
Rygnestad T. Mortality after deliberate self-poisoning: A
Hawton K, Arensman E, Townsend E, Bremner S, Feldman E,
prospective follow-up study of 587 persons observed for 5279
Goldney R, et al.Deliberate self-harm: the efficacy of psychosocial
person years: risk factors and causes of death. Social Psychiatry &
and pharmacological treatments. The Cochrane Library 1998, Issue
Psychiatric Epidemiology 1997;32:443–50.
Issue 3.
Sackett 1997 Hawton 1998b
Sackett DL. The Cochrane Collaboration Handbook. Oxford, UK: Hawton K, Arensman E, Townsend E, Bremner S, Feldman E,
Update Software, 1997. Goldney R, et al.Deliberate self harm: systematic review of the
Schmidtke 1996 efficacy of psychosocial and pharmacological treatments in
Schmidtke A, Bille-Brahe U, Deleo D, et al.Attempted suicide in preventing repetition. BMJ 1998;317:441–7.
Europe: rates, trends, and sociodemograhpic characteristics of ∗
Indicates the major publication for the study

CHARACTERISTICS OF STUDIES
Characteristics of included studies [ordered by study ID]
Allard 1992
Methods Allocation: Subjects were randomly assigned using sealed and numbered envelopes.
Follow up period: 12 months.
N lost to follow up: 24/150 (16%) for repetition data.
Participants Setting: Montreal, Canada.

Inclusion criteria: i) resident in catchment area of hospital; ii) able to speak French or English; iii) no
physical handicap preventing attendance; iv) not already in institutional care; v) capacity to give informed
consent; vi) not sociopathic; vii) attempt more than a week ago.
Numbers: 150 - 76 experimental, 74 control.
Profile: 50% were repeaters. 55% female. 53% - substance abuse diagnosis.
87% - depression diagnosis. 45% - personality disorder diagnosis.
Source of participants: patients presenting to hospital for a suicide attempt.
Interventions Experimental: Intensive intervention: A schedule of visits was arranged including at least one home visit.
Therapy provided where needed. Reminders (telephone or written) and home visits were made in case of
missed appointments).
Control: treatment by another staff team in the same hospital.
Therapist: home visit by made social worker.
Type of therapy offered: various - e.g. psychoanalytic psychotherapy, psychosocial, drug or behavioural
therapy.
Length of treatment: 12 months
Outcomes Included: i) repetition of self-harm.

Excluded: i) compliance
Notes Repetition data from: hospital records, coroners office records plus interview with patients and interview
with other informants e.g. relatives or friends of patient.
Fatal attempts: three suicides in the experimental group, one suicide in the control group.
Risk of bias
Item Authors’ judgement Description
Allocation concealment? Yes A - Adequate
Chowdhury 1973
Methods Allocation: Patients ‘allocated alternately’ to treatment groups. Follow up period: 6 months.
N lost to follow up: none for repetition data.
Participants Setting: Edinburgh, UK.

Inclusion criteria: i) previous episode of deliberate self-harm (DSH); ii) patients with psychiatric diagnoses,

Chowdhury 1973 (Continued)
alcohol and drug addiction were included.

Profile: All were repeaters (had previously self-harmed). Mean age not calculable. 57% female. 53%
alcohol dependency. 76% personality disorder. 14% depression in normal personality. 35% depression in
personality disorder.
Source of participants: patients admitted to a general hospital for deliberate self-harm.
Interventions Experimental: Special aftercare - regular out-patient appointments. Patients could also be seen without
appointment. Patients who missed appointments were visited at home. Emergency telephone access 24
hours (evenings via Samaritans).
Control: Normal aftercare - out-patient appointment to see psychiatrist or social worker. Non-attenders
were not pursued.
Therapist: psychiatrist, psychiatric social worker plus two part time social workers. Type of therapy offered:
not described.

Excluded: i) improvement in problems for those that had housing, financial or employment problems; ii)
improvement in psychiatric state - overt symptoms.
Notes Repetition data from: hospital records.

Fatal attempts: no suicides mentioned.
Repetition data by gender is provided.
Risk of bias
Allocation concealment? No C - Inadequate
Cotgrove 1995
Methods Allocation: Random allocation using open number table.

Participants Setting: Chester, UK.

Inclusion criteria: i) aged 16 years or under.
Profile: Patients aged 12.2 -16.7 years (mean 14.9). % repeaters not given. 85% female. 6% had major
psychiatric disturbance (not specified).
Source of participants: patients admitted to hospital following deliberate self-harm
Interventions Experimental: Standard care plus green card (emergency card): Green card acted as a passport to re-
admission into a paediatric ward in the local hospital.
Control: Standard follow-up or treatment from a clinic or child psychiatry department.
Therapist: not described.
Type of therapy offered: Standard follow-up - not described.

Cotgrove 1995 (Continued)
Length of treatment: one year.

Excluded: i) use of emergency card.
Notes Repetition data from: clinical and hospital notes.

Risk of bias
Draper 1982
Methods See:Hirsch 1982
Participants
Interventions
Outcomes
Notes
Risk of bias
Evans 1999
Methods Allocation: opaque sealed envelopes opened sequentially.

Follow-up period: 6 months for repetition.
N lost to follow-up: 2 in control group for repetition data
Participants Setting: London, UK.

Inclusion criteria: i) personality disturbance (antisocial, disocial, impulsive or borderline; ii) at least one
episode of deliberate self-harm in 12 months preceding entry to trial; iii) not diagnosed with alcohol or
drug dependence, schizophrenia, or organic psychiatric disorder.
Numbers: 34 - 18 experimental, 16 control
Profile: Age range 16-50 years. Mean age not given.
100% were repeaters. 62% female.
100% had a personality disorder.
Source of participants: patients seen after an episode of self-harm in two hospitals in the London area (

Evans 1999 (Continued)
Paddington and Chelsea, Westminster).
Interventions Experimental: Manual Assisted Cognitive Behavioural Therapy (MACT). Six chapters based on therapy
for individuals with personality disorders developed by Davidson and Tyrer (1996) and Linehan et al
(1991). 5 patients did not see a therapist and received all input from the booklets. 1 patient did not have
any intervention.
Control: Standard psychiatric treatment (various - 5 patients had contact with a psychiatrist, 3 saw a
community mental health team, four saw a specialist social worker, 2 saw no mental health professional.
Therapist: 1 psychiatrist, 2 nurses, 2 social workers.
Type of therapy offered: Cognitive behavioural therapy for personality disordered patients. Involving
basic cognitive techniques, problem solving, techniques for managing emotions and thoughts, relapse
prevention plans.
Length of treatment: 2-6 sessions.
Outcomes Included: repetition of self-harm.

Excluded: i) depression and anxiety ii) time until repetition iii) cost of care iv) social functioning
Notes Repetition data from: interview and hospital records.

Risk of bias
Gibbons 1978
Methods Allocation: Patients were ‘randomly assigned’ using sequentially numbered, sealed, opaque envelopes.
Participants Setting: Southampton, UK.

Inclusion criteria: i) over 17 years old; ii) no immediate suicide risk; iii) no formal psychiatric illness.
Profile: Self-poisoning patients. Repeaters and first timers. Mean age not available. 43.6% depressive
neurosis. 2.2% phobic neurosis.
2.4% affective psychosis. 1.1% schizophrenia. 71% females.
Source of participants: patients who presented to an A & E Dept. after deliberate self-poisoning.
Interventions Experimental: Crisis orientated, time limited task-centered social work at home. Problem solving inter-
vention for personal relationships, emotional distress, practical problems etc.
Control:Routine service: 54% were referred to their GP, 33% received a psychiatric referral and 13%
received other kinds of referral - not specified.
Therapist: 2 social workers provided the service. 2 research psychiatrists did assessments.
Type of therapy offered: Experimental: task centered case-work; Control: not specified.

Gibbons 1978 (Continued)
Outcomes Included: i) repetition of self-poisoning.

Excluded: i) depression - no standard deviations, authors contacted who could not provide them; ii) use
of psychiatric and social services; iii) change in social problems; iv) satisfaction with service.

Fatal attempts:no suicides mentioned.
Risk of bias
Harrington 1998
Methods Allocation: series of opaque sealed envelopes which contained either a blank sheet or one bearing the letter
F (for family therapy) were prepared by a researcher. These were opened by a social worker when patients
were assessed.
Follow-up period: 6 months.
N lost to follow-up: 13 for repetition data.
Participants Setting: Manchester, UK.

Inclusion criteria: i) children aged 16 years or younger; ii) living in a family; iii) not in social service care;
iv) no current investigation of physical or sexual abuse; v) not currently in inpatient treatment; vi) not
learning disabled; vii) not seriously suicidal; viii) had not ’self-harmed’ (e.g. cutting or hanging).
Profile: Age range 10-16 years. Mean age 14.5 years.
% who were repeaters not given. 89.5 % female
64.5% major depression, 10.5% conduct disorder.
Source of participants: patients referred to mental health teams in 4 hospitals in Manchester, UK.
Interventions Experimental: Manualised home based family therapy intervention (one assessment session plus 4 home
visits) plus routine care.
Control: Routine psychiatric aftercare - no home visits.
Therapist: Two masters level psychiatric social workers.
Type of therapy offered: Family therapy (Kerfoot, 1986; Kerfoot et al 1995).
Length of treatment: one assessment and four home visits.

Excluded: i) depression ii) suicidal ideation iii) hopelessness
iv) problem solving v) compliance vi) family functioning
vii) satisfaction with treatment; viii) cost-effectiveness; ix) parent GHQ; x) Child social problem solving.
Notes Repetition data from: not stated

Risk of bias

Harrington 1998 (Continued)
Hawton 1981
Methods Allocation: ’Random number method’ (sealed, opaque envelopes used).

Participants Setting: Oxford, UK.

Inclusion criteria: i) over age 16;
ii) not in psychiatric care; iii) not residing outside of study area; iv) not requiring treatment for alcohol or
dug addiction; v) not in need of inpatient psychiatric care; vi) suitable for randomisation e.g. not of no
fixed abode.
Numbers: 96; 48 experimental, 48 control.
Profile: Patients over the age of 16 years (mean age 25.3). 32% repeaters. No information on diagnoses.
70% female.
Source of participants: patients admitted to a general hospital following deliberate self-poisoning.
Interventions Experimental: Domiciliary therapy (brief problem-orientated)as often as therapist felt necessary. Open
telephone access to the general hospital service.
Control: Out-patient therapy once a week in an out-patient clinic in a general hospital.Length of treatment:
up to 3 months.
Therapist: two junior psychiatrists, one psychiatric nurse and 1 social worker.
Type of therapy offered: In both groups brief problem-orientated counselling was used.

Excluded: i) mood; ii) social adjustment; iii) change in problems; iv) suicide ideation;
v) GP questionnaire.
Notes Repetition data from:hospital records, interview with patient and GP questionnaire.
Risk of bias
Hawton 1987
Methods Allocation: ’Randomization procedure’ using sealed opaque envelopes. Follow up period: 12 months.

Hawton 1987 (Continued)
Participants Setting: Oxford, UK.

Inclusion criteria: i) over age of 16; ii) registered with a general practitioner; iii) living up to 15 miles away
from hospital; iv) suitable for out-patient counselling; v) not in need of psychiatric care (day-patient or
in-patient; vi) not in current psychiatric care; vii) willing to accept aftercare offered.
Profile: 31% were repeaters. 66% female. No information on diagnoses.
Source of participants: Patients admitted to a general hospital for self poisoning
Interventions Experimental: Outpatient problem-orientated therapy by non-medical clinicians.

Control: GP (family doctor) care: (e.g. individual support, marital therapy).
Therapist: 5 counsellors from clinical team in the general hospital psychiatric service.
Length of treatment: up to 8 sessions, each lasting on average 54 minutes.
Type of therapy offered: Experimental: problem-solving therapy (Hawton and Catalan, 1987). Control:
various e.g. marital, GP counselling, psychiatric referral.

Excluded: i) social adjustment; ii) Depression; iii) General Health Questionnaire;
iv) improvement in target problems; v) attitudes to treatment; vi) GP interview
Notes Repetition data from:hospital records plus interview with patient and interview with GP of patient.
Fatal attempts: no suicides occurred in the study.
Risk of bias
Hirsch 1982
Methods Allocation: Randomly allocated, double blind, placebo controlled trial.

Follow up period:12 weeks.

Inclusion criteria: i) not taking antidepressant or antipsychotic medication; ii) GHQ score of over 20.
Numbers: 114 - experimental 76, control 38.
Profile: Aged 16 - 65 years,
% repeaters, % female, psychiatric diagnoses not given.
Source of participants: Patients who were admitted to a hospital after deliberate self-poisoning.
Interventions Experimental: Antidepressants: either 30-60 mg mianserin or 75-150 mg nomifensine.

Control: Placebo.
Therapist: n.a.
Type of therapy offered: drug therapy.
Length of treatment: 6 weeks.

Hirsch 1982 (Continued)

Excluded: i) GHQ score;
ii) depression; iii) life events;
iv) compliance with treatment.
Notes Repetition data from: not specified.

Fatal attempts: there were no suicides in this study.
Risk of bias
Liberman 1981
Methods Allocation: ’Assigned randomly’ - method not described. Follow up period: 24 months.
Participants Setting: Los Angeles, USA.

Inclusion criteria: i) patients who were not psychotic; ii) not addicted to drugs and alcohol; iii) without
organic brain syndrome; iv) at least one previous suicide attempt.
Profile: All were repeaters.
Mean age 29.7 years. Age range 18-47 years. 67% female. 100% depressive neurosis. Most met criteria
for personality disorder.
Source of participants: Patients were referred by the psychiatric emergency service or the hospital A & E
Dept. following DSH.
Interventions Experimental: Inpatient treatment with behaviour therapy: social skills training (17 hours); anxiety man-
agement (10 hours) and family work (5 hours). Therapeutic milieu with token economy. Aftercare at
community mental health centre or with private therapist.
Control: Inpatient treatment with insight orientated therapy: individual therapy (17 hours); group therapy
and psychodrama (10 hours) and family therapy (5 hours). Therapeutic milieu with token economy.
Aftercare at community mental health centre or with private therapist.
Therapist: (i) Behaviour therapy: psychologist assisted by 2 bachelor level technicians. (ii) Insight therapy:
experienced social workers and psychologists (N not specified).
Length of treatment: 10 days - 4 hours of therapy for a period of 8 days. 32 hours in total.
Type of therapy offered: see above.

Excluded: i) depression; ii) reinforcement; iii) assertiveness; iv) fear.
Notes Repetition data from: Interview at follow-up at 24 months.

Risk of bias

Liberman 1981 (Continued)
Allocation concealment? Unclear B - Unclear
Linehan 1991
Methods Allocation: Randomized clinical trial (allocation via computer program).

N lost to follow up: 24 patients were deliberately not included in the follow-up.
Participants Setting: Seattle, USA.

Inclusion criteria: i) female; ii) borderline personality disorder diagnosis; iii) at least two attempts in the
last five years - with one in the last 8 weeks; iv) aged 18-45 years; v) agreed to study conditions.
Profile:100% female. 100% Borderline personality disorder. All were repeaters (multiple episodes of self-
harm)at high risk of repetition.
Source of participants: clinically referred patients who had an episode of deliberate self-harm in the last 8
weeks.
Interventions Experimental: Dialectical behavior therapy (individual and group work) for one year. Telephone access
with therapist.
Control: Treatment as usual (alternative therapy referrals).
Length of treatment: 12 months.
Therapist: five psychologists, one clinical psychology graduate, one psychiatrist.
Type of therapy offered: Experimental: dialectial behavior therapy. Control: standard aftercare.
73% individual psychotherapy (42.0% maintained in therapy for whole year)

Excluded: i) maintenance in therapy; ii) psychiatric in-patient treatment; iii) depression; iv) hopelessness;
v) reasons for living; vi) suicide ideation; vii) anger; vii) social adjustment; viii) global functioning.
Notes Repetition data from: parasuicide history interview - 50% were checked with medical records, therapist
records, and observer/nurse/physician ratings.
Risk of bias

McLeavey 1994
Methods Allocation:Patients were assigned on a random basis to the two treatment groups using an open random
number table.
Participants Setting: Cork, Ireland.

Inclusion criteria: i) aged 15-45 years; ii) no history of psychosis, mental retardation, or organic cognitive
impairment; iii) not requiring psychiatric treatment (day care or inpatient).
Numbers: 39; 19 experimental, 20 control.
Profile: Mean age 24.4 years. 74% female. 35.6% were repeaters. 23% dysthymia. 15% dependent per-
sonality disorder. 13% alcohol abuse.
Source of participants: patients admitted to A & E Dept. following self-poisoning.
Interventions Experimental: Interpersonal problem solving skills training. (D’Zurilla & Godfried , 1971). Control: Brief
problem-solving therapy (Hawton & Catalan, 1982).
Therapist: Clinical psychologists and registrars in psychiatry.
Type of therapy offered: Experimental: Interpersonal Problem Solving Therapy.
Control: Problem-solving therapy.
Length of treatment: Experimental: mean no. sessions 5.3. Control: mean no. sessions 4.2

Excluded: i) depression; ii) problem-solving; iii) self-perception; iv) number of problems.
Notes Repetition data from: Hospital records and GP questionnaire.

Risk of bias
Montgomery 1979
Methods Allocation: Random allocation in double blind trial.

N lost to follow up: 7 for repetition data.
Participants Setting: Maidstone, UK.

Inclusion criteria: i) documented history of 2 or more episodes of deliberate self-harm; ii) not suffering
from overt depression or schizophrenia; iii) no organic illness.
Profile: Patients aged 18 - 68 years (mean 35.3). All were repeaters. 70% female. No information on
psychiatric diagnoses.
Source of participants: patients admitted to a general hospital following a suicidal act.

Montgomery 1979 (Continued)
Interventions Experimental: 20 mg intramuscular flupenthixol decanoate 4 weekly for 6 months.

Control: Placebo.
Therapist: n. a.
Type of therapy offered: Drug therapy.

Excluded: none.

Risk of bias
Montgomery 1982
Methods see Montgomery et al 1979 and 1983.
Participants
Interventions
Outcomes
Notes
Risk of bias
Montgomery 1983
Methods Allocation: Patients randomly allocated to treatment under double blind conditions.

Inclusion criteria: i) multiple previous episodes of deliberate self-harm; ii) diagnosis of personality disorder;
iii) not depressed or schizophrenic.
Numbers: 58 - Ns in experimental and control groups are only given for those who completed treatment
(Ex. - 17, Con. - 21).

Montgomery 1983 (Continued)
Profile: Patients with personality disorders (borderline and histrionic). Mean age 35. 7 years for those who
completed treatment. All were multiple repeaters (mean 3.6 attempts). 66% female.
Source of participants: Patients who were admitted to a medical ward following deliberate self-harm.
Interventions Experimental: Mianserin - 30 mg.

Control: Placebo.
Therapist: n.a.
Type of therapy offered: Drug therapy.
Length of treatment: six months

Excluded: i) depression

Risk of bias
Morgan 1993
Methods Allocation: Random selection from a supply of closed envelopes, half of which contained a green card.
Participants Setting: Bristol, UK.

Inclusion criteria: i) no previous episode of deliberate self-harm; ii) resident within Healthcare Trust
catchment area.
Profile: Mean age 30 years. % female not given. No information on diagnoses. No repeaters.
Source of participants: Patients admitted to hospital following first episode of deliberate self-harm.
Interventions Experimental: Standard care plus green card (emergency card indicating that a doctor was available and
how to contact them).
Control: Standard care e.g. referral back to the primary healthcare team, psychiatric inpatient admission.
Therapist: telephone contact/face-to-face interviews conducted by doctor on-call.
Type of therapy offered: crisis intervention.

Excluded: none.
Notes Repetition data from: hospital, psychiatric and GP records.

Fatal attempts: there were no suicides in the study.

Morgan 1993 (Continued)
Risk of bias
Salkovskis 1990
Methods Allocation: Predetermined random allocation (sampling without replacement using envelopes). Follow
up period: 12 months.
N lost to follow up: none for repetition.
Participants Setting: Leeds, UK.

Inclusion criteria: i) Age 16-65; ii) of fixed abode and living within Health Authority boundary; iii)
not requiring immediate psychiatric treatment; iv) non-psychotic; v) no serious organic illness; vi) anti-
depressants taken as part of the overdose; vii) two or more previous attempts; viii) Buglass and Horton
(1974) Risk of Repetition Scale score of at least 4. (Patients had to fulfill at least two of criteria vi-viii to
be included).
Profile: Mean age 27.5 years. All were repeaters with a high risk of further repetition. 50% female. No
information on diagnoses.
Source of participants: patients who were referred by duty psychiatrist following antidepressant self-
poisoning and assessed in an A & E Department.
Interventions Experimental: Domiciliary cognitive-behavioural problem-solving treatment.

Control: Treatment as usual (not described).
Therapist: one CPN.
Length of treatment: five sessions each lasting one hour.
Type of therapy offered: problem-solving therapy.

Excluded: i) suicide ideation;
ii) severity of three main problems; iii) depression; iv) hopelessness.

Risk of bias

Torhorst 1987
Methods Allocation: Patients were randomly offered treatments.

N lost to follow up: 8/141 (5.7%) for repetition data.
Participants Setting: Munich, Germany.

Inclusion criteria: i) non-psychotic.
Profile: all self poisoning patients. 48% were repeaters. 63% female. No information on diagnoses.
Source of participants: patients hospitalised after a suicide attempt.
Interventions Experimental: Short crisis intervention during hospital stay, fixed out patient appointment with same
therapist as saw in hospital. Motivational interview, letter and assessment of motivation towards therapy.
Control: Short crisis intervention during hospital stay, fixed out patient appointment with a different
therapist than was seen in hospital. Motivational interview, letter and assessment of motivation towards
therapy.

Excluded: i) compliance
Notes Repetition data from: follow up interview at one year after index suicide attempt.
Fatal attempts: 3 suicides in experimental group, 2 suicides in control group.
In the first phase of this study the efficacy of standard care was assessed in terms of compliance. Patients
(N=85) were not randomly assigned to this group.
Risk of bias
Torhorst 1988
Methods Allocation: Randomization to treatment (method not specified).

Participants Setting: Munich, Germany.

Inclusion criteria: i) no endogenous psychosis; ii) not already in psychotherapeutic treatment; iii) not in
inpatient psychiatric treatment; iv) not illicit-drug overdose; v) able to understand German; vi) living
within travelling distance of research centre; vii) previous episodes of deliberate self-harm.
Numbers: 80 - experimental 40, control 40.
Profile: All patients were repeaters.
% female not given. Age not given.
Source of participants: patients who had deliberately self-poisoned referred to liaison service of toxicological
ward.

Torhorst 1988 (Continued)
Interventions Experimental: Long term therapy - one therapy session per month over a period of 12 months.
Control: Short term therapy - 12 weekly therapy sessions over a period of three months.
All participants in both groups had brief crisis intervention (3 days) in hospital.
Therapist: 3 psychiatric attendants.
Type of therapy offered: not specified.
Length of treatment: Ex. - 12 months, Con. - 3 months.

Excluded: i) compliance;
ii) depression; iii) complaints.

Risk of bias
Van der Sande 1997
Methods Allocation: Selection from series of opaque, sealed envelopes which contained a number from a list of
random numbers generated by a computer.
Follow up period:12 months.
Participants Setting: Utrecht, The Netherlands.

Inclusion criteria: i) able to understand and write Dutch; ii) living in catchment area of hospital; iii) not
psychiatric inpatient; iv) not in prison; v) no drug or alcohol addiction; vi) no recurrent consultations
with a liaison psychiatrist during a stay of more than 2 days on a somatic ward.
Profile: Over age 15 years. 73% were repeaters. 66% female. 32% mood disorder. 15% adjustment
disorder.
Source of participants: Patients admitted to hospital following a suicide attempt.
Interventions Experimental: Brief psychiatric unit admission, encouraging patients to contact unit on discharge. Out
patient therapy plus 24-hour emergency access to unit.
Control: Usual care: 25% hospitalization, 65% outpatient referral.
Therapist: Experimental: 1 psychiatrist, 2 CPNs and 9 psychiatric nurses. Control: not described.
Type of therapy offered: problem solving treatment used by CPNs with patients in experimental treatment
(Hawton and Catalan, 1982). Control therapy not specified.
Length of treatment: not specified.

Excluded: i) anxiety; ii) depression; iii) hopelessness; iv) phobic anxiety; v) somatisation; vi) obsession-
compulsion; vii) interpersonal sensitivity; viii) hostility; ix) sleep disorder; x) use of clinical services.

Van der Sande 1997 (Continued)
Notes Repetition data from: interview with patient, hospital records.

Risk of bias
van Heeringen 1995
Methods Allocation: Randomisation using open randomisation list was performed by a data nurse who did not
interview patients.
Follow up period:12 months
N lost to follow up:125/516 (24%) for repetition data.
Participants Setting: Gent, Belgium.

Inclusion criteria: i) resident in catchment area; ii) over 15 years old; iii) not in in-patient medical
treatment.
Profile: 15% mood disorder. 2.7% anxiety disorder. 30% repeaters.
43% female.
Source of participants: Patients treated in A & E Dept. after a suicide attempt.
Interventions Experimental: Special care - home visits were made to patients who did not keep outpatient appointments,
the reasons for not attending appointments were discussed and the patient was encouraged to attend.
Control: Outpatient appointments only; non-compliant patients were not visited.
Therapist: community nurse did home visit.
Type of therapy offered: various offered on discharge e.g. out-patient appointment, GP referral, private
psychologist/psychiatrist.
Length of treatment: not specified.

Excluded: i) compliance.
Notes Repetition data from: Interview with patient or with relative/GP if patient could not be contacted.
Fatal attempts: 6 suicides in experimental group, 7 suicides in control group.
Risk of bias

Verkes 1998
Methods Allocation: Participants were randomly assigned to one of two treatment groups.
Follow-up period: 12 months.
N lost to follow-up: none for repetition.
Participants Setting: Leiden and Rotterdam, The Netherlands

Inclusion criteria: i) all were repeaters; ii) aged 18 years and over; iii) without current diagnosis of major
depression (but 5 patients had a past history of this).
Profile: Age range - 18 years and older. Mean age 35.6 years.
100% were repeaters. % female not given. 92% personality disorder, 6.5% dysthymia, 4.4% anxiety
disorder,
8.8% dissociative disoder, 44% alcohol abuse, 20.9% adjustment disorder, 25.3% depressive disorder
(not specified).
16.5% no psychiatric diagnosis.
Source of participants: patients recruited from outpatient departments in A & E wards of University
hospitals in Leiden and Rotterdam.
Interventions Experimental: Paroxetine 40mg/day, plus psychotherapy weekly/fortnightly.

Control: Matched placebo, plus psychotherapy weekly/fortnightly.
Therapist: (type and n) not described.
Type of therapy offered: drugs and psychotherapy.
Outcomes Included: i) repetition of self harm.

Excluded: i) depression; ii) hopelessness; iii) anger; iv) compliance; v) platelet serotonin content; vi) side-
effects.
Notes Repetition data from: not stated.

Risk of bias
Waterhouse 1990
Methods Allocation: Randomisation using sequentially numbered sealed enveloped.

Follow up period: 16 weeks.
Participants Setting: York, UK.

Inclusion criteria: i) patients with no immediate medical or psychiatric treatment needs. (ii) over 16 years
old.
Profile: 36% repeaters. 62% female. Mean age 30.3 years. No information on diagnoses.
Source of participants: Patients admitted to an A & E Department for deliberate self-harm.

Waterhouse 1990 (Continued)
Interventions Experimental: General hospital admission - no additional treatment or counselling.

Control: Discharge from hospital.
On discharge both groups advised to contact their GP if they needed further help.
Length of treatment: The median length of admission was 17 hours.
Therapist: none.
Type of therapy offered: none.

Excluded: i) depression, hopelessness and anxiety; ii) suicidal ideation; iii) social isolation; iv) somatic
concerns; v) daily routine; vi) suicidal behaviour assessment schedule; vi) GP questionnaire; vii) psychiatric
admission; viii) time off work.
Notes Repetition data from:GP interview, hospital records.

Risk of bias
Welu 1977
Methods Allocation: Patients were randomly assigned using a table of random numbers.
N lost to follow up:repetition data not available for one person.
Participants Setting: Pittsburgh, USA.

Inclusion criteria: i) over 16 years old; ii) not a student living in unversity accommodation; iii) not resident
in care giving institution or institutionalized at the time of the attempt.
Numbers:120 -57 control, 63 experimental.
Profile: 60% repeaters. No information on age, sex, or psychiatric diagnoses.
Source of participants: Patients admitted to an A & E Department for deliberate self-harm.
Interventions Experimental: Special outreach programme. Community Mental Health Team (CMHT) contacted patient
immediately after discharge. Home visit arranged as soon as possible. Weekly/bi-weekly contact with
therapist.
Control: Routine treatment program. Psychiatric consultation at request of treating physician. Patients
given appointment for evaluation at the CMHT centre next day. Any further contact after discharge was
up to the patient to decide.
Therapist: 4 nurses, 3 social workers, 2 community workers.
Type of therapy offered: several used e.g. psychotherapy, crisis intervention, family counselling, chemo-
therapy, etc.). Focus on quantity and continuity of care.

Excluded: i) extent of follow up coverage; ii) type and frequency of contacts; iii) purposive accidents; iv)

Welu 1977 (Continued)
excessive use of alcohol; v) drug misuse.
Notes Repetition data from: interview with patient, hospital records, interviews with family and friends.
Fatal attempts: No suicides mentioned.
Risk of bias
Characteristics of excluded studies [ordered by study ID]
Bateson 1989 Non-randomised CCT.

In this study 100 deliberate self-harm patients presenting to the casualty department of a general hospital
in England were assessed by a psychiatric registrar and a psychiatric social worker (experimental group) or
by a psychiatrist alone (control group).
Outcome (repetition)at follow-up: Ex. 12/50 (24%) vs. Con. 11/50 (22%)
Catalan 1980 Non-randomized CCT.

120 deliberate self-harm patients presenting to a casualty department in a general hospital in England.
Patients in the experimental group were assessed by nurses (all Staff Nurse grade of whom 5 had psychiatric
training, 2 were state registered and 1 was state enrolled). Patients in the control group were assessed by
doctors (5 psychiatric trainees of registrar grade and 3 GP trainees). The GP trainees and nurses received
5 weeks of training in the assessment procedure and the psychiatrists’ training lasted 4 weeks.
Outcome (repetition)at follow-up: Ex. 9/75 (12%) vs. Con. 5/45 (11%).
Crawford 1998 Non-randomized CCT - no data on repetition of DSH at follow-up.

252 deliberate self-harm patients presenting to an A & E department in a general hospital in England.
Patients in the experimental group were assessed by A & E staff who had received special training (a one
hour teaching session) in the assessment of deliberate self-harm patients. Patients in the control group were
assessed by members of the same team before they had received the training.
Outcome (repetition)at follow-up: repetition was not assessed.
Deykin 1986 Non-randomised CCT.

319 adolescent deliberate self-harm patients presenting to the casualty department of two general hospitals
in the USA received either Special Care (Boston City Hospital) involving a direct (outreach) social worker
plus an educational program, or Standard Aftercare (Brockton Hospital).
Donaldson 1997 Non-randomised CCT.

23 adolescent deliberate self-harm patients presenting to the casualty department of a general hospital in
the USA received an experimental outreach intervention. The patient and member of his/her family were
encouraged to attend four outpatient therapy sessions and were also telephoned at 1, 2 and 6 weeks. The
control group was 78 adolescents who had participated in a three month follow-up at the same hospital

(Continued)
several years earlier.

Gardner 1977 Original randomization was disrupted in some subjects which affects the analysis of the dependent variable
used in this review (i.e. repetition of DSH).
In this study 276 deliberate self-poisoning patients presenting to a general hospital in England were
randomized to be assessed by a medical team (experimental group) or a psychiatrist (control group).
Included patients who re-presented with DSH were re-randomized, as the main outcome in this trial was
the treatment recommended for each DSH case. Because of this randomization was disrupted for these
patients in terms of repetition data.
Patsiokas 1995 RCT - no data on repetition of deliberate self-harm during follow-up.

15 deliberate self-harm patients were recruited from psychiatric inpatient wards of a University Hospital in
the USA. After randomization patients received either Cognitive Restructuring Therapy, Problem Solving
Therapy or Non-directive Control Treatment
Outcome (repetition)at follow-up: not available.
Rotherham-Borus 1996 Non-randomised CCT - no data on repetition of deliberate self-harm during follow-up.
In this study 140 Latina (female) adolescent deliberate self-harm patients presenting to a Emergency
Room (ER) in a general hospital in the USA. Patients in the experimental group received a specialized
ER programme where staff had undergone special training in care for DSH patients, an educational video
was viewed by patients and their families, a family therapist was on call and an initial therapy session was
provided. Patients in the control group received standard ER care.
After discharge all patients were referred to a specialist centre for six sessions of outpatient family therapy.
Outcome (repetition)at follow-up: not available.
Termansen 1975 Non-randomised CCT.

202 deliberate self-harm presenting to an ER in a general hospital in Canada. Patients were allocated to
one of four treatment groups:
1. ER assessment and follow-up (for three months) by the same mental health professional.
2. ER assessment by a mental health professional, plus follow-up (for three months) by a crisis centre
volunteer with reassessment by the volunteer and mental health worker at three months.
3. ER assessment by a mental health professional, but no follow-up. Reassessment at three months by the
same mental health professional.
4. No ER assessment (identified by medical records). Assessment at three months by mental health pro-
fessional.
Outcome (repetition)at follow-up:
Group 1: 1/45 (2%)
Group 2: 2/33 (6%)
Group 3: 7/25 (28%)
Group 4: 2/16 (13%)
Wullimier 1979 Non-randomised CCT.

In this study 288 deliberate self-harm patients presenting to a general hospital in Switzerland. Patients in the
experimental group were ‘Systematically treated with classical therapeutic interventions’ (e.g. supportive
psychotherapy, psychiatric hospitalization, crisis intervention etc), and were contacted at several points
during the 2 years after the attempt (a few days after discharge, 1 month, 2 months, 6 months, 1 year
and 2 years). Patients in the historical control group received standard aftercare which involved the same
therapeutic interventions, but were were only contacted for follow-up 2 years after their attempt, without

(Continued)
having been told that contact would be made.


DATA AND ANALYSES
Comparison 1. Problem solving therapy vs Standard aftercare
No. of No. of
Outcome or subgroup title studies participants Statistical method Effect size
1 Repetition 5 571 Peto Odds Ratio (Peto, Fixed, 95% CI) 0.71 [0.45, 1.11]
Comparison 2. Intensive intervention plus outreach vs. Standard aftercare
No. of No. of
Comparison 3. Emegency card vs. Standard aftercare
No. of No. of
Comparison 4. Dialectical behavior therapy vs. Standard aftercare
No. of No. of

Comparison 5. Inpatient behavior therapy vs Inpatient insight-orientated therapy
No. of No. of
Comparison 6. Same therapist vs. Different therapist
No. of No. of
Comparison 7. General hospital admission vs. Discharge
No. of No. of
Comparison 8. Flupenthixol vs. Placebo
No. of No. of
Comparison 9. Antidepressants vs. Placebo
No. of No. of

Comparison 10. Long-term therapy vs. Short-term therapy
No. of No. of
Comparison 11. Homebased family therapy vs. Standard aftercare
No. of No. of
Analysis 1.1. Comparison 1 Problem solving therapy vs Standard aftercare, Outcome 1 Repetition.
Review: Psychosocial and pharmacological treatments for deliberate self harm
Comparison: 1 Problem solving therapy vs Standard aftercare
Outcome: 1 Repetition
Study or subgroup Problem solving Standard care Peto Odds Ratio Weight Peto Odds Ratio
n/N n/N Peto,Fixed,95% CI Peto,Fixed,95% CI
Evans 1999 10/18 10/14 10.2 % 0.52 [ 0.13, 2.15 ]
Gibbons 1978 27/200 29/200 64.8 % 0.92 [ 0.52, 1.62 ]
Hawton 1987 3/41 6/39 10.9 % 0.45 [ 0.11, 1.79 ]
McLeavey 1994 2/19 5/20 7.9 % 0.38 [ 0.08, 1.93 ]
Salkovskis 1990 3/12 4/8 6.2 % 0.35 [ 0.06, 2.19 ]
Total (95% CI) 290 281 100.0 % 0.71 [ 0.45, 1.11 ]

Total events: 45 (Problem solving), 54 (Standard care)
Heterogeneity: Chi2 = 2.54, df = 4 (P = 0.64); I2 =0.0%
Test for overall effect: Z = 1.50 (P = 0.13)
0.1 0.2 0.5 1 2 5 10

Favours treatment Favours control

Analysis 2.1. Comparison 2 Intensive intervention plus outreach vs. Standard aftercare, Outcome 1
Repetition.
Comparison: 2 Intensive intervention plus outreach vs. Standard aftercare
Study or subgroup Treatment Control Peto Odds Ratio Weight Peto Odds Ratio
Allard 1992 22/63 19/63 17.3 % 1.24 [ 0.59, 2.61 ]
Chowdhury 1973 17/71 18/84 16.8 % 1.15 [ 0.54, 2.45 ]
Hawton 1981 5/48 7/48 6.6 % 0.69 [ 0.21, 2.29 ]
Van der Sande 1997 24/140 20/134 23.0 % 1.18 [ 0.62, 2.24 ]
van Heeringen 1995 21/196 34/195 29.5 % 0.57 [ 0.32, 1.01 ]
Welu 1977 3/62 9/57 6.8 % 0.30 [ 0.09, 0.99 ]
Total (95% CI) 580 581 100.0 % 0.84 [ 0.62, 1.15 ]

Total events: 92 (Treatment), 107 (Control)
Heterogeneity: Chi2 = 7.46, df = 5 (P = 0.19); I2 =33%
0.1 0.2 0.5 1 2 5 10
Analysis 3.1. Comparison 3 Emegency card vs. Standard aftercare, Outcome 1 Repetition.
Comparison: 3 Emegency card vs. Standard aftercare
Cotgrove 1995 3/47 7/58 36.6 % 0.52 [ 0.14, 1.92 ]
Morgan 1993 5/101 12/111 63.4 % 0.45 [ 0.17, 1.22 ]
Total (95% CI) 148 169 100.0 % 0.48 [ 0.22, 1.05 ]

Heterogeneity: Chi2 = 0.03, df = 1 (P = 0.87); I2 =0.0%
0.1 0.2 0.5 1 2 5 10

Analysis 4.1. Comparison 4 Dialectical behavior therapy vs. Standard aftercare, Outcome 1 Repetition.
Comparison: 4 Dialectical behavior therapy vs. Standard aftercare
Linehan 1991 5/19 12/20 100.0 % 0.26 [ 0.08, 0.92 ]
Total (95% CI) 19 20 100.0 % 0.26 [ 0.08, 0.92 ]

Heterogeneity: not applicable
0.1 0.2 0.5 1 2 5 10
Analysis 5.1. Comparison 5 Inpatient behavior therapy vs Inpatient insight-orientated therapy, Outcome 1
Repetition.
Comparison: 5 Inpatient behavior therapy vs Inpatient insight-orientated therapy
Liberman 1981 2/12 3/12 100.0 % 0.62 [ 0.09, 4.24 ]
Total (95% CI) 12 12 100.0 % 0.62 [ 0.09, 4.24 ]

0.1 0.2 0.5 1 2 5 10

Analysis 6.1. Comparison 6 Same therapist vs. Different therapist, Outcome 1 Repetition.
Comparison: 6 Same therapist vs. Different therapist
Torhorst 1987 12/68 4/73 100.0 % 3.32 [ 1.18, 9.38 ]
Total (95% CI) 68 73 100.0 % 3.32 [ 1.18, 9.38 ]

0.1 0.2 0.5 1 2 5 10
Analysis 7.1. Comparison 7 General hospital admission vs. Discharge, Outcome 1 Repetition.
Comparison: 7 General hospital admission vs. Discharge
Waterhouse 1990 3/38 4/39 100.0 % 0.75 [ 0.16, 3.53 ]
Total (95% CI) 38 39 100.0 % 0.75 [ 0.16, 3.53 ]

0.1 0.2 0.5 1 2 5 10

Analysis 8.1. Comparison 8 Flupenthixol vs. Placebo, Outcome 1 Repetition.
Comparison: 8 Flupenthixol vs. Placebo
Montgomery 1979 3/14 12/16 100.0 % 0.13 [ 0.03, 0.52 ]
Total (95% CI) 14 16 100.0 % 0.13 [ 0.03, 0.52 ]

0.1 0.2 0.5 1 2 5 10
Analysis 9.1. Comparison 9 Antidepressants vs. Placebo, Outcome 1 Repetition.
Comparison: 9 Antidepressants vs. Placebo
Hirsch 1982 16/76 5/38 32.7 % 1.68 [ 0.62, 4.58 ]
Montgomery 1983 8/17 12/21 20.5 % 0.67 [ 0.19, 2.39 ]
Verkes 1998 15/46 21/45 46.8 % 0.56 [ 0.24, 1.29 ]
Total (95% CI) 139 104 100.0 % 0.83 [ 0.47, 1.48 ]

Heterogeneity: Chi2 = 2.88, df = 2 (P = 0.24); I2 =31%
0.1 0.2 0.5 1 2 5 10

Analysis 10.1. Comparison 10 Long-term therapy vs. Short-term therapy, Outcome 1 Repetition.
Comparison: 10 Long-term therapy vs. Short-term therapy
Torhorst 1988 9/40 9/40 100.0 % 1.00 [ 0.35, 2.84 ]
Total (95% CI) 40 40 100.0 % 1.00 [ 0.35, 2.84 ]

0.1 0.2 0.5 1 2 5 10
Analysis 11.1. Comparison 11 Homebased family therapy vs. Standard aftercare, Outcome 1 Repetition.
Comparison: 11 Homebased family therapy vs. Standard aftercare
Harrington 1998 11/74 11/75 100.0 % 1.02 [ 0.41, 2.50 ]
Total (95% CI) 74 75 100.0 % 1.02 [ 0.41, 2.50 ]

0.1 0.2 0.5 1 2 5 10

Favours Treatment Favours Control
WHAT’S NEW
Last assessed as up-to-date: 29 July 1999.
5 November 2008 Amended Converted to new review format.

HISTORY
Protocol first published: Issue 1, 1997
Review first published: Issue 3, 1999
30 July 1999 New citation required and conclusions have changed Substantive amendment
CONTRIBUTIONS OF AUTHORS
KH initiated the study and together with ET and EA designed the original protocol, coordinated the study, and analyzed the data.
ET and EA conducted the electronic searching and the data abstraction. All the authors contributed to revision of the protocol, hand
searching of journals, and the writing of the paper, which was initially drafted by KH and ET. EF, PH, DG, AH, KvH and IS carried
out the quality assessments. KH is guarantor for the review.
DECLARATIONS OF INTEREST
None
SOURCES OF SUPPORT
Internal sources
• University Department of Psychiatry, Warneford Hospital, Oxford., UK.
External sources
• NHS Executive Anglia and Oxford Research and Development Program, UK.
NOTES
This review is in the process of being updated. We hope to publish the updated version in Issue 2, 2008.

INDEX TERMS
Medical Subject Headings (MeSH)

Cognitive Therapy; Self-Injurious Behavior [drug therapy; ∗ therapy]; Social Support
MeSH check words

Humans


Psychosocial and Pharmacological Treatments For Deliberate Self Harm (Review)

Uploaded by

Copyright:

Available Formats

Psychosocial and Pharmacological Treatments For Deliberate Self Harm (Review)

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Psychosocial and Pharmacological Treatments For Deliberate Self Harm (Review)

Uploaded by

Copyright:

Available Formats

Psychosocial and pharmacological treatments for deliberate

self harm (Review)

Hawton KKE, Townsend E, Arensman E, Gunnell D, Hazell P, House A, van Heeringen K

Psychosocial and pharmacological treatments for deliberate self harm (Review)

Psychosocial and pharmacological treatments for deliberate self harm (Review) i

Psychosocial and pharmacological treatments for deliberate

Editorial group: Cochrane Depression, Anxiety and Neurosis Group.

PLAIN LANGUAGE SUMMARY

Psychosocial and pharmacological treatments for deliberate self harm

Psychosocial and pharmacological treatments for deliberate self harm (Review) 3

Types of interventions Data collection and analysis

Psychosocial and pharmacological treatments for deliberate self harm (Review) 4

Psychosocial and pharmacological treatments for deliberate self harm (Review) 5

Psychosocial and pharmacological treatments for deliberate self harm (Review) 6

Psychosocial and pharmacological treatments for deliberate self harm (Review) 7

Termansen 1975 {published data only} Hawton 1992

Wullimier 1979 {published data only} Hawton 1997a

Psychosocial and pharmacological treatments for deliberate self harm (Review) 11

Characteristics of included studies [ordered by study ID]

Participants Setting: Montreal, Canada.

Outcomes Included: i) repetition of self-harm.

Item Authors’ judgement Description

Allocation concealment? Yes A - Adequate

Participants Setting: Edinburgh, UK.

Psychosocial and pharmacological treatments for deliberate self harm (Review) 12

alcohol and drug addiction were included.

Outcomes Included: i) repetition of self-harm.

Notes Repetition data from: hospital records.

Item Authors’ judgement Description

Allocation concealment? No C - Inadequate

Methods Allocation: Random allocation using open number table.

Participants Setting: Chester, UK.

Psychosocial and pharmacological treatments for deliberate self harm (Review) 13

Length of treatment: one year.

Outcomes Included: i) repetition of self-harm.

Notes Repetition data from: clinical and hospital notes.

Item Authors’ judgement Description

Allocation concealment? No C - Inadequate

Methods See:Hirsch 1982

Item Authors’ judgement Description

Allocation concealment? Yes A - Adequate

Methods Allocation: opaque sealed envelopes opened sequentially.

Participants Setting: London, UK.

Psychosocial and pharmacological treatments for deliberate self harm (Review) 14

Paddington and Chelsea, Westminster).

Outcomes Included: repetition of self-harm.

Notes Repetition data from: interview and hospital records.

Item Authors’ judgement Description

Allocation concealment? Yes A - Adequate

Participants Setting: Southampton, UK.

Psychosocial and pharmacological treatments for deliberate self harm (Review) 15

Outcomes Included: i) repetition of self-poisoning.

Notes Repetition data from: hospital records.

Item Authors’ judgement Description

Allocation concealment? Yes A - Adequate

Participants Setting: Manchester, UK.

Outcomes Included: i) repetition of self-harm.

Notes Repetition data from: not stated

Psychosocial and pharmacological treatments for deliberate self harm (Review) 16