paper with references attached that need to be included in the paper

O R I G I N A L C O N T R I B U T I O N

Psychological characteristics, stressful life events and deliberate self-harm: findings from the Child & Adolescent Self-harm in Europe (CASE) Study

Nicola Madge • Keith Hawton • Elaine M. McMahon • Paul Corcoran •

Diego De Leo • Erik Jan de Wilde • Sandor Fekete • Kees van Heeringen •

Mette Ystgaard • Ella Arensman

Received: 7 January 2011 / Accepted: 28 July 2011 / Published online: 17 August 2011

� Springer-Verlag 2011

Abstract There is evidence to suggest that both psy-

chological characteristics and stressful life events are

contributory factors in deliberate self-harm among young

people. These links, and the possibility of a dose–response

relationship between self-harm and both psychological

health and life events, were investigated in the context of a

seven-country school-based study. Over 30,000, mainly 15

and 16 year olds, completed anonymous questionnaires at

secondary schools in Belgium, England, Hungary, Ireland,

the Netherlands, Norway and Australia. Pupils were asked

to report on thoughts and episodes of self-harm, complete

scales on depression and anxiety symptoms, impulsivity

and self-esteem and indicate stressful events in their lives.

Level and frequency of self-harm was judged according to

whether they had thought about harming themselves or

reported single or multiple self-harm episodes. Multi-

nomial logistic regression assessed the extent to which

psychological characteristics and stressful life events dis-

tinguished between adolescents with different self-harm

histories. Increased severity of self-harm history was

associated with greater depression, anxiety and impulsivity

and lower self-esteem and an increased prevalence of all

ten life event categories. Female gender, higher impulsivity

and experiencing the suicide or self-harm of others, phys-

ical or sexual abuse and worries about sexual orientation

independently differentiated single-episode self-harmers

from adolescents with self-harm thoughts only. Female

gender, higher depression, lower self-esteem, experiencing

the suicide or self-harm of others, and trouble with the

police independently distinguished multiple- from single-

episode self-harmers. The findings reinforce the impor-

tance of psychological characteristics and stressful life

events in adolescent self-harm but nonetheless suggest that

some factors are more likely than others to be implicated.

Keywords Adolescence � Cross-cultural � Self-harm � Depression � Anxiety � Impulsivity � Self-esteem � Stressful life events � Psychological characteristics

Background

Deliberate self-harm is a significant problem among the

young. Although hard to predict, and often appearing ‘out

of the blue’, there is evidence to suggest that mental health

problems, impulsivity, self-esteem and stress in young

people’s lives are contributory factors. We examined

N. Madge (&) Centre for Child and Youth Research, Mary Seacole Building,

Brunel University, Uxbridge UB8 3PH, UK

e-mail: [email protected]

K. Hawton

Centre for Suicide Research, University of Oxford, Oxford, UK

E. M. McMahon � P. Corcoran � E. Arensman National Suicide Research Foundation, Cork, Ireland

D. De Leo

Australian Institute for Suicide Research and Prevention,

Griffith University, Brisbane, Australia

E. J. de Wilde

Netherlands Youth Institute, Utrecht, NL, Netherlands

S. Fekete

Department of Psychiatry, University of Pécs, Pécs, Hungary

K. van Heeringen

Unit of Suicide Research, Ghent University, Ghent, Belgium

M. Ystgaard

Centre for Child & Adolescent Mental Health, Oslo, Norway

123

Eur Child Adolesc Psychiatry (2011) 20:499–508

DOI 10.1007/s00787-011-0210-4

associations between self-harm and these factors among a

large sample of (mainly) 15- and 16-year olds participating

in the Child & Adolescent Self-harm in Europe (CASE)

Study and build on earlier findings from this international

research [28, 42, 47].

Considerable evidence links psychological characteris-

tics with self-harm thoughts and behaviour in adolescents

[14]. These problems may be state-dependent (e.g.

depression or anxiety) or trait-dependent (e.g. impulsivity

and self-esteem). Associations with depression are partic-

ularly widely reported [1, 26, 35], including longitudinal

studies [16, 44]. Anxiety symptoms also increase risk of

self-harm, particularly when associated with depression

[14, 18, 41].

Impulsivity has been linked with both self-harm epi-

sodes and suicidal ideation [10, 21, 25, 26]. Brent and

Mann [5] implicate aggressive impulsivity in the interpre-

tation of patterns of suicidal behaviour across generations.

Self-esteem (in relation to peers, school, family, sports/

athletics, body image and global self-worth) has also been

related to suicidal thoughts and attempts [25, 46].

Self-harm shows strong links with stress factors [20],

including difficulties in familial relationships [7, 23, 31],

poor relationships with friends and partners [11, 23] and

perceptions of poor academic performance [30].

In addition, knowing someone who has self-harmed, or

made a suicide attempt, contributes to risk [5, 6, 14, 27, 32].

Gay, lesbian and bisexual young people appear at elevated

risk [3, 39], especially when facing family difficulties [12].

Child abuse, especially sexual abuse, has repeatedly been

associated with self-harming and suicidal behaviour [4, 19,

48], and the link appears direct even though self-esteem [15]

or continued adversity over the life-cycle [19] may play a

mediating role. Both being bullied [8] and fear of bullying [2]

have been linked to an increase in self-harm.

In this paper, we explore links between psychological

characteristics, life events and self-harm history among young

people within a large international dataset. In particular, we

examine the dose–response hypothesis that increasing

adversity, in terms of psychological characteristics and life

events, is associated with increasing level and frequency of

self-harm. Such a relationship between depression and self-

harming behaviour has already been suggested for both fre-

quency [13, 19, 22] and severity [35] of episodes.

Methods

The CASE Study

This paper draws on data from the CASE Study, an inter-

nationally collaborative investigation of self-harm among

young people in seven self-selected countries. These

comprised six European countries—Belgium, England,

Hungary, Ireland, the Netherlands and Norway—and

Australia. The study methodology was similar in each

participating country and is described in detail elsewhere

[28]. School-based surveys were conducted with a total of

30,477 14–17-year-old adolescents, the majority being 15

or 16 years, who consented to provide anonymous self-

report data on self-harm behaviour (e.g. timing, frequency

of episodes, methods used, motives, help-seeking before

and after the episode, hospitalisation, serious thoughts

about self-harm), negative life events, lifestyle and psy-

chological characteristics including symptoms of anxiety,

depression, self-esteem and impulsivity. Schools were

selected to be locally and nationally representative, and

response rates ranged from 81 to 96 per cent in individual

countries.

Measures

Self-harm

To ensure international comparability, strict criteria were

adopted to assess self-harm taking account of whether

young people said they had harmed themselves as well as

descriptions of the last episode. The criterion for self-harm

depended on a report of at least one of the following acts

deliberately undertaken with non-fatal outcome:

• Initiated behaviour (e.g. self-cutting, jumping from a height), which they intended to cause self-harm;

• Ingested a substance in excess of the prescribed or generally recognised therapeutic dose;

• Ingested a recreational or illicit drug that was an act that the person regarded as self-harm;

• Ingested a non-ingestible substance or object.

Judgements of self-harm were in no way dependent on

motives other than that the act in question appeared

deliberate. It was not possible to draw a clear distinction

between suicidal and non-suicidal self-harm from the sur-

vey data. Further information on the assessment of self-

harm is provided elsewhere [28].

Study participants were divided into four groups: ‘no

self-harm’ comprising all those who reported neither self-

harm behaviour meeting the study criteria nor thoughts of

self-harm, in the past year; ‘self-harm thoughts only’ with

self-harm thoughts but no episode of self-harm in the past

year; ‘single self-harm episode’ with one episode in the

previous year; and ‘multiple self-harm episodes’ with an

episode in the previous year as well as at least one other

earlier episode.

500 Eur Child Adolesc Psychiatry (2011) 20:499–508

123

Psychological characteristics

Four psychological scales were included. Anxiety and

depression symptoms were measured by the HADS scale

[49], which includes two 7-item subscales for anxiety and

depression, each scored between 0 and 21, where higher

scores indicated higher levels of anxiety and depression.

Self-esteem was measured using an eight-item abbreviated

version of the Robson [40] self-concept scale, with possible

scores between 8 and 32, and higher scores indicating more

positive self-esteem. A six-item impulsivity scale [37] led

to scores between 6 and 24 with higher scores indicating

greater impulsivity. Raw scores on these measures are used

in the analysis.

Life events

The questionnaire included 20 questions relating to nega-

tive life events in the past 12 months and/or more than a

year ago. For the purposes of analysis, these were reduced

to ten categories of lifetime experience. These were:

‘Difficulties with friends and peers’; ‘Problems with or

between parents’; ‘Serious illness of family or friend’;

‘Physical or sexual abuse’; ‘Suicide or self-harm of family

or friend’; ‘Death of someone close’; ‘Worries about sex-

ual orientation’; ‘Trouble with police’; ‘Bullied’; and

‘Problems with schoolwork’.

Data analysis

All analyses were undertaken for the sample as a whole,

and some analyses were carried out by gender and by

country. Chi-square tests were used to assess the statistical

significance of associations between pairs of categorical

variables such as self-harm history and gender. The

strength of these associations was measured by the Phi

statistic. In line with previous recommendations [9], asso-

ciations were considered very weak if Phi \0.10, weak if \0.30, moderate if \0.50 and strong if 0.50?.

Cronbach’s alpha was used to assess the internal con-

sistency of the scales employed to measure the psycho-

logical characteristics. These measures (depression,

anxiety, impulsivity, self-esteem) and the total number of

stressful life event categories experienced followed a nor-

mal distribution and were summarised by the mean and

standard deviation. Pearson’s correlation coefficient, r, was

used to assess the strength of the linear association between

the psychological measures, and the coefficient of deter-

mination, r2, was used to measure the information in one

measure that could be explained statistically by another.

Between-group comparisons of the psychological mea-

sures and the number of reported stressful life event cate-

gories were carried out using analysis of variance. The

effect size was measured using partial g2, and, following established guidelines [9], the effect size was considered

very small if partial g2 \ 0.01, small if \0.06, medium if \0.14 and large if 0.14?. Post hoc pairwise comparisons were based on Tukey’s honestly significant difference.

Multinomial logistic regression was used to assess the

associations between gender, age, country, psychological

characteristics and reported stressful life event categories

and self-harm history in the past year with a view to

identifying which factors distinguished between adoles-

cents reporting no self-harm, self-harm thoughts only, a

single self-harm episode and multiple self-harm episodes.

The dependent variable comparison group was the group

with no self-harm in the past year. Gender, age and

country were entered into the regression model as the first

block of independent variables. A forward stepwise

approach was adopted for the second block of indepen-

dent variables which contained each reported stressful life

event category, the total number of reported life event

categories and the psychological characteristics. Associa-

tions were reported as odds ratios with 95% confidence

intervals. Nagelkerke’s r2 was used as the estimate of the

proportion of variance explained by the derived regression

model. Wald tests were carried out to identify the factors

distinguishing adolescents with a single self-harm episode

from those with self-harm thoughts only and factors dis-

tinguishing between multiple- and single-episode self-

harmers.

Further multinomial logistic regression models were

estimated in which interaction terms (by country, by gender

and between the psychological measures) were considered

as a third block of factors for entry into the multivariate

regression model described previously. Change in Nage-

lkerke’s r2 was used to quantify the extent to which

interaction terms further distinguished between the four

self-harm history groups.

Results

Self-harm history in past year

Overall, 79.6% of the young people said they had not

experienced thoughts of self-harm in the past year, 14.6%

said they had thought about harming themselves but had

not done so, 2.6% reported single self-harm episodes in the

past year meeting the study criteria, and a further 3.2%

reported multiple episodes. Females were at least twice as

likely as males to report having thoughts of self-harm and

both single- and multiple-episodes of self-harm (Table 1).

While statistically significant at P \ 0.001, the strength of the association between self-harm history and gender was

weak (/ = 0.22).

Eur Child Adolesc Psychiatry (2011) 20:499–508 501

123

Some differences emerged between countries [see also

28]. The Netherlands, in particular, stood out as having low

rates of both thoughts and episodes of self-harm among

both males and females. Hungary also showed an inter-

esting pattern in that, compared with other countries, both

males and females were less likely to report no thoughts or

episodes of self-harm in the previous year, but more likely

to report self-harm thoughts only.

Psychological characteristics

The psychological measures had satisfactory levels of

internal consistency (Cronbach’s alpha: 0.69 for depres-

sion, 0.84 for anxiety, 0.75 for impulsivity and 0.90 for

self-esteem). All intercorrelations between depression,

anxiety, impulsivity and self-esteem were statistically sig-

nificant at P \ 0.001. There was only one strong correla- tion that was between depression and anxiety (Pearson’s

correlation coefficient, r = 0.52; r2 = 0.27). Depression

and anxiety were equally negatively correlated with self-

esteem (r = -0.45; r2 = 0.20). Other correlations were

weak, ranging in magnitude from 0.13 to 0.24

(r2 = 0.02–0.06), indicating that there was limited overlap

between the psychological measures.

There was no gender difference in relation to depression

(Table 1). Female scores were generally higher than male

scores on anxiety and impulsivity and lower on self-

esteem. However, the effect size was small for anxiety and

self-esteem and very small for impulsivity.

Depression, anxiety, impulsivity and self-esteem dif-

fered by country (P \ 0.001 in each case) with partial g2

ranging from 0.03 to 0.04 and indicating a small effect of

country on psychological measures. Mean scores ranged

from 3.6 in Ireland to 5.7 in Hungary for depression,

between 6.0 in Norway and 8.0 in England for anxiety,

between 13.2 in Belgium and 14.8 in Hungary for

impulsivity, and between 21.3 in Hungary and 23.3 in

Norway for self-esteem. Overall Hungarian adolescents

stood out as displaying the highest levels of psychological

difficulties, while Norwegian adolescents showed the

lowest levels.

Stressful life event categories

The reported stressful life event categories varied widely in

prevalence from 59.7% for the death of someone close

down to 5.7% for worries about sexual orientation. On

average, 3.6 categories were reported with female students

Table 1 Psychological characteristics, reported stressful life events and self-harm history in past year by gender

All Male Female P-value Partial g2

Mean (SD) Mean (SD) Mean (SD)

Psychological

characteristics

Depression 4.4 (3.3) 4.4 (3.3) 4.3 (3.3) Not sig. \0.01 Anxiety 6.9 (4.0) 6.1 (3.8) 7.8 (4.1) \0.001 0.04 Impulsivity 13.9 (2.9) 13.8 (2.9) 14.1 (2.9) \0.001 \0.01 Self-esteem 22.4 (3.9) 23.1 (3.8) 21.7 (4.0) \0.001 0.03

Life event category Number of life event categories 3.6 (2.1) 3.3 (2.0) 3.9 (2.1) \0.001 0.02 n (%) n (%) n (%) P-value /

Death of someone close 18203 (59.7%) 8776 (56.3%) 9427 (63.4%) \0.001 0.07 Problems with or between parents 16895 (55.4%) 7844 (50.2%) 9051 (60.8%) \0.001 0.11 Serious illness of family or friend 16753 (55.0%) 8232 (52.8%) 8521 (57.3%) \0.001 0.05 Difficulties with friends and peers 16284 (53.3%) 6999 (44.8%) 9285 (62.3%) \0.001 0.18 Problems with schoolwork 14414 (47.5%) 6943 (44.7%) 7471 (50.4%) \0.001 0.06 Suicide/self-harm of others 9279 (30.4%) 3369 (21.6%) 5910 (39.7%) \0.001 0.20 Bullied 6339 (20.9%) 2955 (19.0%) 3384 (22.9%) \0.001 0.05 Trouble with the police 5318 (17.5%) 3751 (24.2%) 1567 (10.6%) \0.001 0.18 Physical or sexual abuse 3164 (10.4%) 1258 (8.0%) 1906 (12.8%) \0.001 0.08 Worries about sexual orientation 1737 (5.7%) 746 (4.8%) 991 (6.7%) \0.001 0.04

Self-harm history No self-harm 23038 (79.6%) 13020 (88.1%) 10018 (70.7%) \0.001 0.22 Self-harm thoughts only 4237 (14.6%) 1365 (9.2%) 2872 (20.3%)

Single self-harm episode 752 (2.6%) 179 (1.2%) 573 (4.0%)

Multiple self-harm episodes 912 (3.2%) 212 (1.4%) 700 (4.9%)

Effect size measured by partial g2 was very small if \0.01, small if \0.06, medium if \0.14 or large if 0.14? Association measured by / was very weak if \0.10, weak if \0.30, moderate if \0.50 and strong if 0.50?

502 Eur Child Adolesc Psychiatry (2011) 20:499–508

123

reporting a higher number than male students. The preva-

lence of each life event category differed by gender at

P \ 0.001. For all but one stressful life event category (trouble with the police), the prevalence was higher among

female students. However, the strength of the association

between prevalence of stressful life events and gender was

weak or very weak (range of Phi statistic = 0.04–0.20).

The prevalence of each stressful life event category varied

by country (all at P \ 0.001), but the strength of all associations was weak or very weak.

Psychological characteristics and self-harm history

in past year

There were highly statistically significant differences

(P \ 0.001) between the self-harm history groups on each of the psychological characteristics (Table 2). The effect

size was small in relation to impulsivity and medium in

relation to depression, anxiety and self-esteem. There was

evidence of a dose–response or graded relationship

whereby the more severe the self-harm history, the higher

the levels of depression, anxiety and impulsivity and the

lower the level of self-esteem. All pairwise comparisons

among the four self-harm history groups differed signifi-

cantly at P \ 0.001 with three exceptions. The self-harm thoughts only group and the single self-harm episode group

did not differ significantly in relation to depression

(P = 0.131) and self-esteem (P = 0.477) and only differed

marginally in relation to anxiety (P = 0.012).

Figure 1 illustrates the consistency of the dose–

response association between the four psychological

characteristics and self-harm history when examined for

each of the seven countries. The stepped increase in

depression, anxiety and impulsivity and decrease in self-

esteem was evident with increasing self-harm history.

However, in most countries, there was limited or no

difference in level of depression, anxiety and self-esteem

between the adolescents who only thought of self-harm-

ing but did not act and those who engaged in a single

self-harm episode. There was no evidence of gender

modifying the association between the psychological

measures and self-harm history.

Stressful life event categories and self-harm history

in past year

There were highly statistically significant (P \ 0.001) asso- ciations between the reporting of each life event category and

self-harm history in the past year (Table 3). In every case, the

prevalence increased across the groups with increasing self-

harm history. The life event category most strongly related to

self-harm history was experiencing the suicide or self-harm of

others followed by physical or sexual abuse, difficulties with

friends or peers and problems with or between parents. There

was also evidence of a strong dose–response relationship as

the average number of event categories reported varied from

3.1 for adolescents with no self-harm in the past year to 4.8 for

those with self-harm thoughts only to 5.5 for those with a

single self-harm episode and 6.2 for multiple self-harmers.

Figure 1 illustrates this dose–response association.

Gender modified the association between three of the life

event categories (death of someone close (P = 0.001), bullied

(P = 0.013) and worries about sexual orientation (P \ 0.001)) and self-harm history. Country modified the association

between six of the life event categories (difficulties with

friends and peers (P = 0.002), problems with schoolwork

(P = 0.025), suicide/self-harm of others (P \ 0.001), bullied (P \ 0.001), physical or sexual abuse (P \ 0.001) and worries about sexual orientation (P \ 0.001)) and self-harm history. However, none of these interaction effects explained more than

an additional 0.2% of the variation in self-harm history.

Independent associations between gender,

psychological characteristics and stressful life event

categories and self-harm history in past year

Table 4 details the results of the multivariate multinomial

logistic regression analysis carried out to identify the fac-

tors independently distinguishing between adolescents in

the four self-harm history groups. The derived regression

Table 2 Psychological characteristics by self-harm history in past year

No self-harm

Mean (SD)

Self-harm thoughts only

Mean (SD)

Single self-harm episode

Mean (SD)

Multiple self-harm episodes

Mean (SD)

P-value Partial g2

Depression 3.9 (3.0) 5.6 (3.3) 5.9 (3.6) 7.4 (4.2) \0.001 0.08 Anxiety 6.1 (3.7) 9.5 (3.8) 9.9 (4.1) 11.4 (4.2) \0.001 0.11 Impulsivity 13.6 (2.8) 14.8 (3.0) 15.3 (3.1) 15.9 (3.2) \0.001 0.04 Self-esteem 23.1 (3.7) 20.3 (3.7) 20.1 (4.0) 18.4 (4.2) \0.001 0.09

Effect size measured by partial g2 was very small if \0.01, small if \0.06, medium if \0.14 or large if 0.14? All pairwise comparisons among the four self-harm history groups differed significantly at P \ 0.001 with the exception of the comparison between the self-harm thoughts only group and the single self-harm episode group in relation to depression (P = 0.131), anxiety (P = 0.012) and self-esteem (P = 0.477)

Eur Child Adolesc Psychiatry (2011) 20:499–508 503

123

model explained 37.7% of the variation in self-harm his-

tory. Some interactions between the psychological mea-

sures with gender and with country reached statistical

significance but they contributed very little (0.2–0.8%) to

explaining variation in self-harm history.

Female gender strongly distinguished self-harm ideators

and single and multiple self-harmers from those with no

self-harm history. Female gender also distinguished ado-

lescents with a single self-harm episode from those with

self-harm thoughts only (P \ 0.001) but did not distinguish between single and multiple self-harmers (P = 0.523).

Each of the four psychological characteristics independently

contributed to distinguishing between the self-harm groups.

Only impulsivity distinguished single-episode self-harmers

from self-harm ideators (P = 0.005), whereas self-esteem

(P \ 0.001) and depression (P = 0.002) differentiated multi- ple self-harmers from those with a single episode.

All but one (death of someone close) of the ten stressful life

event categories was independently associated with self-harm

history (Table 4). Experience of the suicide/self-harm of

others strongly distinguished adolescents with a single self-

harm episode from those with self-harm thoughts only

(P \ 0.001) and also distinguished multiple- from single- episode self-harmers (P = 0.015). Physical or sexual abuse

(P \ 0.001) and to a lesser extent worries about sexual ori- entation (P = 0.017) differentiated between adolescents with

a single self-harm episode from those with self-harm thoughts

only, while trouble with the police was associated with mul-

tiple self-harmers more than it was with single-episode self-

harmers (P = 0.019).

0

1

2

3

4

5

6

7

8

9

10

No self-harm Self-harm thoughts only Single self-harm episode

Multiple self-harm episodes

D e p re

ss io

n s

co re

Hungary

England

Ireland

Australia

Netherlands

Belgium

Norway

0

2

4

6

8

10

12

14

No self-harm Self-harm thoughts only Single self-harm episode

Multiple self-harm episodes

A n xi

e ty

s co

re

Hungary

England

Ireland

Australia

Netherlands

Belgium

Norway

0

2

4

6

8

10

12

14

16

18

No self-harm Self-harm thoughts only Single self-harm episode

Multiple self-harm episodes

Im p

u ls

iv ity

s co

re

Hungary

England

Ireland

Australia

Netherlands

Belgium

Norway

0

2

4

6

8

10

12

14

16

18

20

22

24

26

No self-harm Self-harm thoughts only Single self-harm episode

Multiple self-harm episodes

S e

lf e

st e

e m

s co

re

Hungary

England

Ireland

Australia

Netherlands

Belgium

Norway

0

1

2

3

4

5

6

7

No self-harm Self-harm thoughts only Single self-harm episode

Multiple self-harm episodes

N u

m b

e r

o f

st re

ss fu

l l ife

e ve

n t

Hungary

England

Ireland

Australia

Netherlands

Belgium

Norway

ca te

g o

ri e

s e

xp e

ri e

n ce

d

Fig. 1 Mean depression, anxiety, impulsivity, self-esteem scores and mean number of stressful life event categories by self-harm history in past year for each country

504 Eur Child Adolesc Psychiatry (2011) 20:499–508

123

Discussion

An earlier paper from the CASE Study [28] addressed

socio-demographic and self-harm characteristics for the

same school-based sample of adolescents. Here, we have

focused in more depth on psychological characteristics and

stressful life events. In particular, we have reported evi-

dence for a dose–response relationship whereby increased

severity of self-harm history was, in general, associated

with higher levels of depression, anxiety and impulsivity

and lower levels of self-esteem, as well as stressful life

events in more areas of young people’s lives. These pat-

terns were consistent across gender and country.

A particularly interesting finding is that thoughts of self-

harm were not always distinguishable from a single self-

harm episode in terms of association with psychological

characteristics and stressful life events. It emerged, how-

ever, that impulsivity and experiencing the suicide or self-

harm of others, physical or sexual abuse and worries about

sexual orientation were the only factors that independently

differentiated single-episode self-harmers from adolescents

with self-harm thoughts only. That few factors distin-

guished between ideators and self-harmers is important for

future research and for prevention. The findings are also in

line with a recently presented model of suicidal behaviour

which maps the relationship between background factors

and trigger events, and the development of suicidal idea-

tion/intent through to suicidal behaviour [34].

The strong associations between psychological charac-

teristics and self-harm on the one hand and stressful life

events and self-harm on the other raise the question of the

associations between psychological health, stressful life

events and self-harm, and the relative significance of each

set of factors. Multivariate models were constructed to

investigate this study objective. Despite some specific

differences, the effects of psychological characteristics and

stressful life events remained significant within the popu-

lation overall and continued to remain significant when

gender and country were considered separately. Our find-

ings of independent associations between four psycholog-

ical characteristics and nine life event categories with

self-harm history support the conclusion of a systematic

review showing that a wide range of factors is linked with

self-harm behaviour [14]. It also confirms the heterogeneity

of young self-harmers among community samples, with

differing frequencies of self-harm and varying levels of

associated difficulties [43].

In any event, it would seem unlikely that a single chain

of events links life events, psychological health and

deliberate self-harm. On the one hand, it is clear that stress

can precede mental health difficulties [36, 38]. On the

other, however, it is apparent that depression can trigger

psychosocial difficulties through poor interpersonal func-

tioning and impaired relationships [42] and that self-harm

or suicidal behaviour may follow life stress in the absence

of depression and hopelessness [2, 29]. Furthermore, self-

harm itself may act as a trigger for depression [17].

Some differences between countries were found. There

was, for instance, some variation in levels of self-harm

behaviour as well as differences in psychological

Table 3 Prevalence of stressful life event categories by self-harm history in past year

No self-harm Self-harm thoughts only Single self-harm

episode

Multiple self-harm

episodes

/a

n (%) n (%) n (%) n (%)

Death of someone close 13249 (57.5%) 2778 (65.6%) 519 (69.7%) 677 (74.2%) 0.09

Problems with or between parents 11263 (49.0%) 3179 (75.2%) 623 (83.2%) 792 (87.3%) 0.24

Serious illness of family or friend 11716 (51.0%) 2838 (67.1%) 520 (69.8%) 685 (75.5%) 0.15

Difficulties with friends and peers 10739 (46.7%) 3180 (75.1%) 599 (79.7%) 776 (85.2%) 0.25

Problems with schoolwork 9573 (41.8%) 2716 (64.7%) 531 (71.5%) 694 (76.5%) 0.21

Suicide/self-harm of others 5281 (22.9%) 2108 (49.8%) 525 (69.9%) 722 (79.3%) 0.32

Bullied 4059 (17.8%) 1267 (30.2%) 263 (35.3%) 394 (43.9%) 0.16

Trouble with the police 3546 (15.5%) 831 (19.7%) 191 (25.7%) 314 (34.6%) 0.10

Physical or sexual abuse 1474 (6.4%) 808 (19.1%) 233 (31.0%) 352 (38.7%) 0.26

Worries about sexual orientation 882 (3.8%) 428 (10.2%) 120 (16.0%) 182 (20.2%) 0.17

Number of life event categories b

3.1 (1.9) 4.8 (1.9) 5.5 (1.9) 6.2 (1.8) 0.16

All associations statistically significant at P \ 0.001 a

Association measured by Phi was very weak if \0.10, weak if \0.30, moderate if \0.50 and strong if 0.50? b

Mean (standard deviation) reported with partial g2 as the effect size measure. Effect size measured by partial g2 was very small if \0.01, small if \0.06, medium if \0.14 or large if 0.14?. All post hoc pairwise comparisons among the four self-harm history groups were statistically significant at P \ 0.001

Eur Child Adolesc Psychiatry (2011) 20:499–508 505

123

T a

b le

4 M

u lt

in o

m ia

l lo

g is

ti c

re g

re ss

io n

a n

a ly

si s

o f

th e

in d

e p

e n

d e n

t a ss

o c ia

ti o

n s

b e tw

e e n

g e n

d e r,

p sy

c h

o lo

g ic

a l

c h

a ra

c te

ri st

ic s

a n

d re

p o

rt e d

st re

ss fu

l li

fe e v

e n

t c a te

g o

ri e s

a n

d se

lf -h

a rm

h is

to ry

in p

a st

y e a r

S e lf

-h a rm

th o

u g

h ts

o n

ly a

S in

g le

se lf

-h a rm

e p

is o

d e

a M

u lt

ip le

se lf

-h a rm

e p

is o

d e s

a G

ro u

p

d if

fe re

n c e s

b

O R

c (9

5 %

C I)

O R

c (9

5 %

C I)

O R

c (9

5 %

C I)

G e n

d e r

F e m

a le

(r e f.

g ro

u p

: M

a le

) 1

.9 4

(1 .7

8 -2

.1 1

) 2

.7 6

(2 .2

8 -3

.3 5

) 3

.0 0

(2 .4

8 -3

.6 4

) A \

B \

C =

D

P sy

c h

o lo

g ic

a l

c h

a ra

c te

ri st

ic s

d D

e p

re ss

io n

1 .0

2 *

* (1

.0 1

-1 .0

4 )

1 .0

4 *

* (1

.0 1

-1 .0

7 )

1 .1

0 (1

.0 7

-1 .1

3 )

A \

B =

C \

D

A n

x ie

ty 1

.1 1

(1 .0

9 -1

.1 2

) 1

.1 0

(1 .0

7 -1

.1 3

) 1

.1 3

(1 .1

1 -1

.1 6

) A \

B =

C =

D

Im p

u ls

iv it

y 1

.0 6

(1 .0

5 -1

.0 8

) 1

.1 0

(1 .0

7 -1

.1 3

) 1

.1 3

(1 .1

0 -1

.1 6

) A \

B \

C =

D

S e lf

e st

e e m

0 .8

9 (0

.8 8

-0 .9

0 )

0 .9

0 (0

.8 7

-0 .9

2 )

0 .8

4 (0

.8 2

-0 .8

6 )

A [

B =

C [

D

L if

e e v

e n

t c a te

g o

ry (r

e f.

g ro

u p

:

th o

se n

o t

re p

o rt

in g

a n

e v

e n

t

in th

e li

fe e v

e n

t c a te

g o

ry )

P ro

b le

m s

w it

h o

r b

e tw

e e n

p a re

n ts

1 .6

4 (1

.5 0

-1 .7

9 )

1 .9

5 (1

.5 8

-2 .4

1 )

2 .1

2 (1

.6 9

-2 .6

5 )

A \

B =

C =

D

S e ri

o u

s il

ln e ss

o f

fa m

il y

o r

fr ie

n d

1 .2

0 (1

.1 0

-1 .3

0 )

1 .1

1 n s

(0 .9

3 -1

.3 2

) 1

.1 9

n s

(0 .9

9 -1

.4 2

) A \

B

D if

fi c u

lt ie

s w

it h

fr ie

n d

s a n

d p

e e rs

1 .6

7 (1

.5 3

-1 .8

2 )

1 .5

4 (1

.2 6

-1 .8

8 )

1 .7

1 (1

.3 8

-2 .1

2 )

A \

B =

C =

D

P ro

b le

m s

w it

h sc

h o

o lw

o rk

1 .4

1 (1

.3 0

-1 .5

4 )

1 .5

4 (1

.2 9

-1 .8

5 )

1 .4

5 (1

.2 1

-1 .7

4 )

A \

B =

C =

D

S u

ic id

e /s

e lf

-h a rm

o f

o th

e rs

1 .8

9 (1

.7 4

-2 .0

4 )

3 .6

9 (3

.0 9

-4 .4

0 )

5 .0

0 (4

.1 5

-6 .0

2 )

A \

B \

C \

D

B u

ll ie

d 1

.1 9

(1 .0

8 -1

.3 0

) 1

.2 5

* (1

.0 5

-1 .4

8 )

1 .4

1 (1

.1 9

-1 .6

6 )

A \

B =

C =

D

T ro

u b

le w

it h

th e

p o

li c e

1 .1

5 *

* (1

.0 3

-1 .2

7 )

1 .3

9 (1

.1 5

-1 .7

0 )

1 .8

6 (1

.5 5

-2 .2

3 )

A \

B =

C \

D

P h

y si

c a l

o r

se x

u a l

a b

u se

1 .4

9 (1

.3 3

-1 .6

7 )

2 .3

0 (1

.9 0

-2 .7

8 )

2 .2

9 (1

.9 1

-2 .7

4 )

A \

B \

C =

D

W o

rr ie

s a b

o u

t se

x u

a l

o ri

e n

ta ti

o n

1 .4

0 (1

.2 1

-1 .6

2 )

1 .8

6 (1

.4 7

-2 .3

6 )

1 .8

3 (1

.4 6

-2 .2

9 )

A \

B \

C =

D

a N

o se

lf -h

a rm

w a s

th e

c o

m p

a ri

so n

c a te

g o

ry o

f th

e d

e p

e n

d e n

t v

a ri

a b

le b

G ro

u p

d if

fe re

n c e s

in d

ic a te

th e

e x

te n

t to

w h

ic h

th e

in d

e p

e n

d e n

t v

a ri

a b

le s

d is

ti n

g u

is h

b e tw

e e n

th e

se lf

-h a rm

g ro

u p

s (A

= N

o se

lf -h

a rm

; B

= se

lf -h

a rm

th o

u g

h ts

o n

ly , C

= S

in g

le se

lf -h

a rm

e p

is o

d e ,

D =

M u

lt ip

le se

lf -h

a rm

e p

is o

d e s)

c T

h e

o d

d s

ra ti

o s

w e re

a d

ju st

e d

fo r

a g

e a n

d c o

u n

tr y

a s

w e ll

a s

th e

fa c to

rs d

e ta

il e d

in th

e ta

b le

. A

ll a ss

o c ia

ti o

n s

ta b

u la

te d

w e re

st a ti

st ic

a ll

y si

g n

ifi c a n

t a t

p \

0 .0

0 1

e x

c e p

t w

h e re

in d

ic a te

d

(n s=

n o

t si

g n

ifi c a n

t; *

p \

0 .0

5 ;

* *

p \

0 .0

1 )

d T

h e se

o d

d s

ra ti

o s

re p

re se

n t

th e

e ff

e c t

o f

a o

n e -p

o in

t in

c re

a se

in th

e sc

o re

o f

th e

sc a le

m e a su

ri n

g th

e p

sy c h

o lo

g ic

a l

c h

a ra

c te

ri st

ic

506 Eur Child Adolesc Psychiatry (2011) 20:499–508

123

characteristics: in this latter respect, Hungarian young peo-

ple reported the highest levels of difficulty, while Norwe-

gians reported the lowest. Despite these differences,

however, the dose–response relationships between psycho-

logical characteristics and history of self-harm, and between

stressful life events and self-harm, were maintained in all

countries. This suggests the universality of the findings.

These findings underline the complexity involved in

specifying the precursors of self-harm among the young.

Although both psychological characteristics and negative

life events are associated with increased levels of self-harm

within the population of young people as a whole, varying

constellations of interrelated factors contribute to individ-

ual risk. Recent research on identifying subgroups of young

self-harmers [43] appears promising, and it is suggested

that knowledge on the aetiology of deliberate self-harm

may develop best in the context of further large-scale

longitudinal research that can look at distinct groups of

young self-harmers over time.

The strengths of this study are its large representative

school-based sample, the clear definition of self-harm and

the standardised methodology used across seven countries.

Whilst we relied on retrospective self-reported information,

and thus were unable to look conclusively at temporal

associations, most of our findings have relied on episodes

of self-harm reported for the previous year, rather than over

the lifetime, to increase accuracy. The cross-sectional

nature of the study, nonetheless, implies that the nature of

the relationships between self-harm, psychological char-

acteristics and life events that we have identified cannot be

assumed to be causal. Additionally, there may have been

other potentially relevant factors, such as substance or

alcohol abuse, that we do not consider. A further limitation

of the study is that we cannot guarantee the representative

nature of all national samples despite attempts to ensure

this was the case. Treatment of life events presents another

possible methodological difficulty in that categories of

stressful events cannot be equated or added. Nonetheless,

the fact that all but one stressful life event category showed

statistically significant associations with a history of self-

harm increases our confidence in the validity of these

measures. Further research exploring the impact of differ-

ent life events, and taking account of severity, occurrence

and timing, could throw more light on this issue.

In clinical terms, the evidence for a dose–response

relationship linking both psychological characteristics and

negative life events with the frequency of self-harm

underlines the need for positive mental health promotion

and early identification of adolescents at risk of self-harm

[24, 33]. Similarities found between adolescents who have

merely thought about harming themselves, and those

reporting a single episode, suggest the salience of taking

intentions as well as behaviour into account.

Conclusion

There is no single pattern of self-harm among young

people, but both psychological characteristics and stressful

life events substantially increase risk. Those developing

prevention and intervention programmes must remain

‘open minded’ to patient characteristics and not neglect

either those who have only thought of harming themselves

or, despite current practice [45], those who do not have

evident signs of depression or mental illness.

Acknowledgments The authors thank the European Commission Daphne Programme for providing funding for the coordination of this

research, as well as the Community Fund in England, the Irish

National Suicide Review Group, the National Office for Suicide

Prevention and the Ireland Funds, QLD-Health and the Common-

wealth of Australia, Department of Health and Ageing, and the

Norwegian Foundation for Health and Rehabilitation, for funding

national studies. We also acknowledge colleagues in the national

centres who contributed to the work undertaken, including Udo

Reulbach and Tony Fitzgerald for statistical advice, and pupils and

schools that participated in the study.

Conflict of interest The authors declare that they have no conflict of interest.

References

1. Andrews JA, Lewinsohn PM (1992) Suicidal attempts among

older adolescents: prevalence and co-occurrence with psychiatric

disorders. J Am Acad Child Adolesc Psychiatry 31:655–662

2. Baldry AC, Winkel WF (2003) Direct and vicarious victimization

at school and at home as risk factors for suicidal cognition among

Italian adolescents. J Adolesc 26:703–716

3. Bearman PS, Moody J (2004) Adolescent suicidality. Am J Pub

Health 94:89–95

4. Bensley LS, Van Eenwyk J, Spieker SJ, Schoder MN (1999) Self-

reported abuse history and adolescent problem behaviours. I:

antisocial and suicidal behaviours. J Adolesc Health 24:163–172

5. Brent D, Mann JJ (2006) Familial pathways to suicidal behavior.

N Engl J Med 355:2719–2721

6. Bridge J, Goldstein T, Brent D (2006) Adolescent suicide and

suicidal behaviour. J Child Psychol Psychiatry 47:372–394

7. Byrne S, Morgan S, Fitzpatrick C, Boylan C, Crowley S, Gahan

H, Howley J, Staunton D, Guerin S (2008) Deliberate self-harm

in children and adolescents: a qualitative study exploring the

needs of parents and carers. Clin Child Psychol Psychiatry

13:493–504

8. Coggan C, Bennett S, Hooper R, Dickinson P (2003) Association

between bullying and mental health status in New Zealand ado-

lescents. Int J Mental Health Promotion 5:16–22

9. Cohen J (1988) Statistical power analysis for the behavioural

sciences. Erlbaum, Hillsdale, NJ

10. Conner KR, Meldrum S, Wieczorek WF, Duberstein PR, Welte

JW (2004) The association of irritability and impulsivity with

suicidal ideation among 15- to 20-year-old males. Suicide Life

Threat Behav 34:363–373

11. Dimmock M, Grieves S, Place M (2008) Young people who cut

themselves—a growing challenge for educational settings. Br J

Spec Edu 35:42–48

Eur Child Adolesc Psychiatry (2011) 20:499–508 507

123

12. Eisenberg ME, Resnick MD (2006) Suicidality among gay, les-

bian and bisexual youth: the role of protective factors. J Adolesc

Health 39:662–668

13. Esposito C, Spirito A, Boergers J, Donaldson D (2003) Affective,

behavioural, and cognitive functioning in adolescents with mul-

tiple suicide attempts. Suicide Life Threat Behav 33:389–399

14. Evans E, Hawton K, Rodham K (2004) Factors associated with

suicidal phenomena in adolescents: a systematic review of pop-

ulation-based studies. Clin Psychol Rev 24:957–979

15. Evans E, Hawton K, Rodham K (2005) Suicidal phenomena and

abuse in adolescents: a review of epidemiological studies. Child

Abuse Negl 29:45–58

16. Fergusson DM, Horwood LJ, Ridder EM, Beautrais AL (2005)

Suicidal behaviour in adolescence and subsequent mental health

outcomes in young adulthood. Psychol Med 35:983–993

17. Flisher AJ (1999) Annotation: mood disorder in suicidal children

and adolescents: recent developments. J Child Psychol Psychiatry

40:315–324

18. Foley D, Goldston D, Costello E, Angold A (2006) Proximal

psychiatric risk factors for suicidality in youth. Arch Gen Psy-

chiatry 63:1017–1024

19. Harrington R, Pickles A, Aglan A, Harrington V, Burroughs H,

Kerfoot M (2006) Early adult outcomes of adolescents who

deliberately poisoned themselves. J Am Acad Child Adolesc

Psychiatry 45:337–345

20. Hawton K, Harriss L (2008) Deliberate self-harm by under-15-

year-olds: characteristics, trends and outcome. J Child Psychol

Psychiatry 49:441–448

21. Hawton K, Cole D, O’Grady J, Osborn M (1982) Motivational

aspects of deliberate self-poisoning in adolescents. Br J Psychi-

atry 141:286–291

22. Hawton K, Kingsbury S, Steinhardt K, James A, Fagg J (1999)

Repetition of deliberate self-harm by adolescents: the role of

psychological factors. J Adolesc 22:369–378

23. Hawton K, Hall S, Simkin S, Bale L, Bond A, Codd S, Stewart A

(2003) Deliberate self-harm in adolescents: a study of charac-

teristics and trends in Oxford, 1990–2000. J Child Psychol Psy-

chiatry 44:1191–1198

24. Horowitz LM, Ballard ED (2009) Suicide screening in schools,

primary care and emergency departments. Curr Opin Pediatr

21:620–627

25. Hull-Blanks EA, Kerr BA, Robinson Kurpius SE (2004) Risk

factors of suicidal ideations and attempts in talented, at-risk girls.

Suicide Life Threat Behav 34:267–276

26. Kerfoot M, Dyer E, Harrington V, Woodham A, Harrington R

(1996) Correlates and short-term course of self-poisoning in

adolescents. Br J Psychiatry 168:38–42

27. Lieb R, Bronisch T, Hofler M et al (2005) Maternal suicidality

and risk of suicidality in offspring: findings from a community

study. Am J Psychiatry 162:1665–1671

28. Madge N, Hewitt A, Hawton K, de Wilde EJ, Corcoran P, Fekete

S, van Heeringen K, De Leo D, Ystgaard M (2008) Deliberate

self-harm within an international community sample of young

people: findings from the Child & Adolescent Self-harm in

Europe (CASE) Study. J Child Psychol Psychiatry 49:667–677

29. Martin G, Bergen HA, Richardson AS, Roeger L, Allison S

(2004) Sexual abuse and suicidality: gender differences in a large

community sample of adolescents. Child Abuse Negl 28:491–503

30. Martin G, Richardson AS, Bergen HA, Roeger L, Allison S

(2005) Perceived academic performance, self-esteem and locus of

control as indicators of need for assessment of adolescent suicide

risk: implications for teachers. J Adolesc 28:75–87

31. McDonald G, O’Brien L, Jackson D (2007) Guilt and shame:

experiences of parents of self-harming adolescents. J Child

Health Care 11:298–310

32. Melhem NM, Day N, Shear MK, Day R, Reynolds CF, Brent D

(2004) Traumatic grief among adolescents exposed to a peer’s

suicide. Am J Psychiatry 161:1411–1416

33. Naylor PB, Cowie HA, Walters SJ, Talamelli L, Dawkins J

(2009) Impact of a mental health teaching programme on ado-

lescents. Br J Psychiatry 194:365–370

34. O’Connor RC (2011) Toward an integrated motivational-voli-

tional model of suicidal behaviour. In: O’Connor RC, Platt S,

Gordon J (eds) International handbook of suicide prevention:

research, policy and practice, 1st edn. John Wiley & Sons Ltd.,

London, pp 181–198

35. Olfson M, Gameroff MJ, Marcus SC, Greenberg T, Shaffer D

(2005) Emergency treatment of young people following delib-

erate self-harm. Arch Gen Psychiatry 62:1122–1128

36. Pelkonen M, Marttunen M, Kaprio J, Huurre T, Aro H (2008)

Adolescent risk factors for episodic and persistent depression in

adulthood. A 16-year prospective follow-up study of adolescents.

J Affect Disord 106:123–131

37. Plutchik R, Van Praag HM (1986) The measurement of suici-

dality, aggressivity and impulsivity. Clin Neuropharmacol

9:380–382

38. Reinherz H, Tanner J, Berger S et al (2006) Adolescent suicidal

ideation as predictive of psychopathology, suicidal behavior, and

compromised functioning at age 30. Am J Psychiatry

163:1126–1232

39. Remafedi G, French S, Story M, Resnick M, Blum R (1998) The

relationship between suicide risk and sexual orientation: results of

a population-based study. Am J Public Health 88:57–60

40. Robson P (1989) Development of a new self-report questionnaire

to measure self-esteem. Psychol Med 19:513–518

41. Ross S, Heath NL (2003) Two models of adolescent self-muti-

lation. Suicide Life Threat Behav 33:277–287

42. Rossow I, Ystgaard M, Hawton K, Madge N, van Heeringen K,

de Wilde EJ, De Leo D, Fekete S, Morey C (2007) Cross-national

comparisons of the association between alcohol consumption and

deliberate self-harm in adolescents. Suicide Life Threat Behav

37:605–615

43. Stanford S, Jones MP (2009) Psychological subtyping finds

pathological, impulsive, and ‘normal’ groups among adolescents

who self-harm. J Child Psychol Psychiatry 50:807–815

44. Steinhausen H-C, Haslimeier C, Winkler Metzke C (2006) The

outcome of episodic versus persistent adolescent depression in

young adulthood. J Affect Disord 96:49–57

45. Storey P, Hurry J, Jowitt S, Owens D, House A (2005) Supporting

young people who repeatedly self-harm. J Royal Soc Promotion

Health 125:71–75

46. Wild LG, Flisher AJ, Lombard C (2004) Suicidal ideation and

attempts in adolescents: associations with depression and six

domains of self-esteem. J Adolesc 27:611–624

47. Ystgaard M, Arensman E, Hawton K, Madge N, van Heeringen

K, Hewitt A, de Wilde EJ, De Leo D, Fekete S (2009) Deliberate

self-harm in adolescents: comparison between those who receive

help following self-harm and those who do not. J Adolesc

32:875–891

48. Ystgaard M, Hestetun I, Loeb M, Mehlum L (2004) Is there a

specific relationship between chilhood sexual and physical abuse

and repeated suicidal behavior? Child Abuse Negl 28:863–876

49. Zigmond AS, Snaith RP (1983) The hospital anxiety and

depression scale. Acta Psychiatr Scand 67:361–370

508 Eur Child Adolesc Psychiatry (2011) 20:499–508

123

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.