An international comparison of adolescent
non-suicidal self-injury (NSSI) and suicide attempts:
Germany and the USA
P. L. Plener
1
*, G. Libal
2
, F. Keller
1
, J. M. Fegert
1
and J. J. Muehlenkamp
3
1
Department of Child and Adolescent Psychiatry and Psychotherapy, University of Ulm, Germany
2
Department of Child and Adolescent Psychiatry, University of Basel, Switzerland
3
Department of Psychology, University of North Dakota, USA
Background. This study examined the prevalence of non-suicidal self-injury (NSSI), suicide attempts, suicide threats
and suicidal ideation in a German school sample and compared the rates with a similar sample of adolescents from
the midwestern USA by using cross-nationally validated assessment tools.
Method. Data were provided from 665 adolescents (mean age 14.8 years,
S.D.=0.66, range 14–17 years) in a school
setting. Students completed the Self-Harm Behavior Questionnaire (SHBQ), the Ottawa Self-Injury Inventory (OSI)
and a German version of the Center for Epidemiological Studies–Depression Scale (CES-D).
Results. A quarter of the participants (25.6 %) endorsed at least one act of NSSI in their life, and 9.5 % of those
students answered that they had hurt themselves repetitively (more than four times). Forty-three (6.5 %) of the
students reported a history of a suicide attempt. No statistically significant differences were observed between the
German and US samples in terms of self-injury or suicidal behaviors.
Conclusions. By using the same validated assessment tools, no differences were found in the prevalence and
characteristics of self-injury and suicidal behaviors between adolescents from Germany and the USA. Thus, it seems
that NSSI has to be understood as worldwide phenomenon, at least in Western cultures.
Received 30 May 2008 ; Revised 19 November 2008; Accepted 11 December 2008; First published online 27 January 2009
Key words : Adolescents, deliberate self-harm (DSH), non-suicidal self-injury (NSSI), prevalence, self-injuring behavior
(SIB).
Introduction
Non-suicidal self-injury (NSSI) represents the direct,
repetitive, intentional injury of one’s own body tissue,
without suicidal intent, that is not socially accepted
(Lloyd-Richardson et al. 2007). To date, the study of
NSSI, especially in relation to its differences from sui-
cidal behavior, has been difficult because of the em-
pirical and conceptual confounding of the variables.
This conflation of suicidal and NSSI behaviors is par-
ticularly true within European countries, where the
construct ‘ deliberate self-harm ’ (DSH) is used as an
umbrella term for self-destructive behaviors regard-
less of suicidal intent (Hawton et al. 2007; Madge et al.
2008). Such general terms make cross-cultural com-
parisons of NSSI difficult, thus hampering research
in this area. A considerable amount of literature has
been published concerning the need to differentiate
between self-injuring behavior (SIB) undertaken with-
out suicidal intent and suicidal behaviors undertaken
with the intent to die (Nock & Kessler, 2006; Posner
et al. 2007 ; Silverman et al. 2007). The relationship
between these entities is still subject to ongoing re-
search and it is necessary to fully understand the
worldwide phenomenon of both sets of behaviors.
Despite the need for a clear nomenclature, it should
not be overlooked that NSSI needs to be understood
as a potential risk factor for future suicide attempts.
A few recent studies have documented the complex
relationship between NSSI and suicide risk. Whitlock
& Knox (2007) presented data from an internet survey
(n=2875, age range 18–24 years) showing that rates of
NSSI were positively correlated with risk for suicidal
behavior, meaning that those who injure themselves
repetitively were at an increased risk for also having
made a suicide attempt. Comparable results were re-
ported by Nock et al. (2006) from an adolescent in-
patient sample (n=89). Suicide attempts were more
common in adolescents with repetitive self-injuries,
* Address for correspondence : Dr P. L. Plener, Department of Child
and Adolescent Psychiatry and Psychotherapy, University of Ulm,
Steinhoevelstr. 5, D-89075 Ulm, Germany.
(Email : paul.plener@uniklinik-ulm.de)
Psychological Medicine (2009), 39, 1549–1558. f 2009 Cambridge University Press
doi:10.1017/S0033291708005114 Printed in the United Kingdom
ORIGINAL ARTICLE
https:/www.cambridge.org/core/terms. https://doi.org/10.1017/S0033291708005114
Downloaded from https:/www.cambridge.org/core. University of Basel Library, on 30 May 2017 at 17:00:38, subject to the Cambridge Core terms of use, available at
who injured themselves for a longer period, used dif-
ferent methods of self-injury, and reported feeling no
pain during the act of self-injury. Both studies are in
accordance with Joiner’s (2005) theory that self-injury
desensitizes individuals and can increase their risk
of later suicide as they habituate to fear and pain. As
Whitlock & Knox (2007) proposed, NSSI should be
understood as a signal that the individual engaging in
the NSSI is under psychological stress, which could
increase risk for suicide attempts.
Prevalence rates
Recent reports have shown high rates of NSSI in the
USA, ranging between 23% and 38% in community
samples of adolescents and young adults (Ross &
Heath, 2002; Gratz, 2006; Whitlock et al. 2006 ; Lloyd-
Richardson et al. 2007; Muehlenkamp & Gutierrez,
2007), and a cross-sectional study identified a lifetime
course of NSSI with high rates in adolescence and de-
clining rates in young adulthood (Young et al. 2007).
Thus, there seems to be consensus that adolescence is
an important period in which to study NSSI. Although
there are plenty of study-specific data on the preva-
lence of NSSI in adolescents from the USA, Canada,
Australia and the UK (Patton et al. 1997 ; Hawton
et al. 2002 ; Muehlenkamp & Gutierrez, 2004 ; Nada-
Raja et al. 2004; Skegg et al. 2004; Laye-Gindhu &
Schonert-Reichel, 2005), data on adolescent NSSI are
scarce from other Western countries. High rates of
NSSI have been found in Turkey, where Zoroglu
et al. (2003) reported a lifetime prevalence rate of
21.4% among 839 students. A prevalence rate of 5.5 %
was reported among a sample of Hungarian ado-
lescents (Csorba et al. 2005), and a rate of 24% was
found among young female adults from Italy (Favaro
et al. 2007). Rodham et al. (2004) reported a rate of 3.7%
in their British school sample of 6020 adolescents
(age range 15–16 years). A Scandinavian study com-
paring rates of DSH ideation and acts at 12 and 15
years of age reported that, at age 12, 2.7% of the girls
and 3.9% of the boys described DSH ideation and acts,
whereas at age 15, 12.6% of the girls and 4.6% of the
boys did so (Sourander et al. 2006). Higher rates were
reported more recently by Lundh et al. (2007), who
stated that 65.9% of their Scandinavian adolescent
community sample ( n=123, mean age 15 years) had
deliberately harmed themselves and 13.8% did so re-
petitively.
The variations in rates of NSSI across countries
could be due to cultural influences, but it may be
premature to suggest that cultural variables are to
blame because there are numerous methodological
variations among the existing studies. Different as-
sessment measures and procedures are used across
studies, which are frequently based on different con-
ceptions or definitions of self-injury. For example, in
the study by Lundh et al. (2007), very high rates of
self-injury were reported, and their questionnaire
(the Deliberate Self-harm Inventory – simplified ver-
sion) assessed a range of minor SIBs, such as inter-
ference with wound healing and sticking needles
under the skin. In comparison, the study of Hungarian
adolescents (Csorba et al. 2005) used the Ottawa Self-
Injury Inventory (OSI) as the questionnaire that as-
sessed more severe forms of self-harm within another
time-frame. In addition, although a broad range of
assessment tools do exist, only a few are validated (for
a review see Cloutier & Humphreys, 2008), and even
fewer have been cross-validated for use with samples
from multiple nationalities.
Data on NSSI from Germany have not been avail-
able until recently. The ‘ Heidelberg school study’
(Brunner et al. 2007) assessed DSH within a large ado-
lescent community sample (n=5759, mean age 14.9
year), and a 1-year prevalence of 18.9% was reported.
Four per cent of the participants reported repeti-
tive DSH (more than four times) within the year pre-
ceding the study. Suicide attempts were reported by
7.9% and suicidal ideation by 14.4% of the sample
(Brunner et al. 2007). From a clinical sample of ad-
missions (n=3694) to a German emergency depart-
ment for child and adolescent psychiatry, rates as
high as 57% have been reported using a German as-
sessment instrument, the BaDo (Kirkcaldy et al. 2006).
However, none of the studies from these countries
assessed their data with internationally validated
questionnaires for self-injury. Furthermore, most of
the studies included suicidal behaviors in their defi-
nitions of DSH, preventing accurate comparisons of
NSSI rates to other nations. Results from the latest
study comparing the prevalence of DSH cross-
nationally (Portzky et al. 2008) show that rates of self-
harm can differ significantly between neighboring na-
tions (Belgium and The Netherlands). A method to
make accurate comparisons regarding NSSI is still
needed. No known study has compared rates of NSSI,
using cross-nationally validated assessment scales,
between differing nations.
The primary aim of this study was to address the
assessment limitations of prior international studies of
NSSI by using a cross-nationally validated assessment
tool to : (1) assess the prevalence of NSSI and suicide
attempts in a community sample of German ado-
lescents, and (2) compare German prevalence rates
of NSSI and suicide attempts with a comparable
sample of adolescents from the midwestern USA (i.e.
Muehlenkamp & Gutierrez, 2007) in order to be able to
describe the phenomenon of NSSI in two different
nations.
1550 P. L. Plener et al.
https:/www.cambridge.org/core/terms. https://doi.org/10.1017/S0033291708005114
Downloaded from https:/www.cambridge.org/core. University of Basel Library, on 30 May 2017 at 17:00:38, subject to the Cambridge Core terms of use, available at
Method
Procedures
We chose to assess rates of NSSI in ninth-grade stu-
dents to ensure a comparable sample to the US sample
previously reported on by Muehlenkamp & Gutierrez
(2007), which we used for comparison in the current
study. All schools (with the exception of schools for
the intellectually handicapped and special schools
for students with disruptive behaviors) within the
midsized city Ulm and surrounding rural areas in
southern Germany were asked to participate in the
current study. Out of 47 eligible schools within this
area, 13 agreed to take part. Once a school indicated
agreement to participate, all students in the ninth-
grade classes were personally informed by P.L.P.
about the study, and forms for active parental consent
and adolescent assent were distributed. Students were
informed that the study focus was on rates of NSSI
and suicidal behavior as well as on depressive symp-
toms, and adolescents were reminded that partici-
pation was voluntary and anonymous. Congruent
with German legislation on research ethics in studies
with minors, participants did not receive any com-
pensation for taking part in the study.
All assessment scales were handed out in the class-
room, in sealed envelopes, to those who presented a
signed parental consent and adolescent assent form.
Only students participating in the study were present
in the room. Because of restrictions from the federal
school authorities, no personalized information could
be obtained from students not participating in the
study. The leader of the study (P.L.P.) was present and
available in every classroom to answer any questions.
After students had completed the assessments they
placed their forms back in the envelopes and sealed
them. Each adolescent’s envelope was then collected
by P.L.P. Completing the packet took about 30–45 min.
The study and its procedures were approved by the
school authorities and the Institutional Review Board
of the University of Ulm, Germany.
Ethical issues
Although it has been argued that filling out a ques-
tionnaire concerning suicidal and self-injuring be-
haviors might lead to suicidal impulses, previous
research has not supported these concerns (Gould et al.
2005; Friedman, 2006). Nevertheless, we chose to im-
plement a direct and indirect way of ensuring partici-
pant safety. Every participating student was handed
out a ‘Contact Card’ in a separate envelope that in-
cluded contact details and telephone hours of the
study coordinator, which could be used whenever
the participants felt they needed to talk to somebody.
As an indirect way of getting in contact with the re-
searchers, students were able to fill out a ‘HELP’ card,
which was provided in a separate envelope, by pro-
viding their email address or telephone number if they
wanted to be contacted by the study team. Out of 670
participants, seven (1.04%) chose to do so. One of the
requests turned out to be a hoax, two requested help
with minor ailments (seeking help for problems with
their friends), and four (0.6%) who requested help
with SIB were provided contact to our out-patient de-
partment.
Participants
A total of 1100 ninth-grade students were available
from the participating schools. On the days of assess-
ment 1034 (94%) students were present, and the rest
were missing from school that day. Approximately
half of the 1034 students were female (n=521, 50.4 %);
513 (49.6%) were male. Six hundred and seventy
(64.8%) students were willing and able to participate
in the study as they brought their signed parental
consent and adolescent assent forms with them. All
of these available students returned their assessment
packets. Upon review of the assessment packets,
one had to be excluded as only age and gender were
filled out, and four packets were eliminated because
of obvious nonsense or joke answers (all by male
participants). This led to the inclusion of 665 partici-
pants (57.1%, n=380 female) for the current analyses.
The mean age of participating students was 14.8 years
(
S.D.=0.66, range 14–17 years). The age of non-
participating students could not be assessed because
of regulations of the school authorities.
Measures
Self-Harm Behavior Questionnaire (SHBQ; Gutierrez
et al. 2001)
The SHBQ is a self-report measure that assesses life-
time prevalence of SHB in four sections: NSSI (e.g.
‘Have you ever hurt yourself on purpose ?’), suicide
attempts (e.g. ‘ Have you ever attempted suicide?’),
suicidal threats (e.g. ‘Have you ever threatened to
commit suicide? ’) and suicidal ideations (e.g. ‘Have
you ever talked or thought about committing sui-
cide? ’). Frequency and onset of these behaviors are
assessed by follow-up questions that elicit further de-
tails about the SHB, such as need for medical attention.
The SHBQ has been recommended as a brief screening
measure for NSSI (Cloutier & Humphreys, 2008).
In the original validation study good internal con-
sistency was shown (Cronbach’s a ranging between
0.89 and 0.96 for the four sections). This measure has
been used in community studies of adolescents in
NSSI and suicide attempts 1551
https:/www.cambridge.org/core/terms. https://doi.org/10.1017/S0033291708005114
Downloaded from https:/www.cambridge.org/core. University of Basel Library, on 30 May 2017 at 17:00:38, subject to the Cambridge Core terms of use, available at
the USA (Muehlenkamp & Gutierrez, 2004, 2007),
demonstrating strong psychometric properties in non-
clinical and in-patient adolescent samples (Gutierrez
& Osman, 2008), and has recently been validated for
use within diverse samples of adolescents (Mue-
hlenkamp et al., unpublished observations). The SHBQ
has been translated into German, using a translation
and retranslation procedure, and validated for use
with German samples by Fliege et al. (2006). They
showed a high internal consistency for the German
version (Cronbach’s a ranging between 0.87 and 0.96
for the four sections). It is important to note that in
the German version of the SHBQ, the open-ended
question inquiring about methods used for NSSI were
omitted to ease completion of the questionnaire. This
change proved to be particularly important for the use
of the questionnaire in schools, as students often ar-
ticulated their fear of being recognized (and thus not
able to report truly) by their handwriting. Thus,
methods of NSSI could only be assessed by the sub-
item of the OSI in the German sample in a standard-
ized way.
We chose to follow the approach of Muehlenkamp
& Gutierrez (2007), dividing participants based on
their responses to the SHBQ into four groups : those
with no self-harm (NoSH), those with non-suicidal
self-injury only (NSSI), those with suicide attempts
only (SA) and those with both NSSI and SA (NSSI+
SA).
Ottawa Self-Injury Inventory (OSI; Nixon et al. 2002)
The OSI is a 21-item questionnaire covering in-depth
information on NSSI with regard to 1- and 6-month
prevalence rates of NSSI and suicidal behavior and
also functions, coping strategies and addictive features
(Heath & Nixon, 2008). Although this measure has
been used in a Canadian and Hungarian study (Nixon
et al. 2002; Csorba et al. 2005), it has not been formally
validated with both German and US samples. To ob-
tain qualitative data on methods of NSSI, we used
a translation–retranslation procedure to generate a
German version of the OSI. As we chose to focus on
reporting and comparing prevalence rates based on
measures that have been validated and used in both
the USA and Germany, only data on methods of NSSI
used were taken from the German-translation OSI.
Statistical analysis
Mean differences were evaluated with an analysis
of variance (ANOVA), with group as the between-
subjects factor (NSSI, NoSH, SA, NSSI+SA) or with
t tests in the case of two groups. To detect signifi-
cant pairwise differences, post-hoc contrasts were
performed using the Student–Newman–Keuls test.
Categorical variables were analysed by means of x
2
test for frequency tables. All statistical analyses were
performed with SAS version 9.1.3 (SAS Institute Inc.,
Cary, NC, USA).
Results
German sample: prevalence of NSSI, suicide
attempts, suicidal threats and suicidal ideation
NSSI was reported by 170 (25.6%) of the 665 students
(0 missing) completing the study. Participants were
asked to indicate how often they injured themselves,
and 44 (6.6%) reported doing so only once, 42 (6.3%)
reported having hurt themselves twice, and 21 (3.2 %)
did so three times. Individuals who reported four
or more acts of NSSI comprised 9.5% of the NSSI
group (n=63). Most of the self-injuring participants
reported that they had started self-injury within
1 year prior to filling out the questionnaire ( n=103),
whereas fewer stated that they first injured themselves
2–3 years before (n=54), 4–5 years before (n=7), or
o6 years before (n=5). When asked for the 12-month
incidence, 161 students (nine missing) with NSSI
answered, out of which 132 (82 %) reported having
hurt themselves within the past year and 29 (18 %)
within the past 24 months. This result means that the
12-month prevalence for NSSI was 19.8% within the
entire sample. Of the 170 participants who reported
injuring themselves, 98 (57.6%) said that they had
talked with someone about their NSSI and 72 (42.4%)
said that they had not.
From the total sample, 664 students answered the
question concerning the history of suicide attempts
(one missing). Forty-three (6.5%) stated that they had
attempted suicide, with 26 reporting having made one
attempt, 12 reporting two attempts, one making three
attempts, and four of the adolescents reported more
than three suicide attempts. The question concerning
suicidal ideation was answered by 656 participants
(nine missing). Two hundred and thirty-nine (36.4 %)
of the students stated that they had talked about or
thought about taking their lives, and 104 (15.6%) re-
ported having verbally threatened to attempt suicide
(661 responded, four missing). Most of them (n=71)
threatened suicide once, 23 threatened two to three
times, and nine reported four or more times (one did
not report the frequency). Of note, only 44 (42.7%) re-
ported really wanting to die at the time they made
their suicide threats.
Prevalence and gender
Significant gender differences were found for rates
of NSSI [50 males versus 120 females, x
2
=16.86,
1552 P. L. Plener et al.
https:/www.cambridge.org/core/terms. https://doi.org/10.1017/S0033291708005114
Downloaded from https:/www.cambridge.org/core. University of Basel Library, on 30 May 2017 at 17:00:38, subject to the Cambridge Core terms of use, available at
p<0.0001, odds ratio (OR) 2.17, 95% confidence inter-
val (CI) 1.49–3.15]. Females were also more likely than
males to report talking to others about their NSSI
(x
2
=13.59, p<0.001, OR 3.56, 95% CI 1.78–7.1). Al-
though more females reported suicide attempts (13
males versus 30 females), the difference was not stat-
istically significant (x
2
=2.95, p=0.09, OR 1.79, 95%
CI 0.91–3.49). However, statistically significant differ-
ences were found for having threatened to attempt
suicide (28 males versus 76 females, x
2
=12.50, p<
0.001, OR 2.28, 95% CI 1.34–3.62) and for reports of
suicidal ideation (72 males versus 167 females, x
2
=
24.20, p<0.0001, OR 2.31, 95% CI 1.65–3.23).
Prevalence in urban versus rural areas
No statistically significant differences were found on
any of the four SHBQ categories between students
from town or country schools (NSSI : 89 town versus 81
country, x
2
=1.01, p=0.31, OR 1.20, 95% CI 0.84–1.7;
suicidal attempts: 24 town versus 19 country, x
2
=0.83,
p=0.36, OR 1.33, 95% CI 0.72–2.49; suicidal threats : 46
town versus 58 country, x
2
=1.2, p=0.27, OR 0.79, 95 %
CI 0.52–1.20 ; suicidal ideation: 113 town versus 126
country, x
2
=0.67, p=0.41, OR 0.88, 95 % CI 0.64–1.20).
Clusters of self-harm
Participants were divided into four self-harm groups
(NoSH, NSSI, SA, NSSI+SA; see Method section),
with frequencies provided in Table 1. There was a
significant association with gender, with girls showing
a 50:50 ratio in the NoSH group and a 70 :30 ratio in
the NSSI, SA and NSSI+SA groups (x
2
=17.93, df=3,
p<0.001); this reflects the gender differences noted
earlier. Suicidal threats were significantly related to
the four clusters of self-harm (x
2
=98.47, df=3, p<
0.0001). There was no significant relationship between
clusters and place of schooling (urban versus rural,
x
2
=1.43, df =3, p=0.70).
Comparison between Germany and US samples
As one of the aims of this paper was to compare rates
of NSSI and suicidal behavior in the USA and
Germany, a closer comparison of the two populations
was necessary (details are provided in Table 2). Both
groups were recruited from schools situated in and
around midsized cities, using similar recruitment
strategies, research methodology, and active parental
consent (see Muehlenkamp & Gutierrez, 2007 for de-
tails on US methods). Although the German sample
was slightly larger in size (665 v. 540 in the US sample),
the gender balance was comparable in both groups
(x
2
=3.19, df=1, p=0.07, OR 1.24, 95 % CI 0.98–1.56).
Differences exist in age, with the US sample being
slightly older (t=11.02, df=726, p<0.0001). When
comparing the frequencies in the four clusters of self-
harm, no statistically significant differences were
Table 1. Clusters of self-harm, gender, place of schooling, suicidal threats and ADS scores
Clusters of self-harm
NoSH
(n=484, 72.9 %)
NSSI
(n=137, 20.6%)
SA
(n=10, 1.5 %)
NSSI+SA
(n=33, 5.0 %)
Gender, n (%)
Male 231 (47.7) 40 (29.2) 3 (30.0) 10 (30.3)
Female 253 (52.3) 97 (70.8) 7 (70.0) 23 (69.7)
Place of schooling, n (%)
Urban 252 (52.1) 67 (48.9) 5 (50.0) 14 (42.4)
Rural 232 (47.9) 70 (51.1) 5 (50.0) 19 (57.6)
Suicidal threats, n (%)
Yes 37 (7.7) 46 (33.8) 3 (30.0) 18 (56.3)
No 446 (92.3) 90 (66.2) 7 (70.0) 14 (43.7)
ADS score, mean (
S.D.)
Female 16.13 (8.66) 24.28 (10.02) 24.86 (7.56) 30.48 (8.78)
Male 10.95 (6.72) 18.23 (10.91) 22 (19.95) 17.60 (6.70)
Total 13.66 (8.21) 22.51 (10.61) 24 (11.34) 26.68 (10.09)
ADS, Depression scale (Allgemeine Depressions Skala) ; NoSH, no self-harm; NSSI, non-suicidal self-injury ; SA, suicide
attempt ;
S.D., standard deviation.
NSSI and suicide attempts 1553
https:/www.cambridge.org/core/terms. https://doi.org/10.1017/S0033291708005114
Downloaded from https:/www.cambridge.org/core. University of Basel Library, on 30 May 2017 at 17:00:38, subject to the Cambridge Core terms of use, available at
