The Mental Health in Austrian Teenagers (MHAT)-Study: preliminary results from a pilot study

Abstract

No epidemiological data on prevalence rates of mental disorders based on a representative sample are available for Austrian adolescents up to now. However, the knowledge of psychiatric disorders, related risk and protective factors is of great significance for treatment and prevention. The purpose of the MHAT-Study (Mental Health in Austrian Teenagers), the first epidemiological study on mental health in Austria, is to obtain prevalence rates of mental disorders and to examine risk factors, protective factors and quality of life in a representative sample of adolescents aged 10-18. Aims of this pilot study were to evaluate the feasibility and acceptability of the screening instruments, pre-estimate the frequency of mental health problems and estimate possible non-responder bias.
Twenty-one schools in eastern Austria were asked to participate. Data on mental health problems were derived from self-rating questionnaires containing standardized screening measures (Youth Self-Report, measuring emotional and behavioral problems and the SCOFF, indicating eating problems. Quality of life as well as related risk and protective factors were also obtained.
Four hundred and eight adolescents of five schools were recruited. The prevalence of mental health problems was 18.9 % [CI 95 %: 14.9-22.7]. Moreover, emotional and behavioral problems were highly correlated with quality of life measures. A Non-Responder Analysis indicated that non-responders (16.7 %) differ from responders with regard of school related problems.
The results demonstrate that mental health problems affect approximately one fifth of the adolescents. A Non-Responder Analysis suggests that the prevalence of behavioral and emotional problems is underestimated.

Full text

original article
198 e Mental Health in Austrian Teenagers (MHAT)-Study: preliminary results from a pilot study 1 3
SCOFF, indicating eating problems. Quality of life as well
as related risk and protective factors were also obtained.
Results Four hundred and eight adolescents of ve
schools were recruited. e prevalence of mental health
problems was 18.9 % [CI 95 %: 14.9–22.7]. Moreover, emo-
tional and behavioral problems were highly correlated
with quality of life measures. A Non-Responder Analy-
sis indicated that non-responders (16.7 %) dier from
responders with regard of school related problems.
Conclusions e results demonstrate that mental
health problems aect approximately one fth of the
adolescents. A Non-Responder Analysis suggests that
the prevalence of behavioral and emotional problems is
underestimated.
Keywords Mental health· Psychiatric disorders· Epide-
miology· Adolescence
Die Mental Health in Austrian Teenagers (MHAT)-
Studie: erste Ergebnisse aus einer Pilotstudie
Zusammenfassung
Grundlagen Bisher sind keine epidemiologischen Daten
zu Prävalenzraten für psychische Störungen für österrei-
chische Jugendliche, basierend auf einer repräsentativen
Stichprobe, verfügbar. Das Wissen über psychiatrische
Störungen sowie Risiko- und Schutzfaktoren ist jedoch
essentiell für erapie und Prävention. Im Rahmen der
MHAT-Studie (Mental Health in Austrian Teenagers, Psy-
chische Gesundheit bei österreichischen Jugendlichen),
der ersten epidemiologischen Studie zur psychischen
Gesundheit in Österreich, sollen Prävalenzraten psychi-
scher Störungen bei einer repräsentativen Stichprobe
von Jugendlichen zwischen 10 und 18 Jahren erhoben
und Risiko- und Schutzfaktoren sowie Lebensqualität
untersucht werden. Zweck der Pilotstudie war die Eva-
luation der Durchführbarkeit und Akzeptanz der Scree-
Summary
Background No epidemiological data on prevalence
rates of mental disorders based on a representative
sample are available for Austrian adolescents up to now.
However, the knowledge of psychiatric disorders, related
risk and protective factors is of great signicance for
treatment and prevention. e purpose of the MHAT-
Study (Mental Health in Austrian Teenagers), the rst
epidemiological study on mental health in Austria, is to
obtain prevalence rates of mental disorders and to exam-
ine risk factors, protective factors and quality of life in a
representative sample of adolescents aged 10–18. Aims
of this pilot study were to evaluate the feasibility and
acceptability of the screening instruments, pre-estimate
the frequency of mental health problems and estimate
possible non-responder bias.
Methods Twenty-one schools in eastern Austria were
asked to participate. Data on mental health problems
were derived from self-rating questionnaires containing
standardized screening measures (Youth Self-Report,
measuring emotional and behavioral problems and the
Neuropsychiatr (2014) 28:198–207
DOI 10.1007/s40211-014-0131-9
The Mental Health in Austrian Teenagers (MHAT)-
Study: preliminary results from a pilot study
Julia Philipp · Michael Zeiler · Karin Waldherr · Martina Nitsch · Wolfgang Dür · Andreas Karwautz ·
Gudrun Wagner
Prof. Dr.A.Karwautz()· Dr.J.Philipp· Mag. Dr.G.Wagner
Department of Child and Adolescent Psychiatry,
Medical University of Vienna,
Währinger Gürtel 18-20,
1090 Vienna, Austria
e-mail: andreas.karwautz@meduniwien.ac.at
Dr.J.Philipp
e-mail: julia.philipp@meduniwien.ac.at
Mag.M.Zeiler· Mag. Dr.M.Nitsch· Priv. Doz. Dr.W.Dür
Ludwig Boltzmann Institute Health Promotion Research,
Vienna, Austria
Mag. Dr.K.Waldherr
Ludwig Boltzmann Institute Health Promotion Research, Ferdinand
Porsche Distance University of Applied Sciences (FernFH),
Vienna, Austria
Received: 7 November 2014 / Accepted: 18 November 2014 / Published online: 28 November 2014
© Springer-Verlag Wien 2014

original article
e Mental Health in Austrian Teenagers (MHAT)-Study: preliminary results from a pilot study 199
1 3
mental health problems. 14.5 % reported mental health
problems associated with severe impairment [10].
e Great Smoky Mountains Study [11] assessed psy-
chiatric disorders in children aged 9, 11 and 13 with a
3-year and 6-year follow up. e 3-month prevalence rate
of all examined disorders was 13.3 %. However, 36.7 % of
the children had at least one psychiatric disorder dur-
ing the study period. Merikangas et al. [12] interviewed
10123 adolescents aged 13 to 18 years within the National
Comorbidity Survey Replication—Adolescent Supple-
ment (NCS-A) in the US. Lifetime prevalence for any dis-
order was 49.5 %. When including only adolescents with
severe impairment, prevalence rates decrease to 22.2 %.
In the last decades, some reviews were published pre-
senting prevalence rates of mental disorders. Robert,
Attkisson and Rosenblatt [13] identied 52 studies con-
ducted in over 20 countries published between 1963 and
1996 estimating the prevalence of psychiatric disorder.
Prevalence estimates ranged from 1 to 51 % with a mean
prevalence of 15.8 %. Ihle and Esser [2] as well compared
19 results from 9 countries between 1970 and 2000, pre-
senting a mean prevalence of 18 % (6.8–37.4 %). A review
concentrating on results from more recent studies in
Great Britain and the United States between 2000 and
2007 including children and adolescents between 5 and
17 years summarized that one person in four suered
from a psychiatric disorder during the past year and even
one out of three throughout the whole life [14]. A review
including 29 studies in Germany between 1949 and 2003
presented an overall prevalence rate of 17.2 % for emo-
tional and behavioral disorders [15]. A very recent review
presented a prevalence rate of 15 % for any disorder [3].
Anxiety disorders are the most common disorders,
followed by behavior (conduct disorders and attention
decit hyperactivity disorder), mood and substance use
disorders [2, 7, 14].
Some studies additionally examined risk, protective
and other associated factors. In general, older adoles-
cents present higher prevalence rates and more prob-
lems than younger ones [4, 7, 13]. Boys report more
behavioral and emotional problems than girls [4]. Behav-
ior disorders and substance use disorders are more com-
mon in boys whereas girls more often suer from eating
disorders and psychosomatic disorders. No dierence is
found for psychotic disorders. Prevalence rates of anxiety
and mood disorders are inconsistent. ey seem to be
more common in boys during childhood and school age
and in girls during adolescence and young adulthood [2].
Children with mental health problems furthermore
show impaired quality of life compared to healthy con-
trols [7].
ere are variations in the ndings of prevalence rates
in epidemiologic studies due to methodological dier-
ences [14], like dierent denitions, criteria, methods,
age groups and sources of information [13]. Still, there is
strong evidence for behavioral and emotional problems
and mental disorders in a large amount of adolescents.
ning-Phase, eine Häugkeitsschätzung von Verhaltens-
auälligkeiten und emotionalen Problemen sowie die
Abschätzung eines möglichen Non-Responder-Bias.
Methodik 21 Schulen im Osten Österreichs wurden
eingeladen, an der Studie teilzunehmen. Daten zur psy-
chischen Gesundheit wurden im Rahmen eines Scree-
nings mithilfe standardisierter Selbstbeurteilungsbögen
wie dem Youth Self-Report erhoben, der emotionale
und Verhaltensprobleme erhebt, und dem SCOFF, der
Hinweise für Essstörungen liefert. Lebensqualität und
Risiko- und Schutzfaktoren wurden ebenfalls erhoben.
Ergebnisse 408 Jugendliche an 5 Schulen wurden in
die Studie eingeschlossen. Die Prävalenzrate für psychi-
sche Probleme lag bei 18,9 % [CI 95 %:14,9–22,7]. Weiters
korrelierten emotionale und Verhaltensauälligkeiten
hoch mit gesundheitsbezogener Lebensqualität. Die
Non-Responder Analyse weist darauf hin, dass sich Non-
Responder (16.7 %) von Respondern hinsichtlich schuli-
scher Probleme unterscheiden.
Schlussfolgerungen Die Ergebnisse weisen darauf hin,
dass jeder fünfte Jugendlichen von einem psychischen
Problem betroen ist. Die Non-Responder Analyse deu-
tet auf eine Unterschätzung der Prävalenzraten hin.
Schlüsselwörter Psychische Gesundheit · Psychische
Störungen· Epidemiologie· Jugendliche
Introduction
Mental disorders tend to develop during adolescence [1,
2], show high comorbidity rates (45 % [1]) and tend to per-
sist into adulthood [3]. erefore, research should particu-
larly focus on adolescents’ mental health. e knowledge
about prevalence rates of psychiatric disorders in child-
hood and adolescence as well as related risk and protec-
tive factors is essential for the development of suitable
prevention strategies and treatment approaches [2].
However, no epidemiological data on prevalence rates
of mental disorders are available for adolescents in Aus-
tria. Due to this lack of epidemiological data in Austria,
international studies and results from other European
countries served as an overview of mental health and
well-being in teenagers so far.
Using only screening questionnaires, Rescorla et al.
[4] investigated rates of behavioral and emotional prob-
lems and compared self-reports from adolescents aged
12–18 (Youth Self-Report [5]) with parental ratings (Child
Behavior Checklist [6]) from 25 societies. With respect to
the total problem score, parents perceived 21.4 % of their
children to have problems, whereas 34.6 % of the chil-
dren themselves reported problems.
In Germany, the BELLA study was established to
assess mental health problems in teenagers [7, 8]. Prob-
lems were assessed by parents, using the Strength and
Diculties Questionnaire (SDQ [9]). 21.9 % of children
and adolescents between 7 and 17 years showed signs of

original article
200 e Mental Health in Austrian Teenagers (MHAT)-Study: preliminary results from a pilot study 1 3
ment, occurring problems and diculties, as well as all
content-related questions by the participants.
During the assessment, teachers were also asked to ll
in a short teacher’s questionnaire to collect basic demo-
graphic data and data on observed behavioral problems
of all pupils in their class as well as their hypotheses on
reasons for non-participation serving as a basis for Non-
Responder Analysis and estimation of possible non-
responder bias. A Non-Responder Analysis is essential
for epidemiologic studies [15] in order to evaluate the
representativeness of the sample. Teachers gave basic
information concerning survey participation, sex and
class repetition and rated all pupils in respect of school
absenteeism, willingness to make an eort during les-
sons, ability to concentrate during lessons, social inte-
gration in class, passivity, disciplinary problems and
making contact to parents or teacher conference because
of behavioral problems.
Subsequent to the assessment, a short interview with
the class teacher was conducted. ey were asked if they
felt well informed about the study and if they would need
any additional information or help for the next time they
would have to moderate this assessment. Data obtained
by documentation of survey process (e.g. duration of data
collection) as well as qualitative data from the teacher’s
interview served as the basis for the evaluation of feasi-
bility and acceptability of the MHAT screening phase and
is part of the process evaluation of the MHAT-Study.
Instruments
e MHAT questionnaire consists of several instruments.
Mental health data were assessed using the Youth
Self-Report (YSR [5], German version: Arbeitsgruppe-
Deutsche-Child-Behavior-Checklist [16, 17]). e YSR
consists of 103 problem items measuring behavioral and
emotional problems in a six-month time period. e
items are answered using a three-point scale (0 = not
true, 1 = somewhat or sometimes true, 2 = very true or
often true) and sum up to three broad-band scales, a
total problem score, internalizing problems, external-
izing problems, as well as eight syndrome scales: with-
drawn, somatic complaints, anxious/depressed, social
problems, thought problems, attention problems, delin-
quent behavior and aggressive behavior. e broad-band
scales show good internal consistency (Cronbach’s alpha
> 0.86). For the syndrome scales, Cronbach’s alphas of
0.56–0.86 were reported. T-Scores are calculated using
German norm data (1991) according to the manual,
whereby higher scores indicate more problems.
As the YSR is lacking items concerning eating disor-
ders, the SCOFF questionnaire was used to determine
signs of disturbed eating habits [18] (German version
[19]). e SCOFF is a very brief questionnaire, consisting
of ve items to be answered with yes or no (Do you ever
make yourself Sick because you feel uncomfortably full?
Do you worry you have lost Control over how much you
eat? Have you recently lost more than One stone (6kg)
Objectives
e Mental Health in Austrian Teenagers (MHAT) – Study
was initiated to collect epidemiological data on mental
health, emotional and behavioral problems, psychiatric
disorders, related risk and protective factors and quality
of life in a representative sample of adolescents between
10 and 18 years in Austria for the rst time. A two-step
design was chosen for the MHAT-Study. Phase 1 (screen-
ing phase) consists of a questionnaire assessing emo-
tional and behavioral problems, social and demographic
correlates, risk and protective factors and quality of life.
In phase 2 (interview phase), positive screened adoles-
cents as well as a sample of negative screened partici-
pants are further contacted for a standardized clinical
interview in order to obtain DSM-5 diagnosis. A pilot
study was conducted in 2013. e aims of the present
pilot study were a) to evaluate the feasibility and accept-
ability of the screening phase b) to pre-estimate preva-
lence rates in order to plan necessary resources for the
interview phase of the MHAT study and c) to estimate
possible non-responder bias.
Methods
For this pilot study, phase 1 (screening phase) was con-
ducted with a small sample of Austrian adolescents. e
MHAT-Study is approved by the ethics committee of the
Medical University of Vienna and the Austrian Federal
Ministry of Education and Women’s Aairs.
Sampling, recruitment and procedure
Adolescents between 10 and 18 were recruited from ve
secondary schools in Lower Austria and Burgenland.
Four age groups were included in the sample: 5th graders
(aged 10–11, resp.), 7th graders (aged 12–13, resp.), 9th
graders (aged 14–15, resp.) and 11th graders (ages 16–17,
resp.).
Information sheets about background and procedure
of the study, as well as a sample questionnaire was sent
to the schools’ administration oce. Adolescents and
parents concerned were given a description of the study.
Written informed consent was obtained from adoles-
cents and parents.
Screening questionnaires were administered either
by paper-pencil administration or by a corresponding
online questionnaire. e assessment was designed for
the duration of one lesson (approximately 50min).
Class teachers were asked to administer the survey
autonomously. erefore, teachers were given detailed
instructions for the procedure, technical instructions for
the online questionnaire as well as predened answers
to possible “FAQs (Frequently Asked Questions)”. A study
member was present in the classroom during data col-
lection and acted as non-participating observer. e
study member documented the duration of the assess-

original article
e Mental Health in Austrian Teenagers (MHAT)-Study: preliminary results from a pilot study 201
1 3
lated in order to check if respondents and non-respon-
dents equally distribute to the categories of the other
variables depicted in Table 2. Item ratings of the YSR
were summed up per scale and transferred into T-scores.
A cut-o score of T > 70 for the syndrome scales and T > 63
for the broad-band scales was used to dene clinically
relevant cases, as suggested in the manual.
Participants in this study are dened as high-risk-
cases with a score above the cut-o for clinical relevance
in at least one YSR syndrome scale or a SCOFF score of
two or more positive answers including at least one of
the following: “Do you ever make yourself sick because
you feel uncomfortably full?”, “Have you recently lost more
than one stone (6kg) in a 3 month period?”
A 2 × 4 ANOVA is conducted to examine the impact of
sex and school grade on YSR sum scores. KIDSCREEN
item rating were recoded if necessary and summed up
for each dimension.
Pearson correlation coecients were calculated to
examine the association between behavioral and emo-
tional problems as obtained by the YSR total problem
score and health-related quality of life as obtained by the
KIDSCREEN scales.
Results
Sample
Figure1 shows the ow diagram of the recruitment pro-
cess. From the 21 schools invited for participation, ve
schools agreed to participate. ese schools provided 27
classes for inclusion in the study: 8 classes of 5th grad-
ers (aged 10–11, resp.), 7 classes of 7th graders (aged
12–13, resp.), 5 classes of 9th graders (aged 14–15, resp.),
in a three month period? Do you believe yourself to be
Fat when others say you are too thin? Would you say that
Food dominates your life?). Item ratings (yes = 1, no = 0)
can be summed up to a total score (0–5) indicating a risk
for an eating disorder at a score of two or more positive
answers. e SCOFF proved to be 100 % sensitive with a
false positive rate of 12.5 % [20]. In a German study, one
out of ve adolescents aged 11–17 reported signs of dis-
ordered eating [19]. For the MHAT-Study, we propose
alternative cut-o criteria that are based on the clinical
relevance of the SCOFF items. As judged by clinical psy-
chologists working in the eld of eating disorders at the
child and adolescent psychiatry of the General Hospital
of Vienna, vomiting (item 1) and weight reduction (item
3) are a stronger indication of an eating disorder than the
other items. erefore, we propose that additionally to
the criteria from the authors (score ≥ 2), at least one of
these two items has to be conrmed by the adolescents.
To measure the socioeconomic status, the Family
Auence Scale (FAS [21]) was used. e FAS was devel-
oped within the WHO-Health Behaviour in School-aged
Children (HBSC) survey and consists of four items ; higher
scores indicating a higher level of family auence. e
FAS has good internal consistency. ree groups can be
described: low family auence, moderate family au-
ence and high family auence.
e KIDSCREEN [22] was used to assess quality of life
in children and adolescents aged 8–18 within the last
week. e following dimensions of the KIDSCREEN-52
version (KS-52) and the KIDSCREEN-27 version (KS-27)
are selected for the purpose of the MHAT-Study: Self-Per-
ception (KS-52), Parent-Relation and Home Life (KS-52),
Social Support and Peers (KS-27), School Environment
(KS-27) and Bullying (KS-52). Additional six items from
the KIDSCREEN questionnaire were included enabling
the calculation of the KIDSCREEN-10 score. Items are
rated on a ve-point scale. Higher scores indicate higher
quality of life. e KIDSCREEN demonstrates good
internal consistency, with a Cronbach’s alpha of 0.77 to
0.89, 0.80 to 0.84 and 0.82 for the three versions (KID-
SCREEN-52, KIDSCREEN-27 and KIDSCREEN-10). An
own KIDSCREEN-questionnaire was composed by single
scales of the original KS-52 and KS-27 versions.
Sociodemographic data (sex, age, migration back-
ground, family and residential environment, school
grade, type of school) were collected as well as several
factors known as risk and protective factors for men-
tal health (including e.g. family-structure, physical and
mental diseases of the participant and near relatives, life-
time-occurrence of traumatic events [23, 24]).
Statistical analyses
Data from the teacher’s questionnaire and the MHAT
questionnaire were entered into and analyzed with IBM
Statistics 22.0 software. 2 × 2 and 2 × 3 contingency tables
with study participation (yes vs. no) and other variables
captured with the teacher’s questionnaire were calcu-
6FKRROVDVNHGIRUSDUWLFLSDWLRQ
Q 
([FOXGHGQ 







'LGQRWUHVSRQGQ 
'HFOLQHGWR SDUWLFLSDWHQ 
)RUJRWWRGLVWULEXWHVWXG\LQIRUPDWLRQQ 
&ODVVHVLQFOXGHGQ 
$GROHVFHQWVLQFOXGHGQ 
5HVSRQGHUVQ 
1RQ5HVSRQGHUVQ 
1RLQIRUPHGFRQVHQWUHWXUQHGQ 
6LFNQHVVQ 
%URNHRIIDVVHVVPHQW Q 
2WKHUUHDVRQV Q 
Fig.1 Flow diagram of participants

original article
202 e Mental Health in Austrian Teenagers (MHAT)-Study: preliminary results from a pilot study 1 3
Acceptability and feasibility
All of the ve schools had a computer lab with a sucient
number of computers. Due to occupancy of computer
lab, concurrent data collection in several classes and
technical problems, a quite small number of participants
completed the online version (n = 81; 19.9 %).
e overall duration of data collection from the begin-
ning of the lesson to the completion of the last question-
naire ranged from 37 to 82min with a median of 46min.
e duration from the beginning of the lesson to the
beginning of completing the questionnaire (including
other activities before starting and reading the instruc-
tion) was quite long (median of 13min, minimum 7min;
maximum 22min). e median net duration for complet-
ing the questionnaire as automatically recorded by the
online questionnaire was 24.6min (minimum 13.5min;
maximum 44.6min).
Twenty-two out of 23 teachers agreed that they would
be able to conduct data collection without any help
of a study member. Some aspects were mentioned as
improvements for the main study: better information
transfer from administration oce to teachers (n = 2),
more time for obtaining informed consent (n = 2), exten-
sion of “FAQs” (n = 2), further information concerning the
procedure (n = 2) and adaptation for 5th graders (n = 1).
Mental health problems
Prevalence rates of behavioral and emotional problems
obtained by the Youth Self-Report and the SCOFF-ques-
tionnaire are depicted in Table2. 15.9 % of the screened
adolescents showed signs of mental health problems
using the Youth Self-Report total problem score. Inter-
nalizing behavioral problems appear more frequent
(18.5 %) than externalizing problems (5.7 %). Regarding
the YSR second-order scales, withdrawn problems and
somatic complaints were most prevalent, in contrast to
aggressive behavior which appeared as the least preva-
lent problem area. 25 % [CI 95 %: 20.7–29.3 %] of the par-
ticipants scored in at least one of the rst or second order
problem scales.
Dierences between sex and school grades according
to YSR problem scores as analyzed by a 2 × 4 ANOVA are
depicted in Table 3. A main eect of sex was observed
for the total problem score, internalizing problems,
withdrawn behavior, somatic complaints and anxious/
depressed mood with larger problem scores for girls
compared to boys. A signicant main eect of school
grade was observed for externalizing problems, thought
problems, attention problems and delinquent behaviors.
Due to signicant interaction eects, the interpretation
regarding the impact of the school grade is not clear in
some cases. However, there is a tendency for larger prob-
lem scores for participants in higher school grades. Sig-
nicant sex*school grade interaction eects occurred in
seven of eleven YSR scales. For the signicant YSR scales,
mean problem scores constantly increased with higher
6 classes of 11th graders (aged 16–17, resp.) and 1 class
dropped out. Altogether, 590 adolescents were recruited
and asked to give informed consent for the study. 408
adolescents and their parents (69.2 %) agreed to partici-
pate. Of the 182 non-responders (30.8 %), 93 (51 %) had
no informed consent, 46 (25 %) were sick, 4 (2 %) broke
o the assessment, 16 (9 %) were absent due to other rea-
sons like sport events and 23 (13 %) were absent due to
unknown reasons.
Sociodemographic information of participants is pro-
vided in Table1.
ere were more female adolescents and more 5th
graders participating in the study. Most of the partici-
pants reported no migration background, high socioeco-
nomic status and occupation of both parents.
Table 1 Adolescents’ demographic information
n%
Total 408 100
Sex
Male 170 41.7
Female 238 58.3
School grade
5th 134 32.8
7th 106 26.0
9th 84 20.6
11th 84 20.6
Migration background
No migration background 365 89.5
1st generationa17 4.2
2nd generationb26 6.4
FAS categoryc
Low 4 1.0
Moderate 71 17.4
High 327 80.1
Missing 6 1.5
Completeness of familiyd
Yes 324 79.4
No 84 20.6
Occupation of parents
No parent 6 1.5
One parent 78 19.1
Both parents 321 78.7
Missing 3 0.7
Diagnosed physical illness
No 322 78.9
Yes 82 20.1
Missing 4 1.0
aOwn birth-place in Austria and birth-place of both parents in foreign
country
bOwn birth-place and birth-place of both parents in foreign country
cFamily affluence scale
dAdolescents living with both biological parents

original article
e Mental Health in Austrian Teenagers (MHAT)-Study: preliminary results from a pilot study 203
1 3
to lower health related quality of life in the assessed
dimensions.
Non-responder analysis
A comparison between responders and non-responders
as derived from the teacher’s questionnaire is shown
in Table 4. Participation rate was signicantly higher
for females. For school absenteeism, signicant dier-
ences could be observed, with responders being absent
from school less often in comparison to non-respond-
ers of the same age. According to the teachers’ rating,
responders made greater eort during lessons than non-
responders and the ability to concentrate during les-
sons was perceived signicantly higher for responders
compared with responders. Participants were also more
likely well integrated in class compared to non-respond-
ers. No signicant dierences between responders and
non-responders were observed in respect of grade rep-
etition, disciplinary problems in school, internalizing
behavioral problems (withdrawn, passive) in school and
contacting parents or teaching sta due to behavioral
problems.
Discussion
e purpose of this pilot study was to determine the
acceptability and feasibility of the screening phase, to
estimate the prevalence of mental health problems and
to perform a Non-Responder Analysis.
school grades for female participants. For males, the pic-
ture is not as clear as for females. Whereas mean scores
remained almost stable for the total problem scale, mean
scores decreased from low to high school grades for
internalizing problems, somatic complaints and social
problems. No clear association between school grade
and problem scores could have been observed in males
for externalizing problems including delinquent behav-
ior and aggressive behavior.
In the SCOFF-questionnaire, 20.8 % of the screened
adolescents scored above the clinical cut-o applying
the criteria (score ≥ 2) proposed by the authors. Since the
SCOFF is known for its very high sensitivity leading to a
high rate of positive screened adolescents, we propose
that additionally to the criteria from the authors (score
≥ 2), at least one of the two clinically relevant items (vom-
iting, weight reduction) has to be conrmed by the ado-
lescents. Applying these new criteria, the percentage of
positive screened adolescents decreased to 6.6 %.
High-risk cases for mental disorders were dened as
scoring above the clinical cut-o in at least one of the YSR
syndrome scales or the SCOFF applying the new criteria
as described above. Following this denition, the over-
all prevalence of mental health problems was 18.9 % [CI
95 %: 14.9–22.7].
YSR total problem scores were signicantly correlated
with health related quality of life measures as derived
by the KIDSCREEN questionnaire (KIDSCREEN-10:
r = − .628; Self Perception: r = − .572; Parent Relation and
Home Life: r = − .484; Bullying: r = − .356; Social Sup-
port and Peers: − .298; School Environment: − .512, all
p-values < .01). Higher YSR problem scores were related
Table 2 Percentage of at-risk cases according to YSR scales and SCOFF and 95 % CIs
Scale 5th grade 7th grade 9th grade 11th grade Overall
YSR Total 12.3 [6.4; 18.2] 14.6 [7.5; 21.7] 13.1 [5.8; 20.4] 25.6 [16.1; 35.1] 15.9 [12.2; 19.6]
YSR Int 14.8 [8.5; 21.1] 16.7 [9.2; 24.2] 15.5 [7.7; 23.3] 29.3 [19.4; 39.2] 18.5 [14.6; 22.4]
YSR Ext 2.5 [0; 5.3] 8.3 [2.8; 13.8] 4.8 [0.2; 9.4] 8.5 [2.4; 14.6] 5.7 [3.4; 8.0]
YSR WD 4.1 [0.6; 7.6] 3.1 [0; 6.6] 7.1 [1.6; 12.6] 7.3 [1.6; 13.0] 5.2 [3.0; 7.4]
YSR SC 4.9 [1.1; 8.7] 4.2 [0.2; 8.2] 3.6 [0; 7.6] 6.1 [0.9; 11.3] 4.7 [2.6; 6.8]
YSR Anx/Dep 4.1 [0.6; 7.6] 2.1 [0; 5.0] 3.6 [0; 7.6] 3.7 [0; 7.8] 3.4 [1.6; 5.2]
YSR SP 1.6 [0; 3.8] 4.2 [0.2; 8.2] 1.2 [0; 3.5] 2.4 [0; 5.7] 2.3 [0.8; 3.8]
YSR TP 2.5 [0; 5.3] 2.1 [0; 5.0] 2.4 [0; 5.7] 4.9 [0.2; 9.6] 2.9 [1.2; 4.6]
YSR AT 0.0 5.2 [0.7; 9.7] 2.4 [0; 5.7] 3.7 [0; 7.8] 2.6 [1.0; 4.2]
YSR Del 0.8 [0; 2.4] 2.1 [0; 5.0] 1.2 [0; 3.5] 6.1 [0.9; 11.3] 2.3 [0.8; 3.8]
YSR Agg 0.8 [0; 2.4] 2.1 [0; 5.0] 0.0 1.2 [0; 3.6] 1.0 [0; 2.0]
SCOFFa17.2 [10.8; 23.6] 24.5 [16.3; 32.7] 16.7 [8.7; 24.7] 26.2 [16.7; 35.7] 20.8 [16.9; 24.7]
SCOFFb8.2 [3.5; 12.9] 6.6 [1.9; 11.3] 4.8 [0.2; 9.4] 6.0 [0.9; 11.1] 6.6 [4.2; 9.0]
YSR and SCOFFc16.8 [10.2; 23.4] 18.6 [10.8; 26.4] 14.3 [6.8; 21.8] 26.8 [17.2; 36.4] 18.8 [14.9; 22.7]
YSR Youth self-report, Int Internalizing, Ext Externalizing, WD Withdrawn, SC Somatic complaints, Anx/Dep Anxious/depressed, SP Social problems, TP Thought
problems, AT Attention problems, Del Delinquent behavior, Agg Aggressive behavior
aSCOFF Score ≥ 2
bSCOFF Score ≥ 2 and at least one of the following SCOFF items is marked as applicable: “Do you make yourself sick because you feel uncomfortably full?”,
“Have you recently lost more than one stone (6 kg) in a 3month period?”
cAbove cut-off score of clinical relevance in at least one YSR syndrome scale OR SCOFF Score ≥ 2 and at least one of the following SCOFF items is marked as
applicable: “Do you make yourself sick because you feel uncomfortably full?”, “Have you recently lost more than one stone (6 kg) in a 3month period?”

original article
204 e Mental Health in Austrian Teenagers (MHAT)-Study: preliminary results from a pilot study 1 3
problems. Compared to another study using the YSR as
screening for emotional and behavioral problems, that
found a total of 34.6 % clinical relevant cases [4] our results
turned out to be lower. Using the SD Q as screening for men-
tal health problems also showed higher rates (21.9 % [8]).
In contrast to Rescorla et al. [4], reporting 11.1 % inter-
nalizing behavioral problems, the results of our study
show a higher rate with 18.5 %. Concerning externalizing
problems, the sample of this pilot study reported lower
rates (5.7 %) compared to 9.7 %.
Our results of the SCOFF questionnaire indicate disor-
dered eating habits in one out of ve participants, which
coincides with a German study [19].
Consistent with the literature [4, 7, 13] problem scores
increase with age, but only for female adolescents. For
males, the problem scores stay rather stable across the
age groups or even decreased by age. Our study reveals
higher problem scores in females, in contrast to the
results of former studies, where problems were more fre-
quently reported by males [4].
Acceptability and feasibility
Online-application as well as paper-pencil-application of
the questionnaire proved to be feasible. A sucient num-
ber of computers were available in all schools. Completing
the questionnaire within one lesson proved to be possible
for both application forms for nearly all of the adolescents.
Providing technical instructions and FAQs turned out
to be helpful for teachers.
Teachers felt that they would be able to conduct the data
collection without the guidance of a study member. Teach-
ers suggested several improvements for the main study: bet-
ter information transfer, more time for obtaining informed
consent, extension of the FAQs, further information con-
cerning the procedure and adaptation for 5th graders.
Mental health problems
According to the Youth Self-Report total problem score,
15.9 % of the adolescents showed signs of mental health
Table 3 Descriptive Results of Youth Self-Report (YSR) Scores and ANOVA results (main effect of sex and school grade, in-
teraction effect sex*school grade)
YSR scale Sex Descriptive statistics (mean, SD) Test statistics
5th grade 7th grade 9th grade 11th grade
Total problem
score
Boys 30.74 (21.58) 30.72 (23.87) 29.16 (15.96) 29.04 (14.68) Sex: F(1,397) = 8.00, p = .005, ηp
2 = .02
Grade: F(3,397) = 2.52, p = .057, ηp
2 = .02
Interaction: F(3,397) = 8.00, p = .005, ηp
2 = .02
Girls 27.30 (17.01) 34.87 (18.79) 35.85 (17.25) 44.48 (18.55)
Internalizing
problems
Boys 9.22 (7.77) 8.48 (7.34) 8.31 (6.90) 7.35 (6.18) Sex: F(1,397) = 28.84, p < .001, ηp
2 = .07
Grade: F(3,397) = 0.99, p = .400, ηp
2 < .01
Interaction: F(3,397) = 3.53, p = .015, ηp
2 = .03
Girls 10.21 (8.14) 12.06 (8.35) 13.40 (8.97) 16.08 (9.12)
Externalizing
problems
Boys 9.87 (7.16) 10.37 (8.90) 8.44 (5.16) 10.70 (5.94) Sex: F(1,397) = 0.07, p = .793, ηp
2 < .01
Grade: F(3,397) = 4.90, p = .002, ηp
2 = .04
Interaction: F(3,397) = 3.75, p = .011, ηp
2 = .03
Girls 6.55 (4.03) 9.42 (5.81) 9.85 (5.84) 12.85 (6.42)
Withdrawn Boys 2.56 (2.36) 2.59 (2.34) 2.66 (2.40) 2.87 (2.42) Sex: F(1,397) = 10.02, p = .002, ηp
2 = .03
Grade: F(3,397) = 1.62, p = .184, ηp
2 = .01
Interaction: F(3,397) = 0.68, p = .568, ηp
2 < .01
Girls 3.21 (2.23) 2.96 (2.35) 3.62 (3.18) 4.33 (3.06)
Somatic com-
plaints
boys 2.38 (2.43) 2.19 (2.72) 2.00 (1.97) 1.48 (1.41) Sex: F(1,396) = 20.06, p <.001, ηp
2 = .05
Grade: F(3,396) = 0.58, p = .631, ηp
2 < .01
Interaction: F(3,396) = 3.41, p = .018, ηp
2 = .03
girls 2.35 (2.59) 3.31 (2.78) 3.56 (2.57) 3.82 (3.29)
Anxious/de-
pressed
Boys 4.51 (4.61) 3.85 (3.89) 3.84 (3.71) 3.13 (3.90) Sex: F(1,397) = 30.24, p < .001, ηp
2 = .07
Grade: F(3,397) = 0.76, p = .520, ηp
2 < .01
Interaction: F(3,397) = 3.89, p = .009, ηp
2 = .03
Girls 5.00 (5.08) 6.31 (5.33) 6.75 (5.53) 8.52 (5.24)
Social problems Boys 2.54 (2.57) 2.54 (2.49) 1.88 (1.70) 1.52 (1.81) Sex: F(1,397) = 0.00, p = .994, ηp
2 < .01
Grade: F(3,397) = 2.57, p = .054, ηp
2 = .02
Interaction: F(3,397) = 1.43, p = .233, ηp
2 = .01
Girls 2.10 (2.18) 2.50 (2.48) 1.50 (1.85) 2.38 (1.92)
Thought prob-
lems
Boys 0.87 (1.56) 0.70 (1.41) 1.25 (1.68) 1.30 (1.46) Sex: F(1,397) = 0.14, p = .713, ηp
2 < .01
Grade: F(3,397) = 5.72, p < .001, ηp
2 = .04
Interaction: F(3,397) = 1.24, p = .296, ηp
2 < .01
Girls 0.89 (1.33) 0.63 (0.95) 0.92 (1.20) 1.92 (2.14)
Attention
problems
Boys 3.97 (2.84) 4.44 (3.66) 4.44 (2.63) 4.61 (2.46) Sex: F(1,397) = 0.04, p = .847, ηp
2 < .01
Grade: F(3,397) = 6.07, p < .001, ηp
2 = .04
Interaction: F(3,397) = 2.22, p = .085, ηp
2 = .02
Girls 2.89 (2.28) 4.48 (3.02) 4.19 (2.47) 5.67 (2.87)
Delinquent
behavior
Boys 2.79 (2.47) 3.11 (2.77) 2.72 (2.07) 3.61 (2.35) Sex: F(1,397) = 0.09, p = .761, ηp
2 < .01
Grade: F(3,397) = 9.64, p < .001, ηp
2 = .07
Interaction: F(3,397) = 4.05, p = .007, ηp
2 = .03
Girls 1.58 (1.56) 2.48 (2.24) 3.13 (2.69) 4.72 (3.10)
Aggressive
behavior
Boys 7.08 (5.31) 7.26 (6.68) 5.72 (3.74) 7.09 (4.06) Sex: F(1,397) = 0.04, p = .840, ηp
2 < .01
Grade: F(3,397) = 2.42, p = .066, ηp
2 = .02
Interaction: F(3,397) = 2.75, p = .042, ηp
2 = .02
Girls 4.97 (2.92) 6.94 (4.22) 6.71 (3.76) 8.13 (4.24)
Significant main and interaction effects are printed bold

original article
e Mental Health in Austrian Teenagers (MHAT)-Study: preliminary results from a pilot study 205
1 3
Non-responder analysis
Non-responders were signicantly more likely to be
male, more often absent from school, showed less eort
and ability to concentrate during lessons and were less
likely to be well integrated in class as per teacher’s opin-
ion. ere were no dierences in the following variables:
signs of disciplinary problems in school, being with-
drawn and passive, repeating grades, making contact
to parents or school sta due to problems. e Non-
Responder Analysis indicates that non-responders may
have more problems in school. As a result prevalence of
behavioral and emotional problems may be underesti-
mated in epidemiological studies conducted in schools.
For this study, high-risk cases for mental disorders
were dened as scoring above the clinical cut-o in
at least one of the YSR syndrome scales or giving two
or more positive answers in the SCOFF questionnaire
including one of the clinically relevant questions (mak-
ing oneself sick, having lost more than 6kg). Using this
denition, the overall prevalence of mental health prob-
lems in this study was 18.9 %, which is in accordance with
several studies indicating a rate of 10–20 % for mental dis-
orders and behavioral problems [2, 3, 13, 15].
Furthermore, this study supports the assumption that
mental health problems highly correlate with impaired
quality of life [7].
Variable Responder observed (and expect-
ed) frequenciesa
Non-Responder observed (and
expected) frequencies
Test statisticbp-value
Sex χ2
(1) = 5.921 .021
Male 155 (166.8) 80 (68.2)
Female 212 (200.2) 70 (81.8)
Grade repetition χ2
(1) = 1.196 .274
Yes 17 (19.6) 11 (8.4)
No 27 (324.4) 137 (139.6)
School absentismcχ2
(2) = 15.561 < .001
Below average 172 (156.8) 52 (67.2)
Average 137 (140.0) 63 (60.0)
Above average 36 (48.3) 33 (20.7)
Effort during lessonscχ2
(2) = 10.342 .006
Below average 79 (93.3) 54 (39.7)
Average 168 (161.3) 62 (68.7)
Above average 98 (90.5) 31 (38.5)
Ability to concentrate during lessonscχ2
(2) = 14.535 .001
Below average 78 (93.2) 55 (39.8)
Average 168 (165.3) 68 (70.7)
Above average 98 (85.5) 24 (36.5)
Good integration in class χ2
(1) = 12.340 <.001
Rather yes 309 (296.6) 114 (126.4)
Rather no 36 (48.4) 33 (20.6)
Behavioral problems in school χ2
(1) = 1.041 .308
Rather yes 49 (52.7) 26 (22.3)
Rather no 294 (290.3) 119 (122.7)
Internalizing behavioral problems (withdrawn and passive in school) χ2
(1) = 0.501 .479
Rather yes 72 (75.0) 35 (32.0)
Rather no 272 (269.0) 112 (115.0)
Making contact with parents or teacher conference due to behavioral problems χ2
(2) = 3.711 .156
Yes 46 (50.4) 26 (21.6)
No 293 (286.5) 116 (122.5)
No, but should be done 5 (7.0) 5 (3.0)
aDue to missing data in the teacher’s questionnaire, sample size of respondents doesn’t correspond necessarily to the sample size of the main analysis
bTest statistic based on 2 × 2 or 2 × 3 contingency table
cRatings in comparison to students of the same age
Table 4 Non-responder analysis

original article
206 e Mental Health in Austrian Teenagers (MHAT)-Study: preliminary results from a pilot study 1 3
References
1. Ihle W, Esser G, Schmidt MH, Blanz B. Prävalenz, Komor-
bidität und Geschlechtsunterschiede psychischer Störun-
gen vom Grundschul- bis ins frühe Erwachsenenalter. Z
Klin Psychol Psych. 2000;29:263–75.
2. Ihle W, Esser G. Epidemiologie psychischer Störungen
im Kindes- und Jugendalter: Prävalenz, Verlauf, Komor-
bidität und Geschlechtsunterschiede. Psychol Rundsch.
2002;53:159–69.
3. Fuchs M, Bösch A, Hausmann A, Steiner H. “e Child is
Father of the Man”. Review von relevanten Studien zur Epi-
demiologie in der Kinder- und Jugendpsychiatrie. Z Kinder
Jug-Psych. 2013;41:45–57.
4. Rescorla LA, Ginzburg S, Achenbach TM, et al. Cross-infor-
mant agreement between parent-reported and adolescent
self-reported problems in 25 societies. J Clin Child Adolesc.
2012;49:1215–24.
5. Achenbach TM. Manual for the Youth Self-Report and 1991
Prole. Burlington: University of Vermont, Departement of
Psychiatry; 1991a.
6. Achenbach TM. Manual for the Child Behavior Check-
list/4–18 and 1991 Prole. Burlington: University of Ver-
mont, Departement of Psychiatry; 1991b.
7. Ravens-Sieberer U, Wille N, Bettge S, Erhart M. Psychische
Gesundheit von Kindern und Jugendlichen in Deutsch-
land. Ergebnisse aus der BELLA-Studie im Kinder- und
Jugendgesundheitssurvey (KiGGS). Bundesgesundheits-
bla. 2007;50:871–8.
8. Ravens-Sieberer U, Kurth BM, KiGGS study group, BELLA
study group. e mental health module (BELLA study)
within the German health interview and examination sur-
vey of children and adolescents (KiGGS): study design and
methods. Eur Child Adoles Psy. 2008a;17(Suppl 1):10–21.
9. Goodman R. e strengths and diculties questionnaire: a
research note. J Child Psychol Psych. 1997;38:581–6.
10. Ravens- Sieberer U, Wille N, Erhart M, Bettge S, Wittchen
HU, Rothenberger A, Herpetz-Dahlmann B, Resch F,
Hölling H, Bullinger M, Barkmann C, Schulte-Markwort M,
Döpfner M, the BELLA study group. Prevalence of mental
health problems among children and adolescents in Ger-
many: results of the BELLA study within the national health
interview and examination survey. Eur Child Adoles Psy.
2008b;17(Suppl 1):22–33.
11. Costello EJ, Mustillo S, Erkanli A, Keeler G, Angold A. Prev-
alence and development of psychiatric disorders in child-
hood and adolescence. Arch Gen Psychiat. 2003;60:837–44.
12. Merikangas KR, He JP, Burnstein M, Swanson SA, Ave-
nevoli S, Cui L, Benjet C, Georigiades K, Swendsen J. Life-
time prevalence of mental disorders in U.S. adolescents:
results from the National Comorbidity Survey Replica-
tion—Adolescent Supplement (NCS-A). J Am Acad Child
Psy. 2010;49:980–9.
13. Roberts RE, Attkisson C, Rosenblatt A. Prevalence of psy-
chopathology among children and adolescents. Am J Psy-
chiat. 1998;155:715–25.
14. Merikangas KR, Nakamura EF, Kessler RC. Epidemiology
of mental disorders in children and adolescents. Dialogues
Clin Neurosci. 2009;11:7–20.
15. Barkmann C, Schulte-Markwort M. Prävalenz psychischer
Auälligkeit bei Kindern und Jugendlichen in Deutsch-
land—ein systematischer Literaturüberblick. Psychiatri
Prax. 2004;31:278–87.
Limitations
As the sample in this pilot study was not selected ran-
domly and due to the small sample size, results and prev-
alence rates cannot be generalized to the population of
adolescents 10 to 18 years. However, the main goal of this
pilot study was to ascertain the acceptability and feasibil-
ity and to conduct a Non-Responder Analysis for which
the described sample is sucient.
Another limitation is the lacking of the second phase
(interview phase). For the pilot study, only the screening
phase was conducted, meaning that no contact informa-
tion was inquired from parents. is anonymity could
probably have an impact on the response rates.
Finally, the presence of a project member during the
data collection could have inuenced teachers’ and ado-
lescents’ behavior.
Conclusion and implications for the MHAT-study
is pilot study was conducted as part of the project risk
management of the MHAT-Study to improve the main
study. e duration as well as the online-application and
paper-pencil-application of the questionnaire were over-
all well accepted and feasible. It is expected that teach-
ers are able to administrate the main study without the
guidance of a study member. Teachers’ suggestions to
improve the information material (i.e. time for obtain-
ing informed consent, extension of the FAQs, further
information concerning the procedure) will be taken
into account, instructions will be upgraded, technical
instructions and FAQs will be adapted.
e results demonstrate that mental health problems
aect a large amount of adolescents from 10 to 18 years.
For the main study, a representative sample of Austrian
adolescents based on age, sex, federal state and school
type will be included.
A Non-Responder Analysis will also be conducted,
because underestimation of prevalence rates can further
be expected in the main study. e screening phase will
be followed by a second phase. e second phase consists
of a structured diagnostic interview to assess psychiat-
ric diagnoses according to the Diagnostic and Statistical
Manual of the American Psychiatric Association, Ver-
sion 5 (DSM-5) criteria with adolescents scoring above
the predened cut-o score described in this study and a
random sample of adolescents scoring below the cut-o.
Acknowledegments
e MHAT study is funded by “Gemeinsame Gesund-
heitsziele aus dem Rahmen-Pharmavertrag” (a coop-
eration between Austrian pharmaceutical industry and
Austrian social insurance).
Conict of interest
Julia Philipp, Michael Zeiler, Karin Waldherr, Martina
Nitsch, Wolfgang Dür, Andreas Karwautz, and Gudrun
Wagner declare that they have no conict of interest.

original article
e Mental Health in Austrian Teenagers (MHAT)-Study: preliminary results from a pilot study 207
1 3
20. Morgan JF, Reid F, Lacey JH. e SCOFF questionnaire. A
new screening tool for eating disorders. Western J Med.
2000;172:164–5.
21. Boyce W, Torsheim T, Currie C, Zambon A. e family
auence scale as a measure of national wealth: valida-
tion of an adolescent self-report measure. Soc Indic Res.
2006;78:473–87.
22. e KIDSCREEN Group Europe. e KIDSCREEN ques-
tionnaires. Qality of life questionnaires for children and
adolescents. Lengerich:Pabst Science Publishers; 2006.
23. Egle UT, Homann SO, Steens M. Psychosoziale Risiko-
und Schutzfaktoren in Kindheit und Jugend als Prädis-
position für psychische Störungen im Erwachsenenalter.
Nervenarzt. 1997;68:683–95.
24. Erhart M, Hölling H, Bettge S, Ravens-Sieberer U, Schlack
R. Der Kinder- und Jugendgesundheitssurvey: Risiken und
Ressourcen für die psychische Entwicklung von Kindern
und Jugendlichen. Bundesgesundheitsbla. 2007;50:800–9.
16. Arbeitsgruppe Deutsche Child Behavior Checklist. Frage-
bogen für Jugendliche; deutsche Bearbeitung der Youth
Self-Report Form der Child Behavior Checklist (YSR). Ein-
führung und Anleitung zur Handauswertung, bearbeitet
von M. Döpfner P. Melchers. Köln: Arbeitsgruppe Kinder-,
Jugend- und Familiendiagnostik (KJFD); 1993.
17. Arbeitsgruppe Deutsche Child Behavior Checklist. Frage-
bogen für Jugendliche; deutsche Bearbeitung der Youth
Self-Report Form der Child Behavior Checklist (YSR).
Einführung und Anleitung zur Handauswertung mit
deutschen Normen, bearbeitet von M. Döpfner, J. Plück, S.
Bölte, K. Lenz, P. Melchers K. Heim (2. Au.). Köln: Arbeits-
gruppe Kinder-, Jugend- und Familiendiagnostik (KJFD);
1998.
18. Morgan JF, Reid F, Lacey JH. e SCOFF questionnaire:
assessment of a new screening tool for eating disorders.
Brit Med J. 1999;319:1467–8.
19. Hölling H, Schlack R. Essstörungen im Jugendalter. Ergeb-
nisse aus dem Kinder- und Jugendgesundheitssurvey
(KiGGS). Ernährungsumschau. 2007;9:514–9.