Adolescent subthreshold-depression
and anxiety: psychopathology, functional
impairment and increased suicide risk
Judit Bala
´
zs,
1,2
Mo
´
nika Miklo
´
si,
3,4
A
´
gnes Kereszte
´
ny,
1,3
Christina W. Hoven,
5,6
Vladimir Carli,
7
Camilla Wasserman,
5,8
Alan Apter,
9
Julio Bobes,
10
Romuald
Brunner,
11
Doina Cosman,
12
Pa
´
draig Cotter,
13
Christian Haring,
14
Miriam
Iosue,
8
Michael Kaess,
11,15
Jean-Pierre Kahn,
16
Helen Keeley,
13
Dragan
Marusic,
17
Vita Postuvan,
17
Franz Resch,
11
Pilar A Saiz,
10
Merike Sisask,
18
Avigal Snir,
9
Alexandra Tubiana,
16
Airi Varnik,
18
Marco Sarchiapone,
8
and
Danuta Wasserman
7
1
Vadaskert Child and Adolescent Psychiatric Hospital, Budapest, Hungary;
2
Institute of Psychology, Eo
¨
tvo
¨
s Lora´nd
University, Budapest, Hungary;
3
Semmelweis University, School of Ph.D. Studies, Budapest, Hungary;
4
Heim Pa´l
Paediatric Hospital, Budapest, Hungary;
5
Department of Child and Adolescent Psychiatry, Columbia University-New
York State Psychiatric Institute, New York, NY, USA;
6
Department of Epidemiology, Mailman School of Public Health,
Columbia University, New York, NY, USA;
7
National Centre for Suicide Research and Prevention of Mental Ill-Health
(NASP), Karolinska Institute, Stockholm, Sweden;
8
Medicine and Health Science Department, University of Molise,
Campobasso, Italy;
9
Feinberg Child Study Center, Schneider Children’s Medical Center, Tel Aviv University, Tel Aviv,
Israel;
10
Department of Psychiatry, School of Medicine, University of Oviedo; Centro de Investigacio´ nBiome´dica en
Red de Salud Mental, CIBERSAM, Oviedo, Spain;
11
Section Disorders of Personality Development, Clinic of Child and
Adolescent Psychiatry, University of Heidelberg, Heidelberg, Germany;
12
Clinical Psychology Department, Iuliu
Hatieganu University of Medicine and Pharmacy, Cluj-Napoca, Romania;
13
National Suicide Research Foundation,
Cork, Ireland;
14
Research Division for Mental Health, University for Medical Information Technology (UMIT),
Innsbruck, Austria;
15
Orygen Youth Health Research Centre, The University of Melbourne, Melbourne, Vic.,
Australia;
16
Department of Psychiatry, Centre Hospitalo-Universitaire (CHU) de NANCY, Universite´ H. Poincare´,
NANCY-France;
17
Mental Health Department, PINT, University of Primorska, Primorska, Slovenia;
18
Estonian-
Swedish Mental Health & Suicidology Institute, Ctr. Behav & Hlth Sci, Tallinn University, Tallinn, Estonia;
Background: Subthreshold-depression and anxiety have been associated with significant impairments
in adults. This study investigates the characteristics of adolescent subthreshold-depression and anxiety
with a focus on suicidality, using both categorical and dimensional diagnostic models. Methods: Data
were drawn from the Saving and Empowering Young Lives in Europe (SEYLE) study, comprising 12,395
adolescents from 11 countries. Based on self-report, including Beck Depression Inventory-II (BDI-II),
Zung Self-Rating Anxiety Scale (SAS), Strengths and Difficulties Questionnaire (SDQ) and Paykel Sui-
cide Scale (PSS) were administered to students. Based on BDI-II, adolescents were divided into three
groups: nondepressed, subthreshold-depressed and depressed; based on the SAS, they were divided
into nonanxiety, subthreshold-anxiety and anxiety groups. Analyses of Covariance were conducted on
SDQ scores to explore psychopathology of the defined groups. Logistic regression analyses were con-
ducted to explore the relationships between functional impairments, suicidality and subthreshold and
full syndromes. Results: Thirty-two percent of the adolescents were subthreshold-anxious and 5.8%
anxious, 29.2% subthreshold-depressed and 10.5% depressed, with high comorbidity. Mean scores of
SDQ of subthreshold-depressed/anxious were significantly higher than the mean scores of the non-
depressed/nonanxious groups and significantly lower than those of the depressed/anxious groups.
Both subthreshold and threshold-anxiety and depression were related to functional impairment and
suicidality. Conclusions: Subthreshold-depression and subthreshold-anxiety are associated with an
increased burden of disease and suicide risk. These results highlight the importance of early identifi-
cation of adolescent subthreshold-depression and anxiety to minimize suicide. Incorporating these
subthreshold disorders into a diagnosis could provide a bridge between categorical and dimensional
diagnostic models. Keywords: Categorical diagnostic model, dimensional diagnostic model, sub-
threshold-depression, subthreshold-anxiety, adolescent, suicide, SEYLE.
Introduction
There is mounting criticism of the current classifi-
cation systems (Diagnostic and Statistical Manual of
Mental Disorders Text Revised (DSM-IV-TR) (American
Psychiatric Association, 2000) and International
Classification of Mental and Behavioral Disorders
(ICD-10) (World Health Organization, 1992), with
increasing evidence for the advantages and disad-
vantages of both categorical and dimensional ap-
proaches (Lecrubier, 2008; Mo
¨
ller, 2008; Okasha,
2009). Individuals requiring psychiatric intervention
may not receive a standard diagnosis based on the
Conflicts of interest statement: No conflicts declared.
Journal of Child Psychology and Psychiatry 54:6 (2013), pp 670–677 doi:10.1111/jcpp.12016
2013 The Authors. Journal of Child Psychology and Psychiatry 2013 Association for Child and Adolescent Mental Health.
Published by John Wiley & Sons Ltd, 9600 Garsington Road, Oxford OX4 2DQ, UK and 350 Main St, Malden, MA 02148, USA
DSM-IV-TR or ICD-10 due to an insufficient number
or duration of symptoms (Johnson, Weissman, &
Klerman, 1992). Patients with substantial functional
impairment who do not meet diagnostic criteria are
regarded as having subthreshold disorders (Judd,
Rapaport, Paulus, & Brown, 1994). Helmchen and
Linden (2000) suggest that subthreshold diagnoses
are not solely artefacts from potentially outdated
definitions, but rather unique conditions demanding
recognition. It has been suggested that implement-
ing a hybrid of categorical and dimensional
approaches in DSM-V would be useful, as both are
important for clinical work and research (Okasha,
2009).
A large number of studies have focused on child and
adolescent subthreshold-depression (Fergusson,
Horwood, Ridder, & Beautrais, 2005; Foley, Goldston,
Costello, & Angold, 2006; Keenan et al., 2008; Klein,
Shankman, Lewinsohn, & Seeley, 2009; Lewinsohn,
Solomon, Seeley, & Zeiss, 2000), showing that sub-
threshold-depression increases the risk of developing
a major depressive episode (MDE) (Pine, Cohen,
Cohen, & Brook, 1999; Shankman et al., 2009). Al-
though the high comorbidity of anxiety [specifically
generalized anxiety disorder (GAD)] and MDE are well
described (Kessler, Chiu, Demler, Merikangas, &
Walters, 2005; Unick, Snowden, & Hastings, 2009;
Wittchen, Zhao, Kessler, & Eaton, 1994), there are
still few studies on subthreshold-GAD among chil-
dren/adolescents (Foley et al., 2006; Guberman &
Manassis, 2011; Nauta et al., 2012).
Epidemiological data on child/adolescent sub-
threshold-depression vary, with 12-month preva-
lence ranging from 3% to 12%, and lifetime
prevalence through late adolescence as high as 26%
(Fergusson et al., 2005; Wittchen, Nelson, & Lach-
ner, 1998). To our knowledge, no study has exam-
ined the prevalence of subthreshold-GAD among
youth, whereas among adults, the 12-month preva-
lence was found to be 3.6–15.7% (Carter, Wittchen,
Pfister, & Kessler, 2001; Rucci et al., 2003).
This variability in epidemiological data may be
explained, in part, by different definitions and diag-
nostic methodologies. Some studies used standard-
ized clinical interviews to screen for subthreshold-
depression and subthreshold-GAD (Carter et al.,
2001; Fergusson et al., 2005; Foley et al., 2006;
Guberman & Manassis, 2011; Keenan et al., 2008;
Rucci et al., 2003; Shankman et al., 2009), others
used self-report (Lewinsohn et al., 2000) or both
(Nauta et al., 2012).
Broad and narrow definitions of child/adolescent
subthreshold-depression and subthreshold-GAD
exist with respect to both the number and the
duration of symptoms and additional criteria (e.g.
presence of distress), but there is no accepted defi-
nitions of these conditions (Angst, Merikangas, &
Preisig, 1997; Fergusson et al., 2005; Foley et al.,
2006; Karsten, Nolen, Penninx, & Hartman, 2011;
Keenan et al., 2008; Kertz & Woodruff-Borden,
2011; Klein et al., 2009; Rucci et al., 2003; Shank-
man et al., 2009).
Psychiatric disorders, especially MDE, are major
risk factors for suicidal behaviour (Gould et al.,
1998). Comorbidity, mainly anxiety disorders,
increases the risk of suicidal behaviour among ado-
lescents (Wunderlich, Bronisch, & Wittchen, 1998).
Bala´zs, Bitter, Lecrubier, Csisze´r, and Ostorharics
(2000) found that almost two thirds of adult suicide
attempters had MDE, half had GAD, one tenth had
subthreshold-depressive episode and one fifth had
subthreshold-GAD. Only a few adolescent studies
have focused on subthreshold mental disorders,
including subthreshold-depressive episodes and
GAD, as a possible risk factor for suicide. Foley et al.
(2006) examined subjects aged 9–16 years and
found that suicidal youth without a full DSM-IV-TR;
psychiatric disorder had significantly higher preva-
lence of subthreshold conditions than nonsuicidal
youth without psychiatric disorders.
The aim of the present study of European adoles-
cents was the examination of the prevalence of
subthreshold-depression and subthreshold-anxiety
and its relationships with psychopathology, func-
tional impairment and suicidal behaviour.
Method
Participants
The sampling procedures of the Saving and Empower-
ing Young Lives in Europe (SEYLE) study were previ-
ously described (Wasserman et al., 2010). SEYLE’s
sample of 12,395 adolescents (aged 14–16 years) is
from 11 European countries: Austria, Estonia, France,
Germany, Hungary, Ireland, Israel, Italy, Romania,
Slovenia and Spain, with Sweden serving as the coor-
dinating centre. Ethical approval was obtained from
each site’s local ethics committee. Local school
authorities granted access to randomly selected
school(s) and informed assent and consent were
obtained, as required.
Data collection
Students were administered a self-report questionnaire
that included well-established measures and items
developed for SEYLE (Wasserman et al., 2010).
Beck Depression Inventory-II (BDI-II) measured
severity of depression by assessing specific symptoms
experienced over the preceding 2 weeks (Beck, Steer,
Ball, & Ranieri, 1996; Byrne, Stewart, & Lee, 2004).
BDI-II item ‘loss of interest in sex’ was excluded from
the SEYLE version because it was considered inappro-
priate in some cultural settings (Byrne et al., 2004).
Symptoms of current anxiety were assessed using
Zung Self-Rating Anxiety Scale (SAS) (Zung, 1971), a
20-item self-report questionnaire. Zung referred to the
scores as an ‘Index score’ (‘normal range’: £44; ‘minimal
to moderate anxiety’: 45–59; ‘marked to severe anxiety’:
60–74; ‘extreme anxiety’: ‡75) (McDowell, 2006).
Psychopathology was evaluated using Strengths and
Difficulties Questionnaire (SDQ), a brief instrument for
doi:10.1111/jcpp.12016 Adolescent subthreshold-depression and anxiety 671
2013 The Authors. Journal of Child Psychology and Psychiatry 2013 Association for Child and Adolescent Mental Health.
screening childhood behaviours, consisting of 25 items
(Goodman, Meltzer, & Bailey, 1998). The extended
version of the SDQ was used, which includes an impact
supplement, a measure of functional impairment.
Frequency of suicidal thoughts/ideations was as-
sessed during the past 2 weeks using Paykel Suicide
Scale (PSS) (Paykel, Myers, Lindenthal, & Tanner, 1974).
Definitions
Adolescents were divided into three groups based on
BDI-II score: ‡20 = depressed (Beck et al., 1996); <20
(BDI-II) and being positive (>0) on items assessing core
symptoms of DSM-IV-TR MDE (sadness or loss of
pleasure) = subthreshold-depression; all others = non-
depressed.
Adolescents were also divided into three groups
based on the SAS: ‡60 = anxious; 45 £ and <60 = sub-
threshold-anxious; <45 = nonanxious.
Analyses
Data were analysed using IBM SPSS Statistic 20 soft-
ware package (SPSS, Inc, 2011). Gender differences
among nondepressed, subthreshold and depressed, as
well as nonanxious, subthreshold and anxious groups
were analysed using chi-squared tests. One-way ANO-
VAs were conducted to explore age differences. Means
and standard deviations, as well as percentages of
borderline and elevated SDQ total and impact scores
are reported for the three groups of anxiety/depression,
using the cut-off defined by Goodman et al. (1998). Due
to cross-cultural variation in cut-off scores of the SDQ
(Vostanis, 2006) continuous scores were used to
explore levels of overall psychopathology. Bivariate
relationships between measures of anxiety and
depression, and SDQ total score were explored using
Spearman correlations. Analyses of Covariance (AN-
COVA) with Tukey post hoc test were conducted sepa-
rately for girls and boys with age as covariate to detect
differences among the three study groups. To differen-
tiate the effects of anxiety and depression, each analysis
was controlled for by depression/anxiety (for BDI-II
total score when exploring the effect of subthreshold
and full anxiety on functional impairments and for SAS
total score when the effects of subthreshold and full
depression were studied). Logistic regression analyses
with dummy-coded SDQ impact score (0 ‘normal’ and
0 < ‘borderline/elevated scores’) and PSS total score (0
or greater) as independents, were conducted to explore
the relationships between functional impairments, as
well as suicidality and subthreshold and full anxiety/
depression. All analyses were adjusted for gender, age
and continuous score of anxiety/depression. Due to
sensitivity of Hosmer and Lemeshow goodness of fit test
for very large sample sizes (Kramer & Zimmerman,
2007), the area under the Receiver operating charac-
teristic (ROC) curve [area under the roc curve (AUC)]
was reported. A p-value of .05 was considered statisti-
cally significant. Effect size measures are also reported
for all analyses.
Multiple imputations (MI) were conducted using the
IBM SPSS Statistics 20 (2011) to account for missing
data. Five imputed datasets were created. Variables in-
cluded in the model: age, country of origin, gender, all
SDQ items, PSS, SAS and BDI-II as target variables and
predictors. Each completed dataset was analysed using
standard methods for assessing differences among non-
depressed/nonanxious, subthreshold-depressed/anx-
ious and depressed/anxious groups. Pooled estimates
were calculated using Rubin’s Rules (Rubin, 1987).
Results
Subjects
Complete data were obtained for 11,109 (89.6%) of
the 12,395 adolescents in SEYLE: 4,506 (40.7%)
boys and 6,565 (59.3%) girls. Mean age: 14.80 years
(SD = .84).
Only 17,652 (1.8%) of the total 961,553 data items
were missing. The proportion of the missing data was
greatest on the sixth item of the SDQ (‘I am usually
on my own. I generally play alone or keep to myself.’)
(10.8%), whereas all other variables had less than
5% missing. Complete cases and incomplete cases
differed significantly in age (t(12310) = )48.356, p <
.001, Cohen’s d = 1.384) and gender (v
2
(1) = 755.363,
p < .001, / = ).248), indicating that adolescents
with incomplete data tended to be older and male
subject.
MI analyses used data of all 12,395 participants:
5,529 males and 6,799 females (67 missing gender
cases). The mean age was 14.91 years (SD = .90).
Anxiety and depression
Among all 12,395 adolescents, 7,476 (60.3%) were
identified as nondepressed, 3,618 (29.2%) sub-
threshold-depressed and 1,301 (10.5%) depressed.
Analysis showed a significant age effect on group
membership (F(2) = 28.321, p < .001). However, the
effect size for this analysis (g
2
= .005) was not found
to exceed Cohen’s (1988) convention for a small ef-
fect (g
2
= .01).
Gender differences of small effect size were found
among groups of nondepressed, subthreshold-de-
pressed and depressed youth (v
2
(2) = 254.956,
p < .001, Cramer’s V = .143). Girls more frequently
were both subthreshold-depressed and depressed
(Table 1) (p < .001, / = .106 and .143 respectively).
Among all subjects, 7,708 (62.2%) were identified
as nonanxious, 3,964 (32.0%) subthreshold-anxious
and 723 (5.8%) anxious.
Among the three levels of anxiety, a significant
effect of group membership was found for age
(F(2) = 44.846, p < .001). Again, the effect size
(g
2
= .007) was not found to exceed Cohen’s (1988)
convention for a small effect.
Gender had a small effect on group membership
across levels of anxiety (v
2
(2) = 290.362, p < .001,
Cramer’s V = .154). Girls more frequently were both
subthreshold-anxious and anxious (Table 1)
(p < .001, / = .115 and .147 respectively).
Results revealed a strong relationship between
SAS and BDI-II scores (r = .503 and .656 for boys
672 Judit Bala´ zs et al. J Child Psychol Psychiatry 2013; 54(6): 670–7
2013 The Authors. Journal of Child Psychology and Psychiatry 2013 Association for Child and Adolescent Mental Health.
and girls respectively). When using a categorical
approach, a strong association was also found be-
tween levels of anxiety and depression
(v
2
(4) = 3,807.565, p < .001, c = .682) (Table 2).
Psychopathology
SDQ total score was found to be strongly related to both
SAS (r = .513, p < .001 and .619, p < .001 for boys and
girls respectively) and BDI-II scores (r = .542, p < .001
and .654, p < .001 for boys and girls respectively). After
adjusting for the scores of depression/anxiety, associ-
ations between SDQ total score and SAS score (r = .287,
p=.004 and .302, p < .001 for boys and girls respec-
tively), as well as between SDQ and BDI-II score
(r = .346, p < .001 and .396, p < .001 for boys and girls
respectively) decreased, but remained significant,
showing small to medium effect.
Descriptive statistics, as well as percentages of bor-
derline and elevated scores of SDQ scales among
groups of nondepressed/nonanxious, subthreshold-
depressed/anxious and depressed/anxious are shown
in Table 3.
Controlling for age and BDI-II score, mean SDQ
problem score differed significantly across the non-
anxious, subthreshold-anxious and anxious groups for
both genders (F(2) = 136.154, p < .001, g
2
= .047 for
boys and F(2) = 213.552, p < .001, g
2
= .059 for girls).
All pair-wise post hoc comparisons were significant
(p < .001), indicating that mean scores of subthreshold-
anxious adolescents were higher than the mean scores
of the nonanxious group (Cohen’s d = .920 and .994 for
boys and girls respectively), but lower than the mean
scores of the anxious group (Cohen’s d = .839 and
1.102 for boys and girls respectively).
Analysis resulted in a significant effect of group
membership across levels of depression on SDQ total
score for both genders (F(2) = 190.553, p < .001,
g
2
= .064 for boys and F(2) = 292.788, p < .001,
g
2
= .079 for girls). Nondepressed adolescents had sig-
nificantly lower scores than subthreshold-depressed
adolescents (p < .001, Cohen’s d = .632 and .776 for
boys and girls respectively), whereas subthreshold-de-
pressed adolescents had significantly lower scores then
depressed adolescents (p < .001, Cohen’s d = 1.126
and 1.171 for boys and girls respectively).
Functional impairment
Logistic regression analysis revealed a significant effect
of age (OR = 1.219; 95% CI = 1.171–1.269) on dummy-
coded SDQ impact score as dependent (0 = no impair-
ment; 1 = borderline/elevated scores), indicating that
for each year of increase in age increases the probability
of having functional impairments with 21.9%. Gender
also had a significant effect (OR = .726; 95% CI = .675–
.780), indicating that boys were predicted to have
functional impairment with lower probability than girls.
Adjusting for the effect of age, gender and BDI-II
score, the odds for a subthreshold-anxious adolescent
having functional impairment was 1.795 (95%
CI = 1.638–1.967) times greater than the odds for a
nonanxious adolescent. Similarly, anxious adolescents
were predicted to have functional impairment with a
2.519 (95% CI = 1.982–3.201) times greater probability
than their nonanxious counterparts (v
2
(5) = 2845.482,
p < .001, Nagelkerke-R
2
= .276, AUC = .773, 95%
CI = .776–.780).
In the second logistic regression model, when the effect
of age, gender and SAS score were controlled, there were
significant main effects of being subthreshold-depressed
(OR = 1.960; 95% CI = 1.795–2.140) and depressed
(OR = 4.102; 95% CI = 3.455–4.871) on having func-
tional impairment (v
2
(5) = 2589.091, p < .001, Nage-
lkerke-R
2
= .254, AUC = .756, 95% CI = .749–.764).
Suicidality
Descriptive statistics of PSS and percentages of positive
responders, by item, in nonanxious/nondepressed,
subthreshold-anxious/depress ed and anxious/depressed
groups are reported in Table 4.
A significant effect of age (OR = 1.174, 95%
CI = 1.126–1.225) was found when using dummy-co-
ded PSS total score (0 or greater) as dependent in the
logistic regression model, indicating that for each year
increase on age increases the probability of suicidality
with 17.4%. Gender had also a significant effect on the
dependent variable (OR = .546, 95% CI = .506–.590),
indicating that boys were predicted to have suicidal
thoughts/ideations with lower probability than girls.
Adjusting for the effect of age, gender and BDI-II score,
the odds for a subthreshold-anxious adolescent for hav-
ing suicidal thoughts/ideations were 1.788 (95%
CI = 1.622–1.971) times greater than the odds for a
nonanxious adolescent. Similarly, anxious adolescents
were predicted to have suicidal thoughts/ideations with a
2.756 (95% CI = 2.159–3.518) times greater probability
than their nonanxious counterparts (v
2
(5) = 3739.359,
Table 1 Gender distribution among nondepressed/nonanx-
ious, subthreshold-depressed/anxious and depressed/anx-
ious groups
Levels of anxiety
Levels of
depression
Boys % Girls % Boys % Girls %
No anxiety/depression 50.35 49.65 50.35 49.65
Subthreshold- anxiety/
depression
38.24 61.76 38.24 61.76
Full anxiety/depression 24.07 75.93 24.07 75.93
N = 12,395.
Table 2 Percents of levels of anxiety and depression
Levels of Anxiety
Totals
NonA % SubA % A %
Levels of
depression
NonD % 46.32 13.39 .60 60.31
SubD % 14.75 12.98 1.46 29.19
D % 1.11 5.61 3.77 10.50
Totals 62.18 31.98 5.83 100.00
v
2
(4) = 3,807.565 p < .001; c = .682.
N = 12,395. NonA, nonanxious group; SubA, subthreshold-
anxious group; A, anxious group; NonD, nondepressed group;
SubD, subthreshold-depressed group; D, depressed group.
doi:10.1111/jcpp.12016 Adolescent subthreshold-depression and anxiety
673
2013 The Authors. Journal of Child Psychology and Psychiatry 2013 Association for Child and Adolescent Mental Health.
p < .001, Nagelkerke-R
2
= .361, AUC = .812, 95%
CI = .805–.819).
When assessing the effect of levels of depression on
suicidality, we found significant main effects of being
subthreshold-depressed (OR = 3.065; 95% CI = 2.792–
3.364) and depressed (OR = 9.210; 95% CI = 7.700–
11.016) when the effect of age, gender and SAS scores
were controlled for (v
2
(5) = 3492,978, p < .001, Nage-
lkerke-R
2
= .340, AUC = .800, 95% CI = .793–.807).
Discussion
Similar to the study of Angst et al. (1997) on a
community sample of individuals ages 19–20, in the
current study approximately half of the adolescents
met the criteria for threshold and/or subthreshold-
depression and/or anxiety. Although our focus was
on subthreshold-depression and subthreshold-anx-
iety, it is noteworthy that based on screening tools an
exceptionally high proportion of this sample was
categorized as depressed (10.5%) and anxious
(5.8%). Similar to prior findings, our results show a
high prevalence of subthreshold-depression and
anxiety among adolescents throughout Europe
(Fergusson et al., 2005; Wittchen et al., 1998).
According to our data, almost one third of adoles-
cents had current subthreshold-depression and one
third had current subthreshold-anxiety. Impor-
tantly, even these less severe cases were associated
with elevated levels of psychopathology, and
increased risk for functional impairment and
suicidality.
No difference in the prevalence of depression
among preadolescent boys and girls has been de-
scribed (Anderson, Williams, McGee, & Silva, 1987;
Kashani et al., 1983). After ages 11–13 and
throughout adulthood, this trend changes and fe-
male subjects are approximately twice as likely as
male subjects to be depressed and the same preva-
lence estimate is true for anxiety disorders (Angold,
Costello, & Worthman, 1998; Mackinaw-Koons &
Vasey, 2000). In the current study, we observed
similar gender distributions in the depressed and
anxiety groups and in the subthreshold-depressed
and subthreshold-anxiety groups as well.
In this large international sample, we found a
strong correlation between depression according to
BDI-II and anxiety according to SAS. While high
comorbidity of both threshold and subthreshold-
depression and anxiety was expected (Kessler et al.,
2005; Unick et al., 2009; Wittchen et al., 1994), it is
still surprising that only one tenth of all adolescents
with threshold-depression or threshold-anxiety had
‘pure forms’ of the disorders. The presence of co-
morbid (even subthreshold) MDE and anxiety is
associated with more severe psychopathology,
greater impairment, increased suicidality and worse
outcome than in noncomorbid conditions (Altamura,
Montresor, Salvadori, & Mundo, 2004; Foley et al.,
Table 3 Mean scores, standard deviations, as well as percentages of borderline and elevated scores of SDQ scales among groups of
nondepressed/nonanxious, subthreshold-depressed/anxious and depressed/anxious.
SDQ Total sample
Levels of anxiety Levels of depression
NonA SubA A NonD SubD D
Problem scale
Borderline % 11.7 5.1 20.6 33.3 5.5 16.6 33.4
Elevated % 5.7 1.5 8.6 34.9 1.5 5.5 30.0
M (SD) 10.74 (5.10) 8.89 (4.22) 13.07 (1.80) 17.77 (4.76) 8.97
(4.30)
12.45 (4.46) 17.07 (4.57)
Impact scale
Borderline % 4.2 4.8 3.0 1.1 5.1 3.6 1.4
Elevated % 37.2 23.9 55.3 83.6 23.7 49.1 82.0
Median (Interquartile Range) 0 (0–3) 0 (0–1) 2 (0–5) 6 (3–10) 0 (0–1) 1 (0–4) 6 (3–10)
N = 12,395. SDQ, Strength and Difficulties Questionnaire; NonA, nonanxious group; SubA, subthreshold-anxious group; A,
anxious group; NonD, nondepressed group; SubD, subthreshold-depressed group; D, depressed group.
Table 4 Descriptive statistics of PSS and per cent of positive responders by item in nondepressed/nonanxious, subthreshold-
depressed/anxious and depressed/anxious groups
Total sample
Levels of anxiety Levels of depression
NonA SubA A NonD SubD D
PSS Median (Intequartile Range) 0 (0–1) 0 (0–0) 1 (0–3) 4 (1–10) 0 (0–0) 0 (0–2) 4 (2–9)
Life not worth living during past 2 weeks % 27.00 14.78 41.95 75.29 12.46 38.93 77.35
Wish that were dead during past 2 weeks % 18.52 8.21 30.00 65.61 6.99 25.08 66.52
Thought of taking own life during past 2 weeks % 16.79 7.86 26.74 57.50 6.74 21.76 60.70
N = 12,395. PSS, Paykel Suicide Scale; NonA, nonanxious group; SubA, subthreshold-anxious group; A, anxious group; NonD,
nondepressed group; SubD, subthreshold-depressed group; D, depressed group.
674 Judit Bala´ zs et al. J Child Psychol Psychiatry 2013; 54(6): 670–7
2013 The Authors. Journal of Child Psychology and Psychiatry 2013 Association for Child and Adolescent Mental Health.