Content uploaded by Anne Mari Sund
Author content
All content in this area was uploaded by Anne Mari Sund on Apr 26, 2019
Content may be subject to copyright.
Research report
Specific depressive symptoms and disorders as associates and
predictors of suicidal acts across adolescence
Latha Nrugham
a,
⁎, Bo Larsson
a
, Anne Mari Sund
a,b
a
Regional Centre for Child and Adolescent Mental Health, Department of Neuroscience, Faculty of Medicine,
Norwegian University of Science and Technology, Trondheim, Norway
b
Department of Child and Adolescent Psychiatry, St.Olav's University Hospital, Trondheim, Norway
Received 25 July 2007; received in revised form 6 February 2008; accepted 7 February 2008
Available online 18 April 2008
Abstract
Objective: To examine the role of depressive symptoms and disorders as associates and predictors of suicidal acts across adolescence.
Method: A representative sample of Norwegian school students (N= 2464, mean age 13.7 years) in grades 8 and 9 was reassessed after
one year (T2) with the same questionnaire. All high scorers of depressive symptoms on the Mood and Feelings Questionnaire (MFQ)
at T2 were defined as cases. One control from low or middle scorers, matched for age and gender, was randomly assigned to every two
cases. This subset (n= 345) was diagnostically assessed by face-to-face K-SADS-PL interviews (mean age = 14.9 years). The same
subset was reassessed after 5 years (T3) by using the same questionnaire (n= 252, mean age = 20.0 years) and telephone K-SADS-PL
interviews (n= 242). The participation rate was 76.9% (n= 265).
Results: Cognitive symptoms dominated the depressive symptom profile among suicide attempters, irrespective of age and time.
Among younger adolescents, suicidal thoughts and acts of self-harm without suicidal intent were associated with suicidal acts.
Recurrent thoughts about death, hopelessness, disturbed concentration and middle insomnia were associates of suicidal acts among
older adolescents.
Worthlessness by 15 years was a significant predictor of suicidal acts between 15 to 20 years. MDD and a depressive episode,
not otherwise specified, continued to be significant associates among younger adolescents, while dysthymia by 15 years remained a
predictor of suicidal acts between 15 to 20 years, even when controlled for depressive symptoms.
Conclusions: Self-harm without suicidal intent, middle insomnia, cognitive depressive symptoms and a formal psychiatric
diagnosis of any depressive disorder should alert professionals in the risk assessment of suicidal adolescents.
© 2008 Elsevier B.V. All rights reserved.
Keywords: Longitudinal; Case-control; School adolescents; Young adults
1. Introduction
The assessment of suicidal youth is one of most com-
mon and demanding emergencies in adolescent mental
health services (Brent, 2001). The ratio of suicide attempts
to suicide completions is higher in adolescence than in
any other age group (King, 1997). The prevalence of
Journal of Affective Disorders 111 (2008) 83–93
www.elsevier.com/locate/jad
⁎Corresponding author. RBUP-INM, Klostergate 46/48, MTFS
N-7489.
E-mail address: latha.nrugham@medisin.uio.no (L. Nrugham).
0165-0327/$ - see front matter © 2008 Elsevier B.V. All rights reserved.
doi:10.1016/j.jad.2008.02.010
suicide attempts starts increasing from the age of 14 years
onwards (Lewinsohn et al., 2001a). Up to 32% of
clinically referred adolescents will attempt suicide at
least once by early adulthood (Kovacs et al., 1993), while
20% will repeat the attempt (Harrington et al., 1994). In
community settings, 70–91% of youth who attempt or
think about suicide have a psychiatric disorder (Gould
et al., 1998). Research on suicidal phenomena among
adolescents has mainly explored for other risk factors
associated with depressive disorders (Brent et al., 1994;
Lewinsohn et al., 2001a; Bridge et al., 2006), since most
persons with mood disorders do not commit suicide and
about half of them never attempt suicide (Rihmer, 2007).
Depressed mood has been found to be the most
common depressive symptom among 11 to 18 year old
adolescent inpatients and outpatients who have attempted
suicide (Bettes and Walker, 1986). Among adolescent
emergency unit inpatients, high levels of self-reported
depressive symptoms predicted suicide attempts in a 2 to
4yearfollow-up(Ivarsson et al., 1998). Among 13 to
17 year olds, hopelessness, negative self-esteem and
violent behaviour were more common among depressed,
suicidal adolescents than among those who are depressed
but non-suicidal (Csorba et al., 2003). A recent review on
sleep and suicidal behaviour among youth has called for
the use of longitudinal studies due to inconsistencies in
findings of cross-sectional studies, wherein the associa-
tions with insomnia often disappeared when controlled
for depression (Liu and Buysse, 2006).
Major depressive disorder (MDD) and dysthymia were
specifically associated with higher rates of suicidal
behaviours than other diagnoses among outpatients aged
8–13 years, followed up for five years (Kovacs et al.,
1993). MDD has also been significantly associated with
suicidal acts among older school adolescents (Lewinsohn
et al., 1994).
Although current empirical research has added con-
siderably to our knowledge on sucidality among
adolescents, gaps continue to exist about specific aspects
which can potentially aid the clinician in the assessment
of suicide risk among adolescents, for example: are there
specific depressive symptoms and disorders related to
suicidal acts? Do these relationships differ with age and
time? To the best of our knowledge, longitudinal studies
exploring the role of specific depressive symptoms and
disorders as associates and predictors of suicidal acts
among adolescents have not yet been reported.
The aim of the current study was, therefore, to ex-
amine relationships between specific depressive symp-
toms and disorders with suicidal acts in a sample of
depressed school adolescents and non-depressed controls
followed for a five-year period into early adulthood. We
first studied the prevalence of depressive symptoms and
disorders in adolescents with high mean scores on
depressive symptoms as compared to those with low to
moderate mean scores as reported in a questionnaire. The
next objective was to examine the specificity of
depressive symptom and diagnostic profiles as associates
of suicidal acts at two developmental phases, i.e., by
15 years and for the next 5-year period as assessed
retrospectively at 20 years of age. Thirdly, we explored
the role of specific depressive symptoms and disorders by
the age of 15 years as potential prospective predictors of
suicidal acts between 15 to 20 years of age. Finally, we
investigated the role of self-harm without suicidal intent
as an associate and predictor of suicidal acts, across age
and time, given its complex relationship with suicidal
acts (Grøholt et al., 2000; Jacobson and Gould, 2007;
Larsson and Sund, 2008).
2. Methods
2.1. Participants and procedures
The participants were selected from a school sample
of adolescents in 8th and 9th grades in 1998 from two
counties in Central Norway. A cluster sampling tech-
nique resulted in a representative sample of 2792 stu-
dents from 22 schools. Based on prevalence estimates
of 5–10% of MDD among adolescents in the gen-
eral population (Lewinsohn et al., 1998; DSM-IV-TR,
2000), we calculated that a sufficient number of
participants with a formal diagnosis of depressive dis-
order would be included in the present study to provide
powerful comparisons with non-depressed adolescents.
A detailed description of the sampling procedure and
the original sample is provided elsewhere (Sund et al.,
2001, 2003).
2.2. Assessment points
At the first assessment, (T1), 2464 (N
1
) adolescents
with a mean age of 13.7 (SD = 0.5) years and a par-
ticipation rate of 88.3% (50.8% females) entered the
project. A questionnaire, including a screening measure
for depressive symptoms, was completed at school. At
the second assessment (T2), 2432 (N
2
) adolescents with
a mean age of 14.9 (SD = 0.5) years, with a participation
rate of 86.7% (50.3% females) completed the same
questionnaire again at school.
2.2.1. Subset selection
On the basis of the mean total scores for depressive
symptoms on the Mood and Feeling Questionnaire
84 L. Nrugham et al. / Journal of Affective Disorders 111 (2008) 83–93
(MFQ) (described below), at T2, the participants were
grouped into three levels: low (0–6), middle (7–24) and
high (25 and above) scorers. All high scorers were
regarded as cases. One control for every two cases was
selected at random from the low and middle scorers,
matched for age and gender. Of the 364 adolescents thus
selected, 345 were diagnostically interviewed face-to-
face by trained interviewers at school. The cases
numbered 224 (64.9%) and the controls, 121 (35%).
The participation rate was 94.7% and included 72.5%
females.
At the third assessment (T3), 265 young adults (N
3
)
with a mean age of 20.0 (SD = 0.5) years and a
participation rate of 76.9% (77% females) were
reassessed. Two hundred and fifty-two young adults
(73%) completed the questionnaire with a mean age of
20.0 (SD = 0.6) years, of whom 77% of them were
females. For the interviews, the participation rate was
70.1% (n= 242) with a mean age of 20.0
(SD = 0.6) years, of whom 76.9% were females. Of the
224 high scorers on the MFQ at T2, 155 were re-
interviewed at T3, with a participation rate of 69.1%.
The non-participants at T3 were significantly more
often males, [χ
2
(1)= 6.5, pb0.01]. While the non-
participants reported significantly higher depressive
symptoms scores on the MFQ at T1 (M= 12.8,
SD= 11.0) than the participants (M=10.3, SD=8.8), t
(343)= −2.0, pb0.05 (effect size of r=.10), no difference
between the participants and non-participants was found
for mean depressive symptom scores on the MFQ, de-
pressive disorders or suicidal acts as assessed at T2.
2.3. Measures
2.3.1. Questionnaire
2.3.1.1. Mood and Feelings Questionnaire (MFQ).
This measure, which was developed for 8–18 year olds
and includes all depressive symptoms based on DSM-III-
R criteria for major depression (Angold, 1989), was
chosen since it is adolescent-friendly in its wording. The
cut-off score for grouping of high scorers as a category
was chosen on the basis of the assessment of the 2464
adolescents with the MFQ one year (T1) before the
interviewing study, when the 90th percentile was at 24.
The participants were asked about their feelings and
behaviours in the last 2 weeks for 34 items rated on a 0–2
scale with a total score range of 0–68. Psychometric
properties of the MFQ in the original sample have been
found to be satisfactory (Sund et al., 2001). A detailed
analysis of MFQ scores and suicidal acts across
adolescence revealed that respondents with suicidal acts
had significantly higher mean scores at all three
assessment points (Nrugham et al., 2008).
2.3.2. Interview
2.3.2.1. The Kiddie –Schedule for Affective Disorders
and Schizophrenia –Present and Lifetime version (K-
SADS-PL). This well established semi-structured
diagnostic interview is built to assess present and past
episodes of psychopathology in children and adolescents
on Axis I according to DSM-III-R and IV-TR criteria
(Kaufman et al., 1997). Each individual symptom is rated
on a 0–3 scale, with a score of 3 representing clinical
threshold.
Experienced clinicians trained in psychopathology
and the use of K-SADS-PL conducted the interviews.
The interviewers were blind to the case/control status
of the participant, at both interview assessments, i.e.,
T2 and T3. The average interval between filling out
the questionnaires and interviews was 20 days at T2
and 21 days at T3. Inter Rater Reliability (IRR) for all
K-SADS symptoms at the end of training and before
initiation of real interviews, was good, with Cohen's
kappa of 0.71 at T2 and 0.70 at T3 with the third
author, an experienced child psychiatrist. Interview in-
tegrity was maintained at T2 and T3 with average
kappas of 0.83 and 0.80, respectively, for all screening
symptoms and affective supplement symptoms.
At T2, 79.8% of the adolescents had at least one parent
as a separate informant. Summary symptom scores and
diagnostic assessments were based on information ob-
tained from both informants. A diagnosis of depressive
disorder and its co-morbidity was investigated for the
lifetime period, i.e., up to15 years of age. At T3, the young
adults were randomly allocated to interviewers.
The interviews at T3 were conducted on telephone
and explored psychopathology for the five-year follow-
up period retrospectively, i.e., from 15 to 20 years of
age. In a comparative study of follow-up interviews of
young adults, the use of telephone interviews as follow-
ups of face-to-face interviews has been found to be as
reliable as face-to-face psychiatric interviews for Axis I
and II psychopathology with substantial economic and
logistic advantages (Rohde et al., 1997). The K-SADS
PL interviews also provide information regarding
functioning levels in three areas: family, school/work
and friends.
2.3.3. Key variables
2.3.3.1. Depressive symptoms. K-SADS-PL provides
a list of 30 items, with guidelines to assess sub-clinical
85L. Nrugham et al. / Journal of Affective Disorders 111 (2008) 83–93
and clinical thresholds, to probe for all DSM-IV-TR
symptoms in a depressive episode, including all sub-
types of major depression and dysthymia. The screening
section for a depressive episode has eight items and a
threshold score in the screening indicates the use of
supplement items. A clinical threshold score on items
assessing various aspects of depressive symptoms was
required for entry into analysis. Medical lethality of
suicidal acts was excluded from the present study since
it is not a DSM-IV-TR criterion for a depressive episode
or a subtype of depressive episode.
2.3.3.2. Depressive disorders. Diagnoses of depres-
sive disorders included Major Depressive Disorder
(MDD) and dysthymia based on DSM-IV-TR criteria.
However, in the present study, a diagnosis of Depressive
Disorder Not Otherwise Specified (DD-NOS) required
at least three depressive symptoms lasting for at least
two weeks. Probes for a diagnosis of any depressive
disorder were limited to the most serious past episode.
Functional impairment, as measured by Childhood-
Global Assessment of Symptoms (C-GAS), was a
requirement for a diagnosis of MDD or dysthymia
during the diagnostic assessments (Shaffer et al., 1983).
All respondents with a formal diagnosis of MDD had
either reduced functioning or a CGAS score below 71.
However, two young adults had a diagnosis of MDD at
T3, with a C-GAS score above 71, in spite of reduced
functioning. For the purpose of analysis, current and
past diagnoses were pooled together, as was done with
depressive symptoms. For example, an episode of MDD
present at anytime up to the age of 15 years was
recorded as MDD, irrespective of whether it was current
or past. However, if, one individual had more than one
diagnosis of depressive disorder, each such diagnosis
was considered (Helzer et al., 2006).
2.3.3.3. Suicidal acts. Positive responses to items on
suicidal acts, either in the questionnaire or the interview
were used. Suicidal acts and acts of self-harm without
suicidal intent were assessed on separate items, both in
the interview and questionnaire. Acts of self-harm
without suicidal intent were differentiated from suicidal
acts, either by the respondent in the questionnaire or the
interviewer using either seriousness of suicidal intent or
lethality or both. Ambiguous instances were defined as
sub-threshold levels of suicidal acts. Acts of self-harm
without suicidal intent and sub-threshold levels of
suicidal acts were excluded from the definition of
suicidal acts.
The item used in the questionnaire was the same as in
a previous, national survey of youth, ‘Young in Norway’
(Wichstrøm, 2000): “Have you ever tried to commit
suicide?”. The response options were: “No, never”;
“Yes, once”;“Yes, several times”. All respondents who
answered positive to this item were defined as at-
empters. Questions on suicidal behaviour were parts of
the screening probes for depression in the K-SADS. A
positive response to the question: “Have you ever (or
since the last interview) tried to kill yourself or done
something which could have killed you?”and assessed
to have reached the clinical threshold level by the
interviewer was defined as a suicidal act. Thus, two
groups were formed: those who reported never to have
attempted a suicidal act as non-attempters and those who
had reported a suicidal act at any one assessment (T1,
T2, or T3), as attempters.
The assessments were approved by the local school
authorities, the school boards and the Regional
Committee for Medical Research Ethics, Central Nor-
way. Based on standards prescribed by The Norwegian
Data Inspectorate, informed consent was obtained from
the participants at all assessment points.
2.4. Statistics
Descriptive methods were used to examine frequencies
of depressive symptoms and disorders. A series of bi-
variate logistic regression models were conducted to
examine crude estimates for potential associates and pre-
dictors. All depressive symptoms and disorders providing
significant Odds Ratios (ORs) at pb.01 level were
subsequently entered into multivariate logistic regression
analyses run separately for age and time. Multicollinearity
diagnostics revealed minor problems for data at T3 with
hypersomnia (.29) and hopelessness (.46) loading to-
gether leading to both being excluded from analyses
(Field, 2005). The multivariate analyses used the standard
method of entry and significance level was set to pb.05.
3. Results
3.1. Frequency of depressive symptoms and disorders
by 15 years of age and between 15 to 20 years of age
Depressed mood was found to be the most frequent
symptom at both assessment points (T2 and T3), (see
Table 1), while the next highest frequencies changed
according to age and time. The frequency and distribu-
tion of depressive diagnoses also differed with age. In
the younger age group, MDD and DD-NOS constituted
76.1% of the depressive diagnoses with a roughly equal
distribution between them, while among older sub-
jects, these two diagnoses accounted for 86.6% of the
86 L. Nrugham et al. / Journal of Affective Disorders 111 (2008) 83–93
depressive diagnoses and MDD dominated the picture.
However, higher mean scores of depressive symptoms
on the MFQ at 15 years were more strongly associated
with suicidal acts by 15 years [OR (95% CI) = 6.5(3.1–
13.5)] than with any depressive disorders by 15 years
[OR (95% CI) = 4.4(2.7–7.4)].
3.2. Associates of suicidal acts performed by 15 years
of age
The results of the first set of bivariate logistic re-
gression analyses which examined depressive symptoms
and disorders assessed at 15 years (T2), retrospectively for
lifetime, (see Tabl e 2), showed that suicidal thoughts, self-
harm without suicidal intent, Circadian reversals and
hopelessness were most strongly associated with suici-
dal acts. Among diagnoses, MDD and DD-NOS were
significant and strong associates of suicidal acts.
3.3. Associates of suicidal acts performed between 15
to 20 years of age
The results of the next set of bivariate logistic regres-
sion analyses which focussed on the role of depressive
Table 1
Percentages of depressive symptoms and diagnoses in a subset of non-depressed and depressed school adolescents as assessed by K-SADS interviews
at two assessments, T2 (at 15 years) and T3 (at 20 years) and grouped into high and low scorers on the Mood and Feelings Questionnaire (MFQ)
T2 (n= 345) T3 (n= 242)
Depressive symptoms/diagnoses MFQ low
scorers
(n= 121)
MFQ high
scorers
(n= 224)
All MFQ low
scorers
(n = 87)
MFQ high
scorers
(n= 155)
All
Depressed mood 24 58 46.1 32.2 57.4 48.3
Irritable mood 14.9 32.6 26.4 13.8 28.4 23.1
Anhedonia 9.1 18.8 15.4 13.8 25.8 21.5
Initial insomnia 3.3 24.1 16.8 14.9 32.9 26.4
Middle insomnia 1.7 8.9 6.4 6.9 19.4 14.9
Terminal insomnia 2.5 7.1 5.5 6.9 9.7 8.7
Circadian reversal 1.7 11.6 8.1 9.2 21.3 16.9
Non-restorative sleep 6.6 22.8 17.1 20.7 31 27.3
Hypersomnia 0.8 10.3 7 12.6 26.5 21.5
Decreased appetite 2.5 13.8 9.9 12.6 19.4 16.9
Weight loss 0 5.4 3.5 3.4 5.8 5
Increased appetite 0.8 5.8 4.1 2.3 7.7 5.8
Weight gain 0 1.8 1.2 0 2.6 1.7
Psychomotor agitation 5.8 5.8 5.8 3.4 9 7
Psychomotor retardation 1.7 9.4 6.7 8 12.3 10.7
Fatigue 11.6 34.8 25.5 25.3 40.6 35.1
Disturbed concentration 8.3 34.8 26.7 16.1 31.0 25.6
Indecisiveness 3.3 14.3 10.4 6.9 11.6 9.9
Worthlessness 15.7 26.3 22.6 20.7 31.6 27.7
Excessive guilt 3.3 13.8 10.1 11.5 18.7 16.1
Hopelessness 9.1 21.9 17.4 11.5 27.1 21.5
Recurrent thoughts about death 10.7 38.4 28.7 11.5 27.1 21.5
Suicidal thoughts 5.8 21.4 15.9 12.6 21.3 18.2
Suicidal acts 3.3 24.1 8.1 1.1 10.3 7
Self-harm without suicidal intent 0.8 7.1 4.9 4.6 12.3 9.5
Non-reactive mood 11.6 25 20.3 19.5 32.9 28.1
Depressed mood qualitatively
different from grief
13.2 33.0 26.1 11.5 24.5 19.8
Diurnal variation —morning 1.7 9.8 7 1.1 3.9 2.9
Diurnal variation —evening 13.2 26.8 22 11.5 21.9 18.2
Increased sensitivity to rejection 9.1 18.3 15.1 5.7 7.7 7
MDD 5.7 25.9 18.8 19.5 38.7 31.8
Dysthymia 5.7 15.1 11.9 3.4 6.4 5.4
DD-NOS 10.7 23.6 19.1 4.5 5.8 5.4
87L. Nrugham et al. / Journal of Affective Disorders 111 (2008) 83–93
symptoms and disorders as assessed retrospectively at
20 years of age (T3) (Table 2), showed that recurrent
thoughts about death, suicidal thoughts, hopelessness
and self-harm without suicidal intent were the strongest
associates of suicidal acts. All sleep disturbances, except
terminal insomnia, were significantly associated with
suicide attempts. Although only MDD was a significant
associate with suicidal acts, the Odds Ratio (OR) for
MDD in the older age group was twice as high for those
in the younger age group.
3.4. Predictors of suicidal acts performed between 15
to 20 years of age
The results of the last set of bivariate logistic regres-
sion analyses which investigated depressive symptoms
and disorders reported by 15 years (T2) (Table 2),
showed that suicidal acts, depressed mood, worthless-
ness and excessive guilt were the strongest prospective
predictors. However, the number of significant depres-
sive symptoms was fewer than those obtained in the
Table 2
Results of bivariate logistic regression of depressive symptoms and diagnoses as associates and predictors of suicidal acts (SA) across adolescence
Depressive symptoms/
diagnoses by 15 years as
associates of suicidal acts
by 15 years
(n= 345) (SA = 86)
Depressive symptoms/diagnoses
between 15–20 years as
associates of suicidal acts
between 15–20 years
(n= 242) (SA = 34)
Depressive symptoms/
diagnoses by 15 years as
predictors of suicidal acts
between 15–20 years
(n= 265) (SA = 36)
Depressive symptoms χ
2
OR (95% CI) χ
2
OR (95% CI) χ
2
OR (95% CI)
Depressed mood 22.3** 3.3(1.9–5.5) 32.8*** 14.6(4.3–49.4) 20.1*** 5.7(2.4–13.4)
Irritable mood 9.1** 2.2(1.3–3.8) 11.1** 3.7(1.7–7.9) 6.1* 2.4(1.2–5)
Anhedonia 6.3* 2.2(1.2–4.1) 23.2*** 6.7(3.1–14.5) 10.7** 3.8(1.7–8.1)
↓Appetite 16.3*** 4.5(2.1–9.3) 5.7* 2.8(1.2–6.4) 1.8 2(.7–5.2)
Weight loss 3.6 3.1(.9–10) 1 2.1(.5–8.3) 2.7 3.8(.8–16.6)
↑Appetite 6.7** 4.2(1.4–12.6) 7.4** 5.3(1.7–16.5) 3.7* 3.7(1.1–12.6)
Weight gain 1.1 3(.4–22) 2.9 6.4(.8–47.3) .6 2.9(.2 –28.1)
Initial insomnia 26.2*** 4.7(2.6–8.5) 10.1** 3.4(1.6–7.2) 8.8** 3.3(1.5–7)
Middle insomnia 4.5 2.6(1.1–6.4) 17*** 5.9(2.6–13.3) .04 .8(.1–3.7)
Terminal insomnia 4.6 2.8(1.1–7.2) 5.6 3.5(1.3–9.6) .6 .4(.06–3.5)
Circadian reversal 17.4*** 5.4(2.4–12.1) 16.8*** 5.5(2.5–12.2) 3.2 2.6(.9 –6.9)
Non-restorative sleep 25.1*** 4.5(2.5–8.1) 14.6*** 4.3(2–9.1) 2.8 2(.9–4.4)
Hypersomnia 2 1.9(.8–4.5) 10.4** 3.6(1.6–7.8) 7.1** 4.1(1.5–10.8)
Agitation 3.9 2.5(1.03–6.4) 5.3 3.8(1.3–11.1) .05 1.2(.2–5.7)
Retardation 5.8 2.9(1.2–6.9) 5.5 3.2(1.2–8.2) .08 .8(.1–3.5)
Fatigue 7.4** 2(1.2–3.5) 20.8*** 5.7(2.6–12.8) 5* 2.3(1.1–4.7)
Disturbed concentration 20.7*** 3.3(2.0–5.6) 36.4*** 10.7(4.7–24.3) 10** 3.1(1.5–6.4)
Indecisiveness 1.4 1.5(.7–3.3) 12.6*** 5.7(2.3–14.4) 2.9 2.3(.9–5.7)
Worthlessness 29.9*** 4.5(2.6–7.8) 28*** 7.7(3.5–17.2) 20.9*** 5.4(2.6–11)
Excessive guilt 12.4*** 3.6(1.8–7.5) 14.8*** 5.1(2.3–11.4) 13.4*** 5.2(2.2–11.8)
Hopelessness 30.3*** 5.2(2.9–9.4) 46.3*** 15.4(6.6–35.6) 6* 2.7(1.2–5.8)
Death thoughts 31.2*** 4.3(2.5–7.2) 58*** 22.7(9.3–55.5) 6.2* 2.4(1.2–4.9)
Suicidal thoughts 49.0*** 8.8(4.7–16.7) 56*** 22.5(9.4–53.8) 3.6* 2.2(1–4.9)
Suicidal acts 14.4*** 6.2(2.6–14.8)
Self-harm —no suicidal intent 16.3*** 8(2.7–23.7) 31.8*** 15.4(5.9–40.2) .4 .5(.06–4)
Non-reactive mood 23.0*** 3.9(2.2–6.9) 23.4*** 6.4(2.9–14) 3.7* 2.1(1–4.5)
Mood different from grief 9.6** 2.3(1.3–3.9) 5.2* 2.6(1.1–5.7) 6.3* 2.5(1.2–5.1)
Diurnal variation —morning 7.3** 3.2(1.4–7.6) 1 2.5(.4–13.6) .9 1.8(.5–5.6)
Diurnal variation —evening 5.2* 1.9(1.1–3.3) 3 2.1(.9–4.8) 2.7 1.9(.9–4)
↑Sensitivity to rejection 8.6** 2.5(1.3–4.7) 2.9 2.8(.9–8.5) 1.4 1.7(.7–4)
Depressive diagnoses
MDD 20*** 3.7(2.1–6.9) 29.4*** 8.2 (3.6–18.4) 14.1*** 4.2(2–8.7)
Dysthymia 4.5* 2.1(1.05–4.2) 5.1* 4.3 (1.3–14) 13.3*** 4.8(2.1–10.6)
DD-NOS 12.2*** 2.8(1.5–4.9) .02 1.1 (.2 –5.2) .8 1.4(.6–3.3)
Model Chi-square, p-values and odds ratios (95% confidence intervals).
Bold = variables entered into multivariate analyses. *pb.05, **pb0.01, ***pb0.001.
88 L. Nrugham et al. / Journal of Affective Disorders 111 (2008) 83–93
cross-sectional analyses. MDD and dysthymia were
both, strong and significant predictors of suicidal acts.
3.5. Depressive symptoms and disorders as associates
and predictors of suicidal acts across adolescence:
results of multivariate analysis
3.5.1. Associates of suicidal acts performed by 15 years
of age
The multivariate logistic regression model for
associations between suicidal acts and statistically sig-
nificant depressive symptoms by 15 (T2), had two
depressive symptoms, i.e., suicidal thoughts and self-
harm without suicidal intent (Table 3). This cross-
sectional model had a χ
2
value of 82.4(df = 19),
pb.0005 and the Nagelkerke's R
2
value was at .31.
The Hosmer and Lemeshow statistic of 3.4 (df = 5) was
non-significant at .63. This model correctly classified
82% of all respondents, and 43% of the attempters.
When these depressive symptoms were compared
with the two depressive diagnoses which had emerged
as significant associates of suicidal acts by 15 years in
the bivariate analyses, a significant multivariate model
was obtained, with a χ
2
value of 63.5 (df = 4), pb.0005
and the Nagelkerke's R
2
value was at .24. The Hosmer
and Lemeshow statistic of 0.6 (df= 3) was non-sig-
nificant at .89. This model correctly classified 79.9% of
all respondents, and 40.7% of the attempters. This final
model of associates of suicidal acts by 15 years (T2) had
four significant variables, two depressive diagnoses:
MDD, OR (95%CI) = 2.6 (1.3–5.2), pb.01; DD-NOS,
OR (95%CI) = 2.3 (1.1–4.5), pb.05; and two depressive
symptoms: self-harm without suicidal intent, OR(95%
CI) = 3.5 (1.1–11.5), pb.05; and suicidal thoughts, OR
(95%CI) = 4.3 (2.1–8.8), pb.0005.
3.5.2. Associates of suicidal acts performed between 15
to 20 years of age
The multivariate logistic regression for associations
between suicidal acts and depressive symptoms between
15 to 20 years, as assessed at 20 years of age (T3), had
six depressive symptoms, three cognitive and three
somatic (Table 3). However, two of the significant
symptoms, i.e., increased appetite and initial insomnia,
were protective in nature. This cross-sectional model
had a χ
2
value of 102.8 (df = 19), pb.0005 and the
Nagelkerke's R
2
value was .62. The Hosmer and
Lemeshow statistic of 17.1 (df =4) was significant at
.002. This model correctly classified 93% of all
respondents and 67.6% of the attempters.
When these significant depressive symptoms were
compared with MDD, the depressive diagnosis which had
emerged as significant associates of suicidal acts between
15 to 20 years of age in the bivariate analyses, a significant
multivariate model was obtained, with a χ
2
value of 90.7
(df=7), pb.0005 and the Nagelkerke's R
2
value was at
.56. The Hosmer and Lemeshow statistic of 2.4 (df = 4)
was non-significant at .65. Overall, this model correctly
classified 85.8% of all respondents, and 89.7% of the
attempters. Although this final model did not retain MDD,
it included the following significant depressive symptoms:
recurring thoughts of death, OR (95%CI)= 14.6 (4.3–
49.1), pb.0005; disturbed concentration, OR (95%CI) =
16.3 (3.8–70.1), pb.0005; hopelessness, OR (95%CI)=
8.3 (2.6–26.6), pb.0005 and initial insomnia, OR(95%
CI)=0.16 (0.0–0.6), pb.01.
3.5.3. Predictors of suicidal acts performed between 15
to 20 years of age
In the third multivariate logistic regression analy-
sis, depressive symptoms by 15 years as prospective
Table 3
Results of multivariate logistic regression as given by Odds Ratios (95% Confidence Intervals) of depressive symptoms as associates and predictors
of suicidal acts (SA) across adolescence
Depressive symptoms Depressive symptoms by
15 years as associates of
suicidal acts by 15 years
(n= 345) (SA = 86)
Depressive symptoms between
15–20 years as associates of suicidal
acts between 15–20 years
(n= 242) (SA = 34)
Depressive symptoms by 15 years
as predictors of suicidal acts between
15–20 years
(n= 265) (SA = 36)
Recurrent thoughts about death 7.9 (1.4–43.1)*
Suicidal thoughts 3.7 (1.6–8.4)**
Self-harm without suicidal intent 4.5 (1.3–16.0)*
Worthlessness 3.3 (1.3–8.6)*
Hopelessness 26.5 (4.1–168.5)***
Disturbed concentration 56.9 (5.6–577.7)***
Increased appetite 0.05 (0.0–0.6)*
Initial insomnia 0.04 (0.0–0.4)**
Middle insomnia 6.0 (1.3–27.8)*
Note. *pb0.05; **pb0.01; ***pb0.001.
89L. Nrugham et al. / Journal of Affective Disorders 111 (2008) 83–93
predictors of suicidal acts performed between 15 to
20 years were examined. This predictive model had only
one depressive symptom: worthlessness (Table 3). It had
aχ
2
value of 37.4 (df = 7), pb.0005. The Nagelkerke's
R
2
value was .24. The Hosmer and Lemeshow statistic
of 38.9(df = 4) was significant at .0005, revealing poor
calibration of the model. This model correctly classified
86.4% of all respondents and 13.9% of the attempters.
When the significant predictors which had emerged in
the bivariate analyses, that is worthlessness, MDD and
dysthymia were contrasted with each other, a significant
multivariate model was obtained, with a χ
2
value of 28.1
(df = 3), pb.0005 and Nagelkerke's R
2
value at .18. The
Hosmer and Lemeshow statistic of 11.1 (df= 3) was
significant at .01. This model correctly classified 86.8% of
all respondents and 16.7% of the attempters. This final
model included dysthymia, OR (95%CI)= 2.4 (1.0–6.2),
pb.05; and worthlessness, OR (95%CI)= 3.3 (1.3–8.2),
pb.01 as significant predictors of suicidal acts between
15 to 20 years.
4. Discussion
In the present study, the role of depressive symptoms
and disorders as associates and predictors of suicidal acts
across was examined in a subset of depressed and non-
depressed adolescents recruited from a representative
sample of Norwegian high school students, followed-
up for a five year period. The main findings were:
(a) cognitive symptoms dominated the depressive psy-
chopathology profile of adolescents with suicidal acts,
regardless of age and time (b) their roles differed with
age, and (c) the role of a formal psychiatric diagnosis of
depressive disorders changed according to time and age.
Among older adolescents, a decline in the frequency
of depressive symptoms was observed, with the ex-
ception of anhedonia, self-harm without suicidal intent,
and somatic symptoms of fatigue, eating, sleeping and
psychomotor disturbances. The frequencies of dysthy-
mia and DD-NOS also declined, while a slight increase
was seen in the frequency of MDD in the older age
group.
Depressed mood, initial insomnia, disturbed concen-
tration, worthlessness and excessive guilt were all sig-
nificantly related to suicidal acts in all the three bivariate
profiles. In contrast, mood worsening in the evenings,
terminal insomnia, weight loss and gain and both aspects
of psychomotor disturbances were conspicuously absent
from the three bivariate profiles.
Depressed mood has been found to be linked to
suicidality among adolescents in earlier studies (Bettes
and Walker, 1986; Ivarsson et al., 1998). To date, initial
insomnia has been related to suicidal acts only among
adults (Bulik et al., 1990). Our findings extend this
knowledge to a sparsely explored but important area, i.e.,
relationships between suicidal acts and initial insomnia
among adolescents. The cross-sectional model of the
older adolescents revealed a clear trend in sleep dis-
turbances, with five out of the six possible sleep
disturbances (except for terminal insomia) highly
significantly associated with suicidal acts. This finding
could be considered important in the light of recent
findings on the close association between insomnia and
completed suicides among adults (McGirr et al., 2007).
However, similar to other reports among adults, further
analysis (not presented here) showed that these sleep
disturbances were primarily explained by the presence of
depressive disorders (Bernert et al., 2005). Although
decreased appetite and weight loss have been reported
among suicidal adults (Bulik et al., 1990), this finding
was not supported by our study.
In the multivariate models, core cognitive depressive
symptoms known to be associates or predictors of suicidal
acts among adults, i.e., recurring thoughts of death,
suicidal thoughts, worthlessness, hopelessness and dis-
turbed concentration, were found to be associates or
predictors of suicidal acts among adolescents. However,
these relationships were not similar or uniform in their
impact, i.e., they were time-limited and age-specific in
their relationship to suicidal acts. Somewhat surprisingly,
increased appetite and initial insomnia between the ages
of 15 to 20 years were found to be protecting against
suicidal acts, with unclear interpretation.
Two longitudinal studies have reported hopelessness
to be an associate, but not a predictor of suicidal phe-
nomena (Pfeffer et al., 1988; Shahar et al., 2006), as
found in the present study. Although higher levels of
hopelessness have predicted suicide among adult in-
patients and outpatients (Beck et al., 1990; Grøholt
et al., 2006), a similar trend was not found in the present
study of attempted suicide among depressed school
adolescents. The role of hopelessness, diagnostically
assessed by interviews, among suicidal adolescents
appears to be different from its role among adults, as
measured by a questionnaire for the preceding week, in
that it was an associate but not a predictor, in the present
study. Hopelessness has also been found to be an even
stronger predictor of suicidal behaviour than depression
and the best predictor of completed suicide in adult
populations (Beck et al., 1974). A study on whether
Beck's cognitive theory of depression and hopelessness
also applied to adolescent community populations found
support for his theory of depression but not for hope-
lessness (Lewinsohn et al., 2001b). Exploratory research
90 L. Nrugham et al. / Journal of Affective Disorders 111 (2008) 83–93
into whether there are three distinct negative thoughts
in the cognitive triad among adults has reported that
there was only a singular one-dimensional negative view
of the self (McIntosh and Fischer, 2000).
Worthlessness, a part of the cognitive triad, by the
age of 15 years, was a significant multivariate predictor
of suicidal acts in the present study. In a study of
children and adolescents with various forms of suicid-
ality, worthlessness has been related to all forms of
suicidal behaviours, after adjustment for other depres-
sive symptoms, comorbid disorders and demographics
(Liu et al., 2006). In the present study, a cross-check
revealed that suicidal acts by 15 years as predictor was
not retained in the model, while worthlessness con-
tinued to be a significant predictor. In the same sample,
that adolescents with suicidal acts had significantly
lower mean scores on self-esteem than participants
without suicidal acts, though the differences were small
and age-dependent (Nrugham et al., 2008). Self-
criticism has been reported to be the cognitive variable
most strongly associated with hopelessness among
adolescent suicide attempters than other depressive
cognitions and perfectionism (Donaldson et al., 2000).
Apart from self-harm without suicidal intent and initial
insomnia, the other depressive symptoms retained in the
final multivariate analyses were all cognitive in nature,
across age and time. These findings underline the
importance of cognitive depressive symptoms as indica-
tors of suicidal acts from adolescence to early adulthood.
Cognitive rigidity is accepted as patently implicated in
suicidal phenomena among adolescents (Beautrais et al.,
1999) and in adults (Arffa, 1983). Although we did not
find suicidal thoughts to be a significant predictor of
suicidal acts, it has been reported that 15-year olds with
suicidal thoughts were 12 times more likely to have
attempted suicide between 15to 30 years of age than those
without suicidal thoughts (Reinherz et al., 2006).
Our finding of self-harm without suicidal intent as a
significant bivariate associate for both age groups and
multivariate associate in the younger age group but not a
bivariate predictor in the older group indicates that acts
of self-harm without suicidal intent decreased with in-
creased age or that acts of self-harm without suicidal
intent had a limited role vis-à-vis suicidal acts in this
sample. Here, it was both, age-specific and time-limited
in its effects. A history of suicidal acts by 15 years,
although an important bivariate predictor, was not
retained in the multivariate model.
While the relationships between dysthymia, DD-
NOS and suicidal acts across adolescence differed with
age and time, MDD was both, a significant associate,
irrespective of age and a significant predictor of suicidal
acts, in the bivariate analyses. However, MDD did not
continue to be significant, neither as an associate nor as
a predictor when compared with depressive symptoms
in the final multivariate analyses of the older adoles-
cents. Dysthymia by 15 years remained a significant
predictor of suicidal acts performed between 15 to
20 years. This is in line with findings among adult
outpatients (Klein et al., 2000). Since cognitive
symptoms dominated the bivariate and multivariate
models it seems that it is not depressive disorders but
specific types of depressive symptoms, i.e., cognitive
depressive symptoms, which are primarily related to
suicidal acts across adolescence. The continuity of
percentages of depressive symptoms and changes in the
distribution of diagnoses as assessed at T3 among the T2
high scorers on the MFQ provide support for the
conceptualization of depression as a continuous variable
and calls to add dimensional criteria to categorical
definitions of diagnosis (Helzer et al., 2006; Pickles and
Angold, 2003; Regier, 2007).
4.1. Limitations
Due to the small number of males in the subset, the
present findings can primarily be generalized to female
adolescents and young adults. At the five-year follow-
up, 30% of the high school adolescents who had been
interviewed at T2 could not be re-interviewed, which
might be an indicator of bias. The analysis of differences
between the participants and non-participants at T3, and
the use of categorical definitions also point to possible
underestimation of the relationships between suicide
attempts and the explanatory variables.
Even though semi-structured diagnostic interviews
were used to obtain information leading to diagnoses
and parents were available as additional informants at
T2, such information is subject to recall bias since it was
retrospective in nature, except for current diagnoses. It
should be noted that information on past and current
symptoms/disorders among school adolescents were
pooled, and was restricted to the most serious episode.
Information regarding other psychiatric symptoms and
disorders at the first interview, T2, was also limited
to symptoms and disorders comorbid with depressive
disorders.
The multivariate models derived in the present study
need to be interpreted with caution. The Hosmer and
Lemeshow statistic was significant for both the multi-
variate predictor models and the final model for older
adolescents indicating poor model calibration. Two of
the associates for the older adolescents had wide
confidence intervals for ORs indicative of low sample
91L. Nrugham et al. / Journal of Affective Disorders 111 (2008) 83–93
size. Overall, the associative models were seen to be
better than the predictive models.
However, this study has substantial strengths in its use
of a longitudinal design covering most of the adolescent
period into early adulthood. The present subset of
adolescents was selected from a school-based representa-
tive sample using a case-control design with an over-
inclusion of depressed adolescents, thus enabling us to
assess the effects of various diagnoses of depression on
suicidality with increased power, in addition to the use of
blind and randomized diagnostic interviews by clinicians
with excellent interview integrity. Together, the over-
sampling of depressed adolescents at 15 years of age and
the inclusion of only clinically significant symptoms,
makes the adolescents in the present study resemble the
adolescent outpatient population.
4.2. Conclusions
Future research should aim to replicate the findings of
the present study, with special focus on cognitive
depressive symptoms in adolescence and integrating
categorical and dimensional approaches. Any depressive
disorder, self-harm without suicidal intent in younger
adolescents, middle insomnia in older adolescents and
cognitive depressive symptoms irrespective of age,
should alert school mental health professionals and
clinicians in their assessment of suicidalilty among
adolescents.
Role of funding source
The funding sources The Norwegian Research Council, Council
for Mental Health, Child and Adolescent Mental Health Clinic,
Trondheim and the Regional Centre for Child and Adolescent Mental
Health, Central Norway have had no role in the planning of the study
design; collection, analysis and interpretation of data; in writing the
report or in the decision to submit the paper for publication.
Conflict of interest
All authors declare that they have no conflict of interest.
Acknowledgement
This research was supported in part by grants from
the Norwegian Research Council, Council of Mental
Health, Regional Centre for Child and Adolescent
Mental Health, Central Norway and the Child and
Adolescent Mental Health Clinic, Trondheim. The
valuable contribution and help from the participating
adolescents, youth, parents and school and project staff,
is gratefully acknowledged. The authors gratefully
acknowledge Prof. Berit Grøholt for her valuable
comments on the manuscript.
References
Angold, A., 1989. Structured assessment of psychopathology in
children and adolescents. In: Thompson, C. (Ed.), The Instruments
of Psychiatric Research. John Wiley, Chichester, pp. 271–304.
Arffa, S., 1983. Cognition and suicide: a methodological review.
Suicide Life-Threat. Behav. 13 (2), 109–122.
Beautrais, A.L., Joyce, P.R., Mulder, R.T., 1999. Personality traits and
cognitive styles as risk factors for serious suicide attempts among
young people. Suicide Life-Threat. Behav. 29 (1), 37–47.
Beck, A.T., Weissmann, A., Lester, D., Trexler, L., 1974. The
measurement of pessimism: the hopelessness scale. J. Consult.
Clin. Psychol. 42 (6), 861–865.
Beck, A.T., Brown, G., Berchick, R.J., Stewart, B.L., Steer, R.A.,
1990. Relationships between hopelessness and ultimate suicide: a
replication with psychiatric outpatients. Am. J. Psychiatry 147 (2),
190–195.
Bernert, R.A., Joiner Jr., T.E., Cukrowicz, K.C., Schmidt, N.B.,
Krakow, B., 2005. Suicidality and sleep disturbances. Sleep 28 (9),
1135–1141.
Bettes, B.A., Walker, E., 1986. Symptoms associated with suicidal
behaviour in childhood and adolescence. J. Abnorm. Child Psychol.
14 (4), 591–604.
Brent, D., 2001. Assessment and treatment of the youthful suicidal
patient. Ann. N.Y. Acad. Sci. 932, 106–131.
Brent, D.A., Perper, J.A., Moritz, G., Baugher, M., Schweers, J., Roth,
C., 1994. Suicide in affectively ill adolescents: a case-control
study. J. Affect. Disord. 31 (3), 193–202.
Bridge, J.A., Goldstein, T.R., Brent, D.A., 2006. Adolescent suicide
and suicidal behaviour. J. Child Psychol. Psychiatry 47 (3–4),
372–394.
Bulik, C.M., Carpenter, L.L., Kupfer, D.J., Frank, E., 1990. Features
associated with suicide attempts in recurrent major depression.
J. Affect. Disord. 18 (1), 29–37.
Csorba, J., Rozsa, S., Gadoros, J., Vetro, A., Kaczvinszky, E., Sarungi,
E., Makra, J., Kapornay, K., 2003. Suicidal depressed vs. non-
suicidal depressed adolescents: differences in recent psychopathol-
ogy. J. Affect. Disord. 74 (3), 229–236.
Donaldson, D., Spirito, A., Farnett, E., 2000. The role of perfectionism
and depressive cognitions in understanding the hopelessness
experienced by adolescent suicide attempters. Child Psychiatry
Hum. Dev. 31 (2), 99–110.
Diagnostic and Statistical Manual of Mental Disorders (DSM) –IV,
Text Revision (TR), 2000. American Psychiatric Association.
Division of Research, 1000 Wilson Boulevard, Arlington.
Field, A., 2005. Discovering Statistics Using SPSS, 2nd ed. Sage
Publications, London.
Gould, M.S., King, R., Greenwald, S., Fisher, P., Schwab-Stone, M.,
Kramer, R., Flisher, A.J., Goodman, S., Canino, G., Shaffer, D.,
1998. Psychopathology associated with suicidal ideation and
attempts among children and adolescents. J. Am. Acad. Child
Adolesc. Psych. 37 (9), 915–923.
Grøholt, B., Ekeberg, Ø., Haldorsen, T., 2000. Adolescents hospita-
lised with deliberate self-harm: the significance of an intention to
die. Eur. Child Adolesc. Psychiatry 9 (4), 244–254.
Grøholt, B., Ekeberg, Ø., Haldorsen, T., 2006. Adolescent suicide
attempters: what predicts future suicidal acts? Suicide Life-Threat.
Behav. 36 (6), 638–650.
Harrington, R., Bredenkamp, D., Groothues, C., Rutter, M., Fudge, H.,
Pickles, A., 1994. Adult outcomes of childhood and adolescent
depression, III: links with suicidal behaviours. J. Child Psychol.
Psychiatry 35 (7), 1309–1319.
92 L. Nrugham et al. / Journal of Affective Disorders 111 (2008) 83–93
Helzer, J.E., Kraemer, H.C., Krueger, R.F., 2006. The feasibility and
need for dimensional psychiatric diagnoses. Psychol. Med. 36,
1671–1680.
Ivarsson, T., Larsson, B., Gillberg, C., 1998. A 2–4 year follow-up of
depressive symptoms suicidal ideation and suicide attempts among
adolescent psychiatric inpatients. Eur. Child Adolesc. Psychiatry 7 (2),
96–104.
Jacobson, C.M., Gould, M., 2007. The epidemiology and phenomenol-
ogy of non-suicidal self-injurious behaviour among adolescents: a
critical review of the literature. Arch. Suicide Res. 11 (2), 129–147.
King, C.A., 1997. Suicidal behaviour in adolescence. In: Maris, R.W.,
Silverman, M.M., Canetto, S.S. (Eds.), Review of suicidology,
pp. 61–95. Guildford, New York.
Kaufman, J., Birmaher, B., Brent, D., Rao, U., Flynn, C., Moreci, P.,
Williamson, D., Ryan, N., 1997. Schedule for Affective Disorders
and Schizophrenia for School-Age Children-Present and Lifetime
Version (K-SADS-PL): initial reliability and validity data. J. Am.
Acad. Child Adolesc. Psych. 36, 980–988.
Klein, D.N., Schwartz, J.E., Rose, S., Leader, J.B., 2000. Five-year
course and outcome of dysthymic disorder: a prospective, natu-
ralistic follow-up study. Am. J. Psychiatry 157 (6), 931–939.
Kovacs, M., Goldston, D., Gatsonis, C., 1993. Suicidal behaviours and
childhood-onset depressive disorders: a longitudinal investigation.
J. Am. Acad. Child Adolesc. Psych. 32 (1), 8–20.
Larsson, B., Sund, A.M., 2008. Prevalence, course, incidence and 1-
year prediction of deliberate self-harm and suicide attempts in early
Norwegian adolescents. Suicide Life-Threat. Behav. 32 (2),
152–165.
Lewinsohn, P.M., Rohde, P., Seeley, J.R., 1994. Psychosocial risk
factors for future adolescent suicide attempts. J. Consult. Clin.
Psychol. 62 (2), 297–305.
Lewinsohn, P.M., Rohde, P., Seeley, J.R., 1998. Major depressive
disorder in older adolescents: prevalence, risk factors and clinical
implications. Clin. Psychol. Rev. 18 (7), 765–794.
Lewinsohn, P.M., Rohde, P., Seeley, J.R., Baldwin, C.L., 2001a.
Gender differences in suicide attempts from adolescence to young
adulthood. J. Am. Acad. Child Adolesc. Psych. 40 (4), 427–434.
Lewinsohn, P.M., Joiner Jr., T.E., Rohde, P., 2001b. Evaluation of
cognitive diathesis-stress models in predicting major depressive
disorder in adolescents. J. Abnorm. Psychology 110 (2), 203–215.
Liu, X., Buysse, D.J., 2006. Sleep and youth suicidal behaviour: a
neglected field. Curr. Opin. Psychiatry 19 (3), 288–293.
Liu, X., Gentzler, A., Tepper, P., Kiss, E., Kothencne, V.O., Zsuzsanna, T.,
Vetro, A., Kovacs, M., 2006. Clinical features of depressed children
and adolescents with various forms of suicidality. J. Clin. Psychiatry
67 (9), 1442–1450.
McGirr, A., Renaud, J., Seguin, M., Alda, M., Benkelfat, C., Lesage,
A., Turecki, G., 2007. An examination of DSM-IV depressive
symptoms and risk for suicide completion in major depressive
disorder: a psychological autopsy study. J. Affect. Disord. 97 (1–3),
203–209.
McIntosh, C.N., Fischer, D.G., 2000. Beck's cognitive triad: one
versus three factors. Can. J. Behav. Sci. 32, 153–157.
Nrugham, L., Larsson, B., Sund, A.M., 2008. Predictors of suicidal
acts across adolescence: Influences of familial, peer and individual
factors. J. Affect. Disord. 109, 35–45.
Pfeffer, C.R., Newcorn, J., Kaplan, G., Mizruchi, M.S., Plutchik, R.,
1988. Suicidal behavior in adolescent psychiatric inpatients. J. Am.
Acad. Child Adolesc. Psych. 27 (3), 357–361.
Pickles, A., Angold, A., 2003. Natural categories or fundamental
dimensions: on carving nature at the joints and the rearticulation of
psychopathology. Dev. Psychopathol. 15, 529–551.
Regier, D.A., 2007. Dimensional approaches to psychiatric classifica-
tion: refining the research agenda for DSM-V: an introduction. Int.
J. Methods Psychiatr. Res. 16 (S1), S1–S5.
Reinherz, H.Z., Tanner, J.L., Berger, S.R., Beardslee, W.R., Fitzmaurice,
G.M., 2006. Adolescent suicidal ideation as predictive of psycho-
pathology, suicidal behaviour, and compromised functioning at age
30. Am. J. Psychiatry 163 (7), 1226–1232.
Rihmer, Z., 2007. Suicide risk in mood disorders. Curr. Opin. Psychiatry
20 (1), 17–22.
Rohde, P., Lewinsohn, P., Seeley, J.R., 1997. Comparability of
telephone and face-to-face interviews in assessing Axis I and Axis
II disorders. Am. J. Psychiatry 154 (11), 1593–1598.
Shaffer, D., Gould, M.S., Brasic, J., Ambrosini, P., Fisher, P., Bird, H.,
Aluwahlia, S., 1983. A children's global assessment scale
(CGAS). Arch. Gen. Psychiatry 40 (11), 1228–1231.
Shahar,G.,Bareket,L.,Rudd,M.D.,JoinerJr.,T.E.,2006.Inseverely
suicidal young adults, hopelessness, depressive symptoms and
suicidal ideation constitute a single syndrome. Psychol. Med. 36 (7),
913–922.
Sund, A.M., Larsson, B., Wichstrøm, L., 2001. Depressive symptoms
among young Norwegian adolescents as measured by the Mood
and Feelings Questionnaire (MFQ). Eur. Child Adolesc. Psychiatry
10 (4), 222–229.
Sund, A.M., Larsson, B., Wichstrøm, L., 2003. Psychosocial correlates
of depressive symptoms among 12–14-year-old Norwegian
adolescents. J. Child Psychol. Psychiatry 44 (4), 588–597.
Wichstrøm, L., 2000. Predictors of adolescent suicide attempts: a
nationally representative longitudinal study of Norwegian adoles-
cents. J. Am. Acad. Child Adolesc. Psych. 39 (5), 603–610.
93L. Nrugham et al. / Journal of Affective Disorders 111 (2008) 83–93