See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/11198652

Screening for Adolescent Depression: Comparison of the Kutcher Adolescent

Depression Scale with the Beck Depression Inventory

Article  in  Journal of Child and Adolescent Psychopharmacology · February 2002

DOI: 10.1089/104454602760219153 · Source: PubMed

CITATIONS READS

114 6,702

4 authors, including:

John C Leblanc Stan Kutcher

Dalhousie University Dalhousie University
93 PUBLICATIONS   3,342 CITATIONS    347 PUBLICATIONS   10,740 CITATIONS   

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

New Initiative Grant, UWI, Mona View project

school mental health literacy View project

All content following this page was uploaded by John C Leblanc on 03 January 2015.

The user has requested enhancement of the downloaded file.

JOURNAL OF CHILD AND ADOLESCENT PSYCHOPHARMACOLOGY
Volume 12, Number 2, 2002
© Mary Ann Liebert, Inc.
Pp. 113–126

Screening for Adolescent Depression:

Comparison of the Kutcher Adolescent
Depression Scale with the Beck Depression Inventory

John C. LeBlanc, M.D., M.Sc., FRCPC,1,2 Anthony Almudevar, Ph.D.,3

Sarah J. Brooks, Ph.D.,1 and Stan Kutcher, M.D., FRCPC1

ABSTRACT

Self-report instruments commonly used to assess depression in adolescents have limited or

unknown reliability and validity in this age group. We describe a new self-report scale, the
Kutcher Adolescent Depression Scale (KADS), designed specifically to diagnose and assess
the severity of adolescent depression. This report compares the diagnostic validity of the full
16-item instrument, brief versions of it, and the Beck Depression Inventory (BDI) against the
criteria for major depressive episode (MDE) from the Mini International Neuropsychiatric
Interview (MINI). Some 309 of 1,712 grade 7 to grade 12 students who completed the BDI had
scores that exceeded 15. All were invited for further assessment, of whom 161 agreed to as-
sessment by the KADS, the BDI again, and a MINI diagnostic interview for MDE. Receiver
operating characteristic (ROC) curve analysis was used to determine which KADS items best
identified subjects experiencing an MDE. Further ROC curve analyses established that the
overall diagnostic ability of a six-item subscale of the KADS was at least as good as that of
the BDI and was better than that of the full-length KADS. Used with a cutoff score of 6, the
six-item KADS achieved sensitivity and specificity rates of 92% and 71%, respectively—a
combination not achieved by other self-report instruments. The six-item KADS may prove to
be an efficient and effective means of ruling out MDE in adolescents.

INTRODUCTION been shown to have weak reliability and/or

validity in this age group (see Brooks and

T HE K UTCHER ADOLESCENT D EPRESSION SCALE

(KADS) was developed for the purpose of
diagnosing and monitoring the severity of de-
pression in adolescents by self-report. Many
self-rated and observer-rated instruments
commonly utilized to measure depression
have had little testing in adolescents or have
Kutcher 2001).
The diagnostic ability of an instrument can
be evaluated against a “gold standard” method
using receiver operating characteristic (ROC)
curve analysis (Metz 1978). An ROC curve de-
picts the trade-off between an instrument’s
true positive rate (its sensitivity; i.e., the pro-

1 Department of Psychiatry, Dalhousie University, Halifax, Nova Scotia, Canada.

2 Department of Pediatrics, Community Health and Epidemiology, Dalhousie University, Halifax, Nova Scotia,
Canada.
3 Department of Mathematics and Computing Science, St. Mary’s University, Halifax, Nova Scotia, Canada.

113
114
portion of positive cases it correctly identifies
as such) and its false positive rate (1 minus its
specificity; i.e., specificity being the proportion
of true negative cases correctly identified as
such by the instrument) for all possible cutoff
scores. By tradition, the x-axis and y-axis dis-
play the false positive and true positive rates,
respectively, with both axes scaled from 0 to 1.
The area under the ROC curve (AUC) is then
an index of the instrument’s ability to identify
whether subjects are cases, and ranges in value
from 0.5 (chance level performance) to 1 (per-
fect identification). Henderson (1993) sug-
gested the following correspondence of AUC
values to diagnostic accuracy level: AUC # 0.7,
low; 0.7 < AUC # 0.9, moderate; AUC > 0.90,
high. ROC analyses have been carried out on
several of the instruments reviewed below.
The 21-item Beck Depression Inventory
(BDI; Beck et al. 1961) was the first formal at-
tempt to use self-ratings to assess the severity
of depression. Its reliability and sensitivity to
change are well documented in adults. How-
ever, it has been claimed that adolescents’
scores on the BDI are generally less stable than
those of depression scales designed specifi-
cally for them (Reynolds 1994). Furthermore,
some authors suggest that BDI scores are not
specific to clinical depression and may instead
index subjects’ dissatisfaction, demoralization,
dysphoria, or anxiety (Gotlib 1984; Kutcher
and Marton 1989). In a community study of
1,704 high school students (47% male, 53% fe-
male), Roberts et al. (1991) used ROC analysis
to assess the diagnostic validity of the BDI
against Diagnostic and Statistical Manual of
Mental Disorders, third edition, revised (DSM-
III-R; American Psychiatric Association 1987)
diagnoses of major depressive disorder
(MDD). The authors reported AUCs of 0.93
and 0.83 for males and females, respectively,
which suggests an AUC of 0.88 assuming
equal numbers of males and females. How-
ever, although Roberts et al.’s (1991) ROC
curve (their Fig. 1, not reproduced in this arti-
cle) suggested it was possible to attain high
sensitivity (93%) and high specificity (87%)
concurrently for males, the ROC curve for fe-
males showed that sensitivity >90% would
come at the price of very low specificity
(<40%).
LEBLANC ET AL.
The Children’s Depression Inventory (CDI;
Kovacs 1992) is a 27-item self-report scale
requiring 10 to 20 minutes to complete. It
was developed as a format- and language-
modified version of the BDI to measure the se-
verity of depression in children age 7 and
older. Although its reliability and internal con-
sistency are satisfactory, and its sensitivity
to change is well established, its ability to
discriminate depressed and nondepressed
youngsters has been questioned (Fundudis et
al. 1991; McCauley et al. 1983; Weissman et al.
1980). In their study of the criterion validity of
the CDI and of Birleson’s (1981) 18-item De-
pression Self-Rating Scale (DSRS) against clini-
cal diagnoses of depression in a clinical sample
of youngsters (age range 12.8 to 16 years), Fun-
dudis et al. (1991) reported minimum misclas-
sification rates of 23% and 26%, respectively
(AUCs were not reported). Our abstraction of
Fundudis et al.’s (1991) graphs (their Figs. 1
and 2, not reproduced here) suggests that a re-
quirement for sensitivity at or >90% would
mean acceptance of specificity <50% with ei-
ther instrument. Of course these results do not
necessarily mean that the CDI and DSRS do
not perform well when restricted to younger
subjects.
The Mood and Feelings Questionnaire
(MFQ; Costello and Angold 1988) is a 32-item
self-report scale designed for use in epidemio-
logical studies of depression in youngsters aged
8 to 18 years. A 13-item version has also been
developed. Both the long and short versions
have satisfactory reliability and internal consis-
tency (Angold et al. 1995a; Costello et al. 1991);
sensitivity to change has yet to be well estab-
lished. The diagnostic accuracy of the long
MFQ was assessed by Wood et al. (1995)
against DSM-III-R diagnoses of MDD made
using the Schedule for Affective Disorders and
Schizophrenia–Child version (K-SADS; Am-
brosini et al. 1989). The authors obtained an
AUC of 0.82 with their sample of 104 outpa-
tients (10 to 19 years) and an AUC of 0.76 with
a subset of 60 of these subjects age 14 and
older. Wood et al.’s (1995) first graph (their Fig.
1, not shown here) shows that the minimum
misclassification rate was 22% for the total
sample, and that when sensitivity reached
90%, specificity was close to 50%. Thapar and
THE KUTCHER ADOLESCENT DEPRESSION SCALE
McGuffin (1998) assessed the performance of
the short MFQ in a community sample of
twins aged 8 to 16 years. The authors used An-
gold et al.’s (1995b) Child and Adolescent Psy-
chiatric Assessment interview to produce the
gold standard diagnoses according to ICD-10
(World Health Organization 1992) and to
DSM-III-R criteria for a major depressive epi-
sode (MDE). The respective AUCs were low
(0.63) and moderate (0.72). Thapar and
McGuffin’s (1998) second pair of graphs (their
Fig. 2, not shown here) indicated that if high
sensitivity ($80%) was required, specificity
would be no better than 52%.
The Reynolds Adolescent Depression Scale
(RADS; Reynolds 1987) is a 30-item self-report
scale that measures the severity of symptoms
of depression in adolescents aged 13 to
18 years. It is designed for use in schools or
clinical settings and takes 5 to 10 minutes to
complete. The RADS has well-documented re-
liability, validity, and normative data (Rey-
nolds 1987), and there is some evidence for its
sensitivity to change (e.g., Reynolds and Coats
1986). It has yet to be subjected to ROC analy-
sis. However, when used with its suggested
cutoff score of 77 and assessed against diag-
noses of depression made using the Hamilton
Depression Rating Scale (Hamilton 1960) or
the Schedule for Affective Disorders and
Schizophrenia (Endicott and Spitzer 1978), its
specificity was high (>90%) but its sensitivity
was fairly low; approximately one third of
depressed teens were misclassified as not de-
pressed (Evert and Reynolds 1986, unpub-
lished manuscript, cited in Reynolds 1987,
pp 27–28).
The 20-item Center for Epidemiological
Studies–Depression scale (CES-D; Radloff
1977) is a self-report instrument designed
to measure symptoms of depression in adult
community studies. However, significant
weaknesses have been observed when the
CES-D is used with adolescents. Roberts et al.
(1991) observed fairly poor test-retest reliabil-
ity (r = 0.61) using a 1- to 4-week intertest in-
terval, and they also reported a very low
concordance (kappa = 0.11) between clinical
diagnoses of MDD and cases of depression
suggested by high CES-D scores. A version
modified for children and adolescents has
115
been developed, but Faulstich et al. (1986)
found that it did not discriminate effectively
between depressed and nondepressed young-
sters. Various authors have claimed that the
scale measures general emotional turmoil
rather than depression (Doerfler et al. 1988;
Faulstich et al. 1986; Garrison et al. 1991).
Roberts et al. (1991) concluded that neither
the BDI nor the CES-D was adequate on its
own for screening adolescents for depression
and that the development of more efficient
screening instruments would require greater
attention to the criterion validity of the items
chosen. Roberts et al. (1989) recommended the
development of depression scales that explic-
itly use items that reflect the diagnostic criteria
of an accepted classification system, such
as DSM-IV (American Psychiatric Association
1994). Today, there is still a need for a time-
efficient screening tool for adolescent depres-
sion whose sensitivity and specificity will
ensure useful positive and negative predictive
values in common clinical situations.
In line with these recommendations and in
view of the aforementioned limitations of
commonly used self-report instruments when
used with adolescents, one of the authors (SK)
developed the 16-item KADS. Its principal fea-
tures are: (a) items address the core symptoms
of adolescent depression, (b) symptoms are
described using both standard and colloquial
terminology, and (c) items measure the fre-
quency of occurrence (in the case of 14 items)
or the severity (in the case of the other two
items) of the specified symptoms.
By focusing attention on the core symp-
toms of depression, the authors expect that
the KADS will correctly screen for depres-
sion in adolescents achieving clinically ac-
ceptable sensitivity and specificity. In this
first study assessing the properties of the
KADS, the authors compared the instrument
against the BDI in a large, randomly selected
school-based cohort of secondary school stu-
dents. The authors refined the instrument by
omitting items with the lowest predictive
power for MDE and compared the abilities of
the full and short versions of the KADS to di-
agnose MDE against that of the BDI. Poten-
tial methods of implementing the KADS are
discussed.
116
METHODS
Subjects and procedures
Subjects were recruited from a school-based
survey designed to measure the prevalence of
a variety of sleep habits and their relationship
to symptoms of depression (LeBlanc et al.
1999). In total, 1,712 students attending
grades 7 through 12 in Halifax Nova Scotia
schools, serving a mixed urban-rural munici-
pality (population 330,000), were selected
using probability-based survey sampling tech-
niques. Subjects (771 males, 920 females,
21 gender unrecorded; mean age = 15.0 years,
SD = 1.8 years, range 12 to 20 years) completed
a sleep habits questionnaire and the BDI. Some
301 subjects had a BDI score of 16 while an ad-
ditional 7 had BDI scores of < 16 but a score
of 2 or 3 on the suicide item. These subjects
(87 males, 217 females, 4 gender unrecorded;
mean age 15.2 years, SD = 1.8 years) and one
additional 16-year-old female who scored
15 on the BDI were invited for a second inter-
view to assess whether they had a depressive
illness.
One hundred sixty-one subjects were inter-
viewed (48 males, 113 females; mean age
15.1 years, SD = 1.8 years, range 12 to 20 years).
Of these subjects, 135 (84%) were Caucasian,
7 (4%) were multiracial, 5 (3%) were African
Canadian, 2 (1%) were Native Canadian,
10 (6%) were of other (non-Latino) races, and
for 2 subjects (1%) this information was not
recorded. One hundred forty-nine subjects
(48% of those invited) either refused interview
or did not show. These noninterviewees did
not differ significantly in age, gender, or eth-
nicity from those who were interviewed. Non-
interviewees also did not differ in their BDI
scores (mean = 24.0, SD = 9.2) from those who
were interviewed (mean = 23.6, SD = 7.3).
During a 1-hour interview, subjects met
with a health professional and completed a
follow-up BDI and the KADS, and were
administered the depression section of the
Mini International Neuropsychiatric Interview
(MINI, version 4.4; Lecrubier et al. 1997). The
MINI is a fully structured diagnostic interview
schedule that covers 15 major axis I psychiatric
disorders using DSM-IV criteria. The module
LEBLANC ET AL.
for diagnosing an MDE consists of nine items
that assess symptoms of depression experi-
enced over the previous 2 weeks. The outcome
is confirmation or otherwise of a current
and/or a past MDE; there are no associated
rating scores. The MINI’s reliability and valid-
ity have been ascertained in a number of stud-
ies (Amorim et al. 1998; Lecrubier et al. 1997;
Sheehan et al. 1997, 1998).
MINI diagnostic assessments were com-
pleted for all 161 subjects who were inter-
viewed. Thirty subjects (18.6%) met the
schedule’s criteria for a current MDE, and
131 (including the female subject who scored
15 on the initial BDI) did not. Nine subjects,
four of whom met the MDE criteria, were ex-
cluded from further analysis because they
omitted 16 or more items on either the KADS
(n = 4) or the BDI (n = 5). These nine subjects
did not differ from the rest of the sample in
terms of their demographic characteristics or
in their original BDI scores. Furthermore,
their scores on the one self-report instrument
that they did complete during the interview
were neither systematically high nor system-
atically low compared to those of the rest of
the sample. This left 152 subjects for data
analysis, 26 (17.1%) of whom met the criteria
for a current MDE. One hundred fifty of the
analyzed subjects completed all items; 2 sub-
jects omitted no more than two items on both
instruments.
The Kutcher Adolescent Depression Scale
The KADS consists of 16 items that cover the
core symptoms of depression (see Appendix).
Fourteen of the items require subjects to indi-
cate the frequency of occurrence of particular
experiences over the previous week by endors-
ing one of four possible responses, coded 0
to 3: “hardly ever,” “much of the time,” “most
of the time,” “all of the time.” Response op-
tions for items 12 and 13 are worded slightly
differently but are also listed in ascending
order of morbidity; four response options are
provided for item 12 (coded 0 to 3) and five for
item 13 (coded 0 to 4). The total score of the in-
strument is the sum of all 16 items. Item 8 ad-
ditionally includes a probe in which subjects
whose appetite has changed can indicate
THE KUTCHER ADOLESCENT DEPRESSION SCALE
whether it has increased or decreased; this re-
sponse does not contribute to the total score.
Data analysis
Data analysis consisted of (a) assessment of
the performance of individual items, (b) cre-
ation of reduced versions of the KADS, and
(c) comparison of the diagnostic validity of the
full-length and reduced versions with that of
the BDI. For (a), the assessment of individual
items, the means and standard deviations of
item scores, and item-total score correlations
were calculated. The diagnostic validity of
individual KADS items was then assessed
against the MINI diagnoses by calculating the
area under each item’s ROC curve (Hanley
and McNeil 1982). For (b), KADS items were
reduced in number to shorten the administra-
tion time of the instrument without unduly
reducing internal consistency or diagnostic
power. Cronbach’s (1951) alpha and ROC
curves were the tools employed. The ROC
AUC for the total score of all original items
was calculated. Items were then removed one
at a time starting with the one that had the
least predictive power (i.e., the smallest AUC).
After each removal, Cronbach’s alpha and the
AUC for the remaining set of items were recal-
culated. This generated a hierarchy of candi-
date subscales where a suitable subscale was
defined as one that exhibited high predictive
power while maintaining acceptably high in-
ternal consistency.
For (c), the ROC curves of two different ver-
sions of the KADS and the BDI were compared
graphically. The statistical significance of dif-
ferences between the magnitudes of their
AUCs was estimated using the following boot-
strap method (see Hamilton 1992 for overview
of bootstrap methods). The study’s 152 sub-
jects (with their relevant KADS and BDI data
and MINI diagnoses) were selected randomly
with replacement (i.e., any given subject could
be selected more than once) to produce a
new study “sample” of 152 subjects. This pro-
cess was carried out 10,000 times to yield
10,000 replications of the original study, each
with 152 subjects. The AUCs of the three in-
struments were recalculated for each of these
new samples. Pairs of instruments were then
117
compared by noting the proportion of the sam-
ples for which the AUC for one instrument ex-
ceeded the AUC for the other. This proportion
of 10,000 replications can be interpreted as a
p value.
RESULTS
Assessment of individual items
Table 1 summarizes various properties of in-
dividual items of the KADS. Columns 2 and 3
show the means and standard deviations of
subjects’ item scores (i.e., response levels).
Item 7 (trouble concentrating, daydreaming)
and item 13 (thoughts/actions regarding sui-
cide/self-harm), respectively, were the most
and least highly endorsed symptoms. Column 4
shows the Pearson correlation coefficients ob-
tained between each item score and the total
score of all other items. Each item correlated sig-
nificantly with the total score of all other items
(p < 0.01). The magnitude each item’s ROC AUC
(an index of diagnostic accuracy) is presented in
column 5. Item 13 (thoughts/actions regarding
suicide/self-harm) ranked highest in this re-
spect, and item 3a (trouble falling asleep, lying
awake) ranked lowest.
Figure 1 presents a visual summary of the
diagnostic ability of each response level of
each KADS item to identify cases of MDE. For
each item, the plots show the proportion of
MDE cases among each subset of subjects who
selected a particular response level. (It should
be noted that these plots do not have a direct
correspondence to ROC curves.) Although
some allowance should be made for variation
due to sampling error, the proportion of cases
of MDE should in general increase with in-
creasing response level.
Item 13 (suicide/self-harm) is notable in
that it yielded the largest individual item
AUC, although its plot in Fig. 1 does not show
the expected increase in proportion of cases of
MDE with higher response levels. This item’s
large AUC is due to the vast majority of sub-
jects endorsing one of its lower three response
levels (n = 143, 94% of subjects) in a manner
appropriate for their diagnostic status (i.e.,
very few MDE cases selected response
118 LEBLANC ET AL.

TABLE 1. PSYCHOMETRIC P ROPERTIES OF INDIVIDUAL ITEMS

Mean Item score Item-total Item Item

item standard score ROC area ROC area
Item Item content score deviation correlation magnitude rank

1 Low mood, depressed, can’t be bothered 0.93 0.87 0.55 0.75 5

2 Irritable, losing temper 1.09 0.89 0.37 0.71 8
3a Trouble falling asleep, lying awake 1.05 1.07 0.50 0.57 16
3b Poor sleep during night, waking up 0.70 0.90 0.51 0.69 9
3c Waking too early in morning 0.59 0.89 0.31 0.61 13
3d Sleeping during day, taking naps 0.68 0.98 0.37 0.58 15
4 Less interest in friends, work, going out 0.59 0.82 0.48 0.74 7
5 Feeling worthless, hopeless 0.78 0.91 0.57 0.78 4
6 Tired, low in energy, low motivation 1.22 0.98 0.60 0.78 3
7 Trouble concentrating, daydreaming 1.41 1.03 0.55 0.59 14
8 Appetite changes 0.84 0.96 0.52 0.68 10
9 Less experience of fun, pleasure 0.66 0.79 0.60 0.80 2
10 Feeling worried, nervous, panicky 0.99 1.00 0.59 0.75 6
11 Headaches, butterflies, nausea, etc. 0.86 0.99 0.56 0.66 11
12 Interest in/thoughts re: sex, sexual arousal 0.59 0.94 0.25 0.64 12
13 Thoughts/actions re: suicide/self-harm 0.54 0.96 0.47 0.81 1

ROC = receiver operating characteristic.

level “0,” but relatively large numbers of cases
selected response levels “1” and “2”). None-
theless, the results provide little support
for having five different response levels for
item 13. For this reason, future versions of the
KADS will be produced with the upper two re-
sponse levels collapsed into one (plans and/or
actions).
results of the bootstrap indicated that the most
reliable difference in AUC magnitude was that
between the six-item and 16-item KADS; of the
two, the six-item KADS’s diagnostic accuracy
was better, with a significance level of 0.053
(two-sided). The six-item KADS also per-
formed better than the BDI, but the difference
could be explained by chance (p < 0.293).

Item selection Criterion scores

Table 2 lists the 16 KADS items in accor-
dance with their individual AUC ranks (see
Table 1) and displays the AUC and Cronbach’s
alpha values for the hierarchical sequence of
candidate instruments. The values in a given
row are those for the instrument comprising
the items in and above that row. If we select
the six-item subscale (comprising items 13, 9,
6, 5, 1, and 10), we maintain a Cronbach’s
alpha coefficient of 0.80, with an AUC of 0.89,
both considered acceptable. This represents a
good balance between the brevity required for
an instrument to be used in a busy clinical set-
ting and the ability to predict MDE status.
Figure 2 shows the ROC curves for the BDI,
six-item KADS, and 16-item KADS. Respec-
tively, their AUCs were 0.86, 0.89, and 0.85. All
are comparable, although the ROC curves of
the 16-item KADS and the BDI are mostly
dominated by that of the six-item KADS. The
Inspection of subjects’ six-item KADS scores
indicated that if minimal overall misclassifica-
tion were desired, a cutoff score of either 10
(misclassification = 13%, sensitivity = 62%,
specificity = 93%) or 11 (misclassification =
13%, sensitivity = 42%, specificity = 97%)
should be used. However, if a high level of
sensitivity is required in conjunction with an
acceptable level of specificity, a considerably
lower cutoff score, such as 6, should be chosen.
This cutoff score yields overall misclassifica-
tion, sensitivity, and specificity rates of 25%,
92%, and 71%, respectively.
When it is known whether or not a random
subject’s score is below the chosen cutoff, the
probability of MDE becomes a function of
the instrument’s specificity and sensitivity and
the prevalence rate for MDE in the sample
from which the subject was drawn. The posi-
tive predictive value (PPV) of the instrument is
THE KUTCHER ADOLESCENT DEPRESSION SCALE 119

FIG. 1. Visual summary of the diagnostic power of each Kutcher Adolescent Depression Scale item. The possible re-
sponse levels for particular items are presented along the horizontal axes. The plots show the proportion of respon-
dents who met the major depressive episode criteria. Although for some items, the number of subjects endorsing high
response levels was relatively low, for most items, the proportion of respondents who met the major depressive epi-
sode criteria can be seen to increase with increases in the response level.

the probability that the disorder is present
given a criterion score (i.e., a six-item KADS
score $ 6, say). The negative predictive value
(NPV) is the probability that the disorder is ab-
sent given a noncriterion score (i.e., a six-item
KADS score < 6, say). These probabilities can
be calculated as follows:
PPV = (Sensitivity 2 Prevalence)/
[(Sensitivity 2 Prevalence) + (1 2 Specificity)
2 (1 2 Prevalence)]
NPV = [Specificity 2 (1 2 Prevalence)]/
[(1 2 Sensitivity) 2 Prevalence + Specificity
2 (1 2 Prevalence)]
Table 3 summarizes the instrument’s sensi-
tivity and specificity rates across a range of
possible cutoff scores. The table also illustrates
the PPV and NPV probabilities for two differ-
ent prevalence rates: one for the community
sample of the current study and another for a
putative clinical setting where, for illustrative
purposes, the probability of an adolescent pre-
senting to clinic with an MDE is set at 10%.
The prevalence rate (preassessment likelihood
of MDE) for the total sample of grade 7 to
grade 12 students of this study (n = 1,712) was
estimated as 30/161 2 309/1,712 = 0.034,
where 30/161 was the fraction of interviewees
confirmed as cases of MDE and 309/1,712 was
120 LEBLANC ET AL.

TABLE 2. ROC AREAS AND CRONBACH’S ALPHA COEFFICIENTS FOR CANDIDATE SUBSCALES OF THE KADS
Subscale Subscale: Subscale:
composition Content of item added to subscale ROC area Cronbach’s alpha

Item 13 alone Thoughts/actions re: suicide/self-harm 0.81 —

Adding item 9 Less experience of fun, pleasure 0.86 0.47
Adding item 6 Tired, low in energy, low motivation 0.88 0.62
Adding item 5 Feeling worthless, hopeless 0.89 0.72
Adding item 1 Low mood, depressed, can’t be bothered 0.89 0.78
Adding item 10 Feeling worried, nervous, panicky 0.89 0.80
Adding item 4 Less interest in friends, work, going out 0.89 0.81
Adding item 2 Irritable, losing temper 0.90 0.82
Adding item 3b Poor sleep during night, waking up 0.90 0.81
Adding item 8 Appetite changes 0.89 0.83
Adding item 11 Headaches, butterflies, nausea, etc. 0.89 0.84
Adding item 12 Interest in/thoughts re: sex, sexual arousal 0.90 0.83
Adding item 3c Waking too early in morning 0.89 0.83
Adding item 7 Trouble concentrating, daydreaming 0.88 0.85
Adding item 3d Sleeping during day, taking naps 0.87 0.85
Adding item 3a Trouble falling asleep, lying awake 0.85 0.87

Items are listed according to their individual ROC area ranks (see Table 1). Values in a given row are for the subscale
comprising items listed in and above that row. KADS = Kutcher Adolescent Depression Scale; ROC = receiver operat-
ing characteristic.

FIG. 2. Receiver operating characteristic curves for the Beck Depression Inventory (BDI), six-item Kutcher Adoles-
cent Depression Scale (KADS), and 16-item KADS. For the most part, the curve of the six-item KADS dominates over
those of the other two instruments.
THE KUTCHER ADOLESCENT DEPRESSION SCALE 121

TABLE 3. D IAGNOSTIC VALIDITY STATISTICS FOR THE 6-ITEM KADS OVER A RANGE OF
C UTOFF SCORES, FOR THE COMMUNITY SAMPLE STUDIED, AND FOR A PUTATIVE CLINICAL
SETTING IN WHICH THE PREVALENCE R ATE IS 0.10

Prevalence = 0.034a Prevalence = 0.10

Cutoff Sensitivity Specificity
score rate (%) rate (%) PPV NPV PPV NPV

6 92 71 0.10 1.00 0.26 0.99

7 81 78 0.11 0.99 0.29 0.97
8 81 86 0.16 0.99 0.39 0.98
9 69 90 0.19 0.99 0.43 0.96
10 62 93 0.23 0.99 0.49 0.96
11 42 97 0.32 0.98 0.60 0.94

KADS = Kutcher Adolescent Depression Scale; MDE = major depressive episode.

a The prevalence rate (preassessment likelihood of MDE) for the total sample of grade 7

to grade 12 students (n = 1,712) of this study was estimated as 30/161 2 309/1,712 = 0.034,
where 30/161 was the fraction of interviewees confirmed as cases of MDE, and 309/1,712
was the fraction of all original subjects invited for interview.

the fraction of all original subjects invited for
interview.
DISCUSSION
The KADS was designed to address limita-
tions of previous instruments used to assess
depression in adolescents. Its items address
the core symptoms of depression, giving it the
potential to have good predictive or criterion
validity as well as good evaluative properties
(i.e., the ability to assess change over time).
This first study of its characteristics compared
the diagnostic ability of the complete 16-item
version and a brief six-item version of the
KADS with the often-used BDI for adolescents
diagnosed with MDE. In this school-based
sample of adolescents, the brief version out-
performed both the full version of the KADS
and the BDI, and bootstrapping results sug-
gest that the six-item KADS would perform at
least as well as the BDI in similar samples of
adolescents. It has the major advantage of
brevity (6 items vs. 21 in the BDI), which al-
lows it to be used as a practical screening tool
in a clinical setting where, for example, it
could be filled in by a patient before he or she
sees a mental health professional. As well, the
small number of items poses a low respondent
burden, thereby allowing it to be combined
with other brief instruments that screen for
different mental health conditions.
The criterion standard for diagnosis used in
this study (i.e., the MINI) will not be in perfect
agreement with other standards (e.g., a lay per-
son using another structured instrument such
as the Diagnostic Interview Schedule for Chil-
dren Version IV; Shaffer et al. 2000), a clinician
using a semistructured instrument such as the
Schedule for Affective Disorders and Schizo-
phrenia for school-age children (Puig-Antich
and Chambers 1978, unpublished manuscript),
a psychiatrist applying DSM-IV criteria without
a standardized interview schedule, and the like.
No two standards will agree completely with
one another on which subjects are cases and
which subjects are not cases; furthermore, there
is no agreed upon gold standard diagnostic
method. The implications of this for our analy-
sis of the KADS are difficult to predict. If, for ex-
ample, another standard were preferred, and
the KADS agreed better with this standard than
did the BDI, the KADS would further outper-
form the BDI, as measured by the area under
the ROC curve. On the other hand, if the BDI
agreed better with another standard, the small
advantage enjoyed by the KADS would de-
crease and perhaps reverse. It would be useful,
in a future study, to compare the instruments
directly to another standard such as DSM-IV di-
agnosis by at least two trained professionals,
using a protocol where interrater disagree-
ments are acknowledged and resolved.
To a small extent, the study design may have
affected the diagnostic validity results of the
122
KADS and of the BDI in that the BDI itself was
used earlier as the screening tool to determine
which subjects to invite for interview. There
may have been a small number of subjects
who were screened out by the initial BDI but
who might have been correctly identified as
cases by the KADS and/or by the second BDI
had they not been screened out. Contrarily,
there may have been a few subjects who were
screened out by the initial BDI but who might
have been wrongly identified as cases by the
KADS and/or by the second BDI had they not
been screened out. We predict that any bias
would likely favor the BDI in a comparison
against the KADS because all of its true posi-
tives have an opportunity for further testing,
whereas some true positives that might have
been identified by the KADS but not the BDI
would not have been assessed further. It is
likely that such a bias is small and clinically
unimportant.
We chose a cutoff score of 6 for the six-item
KADS to enable it to serve as a relatively sensi-
tive screening instrument. A researcher or clin-
ician could choose a different cutoff according
to the sensitivity and specificity desired for a
particular application. As the cutoff score is in-
creased, sensitivity will decrease and speci-
ficity will increase. Our data indicate that if the
overall prevalence of MDE is any value < 10%,
a six-item KADS score < 6 should virtually
rule out MDE, because the NPV would imply
a probability of MDE # 1%. A criterion score
would merely indicate the need for further as-
sessment because the corresponding PPV
would still imply a probability of < 50% (un-
less the prevalence rate is $ 10% and/or the
KADS score is $ 11). Although diagnostic sus-
picion should be raised, it is still most likely
that the individual is not experiencing an
MDE. Low PPVs are common to all existing
self-report depression scales when they are
used with cutoff scores that are chosen to
maintain high sensitivity. In achieving a speci-
ficity rate of > 70% while maintaining very
high sensitivity (>90%), however, the KADS is
less afflicted than other such instruments in
this respect.
The 16-item KADS has excellent content va-
lidity in that the core symptoms of depression
(as specified in DSM-IV) are coded explicitly
LEBLANC ET AL.
in the items. This is intended as a means of
identifying cases of adolescent depression and
of monitoring the severity of these core symp-
toms via self-report. As such it is not intended
to deepen our understanding of adolescent de-
pression itself. It is noteworthy that most of
the questions in the six-item version address
specific symptoms of depression such as anhe-
donia and affect, whereas many of the other
questions in the full version address symp-
toms that occur in mood disorders but are not
specific to them. The six-item KADS per-
formed at least as well as the full-length ver-
sion and would be a worthwhile candidate for
screening in a busy primary care practice set-
ting. Whether this short version has reason-
able evaluative properties remains to be
assessed using longitudinal data. Studies with
longitudinal data are necessary to determine
whether the final KADS will consist of a single
instrument (e.g., six-item version) that has rea-
sonable diagnostic and evaluative properties,
or a package with a six-item version intended
for screening and a longer version (between
7 and 16 questions) that can be used for evalu-
ative purposes. One such study is currently in
progress.
Readers interested in obtaining further in-
formation about the KADS should contact
Stan Kutcher at the Department of Psychiatry,
Dalhousie University, Queen Elizabeth II
Health Sciences Centre, Abbie J. Lane Build-
ing, Suite 9212, 5909 Veterans’ Memorial Lane,
Halifax, Nova Scotia, Canada B3H 2E2,
(e-mail: Stan.Kutcher@dal.ca).
ACKNOWLEDGMENTS
This work was supported in part by a con-
tract from Servier Canada, Ltd. (GlaxoSmith-
Kline, Canada) and by the Designated
Mental Health Research Fund, Province of
Nova Scotia.
REFERENCES
Ambrosini PJ, Metz C, Prabucki K, Lee J-C: Videotape re-
liability of the third revised edition of the K-SADS. J
Am Acad Child Adolesc Psychiatry 28:723–728, 1989.
THE KUTCHER ADOLESCENT DEPRESSION SCALE
American Psychiatric Association: Diagnostic and Statis-
tical Manual of Mental Disorders, 3rd ed., rev. (DSM-
III-R) Washington (DC), American Psychiatric
Association, 1987.
American Psychiatric Association: Diagnostic and Statistical
Manual of Mental Disorders, 4th ed. (DSM-IV). Washing-
ton (DC), American Psychiatric Association, 1994.
Amorim P, Lecrubier Y, Weiller E, Hergueta T, Sheehan
D: DSM-III-R psychotic disorders: Procedural validity
of the Mini International Neuropsychiatric Interview
(MINI): Concordance and causes for discordance with
the MIDI. Eur Psychiatry 13:26–34, 1998.
Angold A, Costello EJ, Messer SC, Pickles A, Winder F,
Silver D: Development of a short questionnaire for
use in epidemiological studies of depression in chil-
dren and adolescents. Int J Methods Psychiatr Res
5:237–249, 1995a.
Angold A, Prendergast M, Cox A, Harrington R, Simonoff
E, Rutter M: The child and adolescent psychiatric as-
sessment (CAPA). Psychol Med 25:739–753, 1995b.
Beck AT, Ward C, Mendelson M, Mock J, Erbaugh J: An
inventory for depression. Arch Gen Psychiatry
4:561–571, 1961.
Birleson P: The validity of depressive disorder in child-
hood and the development of a self-rating scale: A re-
search report. J Child Psychol Psychiatry 22:73–88, 1981.
Brooks SJ, Kutcher S: Diagnosis and measurement of
adolescent depression: A review of commonly uti-
lized instruments. J Child Adolesc Psychopharmacol
11:341–376, 2001.
Costello EJ, Angold A: Scales to assess child and adoles-
cent depression: Checklists, screens, and nets. J Am
Acad Child Adolesc Psychiatry 27:726–737, 1988.
Costello EJ, Benjamin R, Angold A, Silver D: Mood vari-
ability in adolescents: A study of depressed, nonde-
pressed and comorbid patients. J Affect Disord
23:199–212, 1991.
Cronbach LJ: Coefficient alpha and the internal structure
of tests. Psychometrika 16:297–334, 1951.
Doerfler LA, Felner RD, Rowlinson RT, Evans E, Raley
PA: Depression in children and adolescents: A com-
parative analysis of the utility and construct validity
of two assessment measures. J Consult Clin Psychol
56:769–772, 1988.
Endicott J, Spitzer R: A diagnostic interview: A schedule
for affective disorders and schizophrenia. Arch Gen
Psychiatry 35:837–844, 1978.
Faulstich ME, Carey MP, Ruggiero L, Enyart P, Gresham
F: Assessment of depression in childhood and adoles-
cence: An evaluation of the Center for Epidemiologi-
cal Studies Depression Scale for Children (CES-DC).
Am J Psychiatry 143:1024–1027, 1986.
Fundudis T, Berney T, Kolvin I, Famuyiwa OO, Barrett L,
Bhate S, Tyrer SP: Reliability and validity of two self-
rating scales in the assessment of childhood depres-
sion. Br J Psychiatry 11 (Suppl):36–40, 1991.
Garrison CZ, Addy CL, Jackson KL, MeKeown RE,
Waller JL: The CES-D as a screen for depression and
other psychiatric disorders in adolescents. J Am Acad
Child Psychiatry 30:636–641, 1991.
123
Gotlib H: Depression and general psychopathology in
university students. J Abnorm Psychol 93:19–30, 1984.
Hamilton LC: Bootstrap methods. In: Regression with
Graphics. Pacific Grove (California), Brooks/ Cole
Publishing, 1992, pp 313–326.
Hamilton M: A rating scale for depression. J Neurol Neu-
rosurg Psychiatry 23:56–62, 1960.
Hanley JA, McNeil BL: The meaning and use of the area
under a receiver operating characteristic (ROC) curve.
Radiology 143:29–36, 1982.
Henderson AR: Assessing test accuracy and its clinical con-
sequences: A primer for receiver operating characteris-
tic curve analysis. Ann Clin Biochem 30:521–539, 1993.
Kovacs M: The Children’s Depression Inventory. North
Tonawanda (New York), Mental Health Systems, 1992.
Kutcher S, Marton P: Utility of the Beck Depression In-
ventory with psychiatrically disturbed adolescent
outpatients. Can J Psychiatry 34:107–109, 1989.
LeBlanc JC, Morehouse R, Rajda M, Kutcher S, Hayden
D, Thompson S, Rusak B: Prevalence of Delayed Sleep
Phase Syndrome in Adolescents: A School-Based Sur-
vey [abstract]. San Francisco, Society for Pediatric Re-
search, May 1999.
Lecrubier Y, Sheehan DV, Weiller E, Amorim P, Bonora I,
Harnett Sheehan K, Janavs J, Dunbar GC: The Mini In-
ternational Neuropsychiatric Interview (MINI): A short
diagnostic structured interview: Reliability and validity
according to the CIDI. Eur Psychiatry 12:224– 231, 1997.
McCauley E, Mitchell JR, Burke P, Moss S: Cognitive at-
tributes of depression in children and adolescents. J
Consult Clin Psychol 56:903– 908, 1988.
Metz CE: Basic principles of ROC analysis. Semin Nucl
Med 8:283–298, 1978.
Radloff LS: A CES-D scale: A self-report depression scale
for research in the general population. Applied Psy-
chological Measurement 1:385–401, 1977.
Reynolds WM, Coats KI: A comparison of cognitive-be-
havioral therapy and relaxation training for the treat-
ment of depression in adolescents. J Consult Clin
Psychol 54:653–660, 1986.
Reynolds WM: Reynolds Adolescent Depression Scale
Professional Manual. Odessa (Florida), Psychological
Assessment Resources, 1987.
Reynolds WM: Assessment of depression in children and
adolescents by self-report questionnaires. In: Hand-
book of Depression in Children and Adolescents.
Edited by Reynolds W, Johnston H. New York,
Plenum Press, 1994, pp 209–234.
Roberts RE, Vernon SW, Rhoades HM: Effects of lan-
guage and ethnic status on reliability and validity of
the Center for Epidemiologic Studies– Depression
Scale with psychiatric patients. J Nerv Ment Dis 177:
581–592, 1989.
Roberts RE, Lewinsohn PM, Seeley JR: Screening for
adolescent depression: A comparison of depression
scales. J Am Acad Child Adolesc Psychiatry
30:58–66, 1991.
Shaffer D, Fisher P, Lucas CP, Dulcan MK, Schwab-Stone
M: NIHM Diagnostic Interview Schedule for Children
Version IV (NIMH DISC-IV): Description, differences
124 LEBLANC ET AL.

from previous versions, and reliability of some com- Wood A, Kroll L, Moore A, Harrington R: Properties of
mon diagnoses. J Am Acad Child Psychiatry 39:28–38, the Mood and Feelings Questionnaire in adolescent
2000. psychiatric outpatients: A research note. J Child Psy-
Sheehan DV, Lecrubier Y, Sheehan KH, Janavs J, Weiller E, chol Psychiatry 36:327–334, 1995.
Keskiner A, Schinka J, Knapp E, Sheehan MF, Dunbar World Health Organization: The ICD-10 Classification of
GC: The validity of the Mini International Neuropsy- Mental and Behavioral Disorders. Geneva, World
chiatric Interview (MINI) according to the SCID-P and Health Organization, 1992.
its reliability. Eur Psychiatry 12:232–241, 1997.
Sheehan DV, Lecrubier Y, Sheehan KH, Amorim P, Janavs
J, Weiller E, Hergueta T, Baker R, Dunbar GC: The
Address reprint requests to:
Mini-International Neuropsychiatric Interview Stanley C. Kutcher, M.D., M.Sc., FRCPC
(M.I.N.I.): The development and validation of a struc- Department of Psychiatry
tured diagnostic psychiatric interview for DSM-IV and Dalhousie University
ICD-10. J Clin Psychiatry 59 (Suppl 20):22–33, 1998. Queen Elizabeth II Health Sciences Centre
Thapar A, McGuffin P: Validity of the shortened Mood Abbie J. Lane Building
and Feelings Questionnaire in a community sample of Suite 9212
children and adolescents: A preliminary research 5909 Veterans’ Memorial Lane
note. Psychiatry Res 81:259–268, 1998. Halifax, Nova Scotia
Weissman MM, Orvashel H, Padian N: Children’s symp-
Canada B3H 2E2
tom and social functioning self-report scales: Com-
parison of mothers’ and children’s reports. J Nerv
Ment Dis 168:736–740, 1980.
E-mail: Stan.Kutcher@dal.ca
THE KUTCHER ADOLESCENT DEPRESSION SCALE 125

APPENDIX a. hardly ever

THE KUTCHER ADOLESCENT b. much of the time
DEPRESSION SCALE c. most of the time
Over the last week, how have you been “on d. all of the time
average” or “usually” regarding the following 5. Feelings of worthlessness, hopelessness,
items: letting people down, not being a good
person.
1. Low mood, sadness, feeling blah or a. hardly ever
down, depressed, just can’t be bothered. b. much of the time
a. hardly ever c. most of the time
b. much of the time d. all of the time
c. most of the time 6. Feeling tired, feeling fatigued, low in
d. all of the time energy, hard to get motivated, have to
2. Irritable, losing your temper easily, feel- push to get things done, want to rest or
ing pissed off, losing it. lie down a lot.
a. hardly ever a. hardly ever
b. much of the time b. much of the time
c. most of the time c. most of the time
d. all of the time d. all of the time
3. Sleep difficulties: different from your 7. Trouble concentrating, can’t keep your
usual (over the years before you got sick). mind on schoolwork or work, day-
3a.Trouble falling asleep, lying awake in dreaming when you should be working,
bed. hard to focus when reading, getting
a. hardly ever “bored” with work or school.
b. much of the time a. hardly ever
c. most of the time b. much of the time
d. all of the time c. most of the time
3b.Sleeping poorly during the d. all of the time
night,wak-ing up, getting out of bed. 8. Appetite changing from usual (before
a. hardly ever you got sick); not feeling hungry, not
b. much of the time wanting to eat or feeling really hungry,
c. most of the time wanting to eat a lot.
d. all of the time a. hardly ever
3c.Waking up too early in the morning, b. much of the time
at least 1 hour before you want to or c. most of the time
need to get up. d. all of the time
a. hardly ever Note: increased __ or decreased __
b. much of the time 9. Feeling that life is not very much fun, not
c. most of the time feeling good when usually (before get-
d. all of the time ting sick) would feel good, not getting as
3d.Sleep during the day, taking naps, much pleasure from fun things as usual
lying down to rest. (before getting sick).
a. hardly ever a. hardly ever
b. much of the time b. much of the time
c. most of the time c. most of the time
d. all of the time d. all of the time
4. Feeling decreased interest in hanging out 10. Feeling worried, nervous, panicky, tense,
with friends; being with your best friend; keyed up, anxious.
being with your spouse/boyfriend/girl- a. hardly ever
friend; going out of the house; doing b. much of the time
schoolwork or work; doing hobbies, c. most of the time
sports, or recreation. d. all of the time
126
11. Physical feelings of worry like: head-
aches, butterflies, nausea, tingling, rest-
lessness, diarrhea, shakes or tremors.
a. hardly ever
b. much of the time
c. most of the time
d. all of the time
12. Interest in sex, thoughts about sex, sex-
ual arousal (compared to before you
were ill), sexual fantasies.
a. as usual or think about it more
LEBLANC ET AL.
b. occasionally think about it
c. seldom think about it
d. never think about it
13. Thoughts, plans, or actions about suicide
or self-harm.
a. no thoughts or plans or actions
b. occasional thoughts, no plans or actions
c. frequent thoughts, no plans or actions
d. plans
e. actions
This article has been cited by:

1. Ming-Shun Chung, Hsien-Jane Chiu, Wen-Jung Sun, Chieh-Nan Lin, Chien-Cheng Kuo, Wei-Che Huang,
Ying-Sheue Chen, Hui-Ping Cheng, Pesus Chou. 2011. A rapid screening test for depression in junior high
school children. Journal of the Chinese Medical Association . [CrossRef]
2. Teresa Carnevale. 2011. An Integrative Review of Adolescent Depression Screening Instruments:
Applicability for Use by School Nurses. Journal of Child and Adolescent Psychiatric Nursing 24:1, 51-57.
[CrossRef]
3. Pim Cuijpers, Petra Boluijt, Annemieke Straten. 2008. Screening of depression in adolescents through the
Internet. European Child & Adolescent Psychiatry 17:1, 32-38. [CrossRef]
4. Nancy DeRosa, M. Cynthia Logsdon. 2006. A Comparison of Screening Instruments for Depression in
Postpartum Adolescents. Journal of Child and Adolescent Psychiatric Nursing 19:1, 13-20. [CrossRef]
5. Graham J Emslie, Taryn L Mayes, Maryse Ruberu. 2005. Continuation and Maintenance Therapy of Early-
Onset Major Depressive Disorder. Pediatric Drugs 7:4, 203-217. [CrossRef]
6. Sarah Brooks. 2004. The Kutcher Adolescent Depression Scale (KADS). Child and Adolescent
Psychopharmacology News 9:5, 4-6. [CrossRef]
7. HIMANSHU P. UPADHYAYA, KATHLEEN T. BRADY, WEI WANG. 2004. Bupropion SR in
Adolescents With Comorbid ADHD and Nicotine Dependence: A Pilot Study. Journal of the American
Academy of Child & Adolescent Psychiatry 43:2, 199-205. [CrossRef]
8. Lili Penkower, Mary Amanda Dew, Demetrius Ellis, Susan M. Sereika, Julius M. M. Kitutu, Ron Shapiro.
2003. Psychological Distress and Adherence to the Medical Regimen Among Adolescent Renal Transplant
Recipients. American Journal of Transplantation 3:11, 1418-1425. [CrossRef]
9. Sarah J. Brooks, Stanley P. Krulewicz, Stan Kutcher. 2003. The Kutcher Adolescent Depression Scale:
Assessment of Its Evaluative Properties over the Course of an 8-Week Pediatric Pharmacotherapy Trial.
Journal of Child and Adolescent Psychopharmacology 13:3, 337-349. [Abstract] [Full Text PDF] [Full Text
PDF with Links]

View publication stats