Article

Brief case finding tools for anxiety disorders: Validation of GAD-7 and GAD-2 in addictions treatment

Primary Care Mental Health Service, Leeds Community Healthcare NHS Trust, The Reginald Centre, 263 Chapeltown Road, Leeds LS7 3EX, United Kingdom.
Drug and alcohol dependence (Impact Factor: 3.42). 04/2012; 125(1-2):37-42. DOI: 10.1016/j.drugalcdep.2012.03.011
Source: PubMed
ABSTRACT
Anxiety disorders are the most common mental health problems and often co-exist with substance use. Little evidence exists to support the use of brief screening tools for anxiety disorders in routine addictions treatment. This is the first study to test the validity and reliability of GAD-7 and GAD-2 in an outpatient drugs treatment population.
A sample of 103 patients completed brief screening questionnaires and took part in structured diagnostic assessments using CIS-R. A subgroup of 60 patients completed retests after 4 weeks. The results of brief questionnaires were compared to those of gold-standard diagnostic interviews using Receiver Operating Characteristic (ROC) curves. Psychometric properties were also calculated to evaluate the validity and reliability of self-completed questionnaires.
A GAD-7 score ≥ 9 had a sensitivity of 80% and specificity of 86% for any anxiety disorder, also displaying adequate temporal stability at repeated measurements (intra-class correlation=0.85) and high internal consistency (Cronbach's alpha=0.91). A GAD-2 score ≥ 2 had 94% sensitivity and 53% specificity, with adequate internal consistency (0.82).
GAD-7 adequately detected the presence of an anxiety disorder in drug and alcohol users; although this study was limited by sample size to determine its reliability for specific diagnoses. Results in this small sample suggest that GAD-7 may be a useful screening tool in addiction services, although replication in a larger sample is warranted.

Full-text

Available from: Jaime Delgadillo, Jan 31, 2015
This article appeared in a journal published by Elsevier. The attached
copy is furnished to the author for internal non-commercial research
and education use, including for instruction at the authors institution
and sharing with colleagues.
Other uses, including reproduction and distribution, or selling or
licensing copies, or posting to personal, institutional or third party
websites are prohibited.
In most cases authors are permitted to post their version of the
article (e.g. in Word or Tex form) to their personal website or
institutional repository. Authors requiring further information
regarding Elsevier’s archiving and manuscript policies are
encouraged to visit:
http://www.elsevier.com/copyright
Page 1
Author's personal copy
Drug
and
Alcohol
Dependence
125 (2012) 37–
42
Contents
lists
available
at
SciVerse
ScienceDirect
Drug
and
Alcohol
Dependence
jo
u
rn
al
hom
epage:
www.elsevier.com/locate/drugalcdep
Full
length
article
Brief
case
finding
tools
for
anxiety
disorders:
Validation
of
GAD-7
and
GAD-2
in
addictions
treatment
Jaime
Delgadillo
a,
,
Scott
Payne
b
,
Simon
Gilbody
c
,
Christine
Godfrey
c
,
Stuart
Gore
d
,
Dawn
Jessop
e
,
Veronica
Dale
c
a
Primary
Care
Mental
Health
Service,
Leeds
Community
Healthcare
NHS
Trust,
The
Reginald
Centre,
263
Chapeltown
Road,
Leeds
LS7
3EX,
United
Kingdom
b
Western
Health
and
Social
Care
Trust,
Woodlea
House,
Gransha
Park,
Derry
BT47
6TF,
Northern
Ireland,
United
Kingdom
c
Hull
York
Medical
School
and
Department
of
Health
Sciences,
University
of
York,
Alcuin
College
C
Block,
York
YO10
5DD,
United
Kingdom
d
City
and
South
Community
Drugs
Treatment
Service,
St
Anne’s
Community
Services,
66
York
St.
2nd
Floor,
Leeds
LS9
8AA,
United
Kingdom
e
Prison
Healthcare,
HMP
Leeds,
Leeds
Community
Healthcare
NHS
Trust,
Leeds
LS12
2TJ,
United
Kingdom
a
r
t
i
c
l
e
i
n
f
o
Article
history:
Received
18
January
2012
Received
in
revised
form
10
March
2012
Accepted
10
March
2012
Available online 4 April 2012
Keywords:
Screening
Anxiety
Drugs
Alcohol
Addiction
GAD-7
a
b
s
t
r
a
c
t
Background:
Anxiety
disorders
are
the
most
common
mental
health
problems
and
often
co-exist
with
substance
use.
Little
evidence
exists
to
support
the
use
of
brief
screening
tools
for
anxiety
disorders
in
routine
addictions
treatment.
This
is
the
first
study
to
test
the
validity
and
reliability
of
GAD-7
and
GAD-2
in
an
outpatient
drugs
treatment
population.
Methods:
A
sample
of
103
patients
completed
brief
screening
questionnaires
and
took
part
in
structured
diagnostic
assessments
using
CIS-R.
A
subgroup
of
60
patients
completed
retests
after
4
weeks.
The
results
of
brief
questionnaires
were
compared
to
those
of
gold-standard
diagnostic
interviews
using
Receiver
Operating
Characteristic
(ROC)
curves.
Psychometric
properties
were
also
calculated
to
evaluate
the
validity
and
reliability
of
self-completed
questionnaires.
Results:
A
GAD-7
score
9
had
a
sensitivity
of
80%
and
specificity
of
86%
for
any
anxiety
disorder,
also
displaying
adequate
temporal
stability
at
repeated
measurements
(intra-class
correlation
=
0.85)
and
high
internal
consistency
(Cronbach’s
alpha
=
0.91).
A
GAD-2
score
2
had
94%
sensitivity
and
53%
specificity,
with
adequate
internal
consistency
(0.82).
Conclusions:
GAD-7
adequately
detected
the
presence
of
an
anxiety
disorder
in
drug
and
alcohol
users;
although
this
study
was
limited
by
sample
size
to
determine
its
reliability
for
specific
diagnoses.
Results
in
this
small
sample
suggest
that
GAD-7
may
be
a
useful
screening
tool
in
addiction
services,
although
replication
in
a
larger
sample
is
warranted.
© 2012 Elsevier Ireland Ltd. All rights reserved.
1.
Introduction
Anxiety
disorders
are
the
most
common
among
mental
and
brain
disorders
(Wittchen
et
al.,
2011),
with
international
preva-
lence
rates
in
the
region
of
2.4–18.2%
(WHO
World
Mental
Health
Survey
Consortium,
2004).
Commonly
reported
anxiety
problems
include
mixed
anxiety
and
depressive
disorder,
generalised
anxi-
ety
disorder
(GAD),
panic
disorder
with
and
without
agoraphobia,
social
anxiety,
specific
phobias,
post
traumatic
stress
disorder
(PTSD)
and
obsessive
compulsive
disorder
(OCD).
These
problems
can
range
from
mild
presentations
to
very
severely
disabling
con-
ditions
often
associated
with
anxious
apprehension,
autonomic
hyperarousal,
intense
fear
and
avoidance
of
anxiety
provoking
stimuli.
Corresponding
author.
Tel.:
+44
0113
843
4409.
E-mail
address:
jaime.delgadillo@nhs.net
(J.
Delgadillo).
Questions
remain
about
the
differential
classification
of
specific
diagnoses
and
subtypes;
however
recent
advances
in
research
and
clinical
practice
support
the
notion
that
there
are
key
factors
under-
lying
many
anxiety
disorders.
Transdiagnostic
theoretical
models
of
anxiety
emphasise
the
central
role
of
negative
affect
(NA:
a
tendency
towards
worry,
self-criticism
and
negative
self-view),
sensitiv-
ity
towards
NA
inducing
stimuli,
physiological
hyperarousal
and
emotional
avoidance
strategies
(Norton
and
Philipp,
2008).
NA
has
also
been
found
to
play
a
role
in
the
development
and
maintenance
of
problematic
alcohol
and
drug
use
(Shoal
et
al.,
2005;
Mason
et
al.,
2009).
The
dysregulation
of
stress
response
pathways
has
been
well
documented
in
alcoholism,
and
anxi-
ety
sensitivity
is
known
to
be
implicated
in
chronic
addictions
(Kreek
and
Koob,
1998;
Zvolensky
and
Leen-Feldner,
2005).
Alcohol
dependent
subjects
tend
to
have
higher
stress
and
adrenal
sen-
sitivity
compared
to
healthy
subjects,
and
it
is
likely
that
such
hypersensitivity
and
arousal
may
contribute
to
relapse
and
poor
treatment
engagement
(Sinha
et
al.,
2011).
Consistent
with
the
‘self
0376-8716/$
see
front
matter ©
2012 Elsevier Ireland Ltd. All rights reserved.
doi:10.1016/j.drugalcdep.2012.03.011
Page 2
Author's personal copy
38 J.
Delgadillo
et
al.
/
Drug
and
Alcohol
Dependence
125 (2012) 37–
42
medication’
theory
(Khantzian,
1997),
there
is
some
evidence
that
people
with
anxiety
disorders
often
use
substances
to
regulate
affect
or
avoid
emotional
arousal,
and
this
ranges
between
7.9%
(in
social
phobia
cases)
and
35.6%
(in
GAD;
Bolton
et
al.,
2006).
Altogether,
these
findings
suggest
that
there
are
some
common
risk
factors
that
maintain
stress/anxiety
states
and
substance
use
disorders.
Although
there
is
ongoing
debate
about
the
temporal
sequencing
and
interactions
between
these
disorders,
large
scale
epidemiological
studies
suggest
that
anxiety
disorders
most
often
predate
the
onset
of
problematic
substance
use
(Merikangas
et
al.,
1998).
Unsurprisingly,
people
with
anxiety
disorders
are
2–3
times
at
increased
risk
of
comorbid
substance
misuse
(Regier
et
al.,
1990;
Swendsen
et
al.,
1998).
A
UK
based
study,
for
example,
estimated
that
between
19%
and
32%
of
patients
in
drug
and
alcohol
treatment
services
also
met
diagnostic
criteria
for
a
severe
anxiety
disorder;
however
these
problems
often
went
undetected
and
only
a
minor-
ity
of
patients
received
treatment
for
concurrent
mental
disorders
(Weaver
et
al.,
2003).
Such
deficiencies
in
detection
and
access
to
targeted
treatment
are
important,
given
that
comorbid
disor-
ders
are
associated
with
greater
rates
of
functional
impairment
and
suicidality
(Kessler
et
al.,
1999).
Clinical
guidelines
recommend
the
use
of
standardised
case
finding
measures
to
detect
anxiety
disorders
and
to
monitor
treatment
outcomes
(National
Institute
for
Health
and
Clinical
Excellence,
2011a).
However,
such
measures
are
likely
to
be
less
accurate
in
substance
users
because
intoxication
and
withdrawal
symptoms
are
often
similar
to
those
of
certain
anxiety
disor-
ders
(e.g.,
see
alcohol
withdrawal
criteria
in:
American
Psychiatric
Association,
2000).
Previous
diagnostic
validation
studies
have
found
that
measures
such
as
the
Beck
Anxiety
Inventory
(BAI)
and
the
Brief
Psychiatric
Rating
Scale
for
anxiety
(BPRS-A)
do
not
accu-
rately
distinguish
between
respondents
with
or
without
anxiety
disorders
in
clinical
samples
with
comorbid
substance
use
(Lykke
et
al.,
2008).
The
development
and
validation
of
accurate
anxi-
ety
screening
measures
in
substance
users
therefore
remains
an
important
area
of
research.
This
study
aimed
to
test
the
validity
and
reliability
of
the
7
item
questionnaire
for
generalised
anxiety
disorder
(GAD-7),
and
an
ultra-brief
version
of
the
tool
(GAD-2)
in
a
sample
of
outpatients
in
routine
addictions
treatment.
2.
Method
2.1.
Design
Brief
case
finding
questionnaire
results
were
compared
to
a
gold-standard
struc-
tured
diagnostic
interview
in
a
cross-sectional
sample
of
outpatients.
A
prospective
follow-up
phase
of
4–6
weeks
enabled
us
to
assess
temporal
stability,
with
refer-
ence
to
DSM-IV
criteria
for
differential
diagnosis
(American
Psychiatric
Association,
2000).
The
study
was
conducted
as
part
of
the
CCAS
programme
(Case-finding
and
Comorbidity
in
Addiction
Services);
a
wider
research
project
investigating
common
mental
disorders
in
addictions
treatment.
2.2.
Context
and
participants
The
setting
was
a
community
drugs
treatment
service
in
Leeds,
UK.
The
service
offers
access
to
medical
care,
structured
care
co-ordination
and
psychosocial
inter-
ventions
following
national
treatment
guidelines
(Department
of
Health,
2007).
The
service
engages
approximately
640
patients
per
year,
many
of
whom
have
multi-
ple
social,
financial
and
health
problems.
Patients
commonly
seek
treatment
for
heroin,
alcohol,
crack
and
other
substance
dependence.
Participants
were
recruited
via
sequential
contacts
during
a
full
calendar
year.
This
ensured
that
patients
at
various
stages
of
treatment
and
with
a
range
of
needs
had
equal
probability
of
par-
ticipating.
The
study
excluded
patients
with
severe
mental
illness
such
as
psychotic
disorders
identified
in
clinical
records.
2.3.
Measures
2.3.1.
Generalised
anxiety
disorder
scale
(GAD-7).
GAD-7
is
a
questionnaire
initially
developed
to
diagnose
generalised
anxiety
disorder
and
to
measure
the
severity
of
symptoms
following
DSM-IV
criteria
(Spitzer
et
al.,
2006).
This
is
a
7
item
measure
in
which
each
item
is
rated
on
a
0–3
scale
relating
to
the
frequency
of
anxiety
symp-
toms
over
the
last
two
weeks
(0
=
‘not
at
all’
to
3
=
‘nearly
every
day’).
Scores
range
from
0
to
21
with
higher
scores
indicating
a
greater
severity
of
anxiety.
Some
sample
items
are:
“Feeling
nervous,
anxious
or
on
edge?”;
“Not
being
able
to
stop
or
con-
trol
worrying?”.
Scores
of
5,
10,
and
15
are
taken
to
represent
mild,
moderate,
and
severe
levels
of
anxiety.
This
measure
can
be
self-administered
in
less
than
5
min,
or
administered
by
an
interviewer.
The
original
validation
study
proposes
that
a
cut-off
score
of
10
provides
an
optimal
trade-off
between
sensitivity
(89%)
and
specificity
(82%)
for
a
diagnosis
of
GAD.
The
measure’s
reliability,
construct
validity,
and
factorial
validity
have
been
established
in
the
general
population
(Löwe
et
al.,
2008).
The
capacity
of
the
GAD-
7
to
detect
other
anxiety
disorders
including
social
phobia,
post-traumatic
stress
disorder
and
panic
disorder
has
also
been
established
(Kroenke
et
al.,
2007,
2010).
A
cut-off
score
of
8
or
above
has
been
recommended
to
detect
cases
that
reliably
meet
criteria
for
any
anxiety
disorder.
GAD-2
is
an
ultra-brief
version
of
the
above
measure
and
has
been
recommended
for
use
in
primary
care
as
a
starting
point
in
the
detection
and
assessment
process
for
anxiety
disorders
using
a
cut-off
score
of
3
or
above
(National
Institute
for
Health
and
Clinical
Excellence,
NICE,
2011a).
2.3.2.
Revised
clinical
interview
schedule
(CIS-R).
CIS-R
is
a
gold
standard
structured
diagnostic
interview
that
can
be
administered
by
suitably
trained
lay
interview-
ers
with
the
help
of
a
computer
interface
(Lewis
et
al.,
1992).
This
test
generates
up
to
2
diagnoses
based
on
ICD-10
criteria
(World
Health
Organization,
1992)
for
non-psychotic
affective
and
anxiety
disorders
including:
generalised
anxiety
disor-
der,
mixed
anxiety
and
depressive
disorder,
depressive
episode,
phobias,
obsessive
compulsive
disorder,
and
panic
disorder.
CIS-R
also
generates
an
overall
severity
score
based
on
the
severity
rating
and
frequency
of
mental
health
symptoms
and
problems
covered
in
the
interview.
A
CIS-R
severity
score
of
12
or
more
indicates
a
clinically
significant
diagnosis
and
a
score
above
18
is
indicative
of
a
severe
disorder
warranting
treatment.
CIS-R
has
been
used
as
a
diagnostic
measure
in
the
national
Psychiatric
Morbidity
Surveys
conducted
in
the
UK
(McManus
et
al.,
2009;
Meltzer
et
al.,
1995),
and
also
in
addictions
research
(Gilchrist
et
al.,
2005).
2.3.3.
Secondary
measures.
The
Severity
of
Dependence
Scale
(SDS)
is
a
brief,
5-item
measure
of
compulsive
substance
use
and
it
covers
aspects
of
the
respondent’s
con-
cern
and
degree
of
control
over
his
or
her
consumption
(Gossop
et
al.,
1995).
This
measure
has
been
extensively
validated
in
adult
and
adolescent
samples
(Martin
et
al.,
2006),
with
empirically
derived
diagnostic
cut-off
scores
being
reported
for
dependence
to
alcohol,
heroin,
crack,
cannabis
and
several
other
illicit
and
prescrip-
tion
drugs
(Lawrinson
et
al.,
2007;
Castillo
et
al.,
2010;
Kaye
and
Darke,
2002;
Swift
et
al.,
1998).
Drug
and
alcohol
use
was
assessed
using
the
Treatment
Outcomes
Profile
(TOP).
The
TOP
is
a
validated
composite
20-item
measure
covering
four
domains:
sub-
stance
use,
injecting
risk
behaviour,
crime,
health
and
social
functioning
(Marsden
et
al.,
2008).
This
questionnaire
is
routinely
used
as
an
outcome
measure
in
drugs
treatment
services
in
England
as
part
of
a
drugs
treatment
monitoring
system
led
by
the
National
Treatment
Agency
(NTA).
2.4.
Procedure
Following
approval
from
an
independent
research
ethics
committee,
patients
were
recruited
using
a
standard
study
information
leaflet.
Consenting
participants
were
invited
to
self-complete
brief
questionnaires
in
a
confidential
interview
room,
and
staff
support
was
provided
for
people
who
could
not
do
so
unassisted.
Poor
literacy
was
a
problem
for
nearly
a
quarter
of
all
participants,
and
therefore
staff
assistance
was
an
important
facilitator
to
acceptable
screening
(see
Delgadillo
et
al.,
2012).
Participants
were
then
asked
to
partake
in
a
diagnostic
interview
(CIS-R)
conducted
by
trained
interviewers
who
were
blind
to
self-completed
question-
naire
results
until
the
end
of
the
diagnostic
interview.
This
was
followed
by
an
opportunity
to
discuss
the
test
results
and
any
necessary
assistance
or
treatment.
Participants
were
also
invited
to
participate
in
a
retest
appointment
4–6
weeks
later.
Supermarket
vouchers
valued
at
£
10
were
offered
at
both
baseline
and
follow-up
appointments
to
incentivise
participation
in
the
study.
2.5.
Data
analysis
We
evaluated
the
diagnostic
accuracy
of
GAD-7
and
GAD-2
with
comparison
to
ICD-10
psychiatric
diagnoses
assessed
using
CIS-R
as
a
gold
standard.
GAD-2
results
were
extracted
from
responses
to
the
GAD-7
questionnaire.
Brief
questionnaires
were
evaluated
as
case
finding
tools
for
any
anxiety
disorder
and
also
specifically
for
generalised
anxiety
disorder
(GAD).
Receiver
Operating
Characteristic
(ROC)
curves
were
used
to
assess
the
overall
performance
of
tests
with
reference
to
the
Area
Under
the
Curve
(AUC)
statistic.
We
calculated
a
minimum
sample
of
60
cases
and
43
controls
to
reliably
conduct
ROC
curve
analyses,
according
to
the
sampling
method
proposed
by
Flahault
et
al.
(2005).
This
was
based
on
an
expected
sensitivity
value
of
0.80
informed
by
the
original
validation
of
GAD-7
as
a
screening
tool
for
any
anxiety
disorder
(Kroenke
et
al.,
2007),
a
minimal
acceptable
lower
confidence
limit
of
0.60,
and
an
expected
prevalence
rate
of
approximately
58%
based
on
the
figures
reported
by
Strathdee
et
al.
(2002)
in
UK
substance
use
services.
Sensitiv-
ity,
specificity,
predictive
power
and
likelihood
ratios
were
computed
as
indicators
Page 3
Author's personal copy
J.
Delgadillo
et
al.
/
Drug
and
Alcohol
Dependence
125 (2012) 37–
42 39
Fig.
1.
ROC
curves:
brief
case
finding
tools
for
any
anxiety
disorder.
of
diagnostic
validity
for
the
optimal
cutpoints
identified
using
the
ROC
method.
Validity
was
further
examined
by
correlating
the
measures
against
the
gold
stan-
dard
CIS-R
measure
(convergent
validity)
and
the
SDS
which
theoretically
measures
a
distinct
construct
(discriminant
validity).
The
reliability
of
these
measures
was
evaluated
using
Cronbach’s
alpha
to
test
internal
consistency.
Intra-class
correla-
tions
were
used
to
assess
temporal
stability
between
test
and
retest
after
4
weeks.
Finally,
Youden’s
index
was
calculated
as
a
single
summary
measure
of
overall
test
accuracy;
where
an
upper
index
of
1
represents
a
perfect
test,
and
a
lower
index
of
1
represents
a
flawed
test
(Biggerstaff,
2000).
3.
Results
We
invited
162
patients
to
partake
in
the
study.
A
total
of
103
participated,
60
of
whom
also
completed
retests
after
4
weeks.
Only
3
patients
expressly
declined
participation
and
a
further
56
(35%)
failed
to
attend
their
planned
appointments.
This
was
higher
than
the
28%
attrition
rate
reported
in
outcomes
research
in
compara-
ble
mainstream
addiction
services
in
the
UK
(Gossop
et
al.,
2000).
Most
participants
were
unemployed
(n
=
86,
84%
of
total),
males
(n
=
79,
77%)
and
of
white
British
background
(n
=
96,
93%)
with
a
mean
age
of
35
(SD
=
7.09;
range
=
23–54).
Only
5
(5%)
respondents
were
currently
in
training
or
education.
The
most
frequently
used
substances
were
alcohol,
heroin,
cannabis
and
crack,
with
63
(61%)
respondents
reporting
polysubstance
use
and
only
9
(9%)
report-
ing
abstinence
in
the
last
month.
More
detailed
demographic
and
clinical
characteristics
of
the
participating
sample
are
reported
in
Delgadillo
et
al.
(2011).
Sixty
seven
participants
(65%
of
sample)
met
diagnostic
criteria
for
an
anxiety
disorder
according
to
CIS-R
test
results.
The
most
commonly
diagnosed
anxiety
disorders
were
GAD
(n
=
31,
30%),
mixed
anxiety
and
depressive
disorder
(n
=
27,
26%),
panic
disorder
(n
=
7,
7%)
and
social
phobia
(n
=
7,
7%).
Comorbidity
was
common
in
this
patient
population;
with
40
participants
(39%)
meeting
diag-
nostic
criteria
for
major
depression
and
a
secondary
severe
anxiety
disorder
(this
proportion
excludes
mixed
anxiety
and
depressive
disorder).
Receiver
operating
characteristic
(ROC)
curves
displayed
in
Fig.
1
summarise
the
diagnostic
accuracy
of
GAD-7
and
GAD-2
as
screening
tools
for
any
anxiety
disorder.
Fig.
2
presents
compar-
ative
ROC
curves
for
the
specific
detection
of
generalised
anxiety
disorder.
ROC
curves
represent
the
trade-off
between
sensitivity
and
specificity
across
the
full
range
of
values
for
each
measure;
where
figures
curving
closest
to
the
upper
left
corner
are
indica-
tive
of
good
diagnostic
accuracy.
A
visual
inspection
of
Figs.
1
and
2
reveals
that
both
measures
performed
better
as
broad
case
find-
ing
tools
for
anxiety
disorders
rather
than
specifically
for
GAD.
This
Fig.
2.
ROC
curves:
brief
case
finding
tools
for
generalised
anxiety
disorder.
was
further
confirmed
by
the
comparatively
higher
area
under
the
curve
(AUC)
values
displayed
in
Table
1.
Table
1
describes
the
operating
characteristics
of
case
finding
tools
at
different
cut-off
scores.
A
limited
range
of
alternative
cut-
points
is
presented
for
simplicity,
illustrating
the
relative
trade-off
between
sensitivity
and
specificity.
We
selected
optimal
cut-off
scores
which
maximised
specificity
whilst
maintaining
a
minimum
sensitivity
standard
of
75%.
We
then
calculated
detailed
psychome-
tric
properties
based
on
optimal
cutpoints
displayed
in
Table
2.
GAD-7
as
a
case
finding
tool
for
anxiety
disorders
had
a
sig-
nificant
AUC
value
of
.88
(.81–.95)
and
the
best
trade-off
between
sensitivity
(80%)
and
specificity
(86%)
at
a
cutpoint
of
9
points
and
above.
GAD-2
had
an
AUC
value
of
.86
(.79–.93)
and
was
highly
sen-
sitive
(94%),
but
its
specificity
was
considerably
lower
(53%)
at
an
optimal
cutpoint
of
2
and
above.
These
results
indicate
that
both
measures
are
clinically
useful
based
on
conventional
guidelines
suggesting
that
AUC
values
.70
and
<.90
have
moderately
good
Table
1
Area
under
the
curve,
sensitivity
and
specificity
at
different
cut-off
scores.
AUC
(95%
CI)
Cut-off
Sensitivity
Specificity
Any
anxiety
disorder
a
GAD-7
.882
(.814–.949)
7
0.833
0.694
8
0.818
0.833
9
0.803
0.861
10
0.727
0.917
GAD-2
.858
(.785–.931)
2
0.939
0.528
3
0.667
0.917
4
0.576
0.917
5
0.273
1.000
Generalised
anxiety
disorder
(GAD)
GAD-7
.730
(.631–.828)
7
0.871
0.451
8
0.871
0.535
9
0.839
0.549
10
0.710
0.592
GAD-2
.751
(.657–.846)
2
1.000
0.324
3
0.742
0.662
4 0.613
0.690
5
0.323
0.887
a
Includes:
GAD,
mixed
anxiety
and
depressive
disorder,
panic
disorder,
agora-
phobia,
social
phobia,
specific
phobias,
obsessive
compulsive
disorder.
Page 4
Author's personal copy
40 J.
Delgadillo
et
al.
/
Drug
and
Alcohol
Dependence
125 (2012) 37–
42
Table
2
Psychometric
properties
of
anxiety
disorder
screening
tools
at
the
optimal
cut-off
scores.
Cronbach’s
ICC
Cut-off
Sensitivity
Specificity
+PV
PV
+LR
LR
Youden’s
index
Any
anxiety
disorder
a
GAD-7 0.914
0.854
9 0.803
0.861
0.914
0.689
5.696
0.243
0.652
GAD-2 0.820
0.811
2
0.939
0.528
0.785
0.826
1.989
0.115
0.467
Generalised
anxiety
disorder
GAD-7
0.914
0.854
9
0.839
0.549
0.448
0.889
1.887
0.290
0.394
GAD-2
0.820
0.811
3
0.742
0.662
0.489
0.855
2.195
0.390
0.404
ICC,
intra-class
correlation;
PV,
predictive
value;
LR,
likelihood
ratio.
a
Includes:
GAD,
mixed
anxiety
and
depressive
disorder,
panic
disorder,
agoraphobia,
social
phobia,
specific
phobias,
obsessive
compulsive
disorder.
accuracy
and
AUC
values
.90
are
highly
accurate
(Swets,
1988;
Greiner
et
al.,
2001).
GAD-7
was
accurate
in
91%
of
cases
with
a
positive
screen
and
69%
of
cases
that
did
not
screen
positive.
The
corresponding
positive
and
negative
predictive
values
for
GAD-2
were
78%
and
83%
respectively.
Although
GAD-7
was
most
accu-
rate
overall,
it
was
more
accurate
in
confirming
cases
(+PV
=
0.91)
whilst
the
GAD-2
measure
was
more
accurate
in
confirming
non-
cases
(PV
=
0.83).
Positive
and
negative
likelihood
ratios
(LR)
are
also
displayed
in
Table
2.
These
indicated
that
patients
with
an
anx-
iety
disorder
were
5.7