Full text of DOI:10.1002/mpr.103

8
6
International Journal of Methods in Psychiatric Research, Volume 10, Number 2
Screening for mental disorders in cancer
patients – discriminant validity of HADS and
GHQ-12 assessed by standardized clinical
interview
KATRIN REUTER, MARTIN HÄRTER, Freiburg University Medical Center, Freiburg
ABSTRACT The detection and classification of comorbid mental disorders has major implications in cancer care. Valid
screening instruments for different diagnostic specifications are therefore needed. This study investigated the discriminant
validity of the German versions of the Hospital Anxiety and Depression Scale (HADS) and the General Health
Questionnaire (GHQ-12). A total of 188 cancer patients participated in the examination, consisting first of the assessment
of psychological distress and, second, of the diagnosis of mental disorders according to DSM-IV by clinical standardized
interview (CIDI). Discriminant validity of the two instruments regarding the diagnosis of any mental disorder, anxiety,
depression and multiple mental disorders was compared using ROC analysis.
Overall, the total HADS scale shows a better screening performance than the GHQ-12, especially for the detection of
depressive and anxiety disorders. Best results are achieved for depressive disorders with an area under the curve (AUC) of
0.80, a sensitivity of 79% and a specificity of 76% (cut-off point = 17). For the ability of the instruments to detect patients
with mental disorders in general, the GHQ-12 (AUC: 0.68) shows a similar overall accuracy to the HADS (AUC:
0.70). The screening performance of both scales for comorbid mental disorders is comparable.
The HADS is a valid screening instrument for depressive and anxiety disorders in cancer care. The GHQ-12 can be
considered as an alternative to the HADS when diagnostic specifications are less detailed and the goal of screening proce-
dures is to detect patients with single or multiple mental disorders in general. Limitations of conventional screening
instruments are given through the differing methodological approaches of screening tests (dimensional approach) and diag-
nosis according to DSM-IV (categorical approach).
Key words: cancer, screening, mental disorder, comorbidity, anxiety, depression
Introduction
In the past numerous studies have drawn attention to
psychiatric distress that patients who are suffering from
a tumour disease have to deal with. The majority of
these patients can cope with the diagnosis, the necessary
treatment and the resulting change in their lives. A sig-
nificant number, however, develop clinically relevant
psychiatric syndromes that must be classified by the cur-
rent classification systems (DSM-IV and ICD-10) at
least as an adjustment disorder, frequently as a manifest
psychiatric disorder (Massie and Holland, 1989; Strain,
patients with cancer were set at up to 53% depending on
tumour type, treatment setting and methods (study
design, diagnostic instruments etc.) (Massie and Popkin,
1998; Noyes et al., 1998; Van’t Spijker et al., 1997).
Depressive syndromes are the most frequent (up to 50%)
often fulfilling the criteria of major depression
(McDaniel et al., 1995). Although clinically relevant,
comorbid mental disorders (such as anxiety and affec-
tive disorders) have received only little scientific
attention up to now.
1
998). Additionally, psychiatric disorders may appear
For the affected patients as well as for the attending
teams, it is of great importance to recognize psycholog-
ical symptoms early and to introduce psychological or
psychiatric intervention in time, thus preventing the
that existed premorbidly, the symptoms either reappear-
ing or being intensified by the tumour disease (Breitbart,
1995). The prevalence rates of mental disorders in

Screening for mental disorders in cancer patients
87
chronification of disorders (Zabora, 1998). Besides the
considerable influence of mental impairment regarding
the course of tumour disease and the reduction of life
quality, undetected psychological disorders can lead to
an extended hospital stay and additional costs (Payne
et al., 1999; Zabora, 1998). On account of the specific
problems involved in diagnosing psychiatric disorders
in cancer patients (such as overlap of depression crite-
ria with symptoms of disease and treatment), efficient
instruments are necessary to reliably identify patients
in urgent need of psychological and psychiatric mea-
sures (Payne et al., 1999). Screening instruments can
therefore contribute to the effective use of scarce
resources in health care. However, the available instru-
ments must be compared regarding their validity and
accuracy of detection (Hall et al., 1999; Ibbotson et al.,
Likewise, the General Health Questionnaire in its
12-item version is a short self-report questionnaire,
excluding somatic items and assessing global burden of
disease (Goldberg and Williams, 1988). Conceived for
the detection of psychiatric ‘cases’ in epidemiological
studies, up to now it has mainly been used in large
mental health studies in the community and in prima-
ry care (Goldberg et al., 1998; Jacob et al., 1997;
Schmitz et al., 1999; Üstün and Sartorius, 1995).
Earlier versions of the GHQ in its 30- and 28-item ver-
sions have been implemented with the HADS among
cancer patients (Ford et al., 1995; Ibbotson et al.,
1994). Le Fevre and colleagues (1999) compared the
screening ability of the GHQ-12 and the HADS for
patients in palliative care. All the studies investigating
the discriminant validity of self-report questionnaires
in cancer care used either the diagnostic results from
clinical or structured clinical interviews (such as
SKID) as criteria (Costantini et al., 1999; Ibbotson et
al., 1994; Razavi et al., 1990; Kugaya et al., 1998; Le
Fevre et al., 1999). The CIDI has been implemented as
diagnostic measure in cancer care twice up to now
(Grassi et al., 1997; Jenkins et al., 1991) but so far no
study has carried out a validation of two short screen-
ing instruments using standardized clinical interview
diagnoses as the ‘gold standard’ for validity.
1
994; Chochinov et al., 1997; Hopwood et al., 1991).
Today the Hospital Anxiety and Depression Scale
(
HADS) is the most frequently used self-report ques-
tionnaire for psychological distress in cancer care
Carroll et al., 1993; Herrmann, 1997; Zigmond and
(
Snaith, 1983). As the original English version has been
translated into and validated in many languages, good
psychometric properties have been demonstrated
repeatedly (Chaturvedi et al., 1994; Costantini et al.,
1999; Herrmann, 1997; Kugaya et al., 1998; Malasi et
al., 1991; Razavi, 1990). The HADS was specifically
designed for patients with physical illness; therefore,
the items do not include somatic symptoms of psycho-
logical distress. The bidimensional structure of the
inventory, containing both an anxiety and a depression
subscale, has been confirmed by most studies
The aim of this study was to implement both screen-
ing instruments in the same sample of patients to com-
pare their accuracy of detection and utility for cancer
care. The instruments were compared regarding:
• the ability to detect mental disorders in general;
• the ability to detect depressive disorders; and
• anxiety disorders; as well as
• the ability to identify mental illness in patients
with comorbid psychiatric disorders.
(Costantini et al., 1999; Herrmann, 1997; Moorey et
al., 1991; Spinhoven et al., 1997). Nevertheless, some
authors failed to recognize the two dimensions. It seems
unclear, whether the subscales really measure two dis-
tinct mood states (anxiety/depression), which often
coexist, as shown in epidemiological general population
studies (Angst et al. 1997; Kendler et al. 1993; Magee
et al. 1996; Merikangas 1990) and by clinical experi-
ence (Costantini et al., 1999; Dunbar et al., 2000; Le
Fevre et al., 1999). To that end, several authors decid-
ed to use the 14-item total score as a global measure of
emotional distress, sustaining the original theoretical
framework of the HADS, which was conceived for the
evaluation of milder and mixed features of mood disor-
ders (Costantini et al., 1999; Hopwood et al., 1991;
Ibbotson et al., 1994; Kugaya et al., 2000; Razavi et al.,
Methods
Sample
The study was carried out within the first German clin-
ical epidemiological study on the prevalence of mental
disorders and associated factors in outpatient and inpa-
tient cancer care (Härter et al., 2000; Härter et al., in
press). Between May 1998 and June 2000 patient sam-
ples were recruited from four different inpatient
clinics, one outpatient clinic and two specialized prac-
tices for oncology and haematology in southern
Germany. During a specific period of time, all patients
1990; Van’t Spijker et al., 1997).

8
8
Reuter and Härter
were included consecutively, thus minimizing selection
biases. The survey was based on a two-stage procedure.
First, patients were administered the HADS (Herrmann
et al., 1995), the GHQ-12 (Linden, 1990) and the
SF-36 Health Survey (Bullinger and Kirchberger, 1998)
for life quality. To record mental disorders every second
patient was asked to participate in a clinical interview
using the German version of the CIDI (Wittchen and
Pfister, 1997), a standardized interview, which is well
established in epidemiological research (Wittchen et al.,
referred to as cases in the third analysis. Because pho-
bias do not necessarily impair patients’ daily lives only
those phobias with a reported degree of severity that
indicated a need of treatment were included. Patients
with more than one psychiatric diagnosis from differ-
ent diagnostic categories (such as major depression and
panic disorder) entered the fourth analysis.
According to recent findings that question the two
subscales as separate measures of anxiety and depres-
sion (Costantini et al., 1999; Hopwood et al., 1991;
Ibbotson et al., 1994; Kugaya et al., 2000; Razavi et al.,
1990; Van’t Spijker et al., 1997), the global score
(0–42) of the HADS was used. The differences in
mean scores of HADS and GHQ-12 between the
patients with current mental disorders and no diagno-
sis were analysed with t-test. An ANOVA was
performed for the detection of differences in mean
scores between the patients without diagnosis and the
patients with depressive disorders, anxiety disorders
and comorbid mental disorders. Normality of variances
had been explored with the Kolmogorov-Smirnov test
and the equality of variances had been tested with the
Levene-test adjusting the p-values for the comparison
of mean scores according to its result.
Receiver operating characteristics (ROC) analysis
(Murphy et al., 1987; Zweig and Campbell, 1993) was
used to evaluate the diagnostic accuracy of the HADS
and the GHQ-12 (score 0–12) to detect the cases in
each of the four diagnostic categories. The results of
the CIDI-procedure concerning current (four-week
prevalence) psychiatric diagnosis were used as criteria.
The ROC-curve expresses the sensitivity and specifici-
ty for each score and represents an index of the overall
ability of the instrument to discriminate between cases
and non-cases (Zweig and Campbell, 1993). The esti-
mation of the area under the curve (AUC) quantifies
this accuracy. The values range between 0.5 (no dis-
criminatory ability) and 1.0 (perfect discrimination).
Furthermore it expresses an estimated probability that
a randomly selected case scores higher on the screen-
ing instrument than a randomly selected non-case
(Dunn, 2000). For each diagnostic group the ROC
curves and the corresponding AUCs for both screening
instruments were compared. The cut-off points that
minimize false positive and false negative test results
for each diagnostic category are reported, as well as the
corresponding sensitivity, specificity and the predictive
values (PPV and NPV).
1994, 1998). This assessment system enables the diag-
nosis of mental disorders reliably and efficiently for
specific periods of time (for example, four-weeks, preva-
lence, lifetime prevalence) using the scientific
DSM-IV-criteria. The evaluation is computerized and
for that reason objective regarding the evaluator. The
main advantage of this procedure is the high inter-rater
reliability. All the interviewers were psychologists or
medical doctors with basic clinical experience and had
been trained in this method. In all patients entering the
second stage and being interviewed within two to five
days after the screening procedure the CIDI led to psy-
chiatric diagnosis according the criteria of DSM-IV or
to an absence of diagnosis. The subsample of 188
patients (113 inpatients, 75 outpatients) having com-
pleted the questionnaires and the CIDI represented the
database for this validation study. Their mean age is 54
years (SD 12.1). Because an important part of the study
was conducted in oncological wards of a gynaecological
department, the percentage of women (73%) is nearly
three times higher than that of men (27%). The cancer
sites are breast cancer (42%), gastrointestinal (15%),
gynaecological (14%), haematological (12%), urologi-
cal (4%) and other tumours (13%). Of the 188 patients,
64% have metastases and the mean time since first diag-
nosis is 2.7 years (SD 4.1).
Data analysis
In order to compare the discriminant validity of the
screening instruments, four analyses were performed.
In the first analysis all patients with a current psychi-
atric diagnosis (four-weeks’ prevalence) were selected
as cases. For the following analysis this group of
patients was divided into subgroups. The second analy-
sis entered all patients with a current depressive
disorder only (such as major depression and dys-
thymia). Patients with exclusively anxiety disorders
(panic disorder, simple phobias and so forth) were

Screening for mental disorders in cancer patients
89
Results
one or more anxiety disorder (N = 11), and patients
with more than one mental disorder from different
diagnostic categories (N = 8). Seven patients fulfilled
exclusively the criteria for substance dependence,
somatoform or eating disorder and therefore entered
only the first analysis. The patients of each group
entered as cases the corresponding analysis. All other
patients were used as non-cases for each analysis.
The mean scores of HADS and GHQ-12 are signif-
icantly higher in patients with a current psychiatric
diagnosis than in patients with no mental disorder. In
addition, the patients with depressive disorders and
with comorbid mental disorders differ significantly
from the patients without mental disorders. The
patients with anxiety disorders also have a higher
mean score than patients without a diagnosis even
though this difference is not significant. The differ-
ences in mean scores between the three diagnostic
groups are not significant.
Patients
More than half of the 188 patients have a lifetime
diagnosis of mental disorder (Table 1). There are 21%
who fulfil the criteria for a current DSM-IV diagnosis
(four-weeks’ prevalence). Affective disorders are pre-
sent in 10%, and anxiety disorders in 7% of the
patients. For affective disorders due to medical condi-
tion and anxiety disorders not otherwise specified the
symptomatology is related to the cancer disease and
does not fulfil the criteria for major depression or gen-
eralized anxiety disorder. The diagnostic procedures of
the CIDI do not include adjustment disorders as a diag-
nosis. The prevalence rate of anxiety disorders includes
only those phobias with need of treatment opera-
tionalized through the reported degree of severity.
Another 7% of the patients have other mental disor-
ders either exclusively or additionally.
The ability to identify patients with mental disor-
ders from the corresponding diagnostic groups are
reported in Figures 1 to 4. The overall accuracy of the
two scales in detecting patients with mental disorders
does not differ significantly (p = 0.61, test for equalities
of area). Performing an ROC-analysis for the detection
of depressive disorders in specific, the AUCs of both
scales are higher. The AUC of the HADS surpasses the
AUC of the GHQ-12 even though the difference is
not significant (p = 0.50, test for equalities of area).
Similarly, the analysis shows a better result for the
HADS than the GHQ-12 in identifying anxiety disor-
ders (p = 0.26, test for equalities of area). The fourth
ROC-analysis tested the ability of both instruments to
identify cases with different comorbid mental disor-
ders. In this analysis, the AUCs of HADS and GHQ-
12 are equally good (p = 0.82, test for equalities of
area).
Screening performance
For the comparison of the instruments’ ability to detect
mental disorders in general, depressive and anxiety dis-
orders in specific and comorbid mental disorders, the
patients with a current psychiatric diagnosis (N = 40)
were divided into three groups: a group of patients with
only depressive disorders (N = 14), patients with only
Table 1: Psychiatric Diagnoses According to DSMAV,
N= 188 Cancer Patients (in %)
Lifetime prevalence
52
23
Current prevalence (four-week)
Affective disorder
10
5
3
Major depression
Dysthymia
Affective disorder due to medical condition
For three of the four diagnostic groups (mental dis-
order in general, depressive disorder, comorbid mental
disorder) the best cut-off point is 16/17 for the HADS-
score (table 2). For anxiety disorders the cut-off point
has to be lowered for sufficient sensitivity. The most
balanced values for sensitivity and specificity (79%,
76%) as well as the best positive predictive value
2
Anxiety disorder
7
3
2
1
1
1
Phobia and social phobia
Panic disorder with/without agoraphobia
PTSD
Anxiety disorder not otherwise specified
Obsessive compulsive disorder
(69%) are found for depressive disorders. For mental
disorders in general and comorbid mental disorders, a
sensitivity of 60% and 87% respectively and a speci-
ficity of 79% and 67% respectively, are found. The low
positive predictive value for comorbid mental disorders
Substance dependence (alcohol, tobacco)
Pain disorder
Eating disorder
3
3
1

9
0
Reuter and Härter
Table 2: Mean scores of HADS and GHQ-12
HADS
mean score
SD
6.8
8.5
6.9
8.2
7.3
GHQ-12
mean score
SD
3.6
4.2
3.6
4.4
3.6
a) No diagnosis
N 148
12*a
17.6*
20.9a
17
3.6*a
6.1*
7.1a
5.8
b) Any mental disorder
N=40
c) Depressive disorders
N= 14
d) Anxiety disorders
N = 11
e) Comorbid mental disorders
N=8
21a
7.6a
*
t-test, p<.000; T=4; df--6 1, 1; a < b
* t-test, p<.000; T=33; df--64; a < b
ANOVA, F=8. 1; df=3; p<.007; c,e > a
a
a
ANOVA, F=1 1.8; df--3; p <.000; c, e > a
Figure 1: Receiver Operating Characteristics (ROC) Curves
for the Detection of Mental Disorders. The areas under the
curve (AUC) (95% CI) are 0.70 (0.63-0.77) for the HADS,
Figure 2: Receiver Operating Characteristics (ROC) Curves
for the Detection of Depressive Disorders. The areas under
the curve (AUC) (95% CI) are 0.80 (0.73-0.85) for the
HADS, 0.75 (0.67-0.81) for the GHQ-12; p = 0.53 (test for
equality of areas).
0.68 (0.61-0.75) for the GHQ-12; p = 0.61 (test for equality
of areas).

Screening for mental disorders in cancer patients
91
Figure 3: Receiver Operating Characteristics (ROC) Curves
for the Detection of Anxiety Disorders. The areas under the
curve (AUC) (95% CI) are 0.76 (0.70-0.83) for the HADS,
Figure 4: Receiver Operating Characteristics (ROC) Curves
for the Detection of Comorbid Mental Disorders. The areas
under the curve (AUC) (95% CI) are 0.78 (0.72-0.84) for
the HADS, 0.77 (0.70-0.82) for the GHQ-12; p = 0.82 (test
for equality of areas).
0.68 (0.60-0.75) for the GHQ-12; p = 0.26 (test for equality
of areas).
(
35%) is related to the lower number of patients with
week prevalence) according to DSM-IV criteria.
Therefore, specific analyses regarding the screening
ability of HADS and GHQ-12 for different diagnostic
categories could be performed. The Hospital Anxiety
and Depression Scale showed a better screening per-
formance than the General Health Questionnaire in
its 12-item version for all four diagnostic specifica-
tions. Its superiority was especially marked for the
detection of depressive and anxiety disorders.
Because recent findings have questioned the dis-
criminant validity of the two HADS subscales, the
suggestion of several authors to use the total scale as a
single measure of emotional disturbance according to
the concept of mixed emotions was followed. The best
screening performance was found for depressive disor-
ders (AUC = 0.80, 79% sensitivity; 76% specificity).
For the other diagnostic categories, sensitivity and
specificity values were either lower or less balanced,
meaning that higher percentages of false positive and
false negative cases respectively, have to be considered.
comorbid mental disorders in the sample, as this value
in particular is dependent on prevalence rates of cases
in the study population.
The optimal cut-off points for the GHQ-12 vary
between 2 and 6. Sensitivity is lower than for the
HADS in all diagnostic categories, except for depres-
sive disorders (93%). However, this high sensitivity
value can only be reached through a low cut-off and
therefore low specificity (49%). The specificity for the
GHQ-12 is lower than for the HADS in all analyses
except for anxiety disorders (76%), in which group the
GHQ-12 also reached the best positive predictive
value (60%).
Discussion
Clinical diagnostic interviews with a standardized
assessment system (CIDI) were carried out with 188
inpatients and outpatients and enabled reliable and
valid diagnosis of mental disorders (lifetime and four-

9
2
Reuter and Härter
Table 3: Sensitivity, Specificity, Positive and Negative Predictive Value (PPV 1 NPV) for the Optimal Cut-off Points
(N = 188 Cancer Patients)
HADS score
GHQ-12 score
Cut-
off
Sensitivity Specificity PPV
NPV
(%)
Cut-
off
Sensitivity
(in %)
Specificity
(in %)
PPV
(%)
NPV
(%)
(in %)
60
(in %)
79
(%)
47
Any
Mental
Disorder
16
86
5
55
73
39
84
Base rate:
23.8%
Depressive
Disorder
17
13
16
79
88
87
76
57
67
69
53
35
84
90
96
2
6
3
93
67
75
49
76
55
55
60
25
91
81
91
Base rate:
7.4%
Anxiety
Disorder
Base rate:
6%
Comorbid
Mental
Disorder
Base rate:
4.2%
The reported values for the HADS score vary between
0% and 91% for sensitivity (Razavi et al., 1990;
Kugaya et al., 1998) and 66% and 96% for specificity
Razavi et al., 1992; Kugaya et al., 1998) depending on
through the use of structured or/and standardized
assessment methods for mental disorders (such as
SKID, CIDI). On the other hand, the results of this
study might be due to the fact that subthreshold disor-
ders had not been included and therefore the preva-
lence of cases in the four analyses was rather low. It had
been demonstrated before that the inclusion of adjust-
ment disorders, being less severe and specific in their
symptomatology had a positive effect on screening per-
formance (Razavi et al., 1990; Kugaya et al., 1998).
The cut-off points varied between 16 and 17. Only
for anxiety disorders did a lower cut-off point of 13
maximize sensitivity and specificity. Compared to the
range of cut-off points between 10/11 (Razavi et al.,
1992, Kugaya et al., 1998; Costantini et al., 1999) and
18/19 (Hopwood et al., 1991; Ibbotson et al., 1994;
Malasi et al., 1991; Razavi et al., 1990) that have been
reported for the overall scale, the optimal cut-off
7
(
the homogeneity of the sample, the prevalence of dis-
orders in the sample, the measure used as criterion (for
example structured or clinical interview) and the diag-
nosis included (for example major depression
and/without adjustment disorder). Within this spec-
trum the sensitivity and specificity values of the
present study are situated among the lower results.
Some authors have recently suggested questioning
the suitability of the HADS as a screening instrument
because of poor validation results (Hall et al., 1999;
Ramirez et al., 1995). On the one hand, it is possible
that the HADS’s capacity for detection of mental dis-
orders in cancer patients has been overestimated for a
long time and therefore more realistic values appear

Screening for mental disorders in cancer patients
93
points for the four diagnostic groups showed relative
consistency. Even though this means that similar
severity of case definition on the scale is needed for the
best trade-off between sensitivity and specificity in all
four diagnostic groups, the scores are not necessarily
the best. The cut-off points have to be chosen in rela-
tion to the screening strategies required by specific
clinical and economic situations (for example, choos-
ing a high cut-off for high specificity in order to
minimize false positive results).
Compared to the well-implemented HADS, little
experience is available at this point about the use of the
GHQ-12 in clinical settings. For assessing unspecific
psychological distress, it has been well validated for the
detection of mental disorders in community-based and
cross-cultural studies. The results demonstrated good
overall ability of detection (AUC = 0.88 to 0.96) as
well as satisfying values for sensitivity (83% to 97%)
and specificity (76% to 90%) (Goldberg et al., 1997,
surveys (Sandanger et al., 1998; Goldberg et al., 1998).
As in comorbid mental disorders psychopathology vary
more and might be more severe than in single disorders
the consideration of comorbidity for screening in clini-
cal samples is important. Similar to the results for single
mental disorders, the overall accuracy of HADS and
GHQ-12 doesn’t differ as much as for the specific diag-
nosis of anxiety and depressive disorders. However, as
in most patients with comorbid mental disorders, anxi-
ety and/or depressive disorders are present, sensitivity
and specificity of the HADS are still considerably bet-
ter than of the GHQ-12.
The results of this study have to be looked at under
the restriction of relatively small sample sizes of the
four diagnostic groups, heterogeneous patients regard-
ing duration and severity of illness, and the fact that
rather severe criteria for mental disorders were applied,
excluding adjustment and subthreshold disorders.
Further studies are needed, which investigate the per-
formance of screening instruments for specific
questions regarding diagnostic subgroups, because the
continuous comparison of existing instruments today is
seen as one important condition for the improvement
of screening procedures (Payne et al., 1999; Hall et al.,
1999; Hopwood et al., 1991; Chochinov et al., 1997).
Furthermore, one must realize that like most screen-
ing inventories, the HADS and the GHQ-12 measure
emotional and psychological distress with a dimen-
sional approach, whereas the diagnosis of mental
disorders is based on a categorical approach through
the assessment of psychopathological syndromes.
Therefore, the validity, especially the specificity, of
dimensional instruments for the screening of psychi-
atric disorders will always be limited, because tests and
criteria are based on different methodologies.
Symptom-oriented instruments that screen for mental
disorders following the criteria of DSM-IV could close
this gap in the future. New and promising instruments
in that sense are the Anxiety and Depression
Screening Questionnaires (ASQ-15 and DSQ-15)
developed within the German CIDI-version
(Wittchen and Boyer, 1998; Wittchen and Pfister,
1997), as well as the Patient Health Questionnaire
(PHQ) developed and validated within the Prime-
MD, an interview for mental disorders in primary care
(Spitzer et al., 1994, 1999; Loerch et al., 2000). Equal
to conventional screening inventories, their use should
be investigated for cancer care and the field of somatic
medicine in general, comparing its screening validity
1998; Jacobs et al., 1997; Schmitz et al., 1999). This
study with a clinical sample shows less satisfying results
for the GHQ-12. The highest overall ability of detec-
tion was reached for comorbid mental disorders (AUC
=
0.77). Sensitivity and specificity were, except for
depressive disorders, in all diagnostic groups far below
0%. The wide range of cut-off points reflects the vari-
8
ety of thresholds between 1 to 6 that have been
reported before (Goldberg et al., 1998). In contrast, the
findings from the cross-cultural WHO-study, relating
higher cutoffs with those samples having higher per-
centages of multiple diagnoses (comorbid mental
disorders) were not confirmed here (Goldberg et al.,
1
998).
Corresponding to Le Fevre et al. (1999) who com-
pared the screening ability of the HADS with the
GHQ-12 in a palliative care setting, the HADS is supe-
rior to the GHQ-12 for cancer care. Even though the
screening performance of the HADS could not confirm
the outstanding results of earlier studies, the scale
proved to be a valid screening instrument for the detec-
tion of depressive and anxiety disorders as specific
diagnostic categories. When other, less prevalent men-
tal disorders are included, the superiority of the HADS
is less evident. This means that the GHQ-12, based on
a non-specific concept of psychological distress, might
be considered as an alternative instrument to the HADS
for the screening of mental disorders in general.
The validity of screening for comorbid mental disor-
ders has until now only been investigated in population

9
4
Reuter and Härter
with implemented instruments on the basis of relevant
diagnostic questions. This allows detection of incre-
mental screening abilities of new instruments that
might be useful in different medical care settings.
Ford S, Lewis S, Fallowfield L. Psychological morbidity in
newly referred patients with cancer. J Psychosom Res
1
995; 39: 193–202.
Goldberg D, Williams P. A user´s guide to the General
Health Questionnaire. London: NFER-Nelson, 1988.
Goldberg DP, Gater R, Sartorius N, Üstun TB, Piccinelli M,
Gureje O, Rutter C. The validity of two versions of the
GHQ in the WHO study of mental illness in general
health care. Psychol Med 1997; 27: 191–7.
Acknowledgements
This work resulted from a research project (grant: 01
GD 9802/4) of the Freiburg/Bad Saeckingen
Rehabilitation Research Network. The project is sup-
ported by the Federal Ministry of Education and
Research (BMBF) and the Federation of German
Pension Insurance Institutes (VDR). It was also sup-
ported in part by grants from Hoffmann LaRoche and
Aventis. We sincerely thank the following clinics and
practices for their cooperation in the study and for
their support in collecting the data: Clinic for Tumour
Biology, Freiburg (Prof Dr. Unger, Prof Dr. Bartsch);
Department of Gynaecology and Obstetrics,
University Clinic Freiburg (Prof Dr. Kieback); Portens
Private Clinic St Georg, Höchenschwand (Dr
Kornotzki); Dr Reiber, Freiburg; Dr Marschner,
Freiburg. Special thanks are due to Katrin Groß-Hardt
for her statistical support.
Goldberg DP, Oldehinkel T, Ormel J. Why GHQ threshold
varies from one place to another. Psychol Med 1998; 28:
9
15–21.
Grassi L, Malacarne P, Maestri A, Ramelli E. Depression,
psychosocial variables and occurrence of life events
among patients with cancer. J Affect Disorders 1997; 44:
2
1–30
Hall A, A’Hern R, Fallowfield L. Are we using appropriate
self-report questionnaires for detecting anxiety and
depression in women with early breast cancer? Eur J
Cancer 1999; 35: 79–85.
Härter M, Reuter K, Schretzmann B, Hasenburg A,
Aschenbrenner A, Weis J. Komorbide psychische
Störungen bei Krebspatienten in der stationären
Akutbehandlung und medizinischen Rehabilitation. Die
Rehabilitation 2000; 39: 1–7.
Härter M, Reuter K, Aschenbrenner A, Schretzmann B,
Marschner N, Hasenburg A, Weis J. Psychiatric disorders
and associated factors in cancer: results of an interview
study with patients in inpatient, rehabilitation and outpa-
tient treatment. Eur J Cancer (in press).
Herrmann C. International experiences with the Hospital
Anxiety and Depression Scale - a review of validation
data and clinical results. J Psychosom Res 1997; 42:
References
Angst J, Merikangas KR, Preisig M. Subthreshold syndromes
of depression and anxiety in the community. J Clin
Psychiatry 1997; 58; suppl 8: 6–10.
Breitbart W. Identifying patients at risk for, and treatment of
major psychiatric complications of cancer. Support Care
Cancer 1995; 3: 45–60.
1
7–41.
Bullinger M, Kirchberger I. SF-36. Fragebogen zum
Gesundheitszustand. Göttingen: Hogrefe, 1998.
Carroll BT, Kathol RG, Noyes R Jr., Wald TG, Clamon GH.
Screening for depression and anxiety in cancer patients
using the Hospital Anxiety and Depression Scale. Gen
Hosp Psychiatry 1993; 15: 69–74.
Chaturvedi SK, Chandra PS, Channabasavanna SM, Beena
MB, Pandian RD. Detection of anxiety and depression in
cancer Patients. NIMHANS Journal 1994; 12: 141–4.
Chochinov HM, Wilson KG, Enns M, Lander S. ‘Are you
depressed?’ Screening for depression in the terminally ill.
Am J Psychiatry 1997; 154: 674–6.
Costantini M, Musso M, Viterbori P, Bonci F, Del Mastro L,
Garrone O, Venturini M, Morasso G. Detecting psycho-
logical distress in cancer patients: validity of the Italian
version of the Hospital Anxiety and Depression Scale.
Support Care Cancer 1999; 7: 121–7.
Dunn G. Statistics in Psychiatry. London: Arnold, 2000.
Dunbar M, Ford G, Hunt K, Der G. A confirmatory factor
analysis of the Hospital Anxiety and Depression scale:
comparing empirically and theoretically derived struc-
tures. Br J Clin Psychol 2000; 39 (Pt 1):79–94.
Herrmann Ch, Buss U, Snaith RP. Hospital Anxiety and
Depression Scale-Deutsche Version. Bern: Hans Huber,
1
995.
Hopwood P, Howell A, Maguire P. Screening for psychiatric
morbidity in patients with advanced breast cancer: vali-
dation of two self-report questionnaires. Br J Cancer
1
991; 64: 353–6.
Ibbotson T, Maguire P, Selby P, Priestman T, Wallace L.
Screening for anxiety and depression in cancer patients:
the effects of disease and treatment. Eur J Cancer 1994;
3
0A: 37–40.
Jacobs KS, Bhugra D, Mann AH. The validation of the 12-
item General Health Questionnaire among ethnic Indian
woman living in the United Kingdom. Psychol Med 1997;
2
7: 1215–17.
Jenkins PL, May VE, Hughes LE. Psychological morbidity
associated with local recurrence of breast cancer. Int J
Psychiatry Med 1991; 21: 149–55.
Kendler KS, Neale MC, Kessler RC, Heath AC, Eaves LJ.
Major depression and phobias: the genetic and environ-
mental sources of comorbidity 1993; 23: 361–71.

Screening for mental disorders in cancer patients
95
Kugaya A, Akechi T, Okuyama T, Okamura H, Uchitomi Y.
Screening for psychological distress in Japanese cancer
patients. Jpn J Clin Oncol. 1998; 28: 333–8.
Kugaya A, Akechi T, Okuyama T, Nakano T, Mikami I,
Okamura H, Uchitomi Y. Prevalence, predictive factors,
and screening for psychologic distress in patients with
newly diagnosed head and neck cancer. Cancer 2000; 88:
Ramirez AJ, Richards MA, Jarrett SR, Fentiman IS. Can
mood disorder in women with breast cancer be identified
preoperatively? Br J Cancer 1995; 72: 1509–12.
Razavi D, Delvaux N, Farvacques C, Robaye E. Screening for
adjustment disorders and major depressive disorders in
cancer in-patients. Brit J Psychiat 1990; 156: 79–83.
Razavi D, Delvaux N, Bredart A, Paesmans M, Debusscher L,
Bron D, Stryckmans P. Screening for psychiatric disorders
in a lymphoma out-patient population. Eur J Cancer
1992; 28A: 1869–72.
Sandanger I, Moum T, Ingebrigtsen G, Dalgard OS,
Sorensen T, Bruusgaard D. Concordance between symp-
tom screening and diagnostic procedure: the Hopkins
Symptom Checklist-25 and the Composite International
Diagnostic Interview I. Soc Psychiatry Psychiatr
Epidemiol 1998; 33: 345–54.
Schmitz N, Kruse J, Heckrath C, Alberti L, Tress W.
Diagnosing mental disorders in primary care: the General
Health Questionnaire (GHQ) and the Symptom Check
List (SCL-90-R) as screening instruments. Soc Psychiatry
Psychiatr Epidemiol 1999; 34: 360–6.
Spinhoven P, Ormel J, Sloekers PP, Kempen IJ, Speckens AE,
Van Hemert AM. A validation study of the Hospital
Anxiety and Depression Scale (HADS) in different
groups of Dutch subjects. Psychol Med 1997; 27: 363–70.
Spitzer RL, Williams JB, Kroenke K, Linzer M, Verloin
deGruy F, Hahn SR, Brody D, Johnson JG. Utility of a
new procedure for diagnosing mental disorders im prima-
ry care. JAMA 1994; 272: 1749–56.
2
817–23.
Le Fevre P, Devereux J, Smith S, Lawrie SM, Cornbleet M.
Screening for psychiatric illness in the palliative care
inpatient setting: a comparison between the Hospital
Anxiety and Depression Scale and the General Health
Questionnaire-12. Palliat Med 1999; 13: 399–407.
Linden M. Fragebogen zum gesundheitlichen Befinden
(
General Health Questionnaire, 12-Itemversion).
Deutsche Übersetzung. Unveröffentlicht 1990.
Loerch B, Szegedi A, Kohnen R, Benkert O. The primary
care evaluation of mental disorders (PRIME-MD),
German version: a comparison with the CIDI. J Psychiatr
Res 2000; 34: 211–20.
Magee WJ, Eaton WW, Wittchen HU, McGonagle KA,
Kessler RC. Agoraphobia, simple phobia, and social pho-
bia in the national comorbidity survey. Arch Gen
Psychiatry 1996; 159–68.
Malasi TH, Mirza A, El-Islam MF. Validation of the Hospital
Anxiety and Depression Scale in Arab patients. Acta
Psychiat Scand 1991; 84: 323–6.
Massie JM, Holland JC. Overview of normal reactions and
prevalence of psychiatric disorders. In: Holland JC,
Rowland JH, eds. Handbook of Psychooncology. New
York: Oxford University Press, 1989, pp. 273–82.
McDaniel S, Musselmann DL, Porter MP, Reed DA,
Nemeroff CB. Depression in patients with cancer:
Diagnosis, biology, and treatment. Arch Gen Psychiatry
Spitzer RL, Kroenke K, Williams JB. Validation and utility of
a self-report version of PRIME-MD. JAMA 1999; 282:
1737–44.
Strain JJ. Adjustment Disorders. In: Holland JC, ed. Psycho-
oncology. New York: Oxford University Press, 1998, pp.
509–17.
1
995; 52: 89–99.
Merikangas KR. Comorbidity for anxiety and depression:
review of family and genetic studies. In: Maser JD, CR
Cloninger, eds. Comorbidity of Mood and Anxiety disor-
ders. Washington DC: American Psychiatric Press, 1990,
pp. 331–48.
Moorey S, Greer S, Watson M, Gorman C, Rowden L,
Tunmore R, Robertson B, Bliss J. The factore structure
and factor stability of the Hospital Anxiety and
Depression Scale in patients with cancer. Brit J Psychiat
Üstün TB, Sartorius N (eds) Mental Illness in General
Health Care. Chichester: John Wiley & Sons, 1995.
Van’t Spijker A, Trijsberg RW, Duivenvoorden HJ.
Psychological sequelae of cancer diagnosos: a meta-ana-
lytical review of 59 studies after 1980. Psychosom Med
1997; 59: 280–93.
Wittchen HU. Reliability and validity studies of the WHO-
Composite International Diagnostic Interview (CIDI): a
critical review. J Psychiatr Res 1994; 28: 57–84.
Wittchen HU, Pfister H. DIA-X Interview. Frankfurt: Swets
Test Services Deutschland, 1997.
Wittchen HU, Boyer P. Screening for anxiety disorders.
Sensitivity and specificity of the Anxiety Screening
Questionnaire (ASQ-15). Brit J Psychiat 1998; suppl 34:
10–17.
Wittchen HU, Lachner G, Wunderlich U, Pfister H. Test-
retest reliability of the computerized DSM-IV version of
the Munich-Composite International Diagnostic
Interview (M-CIDI). Soc Psychiatry Psychiatr Epidemiol
1998; 33: 568–78.
1
991; 158: 255–9.
Murphy JM, Berwick DM, Weinstein MC, Borus JF, Budman
SH, Klerman GL. Performance of screening and diagnos-
tic tests. Arch Gen Psychiatry 1987; 44: 550–5.
Noyes R, Holt CS, Massie MJ. Anxiety Disorders. In:
Holland JC, ed. Psycho-oncology. New York: Oxford
University Press, 1998, pp. 548–63.
Payne DK, Hoffman RG, Theodoulou M, Dosik M, Massie
MJ. Screening for anxiety and depression in women with
breast cancer. Psychiatry and medical oncology gear up for
managed care. Psychosomatics 1999; 40: 64–9.

9
6
Reuter and Härter
Zabora JR. Screening procedures for psychological distress.
In: Holland JC (ed.) Psycho-oncology. New York: Oxford
University Press, 1998, pp. 653–62.
Zigmond AS, Snaith RP. The Hospital Anxiety and
Depression Scale. Acta Psychiat Scand 1983; 67: 361–70.
Zweig MH, Campbell G. Receiver-Operating Characteristics
Correspondence: Katrin Reuter, Department of Psychiatry
and Psychotherapy, Freiburg University Medical Center,
Hauptstrasse 5, D-79104 Freiburg.
Phone (+49) 761 290-9538.
Fax (+49) 761 290-9527.
Email Katrin_Reuter@psyallg.ukl.uni-freiburg.de
(ROC) Plots: A Fundamental Evaluation Tool in Clinical
Medicine. Clin Chem 1993; 39: 561–77.