Journal of Affective Disorders 133 (2011) 281–293

Contents lists available at ScienceDirect

Journal of Affective Disorders

j o u r n a l h o m e p a g e : w w w. e l s ev i e r. c o m / l o c a t e / j a d

Research report

Screening for perinatal common mental disorders in women in the north of

Vietnam: A comparison of three psychometric instruments
Thach D. Tran a,b,c,⁎, Tuan Tran a, Buoi La a,d, Dominic Lee e, Doreen Rosenthal c, Jane Fisher b,c
a
Research and Training Centre for Community Development, Hanoi, Viet Nam
b
Jean Hailes Research Unit, School of Public Health and Preventive Medicine, Monash University, Clayton, Victoria 3008, Australia
c
Centre for Women's Health, Gender and Society, Melbourne School of Population Health, University of Melbourne 3010, Australia
d
TUNA Clinic, Hanoi, Viet Nam
e
School of Public Health, The Chinese University of Hong Kong, Shatin, NT, Hong Kong, China

a r t i c l e i n f o a b s t r a c t

Article history: Background: There is increasing recognition that Perinatal Common Mental Disorders (CMDs)
Received 14 January 2011 are a major public health problem for women in resource-constrained countries. There is an
Received in revised form 25 March 2011 urgent need for screening tools suitable for use by community based health workers to assist in
Accepted 25 March 2011 the identification of people with compromised mental health. The aim of this study was to
Available online 6 May 2011
establish the validity of three widely used psychometric screening instruments in detecting
CMDs in women in northern Viet Nam.
Keywords: Methods: Translated and culturally verified versions of the Edinburgh Postnatal Depression
Screening
Scale (EPDS), General Health Questionnaire 12 items (GHQ-12), Zung's Self-rated Anxiety Scale
Depression
(Zung SAS) and a gold-standard diagnostic tool, the Structured Clinical Interview for DSM IV,
Anxiety
Validation were administered to a community-based representative cohort of 364 Vietnamese women in
Perinatal the perinatal period. Post-hoc analyses, Cronbach's alpha, and Receiver Operating Character-
Developing countries istic (ROC) analyses were performed to identify the optimal cut-off points and to compare the
validity of three scales.
Results: The Areas under the ROC Curve were: EPDS 0.77 (95%CI 0.72–0.82); Zung SAS 0.79 (95%
CI 0.74–0.84) and GHQ-12 0.72 (95%CI 0.67–0.78). The optimal cut-off point for the EPDS was
3/4 (Se 69.7%; Sp 72.9%). The corresponding value for Zung SAS was 37/38 (Se 67.9%; Sp 75.3%)
and for GHQ-12 was 0/1 (Se 77.1%; Sp 56.6%). The internal reliability Cronbach's alpha for EPDS
was 0.75, for Zung SAS was 0.76, and for GHQ-12 was 0.64.
Conclusions: These instruments are suitable for use as screening tools for CMDs in women in
northern Viet Nam, but probably because of differences in emotional literacy, familiarity with
test-taking and the effects of chronic social adversity require much lower cut off scores to
detect clinically significant symptoms than in other settings.
© 2011 Elsevier B.V. All rights reserved.

1. Introduction pregnant women and mothers of infants in resource

There is increasing recognition that non-psychotic com-
mon mental disorders (CMD) (Goldberg and Huxley, 1992) in
⁎ Corresponding author at: Research and Training Centre for Community
Development, 39 lane 255 Vong Street, Hanoi, Viet Nam. Tel.: + 84
436280350; fax: + 84 436280200.
E-mail address: indthach@yahoo.com (T.D. Tran).
constrained settings are a major public health problem. The
prevalence of perinatal depression in women in high income
countries ranges from 10 to 13% (Hendrick, 1998; O'Hara and
Swain, 1996). A recent series of studies provide evidence
that rates of perinatal CMD in women in low and middle
income (LAMI) countries may be at least twice those in
high income countries (Chandran et al., 2002; Cooper et al.,
1999; Patel et al., 2003) (Ekuklu et al., 2004; Rahman et al.,
2003a, 2003b; Fisher et al., 2004; Iranfar et al., 2005; Fisher

0165-0327/$ – see front matter © 2011 Elsevier B.V. All rights reserved.
doi:10.1016/j.jad.2011.03.038
282 T.D. Tran et al. / Journal of Affective Disorders 133 (2011) 281–293

et al., 2010). While the predominant focus has been on
perinatal depression, anxiety disorders either alone or
occurring co-morbidly with depression are also problematic
(Fisher et al., 2009).
In addition to constituting a serious burden to women's
health, there is increasing evidence that maternal mental
health problems affect the health and development of
newborns and infants, in particular in the context of poverty
and chronic social adversity (Engle, 2009; Stewart, 2007). In
low income countries, maternal depression has been linked
directly to low infant birthweight and under-nutrition in the
first year of life, higher rates of diarrhoeal diseases, reduced
completion of recommended schedules of immunisation and
poorer cognitive development in the children (Harpham
et al., 2005; Patel et al., 2003; Rahman et al., 2007; Rahman
et al., 2004; Rahman et al., 2003a, 2003b).
The World Health Organization recommends that in
addition to the provision of appropriate and effective local
interventions, early detection of mental health problems is
essential to improving prognosis and reducing disability
(World Health Organization, 2001). In order to strengthen
the mental health care capacity of low income countries,
there is a clear need for screening tools which primary
health care staff can use to identify people with CMDs,
including pregnant women and mothers of infants in the
community.
There are a number of self-reporting or interviewer-
administered scales to detect symptoms of CMDs devel-
oped in English and translated into other languages for use
in resource-constrained countries. The Edinburgh Postna-
tal Depression Scale (EPDS) is widely used as a screening
tool to detect perinatal depressive symptoms (Cox et al.,
1987a, 1987b). The General Health Questionnaire 12 items
(GHQ-12) (Goldberg and Williams, 1988), and Zung's Self-
rated Anxiety Scale (Zung SAS) (Zung, 1971) are com-
monly used for detection of CMDs in primary health care
settings.
There is a debate about whether psychiatric instruments
have the same sensitivity (Se) and specificity (Se) in
different settings. The concerns focus on whether psychiatric
constructs are equivalent between cultures and nations, the
potential limitations in translation of instruments developed
in vernacular English (Kleinman, 1987), and differences in
response patterns between settings. Therefore, evidence to
establish the validity of an instrument is always required
before its use in a new setting. The EPDS and the GHQ-12
have been validated in some low income countries and the
varied cut-off scores to detect depression or CMDs in these
settings established (see Table 1). However, no validation
studies of the Zung SAS in low income countries have been
published.
Some investigators also suggest that screening instru-
ments perform differently among groups within a popu-
lation. It is argued that as specific concerns about the
health of the foetus and childbirth are elevated in most
women during pregnancy, transient elevated anxiety is
normal and should not be regarded as diagnostic of an
anxiety disorder (Ross et al., 2004). Matthey et al. (2006)
suggest therefore that a higher EPDS cut-off score should
be used to detect clinically significant symptoms in
English-speaking pregnant women than in those who
have recently given birth. Others have found that the
EPDS performs differently in urban than in rural popula-
tions. Tesfaye et al. (2010) found in Ethiopia that the EPDS
performed better in detecting common mental disorders in
relatively well educated urban-dwelling women than in
those living in rural areas.
The primary aim of this study was to validate the
Edinburgh Postnatal Depression Scale, the General Health
Questionnaire 12 and the Zung Self-rated Anxiety Scale for
use in identifying the common mental disorders of depres-
sion and anxiety in pregnant women and mothers of infants
in northern Viet Nam. The secondary aims of the present
study were (1) to clarify whether or not the optimal cut-off

Table 1
Previous validation studies in EPDS and GHQ12 for screening CMD during the perinatal period in developing countries.

Country/author (year) Detection Cut-off Se* Sp** Diagnostic instrument*** diagnostic criteria****

EPDS
Thailand/(Pitanupong et al., 2007) Postpartum depression 6/7 74% 74% Semistructured interview DSM-IV of depression
Ethiopia/(Hanlon et al., 2008) Postpartum CMD 6/7 53% 61% CPRS DSM-IV of CMD
Nigeria/(Uwakwe and Okonkwo, 2003) Postpartum CMD 8/9 75% 97% CIDI ICD10 of CMD
Bangladesh/(Gausia et al., 2007) Postpartum depression 9/10 89% 87% SCID DSM-IV of depression
Nigeria/(Adewuya et al., 2006) Prenatal depression 9/10 87% 92% CIDI DSM-IV of depression
Ghana/(Weobong et al., 2009) Postpartum CMD 10/11 78% 73% CPRS DSM-IV of CMD
Malaysia/(Rushidi et al., 2003) Postpartum depression 11/12 100% 98.2% CIDI ICD10 of depression
Nepal/(Regmi et al., 2002) Postpartum major Depression 12/13 100% 92.6% SCID DSM-IV of major depression
Taiwan/(Teng et al., 2005) Postpartum depression 12/13 96% 85% MINI DSM-IV of depression

GHQ12
India/(Patel et al., 2008) CMD 5/6 73% 90% CIS-R ICD10 of CMD
Chile/(Araya et al., 1992) CMD 4/5 73% 76% CIS-R ICD10 of CMD

*Se : Sensitivity ** Sp : Specificity.

(***)CPRS: Comprehensive Psychopathological Rating Scale; CIDI: Composite International Diagnostic Interview; SCID: Structured Clinical Interview for DSM;
MINI: Mini International Neuropsychiatric Interview; CIS-R: Revised Clinical Interview Schedule.
(****)DSM-IV: Diagnostic and Statistical Manual for Mental Disorders—Fourth Edition; ICD10: WHO International Classification of Diseases.
 T.D. Tran et al. / Journal of Affective Disorders 133 (2011) 281–293
points of the scales differ between pregnant and postpartum
women, and (2) to establish whether or not there were urban
rural differences of the performance in each of the three
scales.
2. Methods
The study used the standard validation method of translation
and cultural verification of the scales and comparison of
responses on these scales with the gold standard of an
independently administered structured clinical interview in a
representative cohort.
2.1. Setting
In order to obtain a sample in which women from both
urban and rural areas were represented, the study was
undertaken in randomly selected study sites in Ha Nam, a
typical Red River delta rural province and Hanoi the
national capital and a major urban centre in northern Viet
Nam. The language in these provinces is Vietnamese and
99% of the population is Kinh or ethnically Vietnamese
(GSO, 2009).
2.2. Sample
It was estimated that with the expected prevalence of 25%
a sample of 350 women would include at least 80 women
meeting diagnostic criteria for a PCMD. A sample of this
size would permit Se and Sp of 70% to be established with a
95% CI ± 10%. A two-stage sampling protocol was used to
select commune health centres. Each district in the two
provinces was assigned a number, and using tables of
random numbers, four in Hanoi (Hai Ba Trung, Dong Da,
Thanh Xuan and Hoang Mai) and six in Hanam (Binh Luc, Duy
Tien, Phu Ly, Ly Nhan, Thanh Liem and Kim Bang) were
selected randomly. Commune health centres have similar
characteristics within districts and using the same tech-
nique, one commune health centre was selected in each
district (Fisher et al., 2010). All women meeting the criteria
of being at least 28 weeks pregnant or mothers of 4–6 week
old babies and registered for pregnancy or newborn health
care at the commune health station were eligible to
participate.
2.3. Materials
The Edinburgh Postnatal Depression Scale (EPDS) is a
widely used 10-item self-reported questionnaire about
feelings experienced over the past seven days (Cox et al.,
1987a, 1987b). Each item has four short statements to
reflect degree of agreement with an item and scored 0–3.
The total score ranges from 0 to 30. Cox et al. (1987a,
1987b) found in women in the north of England that a
threshold of 12/13 detected depressive illness (Se 86%;
Sp 78%) and a threshold of 9/10 for routine use in detecting
clinically significant symptoms at primary health care
level.
Zung's Self-rated Anxiety Scale (Zung SAS) is a 20-item self-
rating scale (5 affective and 15 somatic symptoms) for detec-
tion of anxiety disorders. The scoring method is derived by
283
dividing the total sum of raw scores (obtained on the 20
items) by the maximum possible score, converted to a
decimal and multiplied by 100 (Zung, 1971). A cut-off 44/45
is suggested to detect clinical anxiety (Zung, 1980) and also
detects co-morbidity of anxiety and depression in adults (Zung
et al., 1990).
General Health Questionnaire 12 items (GHQ12) is designed
to detect psychiatric disorders in community settings and
non-psychiatric clinical settings. GHQ12 is a short version
using a four-level alternative response scale. There are two
scoring methods: bi-modal (0-0-1-1) and Likert scoring styles
(0-1-2-3) (Goldberg and Williams, 1988). This study applied
the first method. A cut-off of 1/2 was initially suggested to
detect mental illness at primary health care in England
(Goldberg, 1972).
2.3.1. Structured clinical interview
All participants completed individual psychiatrist-adminis-
tered Structured Clinical Interviews for DSM IV Axis 1
Diagnoses (SCID) modules for depression, generalised anxiety,
and panic disorder (First et al., 1996). This is the demonstrated
gold standard for ascertainment of perinatal mood disorder in
women in diverse cultural settings and countries (Gorman
et al., 2004).
2.3.2. Vietnamese translations and cultural verification
The scales were translated from English into Vietnam-
ese, reviewed by a group of health professionals and
research workers for meaning, comprehensibility and
cultural appropriateness and back-translated to English
for verification (Laungani, 2000). The EPDS had been
modified after a similar process for use in an earlier study
(Fisher et al., 2004) to include appropriate local linguistic
expressions. In particular, item 6 “Things have been getting
on top of me” translated literally would indicate either that
expected tasks exceeded intellectual capacity or in concrete
terms that a flood or natural disaster had put things on top
of the individual. It was therefore altered to: “Do you feel
that you have too many tasks to manage?” Similarly, to
avoid misinterpretation, item 10 “I have had thoughts of
harming myself” was altered to “Have you had thoughts
that you do not want to live any more, and if so, how
often?”.
The pilot-testing of the Vietnamese versions was
undertaken with a group of 20 women who were either
pregnant or had recently given birth at the study sites by Dr
Buoi La a senior Vietnamese psychiatrist and research
officers. During the pilot testing, the researchers inter-
viewed participants using the scales and discussed their
understandings of the questions. The results of pilot testing
were used to reword and amend the scales before use in the
validation studies.
2.4. Procedure
In Viet Nam self-report questionnaire completion is
unfamiliar and data were collected in interviews which are
preferred (Fisher et al., 2004), thus all questionnaires were
administered as individual structured interviews. These
were conducted at commune health stations or, infrequent-
ly, at women's homes. SCIDs were administered by a senior
284 T.D. Tran et al. / Journal of Affective Disorders 133 (2011) 281–293
Vietnamese psychiatrist and the study-specific interviews
by Vietnamese health research workers. The two interviews
were conducted on the same day and both the psychiatrist
and the research workers were blinded to the data
generated in each other's interviews. Data collection was
carried out in Ha Nam in November 2006 and in Hanoi in
February and March 2007.
2.5. Statistical analysis
Sociodemographic characteristics were summarised by
descriptive statistics. Mean scores of EPDS, Zung SAS and
GHQ12 were calculated by diagnostic groups and com-
pared by post hoc analyses with Bonferroni multiple-
comparison tests. To select the optimal cut-offs, Se, Sp
and correctly classified proportions for different cut-off
points were computed. We performed the comparisons of
the three scales in terms of reliability (using Cronbach's
alpha coefficient), overall performance (the Area Under
the ROC Curves (AUROC)), the performance at the optimal
thresholds (Se, Sp, positive predictive value, negative
predictive value, likelihood ratio and kappa). Cohen's
Kappa (often simply called Kappa) is a measure of
agreement between two binary variables which are
attempts to measure the same thing (Cohen, 1960). That
statistic is commonly used to measure the inter-rater
reliability of a scale (Cook and Beckman, 2006). However,
it can also be used to measure validity (the agreement
between a scale's results and gold standard results). In our
paper, we calculated kappa coefficients between the SCID
(mental disorders or not) and each of the three scales
(above or below suggested cutoff point). Data were
analysed in STATA version 9.
2.6. Ethics
The study was approved by the University of Melbourne's
Human Research Ethics Committee and the Viet Nam Medical
Association's Scientific Committee.
3. Results
3.1. Description of the sample
A sample of 364/392 (93%) eligible women was recruited
systematically from randomly selected health services and
Table 2
Mean, standard deviation and range of EPDS, Zung SAS and GHQ12 scores by diagnostic groups.
EPDS
Mean(SD) Range
Controls (n = 255) 2.3 (2.7) 0–11
ANX (n = 43) 3.9 (3.5) 0–14
DEP (n = 53) 7.6 (4.3) 0–21
Anx-Dep (n = 13) 9.2 (6.0) 0–17
Post hoc comparisons* Anx-Dep; DEP N ANX N controls
DEP: Pure depression; ANX = Pure Anxiety; Anx-Dep: Co-morbid Anxiety Disorder and Depression.
*All post hoc comparisons are significant at p b 0.05.
provided data. Of these 199 (54.7%) were pregnant and the
remainder were mothers of newborns; 36% (130/364) were
living in urban areas (in Hanoi) and almost all participants
(360/364, 99%) were married. They were on average
27 years old (SD = 5.3 years); 10.7% of participants had not
completed primary school and 76% (277/364) generated
income through agricultural labour or manual work. There
were no differences in sociodemographic characteristics
between pregnant women and mothers of infants (Fisher
et al., 2010).
Following the standard SCID interview, 109 participants
(30%) were diagnosed as having current symptoms meeting
criteria for a CMD. Specifically, depression was the most fre-
quent diagnosis (53/364, 14.6%), followed by anxiety disor-
ders (43/364, 11.8%). Co-morbid depression and an anxiety
disorder were diagnosed in 13 participants (3.6%) (Fisher
et al., 2010).
The means, SD and ranges of scores, and post hoc
comparisons for the three scales by diagnostic groups are
presented in Table 2. Post hoc analyses showed the mean
EPDS scores of the pure depression and co-morbid anxiety
disorder and depression groups were statistically signifi-
cantly higher than that of the pure anxiety group (p b 0.01).
GHQ12 scores were similar to those generated from the
EPDS, however the mean score of the pure anxiety group
was not different to that of the no-diagnosis group. The
analyses for the Zung SAS revealed differences between the
mean score of co-morbid anxiety disorder and depression
group and those of pure depression and pure anxiety groups
(p b 0.01).
3.2. Optimal cut-off points of the three scales in pregnant and
postpartum women
Table 3 shows the sensitivity (probability of a score
detecting the test outcome in a person with a CMD),
specificity values (probability of a negative test outcome
in a non-CMD individual), and correct-classification rates
(predicted probabilities) in pregnant women, postpartum
women and both (perinatal) for different cut-off points of
the tests. For each scale, the optimal threshold was chosen
as the score that gives the optimal Se and Sp and maximises
the correct-classification rate. Although the specific values
of Se and Sp were slightly different between pregnant and
postpartum women, the optimal cut-off points were not
different for each of the scales. As similar results were found
Zung SAS GHQ12
Mean(SD) Range Mean(SD) Range
33.5 (5.5) 25–55 0.7 (1.1) 0–7
38.5 (6.7) 29–53 1.2 (1.3) 0–4
41.8 (7.4) 29–63 2.4 (1.9) 0–9
47.7 (8.7) 33–63 2.8 (2.6) 0–9
Anx-Dep N DEP; ANX N controls Anx-Dep; DEP N ANX, controls
 T.D. Tran et al. / Journal of Affective Disorders 133 (2011) 281–293 285

Table 3
Sensitivity, specificity and correct classification of the EPDS, Zung SAS and GHQ12 (%).

Pregnant women Postpartum women Perinatal women

Se Sp Correctly classified Se Sp Correctly classified Se Sp Correctly classified

EPDS cut-off
2/3 67.2 63.8 64.8 78.4 64.9 69.1 72.5 64.3 66.8
3/4 65.5 71.6 69.7 74.5 74.6 74.6 69.7 72.9 72.0
4/5 58.6 78.7 72.7 68.6 84.1 79.4 63.3 81.2 75.8
5/6 53.5 85.1 75.9 47.1 87.7 75.2 50.5 86.3 75.6
Zung cut-off
35/36 77.6 59.6 64.8 70.6 73.7 72.7 74.3 65.9 68.4
37/38 72.4 71.6 71.9 62.8 79.8 74.6 67.9 75.3 73.1
38/39 69.0 78.0 75.4 54.9 86.0 76.4 62.4 81.6 75.8
39/40 65.5 82.3 77.4 47.1 87.7 75.2 56.9 84.7 76.4
GHQ12 cut-off
0/1 81.0 58.2 64.8 72.6 54.4 60.0 77.1 56.5 62.6
1/2 55.2 85.1 76.4 43.1 86.8 73.3 49.5 85.9 75.0
2/3 34.5 92.2 75.4 29.4 95.6 75.2 32.1 93.7 75.3

in the AUROCs of the three scales in pregnant women and
mothers of infants, we suggest using the same cut-off points
for both groups (see Table 4). The optimal cut-off point for
the EPDS was 3/4, the Zung SAS was 37/38 and the GHQ12
was 0/1.
than that of GHQ12. Similar differences were observed among
Se and kappa values at the optimal cut-off scores. GHQ12 at cut-
off 0/1 detected half of CMD women, whereas EPDS at cut-off 3/
4 and Zung SAS at cut-off 37/38 detected more than two thirds
of women with CMD.

3.3. Comparisons of the performance of the scales between 4. Discussion

urban and rural
AUROC values and 95% CI of EPDS, Zung SAS and GHQ12 in
urban and rural groups are shown in Table 4. In this popula-
tion there were no statistically significant differences between
urban and rural populations.
3.4. Comparisons of the validation of the three scales
The performances of the tests are summarised in Fig. 1
by the ROC curves (true-positive rate (Se) is plotted against
the false–positive rate (1-Sp) for various cut-offs of the
test) and in Table 5 by validation statistics. Overall, there
were no statistically significant differences among the Area
Under the ROC Curves (AUCs) of EPDS, Zung SAS and
GHQ12. The three AUCs were in the range of moderate
accuracy (Greiner et al., 2000) (0.77, 0.79 and 0.72 for
EPDS, Zung SAS and GHQ12, respectively). The internal
reliability values of EPDS and Zung SAS (measured by
Cronbach's α), however, were statistically significantly higher
The aim of this study was to validate and compare three
scales: EPDS, Zung SAS, and GHQ12, as screening tools for
detection of common non-psychotic perinatal mental disor-
ders in women in northern Vietnam. The study had a number
of strengths. The sample was systematically recruited from
randomly selected urban and rural health services and there
was a high recruitment fraction. We used a gold-standard
design in which responses to culturally validated translations
of the three scales were compared to an independently
administered psychiatric diagnostic interview. We acknowl-
edge that as we validated three instruments at one time there
was potential for women to experience fatigue as some items
on each of the measures are similar and this might have led to
lower levels of endorsement of symptoms on these scales. In
order to minimise this effect, we rotated the order of admin-
istration. We believe that these findings can be generalised
with confidence and that the validated versions of these
scales are appropriate for use with women in the North of
Vietnam.

Table 4
Area Under the ROC Curve in urban and rural, and pregnant and postpartum women for the three scales.

EPDS Zung SAS GHQ12

Type of participant
Pregnant (N = 199) 0.75 (95%CI:0.67–0.83) 0.81 (95%CI:0.75–0.88) 0.76 (95%CI:0.69–0.83)
1Postpartum (N = 165) 0.79 (95%CI:0.71–0.87) 0.78 (95%CI:0.70–0.85) 0.69 (95%CI:0.60–0.78)
Setting
Rural (N = 234) 0.79 (95%CI:0.73–0.86) 0.83 (95%CI:0.78–0.89) 0.72 (95%CI:0.66–0.79)
Urban (N = 130) 0.72 (95%CI:0.61–0.82) 0.71 (95%CI:0.60–0.82) 0.74 (95%CI:0.63–0.84)
286 T.D. Tran et al. / Journal of Affective Disorders 133 (2011) 281–293

1.00 cut-off point of 9/10 was the most accurate for detecting
combined major and minor depression (Gibson et al., 2009).
However, in assessments in East Asian countries, a range of
0.75

cut-off scores for detecting major and minor postpartum

Sensitivity

depression has been found including from 6/7 in Thailand

0.50

to 12/13 in Taiwan (Table 1). This provides further support

to the conclusions of Gibson et al. (2009) that responses to
this questionnaire are governed to some extent by cultural
0.25

factors.
The Zung SAS is widely used in Vietnam, however to date
0.00

there has been no validation study of the scale published in

0.00 0.25 0.50
1-Specificity
Zung_SAS ROC area: 0.7925 EPDS ROC area: 0.7709
GHQ12 ROC area: 0.7296 Reference
Fig. 1. The Receiver Operating Characteristic curves for EPDS, Zung SAS and
GHQ12 in screening for CMD in perinatal women.
It is common in validation studies to identify a cut-off
point for screening purposes in which Se is higher than Sp
in order to detect more potential cases (Warner, 2004). In
this study we sought a cut-off point which maximised both
Se and Sp by minimising the value of the square of (Se–Sp).
At our suggested cut-off points Se is lower then Sp. A
higher Se means that more people are identified as having
disease, but this involves a higher false positive rate and
imposes a higher burden on the health care system,
including wasted resources. Our suggested threshold
balances the false positive and false negative rates to
provide an optimal overall rate of correct classification.
This contributes to more effective use of health care
resources which is essential in resource-constrained
settings and increases acceptability to health policy makers
including in Viet Nam.
We found that a cut-off score of 3/4 on the Edinburgh
Postnatal Depression Scale yielded optimal Se (69.7%) and Sp
(72.9%) in this setting. Gibson reviewed thirty-seven studies
validating the EPDS to detect minor and major depression in
prenatal and post partum periods in both English speaking
and non English speaking populations. They concluded that a
0.75 1.00 this country or any other developing country. Zung (1980)
suggested a cut-off 44/45 to detect clinically significant
anxiety and he also concluded that this cut-off was the
optimal one to detect co-morbid anxiety and depression in
adults (Zung et al., 1990). In our study a much lower cut-off
point (37/38) was established to detect CMD (anxiety,
depression, and co-morbid anxiety and depression) in
women during pregnancy or after childbirth.
The GHQ12 has been validated and confirmed as a useful
screening tool to detect CMD in general population in many
settings (Goldberg et al., 1997). However, there has been only
one validation study of GHQ12 for screening psychiatric
morbidity in prenatal and postnatal periods (Navarro et al.,
2007). That study conducted in Spain suggested a cut-off
point 4/5 for detecting anxiety and depression in postpartum
women with 80.6% Se and 80.4% Sp. In our study, the GHQ12
yielded a low Sp at the cut-off point 0/1 (56.5%) and
unacceptably low Se with a 1/2 cut-off (49.5%) for screening
PCMDs in women.
The items in the EPDS and GHQ12 are designed to
detect psychiatric symptoms and exclude the somatic
symptoms of mental disorders like appetite and sleep
disturbance. Vietnamese people tend to report somatic
symptoms more openly than psychiatric symptoms (Kinzie
et al., 1982). Viet Nam has a collective culture in which the
functioning of the family or social group is a focus rather
than the needs, hopes, wishes and experiences of in-
dividuals. The expression of emotion either positive or
negative is generally socially proscribed and it is possible
that people in this setting do not have a wide emotional
vocabulary and therefore the language to describe separate
emotional states or differences in these states over time.
This might make questions about related, but separate

Table 5
Validation statistics of the three scales.

EPDS Zung SAS GHQ12

Area Under the ROC curve 0.77 (95%CI:0.72–0.82) 0.79 (95%CI:0.74–0.84) 0.72 (95%CI:0.67–0.78)
Internal reliability* 0.75 (95%CI:0.71–0.78) 0.76 (95%CI:0.72–0.80) 0.64 (95%CI:0.58–0.69)
Optimal cut-off score 3/4 37/38 0/1
Sensitivity 69.7% (95%CI:60.1–78.0) 67.9% (95%CI:58.1–76.3) 77.1% (95%CI:67.8–84.3)
Specificity 72.9% (95%CI:66.9–78.2) 75.3% (95%CI:69.4–80.4) 56.6% (95%CI:50.1–62.6)
Positive predictive value 69.7% 54.0% 43%
Negative predictive value 72.9% 84.6% 85%
Positive likelihood ratio 2.6 2.7 1.8
Kappa 0.39 0.4 0.27

*Cronbach's α.
 T.D. Tran et al. / Journal of Affective Disorders 133 (2011) 281–293
emotions (fear, anxiety, and sadness) confusing and
difficult to respond to.
Neither the EPDS nor the GHQ12 are sensitive to
detecting chronic suffering or emotional distress. The
EPDS asks respondents to assess whether their emotional
experiences over the past seven days are different to a usual
state e.g. ‘As much as I always could’ or ‘Definitely less than I
used to’. Goldberg suggests that this form of questioning is
insensitive to people who are living in the context of
poverty and sustained adversity and might be experiencing
chronic low mood that does not vary from week to week
(Goldberg et al., 1998). The GHQ12 questions ask whether
the respondent has experienced a particular symptom or
behaviour recently and each item is rated on a four-point
scale: less than usual, no more than usual, rather more than
usual, or much more than usual. Again in the context of
chronic socioeconomic adversity asking people about
change from a usual state is unlikely to detect symptoms
that are experienced most of the time. Therefore, the rate of
chronic mental disorders could influence the cut-off points
and overall performance of the instruments in these
populations.
The comparison of the three tools in this study revealed
that the Zung SAS was slightly more powerful than the EPDS,
whereas GHQ12 was much less powerful than the other two
tools according to the area Under the ROC curve and internal
reliability statistics. Zung SAS scores were able to distin-
guish different disorders and scores were higher for more
severe symptoms. Zung SAS may be better than GHQ12 in
detecting non-psychotic psychiatric morbidity in the Viet-
namese context because it contains both psychiatric and
somatic symptoms. The WHO's SRQ 20 which also contains
both psychiatric and somatic symptoms has already been
established as a suitable and powerful screening tool in a
validation study in Vietnamese women (Tuan et al., 2004).
The EPDS excluded somatic symptoms because it was
specifically designed for screening depression in postpar-
tum women in whom weight loss and sleep disturbance are
normal. We found that despite differences in cut off scores
and lower Se and Sp than has been reported in international
studies, the EPDS is still a powerful tool for screening CMD in
perinatal women in Vietnam. All three tools detected co-
morbid anxiety disorder and depression, and pure depres-
sion better than detecting pure anxiety. Among them, not
surprisingly, the Zung SAS was the most powerful in
screening for pure anxiety, as it was designed for this
purpose. It was surprising thought that the Zung SAS, was
better in detecting depression than detecting anxiety,
perhaps because it does not ask about change from usual
state, but just ascertains symptoms that are being experi-
enced currently.
There has only been recent recognition in Viet Nam that
women who are pregnant or who have recently given birth
287
experience CMD. However, like most resource-constrained
countries mental health services have focused predominantly
on the needs of people with severe psychotic illnesses. There
are few services for people experiencing CMD in particular in
the rural areas where most people live. None of the
participants in this study had ever received mental health
care (Fisher et al., 2010). Viet Nam has a very well developed
primary health care system with health workers available in
commune health stations and established systems of referral
to district and provincial hospitals. The availability of
screening instruments to identify women with heightened
psychological needs during the perinatal period is the first
step in the development of an integrated primary health care
approach to addressing perinatal mental health problems
informed by the strategies used in other low income settings
(Rahman et al., 2008).
The high prevalence of CMD in Vietnamese perinatal
women confirms the need to identify appropriate screening
instruments for the Vietnamese context. Based on this
validation study, we propose Zung SAS and EPDS for use as
screening tools at the primary health care level to detect
PCMD in women. At present we can only speculate about the
reasons why there were low cut-off scores to detect clinically
significant symptoms compared with others and those that
need to be confirmed and elaborated by further quantitative
and qualitative research. Our data suggest that there might
be value in developing a Vietnamese-specific scale to detect
PCMDs which is derived from the EPDS and Zung SAS and
incorporates culturally specific somatic symptoms of emo-
tional distress.
Role of funding source
Funding for this study was provided by Myer Foundation under its
Beyond Australia; Myer Foundation had no further role in study design; in
the collection, analysis and interpretation of data; in the writing of the
report; and in the decision to submit the paper for publication.
Conflicts of interest
The authors declare that they have no conflicts of interest.
Acknowledgments
The authors are grateful to the Myer Foundation who
funded the study under its Beyond Australia scheme. They are
also very grateful to Professor Margot Prior who made
valuable contributions to the study design; and the research
staff from the Research and Training Centre for Community
Development in Hanoi who collected the data, the commune
health workers in Hanoi and Ha Nam who assisted with
recruitment and the participants who contributed their time
and experiences.
288 T.D. Tran et al. / Journal of Affective Disorders 133 (2011) 281–293

Appendix 1. EPDS Vietnamese version

 T.D. Tran et al. / Journal of Affective Disorders 133 (2011) 281–293 289

8 Chị có cảm giác buồn hay khổ sở không?

+ Có, hầu hết mọi lúc [ ] 1

+ Có, khá thường xuyên [ ] 2

+ Hiếm khi [ ] 3

+ Không, không bao giờ [ ] 4

9 Chị có cảm giác buồn rầu đến mức phải khóc không?

+ Có, hầu hết mọi lúc [ ] 1

+ Có, khá thường xuyên [ ] 2

+ Chỉ thỉnh thoảng [ ] 3

+ Không, không bao giờ [ ] 4

10 Chị có cảm nghĩ không muốn sống nữa không?

+ Có, khá thường xuyên [ ] 1

+ Thỉnh thoảng [ ] 2

+ Hiếm khi [ ] 3

+ Không bao giờ [ ] 4

290 T.D. Tran et al. / Journal of Affective Disorders 133 (2011) 281–293

Appendix 2. GHQ12 Vietnamese version

T.D. Tran et al. / Journal of Affective Disorders 133 (2011) 281–293 291
292 T.D. Tran et al. / Journal of Affective Disorders 133 (2011) 281–293
Appendix 3. Zung SAS Vietnamese version
References
Adewuya, A.O., Ola, B.A., Dada, A.O., Fasoto, O.O., 2006. Validation of the
Edinburgh Postnatal Depression Scale as a screening tool for depression
in late pregnancy among Nigerian women. Journal of Psychosomatic
Obstetrics and Gynaecology 27 (4), 267–272.
Araya, R., Wynn, R., Lewis, G., 1992. Comparison of two self administered
psychiatric questionnaires (GHQ-12 and SRQ-20) in primary care in
Chile. Social Psychiatry and Psychiatric Epidemiology 27 (4), 168–173.
Chandran, M., Tharyan, P., Muliyil, J., Abraham, S., 2002. Post-partum
depression in a cohort of women from a rural area of Tamil Nadu
India. The British Journal of Psychiatry 181, 499–504.
Cohen, J., 1960. A coefficient of agreement for nominal scales. Educational
and Psychological Measurement XX (1), 37–46.
Cook, D.A., Beckman, T.J., 2006. Current concepts in validity and reliability for
psychometric instruments: theory and application. The American
Journal of Medicine 119 (2), 166 e7-16.
Cooper, P., Tomlinson, M., Swartz, L., Woolgar, M., Murray, L., Molteno, C., 1999.
Postpartum depression and the mother infant relationship in a South African
peri-urban settlement. The British Journal of Psychiatry 175, 554–558.
Cox, J., Holden, J., Sagovsky, R., 1987a. Detection of postnatal depression.
Development of the 10-item Edinburgh Postnatal Depression Scale. The
British Journal of Psychiatry 150, 782–786.
Cox, J.L., Holden, J.M., Sagovsky, R., 1987b. Detection of postnatal depression.
Development of the 10-item Edinburgh Postnatal Depression Scale. The
British Journal of Psychiatry 150, 782–786.
Ekuklu, G., Tokuc, B., Eskiocak, M., Berberoglu, U., Saltik, A., 2004. Prevalence
of postpartum depression in Edirne, Turkey, and related factors. The
Journal of Reproductive Medicine 49 (11), 908–914.
Engle, P., 2009. Maternal mental health: program and policy implications.
The American Journal of Clinical Nutrition 89, 963S–966S.
First, M., Gibbon, M., Spitzer, R., Williams, J., 1996. User's Guide for the
Structured Interview for DSM-IV Axis 1 Disorders—Research Version.
Biometrics Research, New York.
Fisher, J., Morrow, M., Ngoc, N., Anh, L., 2004. Prevalence nature, and
correlates of postpartum depression in Vietnam. BJOG An International
Journal of Obstetrics and Gynaecology 111, 1353–1360.
Fisher, J., Cabral de Mello, M., Isutzu, T., 2009. Pregnancy, childbirth and the
postpartum year. In: Fisher, J., Astbury, J., Cabral de Mello, M., Saxena, S.
(Eds.), Mental Health Aspects of Women's Reproductive Health. A Global
Review of the Literature. World Health Organization and United Nations
Population Fund, Geneva, pp. 8–43.
Fisher, J., Tran, T., La, B.T., Kriitmaa, K., Rosenthal, D., Tuan, T., 2010. Common
perinatal mental disorders in northern Viet Nam: community prevalence
and health care use. Bulletin of the World Health Organization 88 (10),
737–745.
 T.D. Tran et al. / Journal of Affective Disorders 133 (2011) 281–293
Gausia, K., Fisher, C., Algin, S., Oosthuizen, J., 2007. Validation of the Bangla
version of the Edinburgh Postnatal Depression Scale for a Bangladeshi
sample. Journal of Reproductive and Infant Psychology 25 (4), 308–315.
Gibson, J., McKenzie-McHarg, K., Shakespeare, J., Price, J., Gray, R., 2009. A
systematic review of studies validating the Edinburgh Postnatal
Depression Scale in antepartum and postpartum women. Acta Psychia-
trica Scandinavica 119 (5), 350–364.
Goldberg, D.P., 1972. The detection of psychiatric illness by questionnaire.
Maudsley Monograph No. 21. Oxford University Press, Oxford.
Goldberg, D., Huxley, P., 1992. Common Mental Disorders: A Biosocial Model.
Tavistock/Routledge, London.
Goldberg, D., Williams, P., 1988. A User's Guide to the General Health
Questionnaire. NFER-Nelson Publishing Company Ltd, Oxford.
Goldberg, D., Gater, R., Sartorius, N., Ustun, T., Piccinelli, M., Gureje, O., Rutter,
C., 1997. The validity of two versions of the GHQ in the WHO study of
mental illness in general health care. Psychological Medicine 27,
191–197.
Goldberg, D., Oldehinkel, T., Ormel, J., 1998. Why GHQ threshold varies from
one place to another. Psychological Medicine 28, 915–921.
Gorman, L., O'Hara, M., Figueirdo, B., Hayes, S., Jacquemain, F., Kammerer, M.,
Klier, C., Rosi, S., Seneviratne, G., Sutter-Dallay, A.-L., TCS-PND Group,
2004. Adaptation of the Structured Clinical Interview for DSM-IV
Disorders for assessing depression in womrn during pregnancy and
post-partum across countries and cultures. The British Journal of
Psychiatry 184 (s17), s17–s23.
Greiner, M., Pfeiffer, D., Smith, R.D., 2000. Principles and practical application
of the receiver-operating characteristic analysis for diagnostic tests.
Preventive Veterinary Medicine 45 (1–2), 23–41.
GSO, 2009. Statistical Yearbook of Vietnam. General Statistics Office of
Vietnam, Hanoi.
Hanlon, C., Medhin, G., Alem, A., Araya, M., Abdulahi, A., Hughes, M., Tesfaye,
M., Wondimagegn, D., Patel, V., Prince, M., 2008. Detecting perinatal
common mental disorders in Ethiopia: validation of the self-reporting
questionnaire and Edinburgh Postnatal Depression Scale. Journal of
Affective Disorders 108 (3), 251–262.
Harpham, T., Huttly, S., De Silva, M., Abramsky, T., 2005. Maternal mental
health and child nutritional status infour developing countries. Journal of
Epidemiology and Community Health 59, 1060–1064.
Hendrick, V., 1998. Evaluation of mental health and depression during
pregnancy. Psychopharmacology Bulletin 34 (3), 297–299.
Iranfar, S., Shakeri, J., Ranjbar, M., NazhadJafar, P., Razaei, M., 2005. Is
unintended pregnancy a risk factor for depression in Iranian women?
Eastern Mediterranean Health Journal 11 (4), 618–624.
Kinzie, J.D., Manson, S.M., Vinh, D.T., Tolan, N.T., Anh, B., Pho, T.N., 1982.
Development and validation of a Vietnamese-language depression rating
scale. The American Journal of Psychiatry 139 (10), 1276–1281.
Kleinman, A., 1987. Anthropology and psychiatry. The role of culture in cross-
cultural research on illness. The British Journal of Psychiatry 151, 447–454.
Laungani, P., 2000. Postnatal depression across cultures:conceptual and
methodological considerations. International Journal of Health Promo-
tion and Education 38 (3), 86–94.
Matthey, S., Henshaw, C., Elliott, S., Barnett, B., 2006. Variability in use of cut-
off scores and formats on the Edinburgh Postnatal Depression Scale:
implications for clinical and research practice. Archives of Women's
Mental Health 9 (6), 309–315.
Navarro, P., Ascaso, C., Garcia-Esteve, L., Aguado, J., Torres, A., Martin-Santos,
R., 2007. Postnatal psychiatric morbidity: a validation study of the GHQ-
12 and the EPDS as screening tools. General Hospital Psychiatry 29 (1),
1–7.
O'Hara, M.W., Swain, A.M., 1996. Rates and risk of postpartum depression: a
meta-analysis. Int Rev Psychiatry 8, 37–54.
Patel, V., Desouza, N., Rodrigues, M., 2003. Postnatal depression and infant
growth and development in low income countries: a cohort study from
Goa, India. Archives of Disease in Childhood 88 (1), 34–37.
293
Patel, V., Araya, R., Chowdhary, N., King, M., Kirkwood, B., Nayak, S., Simon, G.,
Weiss, H.A., 2008. Detecting common mental disorders in primary care
in India: a comparison of five screening questionnaires. Psychological
Medicine 38 (2), 221–228.
Pitanupong, J., Liabsuetrakul, T., Vittayanont, A., 2007. Validation of the Thai
Edinburgh Postnatal Depression Scale for screening postpartum depres-
sion. Psychiatry Research 149 (1–3), 253–259.
Rahman, A., Iqbal, Z., Harrington, R., 2003a. Life events, social support and
depression in childbirth: perspectives from a rural community in the
developing world. Psychological Medicine 33, 1161–1167.
Rahman, A., Lovel, H., Bunn, J., Iqbal, Z., Harrington, R., 2003b. Mothers'
mental health and infant growth: a case-control study from Rawalpindi,
Pakistan. Child: Care, Health and Development 30, 21–27.
Rahman, A., Bunn, J., Lovel, H., Harrington, R., 2004. Impact of maternal
depression on infant nutritional status and illness: a cohort study.
Archives of General Psychiatry 61, 946–952.
Rahman, A., Bunn, J., Lovel, H., Creed, F., 2007. Depression and low birthweight
in a developing country. Acta Psychiatrica Scandinavica 115, 481–486.
Rahman, A., Malik, A., Sikander, S., Roberts, C., Creed, F., 2008. Cognitive
behaviour therapy-based intervention by community health workers for
mothers with depression and their infants in rural Pakistan: a cluster-
randomized controlled trial. Lancet 372, 902–909.
Regmi, S., Sligi, W., Carter, D., Grut, W., Seear, M., 2002. A controlled study of
postpartum depression among Nepalese women: validation of the
Edinburgh Postpartum Depression Scale in Kathmandu. Troipcal Med-
icine and International Health 7 (4), 378–382.
Ross, L.E., Sellers, E.M., Gilbert Evans, S.E., Romach, M.K., 2004. Mood changes
during pregnancy and the postpartum period: development of a
biopsychosocial model. Acta Psychiatrica Scandinavica 109 (6), 457–466.
Rushidi, W.M.W.M., Awang, A., Mohamed, M.N., 2003. Revalidation of the
Malay version of the Edinburgh Postnatal Depression Scale (EPDS)
among Malay Postpartum women attending the Bakar Bata health center
in Alor Setar, Kedah, North West of Peninsular Malaysia. Malaysian
Journal of Medical Sciences 10 (2), 71–75.
Stewart, R., 2007. Maternal depression and infant growth—a review of recent
evidence. Maternal & Child Nutrition 3, 94–107.
Teng, H.W., Hsu, C.S., Shih, S.M., Lu, M.L., Pan, J.J., Shen, W.W., 2005. Screening
postpartum depression with the Taiwanese version of the Edinburgh
Postnatal Depression scale. Comprehensive Psychiatry 46 (4), 261–265.
Tesfaye, M., Hanlon, C., Wondimagegn, D., Alem, A., 2010. Detecting postnatal
common mental disorders in Addis Ababa, Ethiopia: validation of the
Edinburgh Postnatal Depression Scale and Kessler Scales. Journal of
Affective Disorders 122 (1–2), 102–108.
Tuan, T., Harpham, T., Huonh, N., 2004. Validity and reliability of the self-
reporting questionnnaire 20 items in Vietnam. Hong Kong Journal 14 (3),
7–10.
Uwakwe, R., Okonkwo, J.E., 2003. Affective (depressive) morbidity in
puerperal Nigerian women: validation of the Edinburgh Postnatal
Depression Scale. Acta Psychiatrica Scandinavica 107 (4), 251–259.
Warner, J., 2004. Clinicians' guide to evaluating diagnostic and screening
tests in psychiatry. Advances in Psychiatric Treatment 10, 446–454.
Weobong, B., Akpalu, B., Doku, V., Owusu-Agyei, S., Hurt, L., Kirkwood, B.,
Prince, M., 2009. The comparative validity of screening scales for
postnatal common mental disorder in Kintampo, Ghana. Journal of
Affective Disorders 113 (1–2), 109–117.
World Health Organization, 2001. World Health Report 2001. Mental Health:
New Understanding, New Hope. World Health Organization, Geneva.
Zung, W., 1971. A rating instrument for anxiety disorders. Psychosomatics 12
(6), 371–379.
Zung, W., 1980. How normal is anxiety?. vol. s The Upjohn Company,
Kalamazoo.
Zung, W.W., Magruder-Habib, K., Velez, R., Alling, W., 1990. The comorbidity
of anxiety and depression in general medical patients: a longitudinal
study. The Journal of Clinical Psychiatry 51, 77–80 Suppl, discussion 1.