A systemic review of studies validating the Edinburgh Postnatal Depression Scale in antepartum and postpartum women

ArticleinActa Psychiatrica Scandinavica 119(5):350-64 · June 2009with678 Reads
Impact Factor: 5.61 · DOI: 10.1111/j.1600-0447.2009.01363.x · Source: PubMed
Abstract

The Edinburgh Postnatal Depression Scale (EPDS) is the most widely used screening tool for postpartum depression (PPD). We systematically reviewed the published evidence on its validity in detecting PPD and antepartum depression (APD) up to July 2008. Systematic review of validation studies of the EPDS included 1987-2008. Cut-off points of 9/10 for possible PPD, 12/13 for probable PPD and 14/15 for APD were used. Thirty-seven studies met the inclusion criteria. Sensitivity and specificity of cut-off points showed marked heterogeneity between different studies. Sensitivity results ranged from 34 to 100% and specificity from 44 to 100%. Positive likelihood ratios ranged from 1.61 to 78. Heterogeneity among study findings may be due to differences in study methodology, language and diagnostic interview/criteria used. Therefore, the results of different studies may not be directly comparable and the EPDS may not be an equally valid screening tool across all settings and contexts.

Full-text

Available from: Kirstie McKenzie-McHarg, Jan 10, 2014
Review
A systematic review of studies validating the
Edinburgh Postnatal Depression Scale in
antepartum and postpartum women
Introduction
Postpartum depression (PPD) is a non-psychotic
depressive episode of mild to moderate severity,
beginning in or extending into the first postnatal
year (1). A meta-analysis of 59 studies
(n = 12 810) found an average prevalence rate of
non-psychotic PPD of 13% (95% CI 12.3–13.4)
(2). PPD is an important public health problem,
having a significant impact on the mother, the
family (3), her partner (4), mother–baby interac-
tion (5) and on the long-term emotional (6) and
cognitive development of the baby (7).
Antepartum depression (APD) is defined as a
non-psychotic depressive episode of mild to mod-
erate severity, beginning in or extending into
pregnancy. APD is less well documented than
PPD; however, it is at least as common as PPD (8)
Gibson J, McKenzie-McHarg K, Shakespeare J, Price J, Gray R. A
systematic review of studies validating the Edinburgh Postnatal
Depression Scale in antepartum and postpartum women.
Objective: The Edinburgh Postnatal Depression Scale (EPDS) is the
most widely used screening tool for postpartum depression (PPD). We
systematically reviewed the published evidence on its validity in
detecting PPD and antepartum depression (APD) up to July 2008.
Method: Systematic review of validation studies of the EPDS included
1987–2008. Cut-off points of 9 10 for possible PPD, 12 13 for
probable PPD and 14 15 f or APD we re used.
Results: Thirty-seven studies met the inclusion criteria. Sensitivity and
specificity of cut-off points showed marked heterogeneity between
different studies. Sensitivity resultsrangedfrom34to100%and
specificity from 44 to 100%. Positive likelihood ratios ranged from 1.61
to 78.
Conclusion: Heterogeneity among study findings may be due to
differences in study methodology, language and diagnostic
interview criteria used. Therefore, the results of different studies may
not be directly comparable and the EPDS may not be an equally valid
screening tool across all settings and contexts.
J. Gibson
1
, K. McKenzie-McHarg
1
,
J. Shakespeare
2
, J. Price
3
,
R. Gray
1
1
National Perinatal Epidemiology Unit, University of
Oxford,
2
Summertown Health Centre and
3
Department
of Psychiatry, University of Oxford, Oxford, UK
Key words: postpartum depression; ante partum
depres sion; Edinburgh Post natal Depression Scale;
systematic re view; validation
Dr Jessica Gibson, National Perinatal Epidemiology
Unit, University of Oxford, Old Road Campus, Oxford
OX3 7LF, UK.
E-mail: Jessica.gibson@obmh.nhs.uk
Accepted for publication January 23, 2009
Summations
There is a wide variation across different settings in sensitivity, specificity, positive and negative
likelihood ratios in validation studies of the Edinburgh Postnatal Depression Scale (EPDS).
Differences in methodology of the included studies are likely to account for the variation in results.
There are only three studies evaluating the EPDS for use in antenatal populations.
Considerations
Because of considerable differences in the methodology of included studies, quantitative approaches
using meta-analysis were not performed.
Acta Psychiatr Scand 2009: 119: 350–364
All rights reserved
DOI: 10.1111/j.1600-0447.2009.01363.x
Copyright 2009 The Authors
Journal Compilation 2009 Blackwell Munksgaard
ACTA PSYCHIATRICA
SCANDINAVICA
350
Page 1
and for some, but by no means all women, it may
continue into the postnatal period (9).
The Edinburgh Postnatal Depression Scale
(EPDS) is the most widely used screening question-
naire for PPD (10). It is a 10-item self-report
questionnaire in which women are asked to rate
how they have felt in the previous 7 days. Each
question is scored 0–3 (resulting range 0–30) and
completion takes around 5 min. In many areas of
the UK, the EPDS is routinely administered to
women at 6–8 weeks postnatally by their health
visitor. A cut-off for Ôprobable depressionÕ has been
suggested at 12 13, and for Ôpossible depressionÕ at
9 10 (11). However, the EPDS is only a screening
instrument and a subsequent clinical diagnosis must
be made by an appropriately trained health profes-
sional. The cut-off points of 9 10 and 12 13 are also
used as markers of possible minor and major
depression respectively. The EPDS has been trans-
lated into, and validated in, many languages other
than English. The EPDS has also been validated as a
screening tool for APD in pregnant women, with a
cut-off point of 14 15 for Ôprobable depressionÕ (12).
There is as yet no research tool which can serve as a
universally agreed Ôgold standardÕ for the diagnosis
of PPD or APD. Nevertheless, many structured and
semistructured psychiatric interviews have been
used as comparators including the Structured Clin-
ical Interview for DSM disorders (SCID) (13), the
Mini International Neuropsychiatric Interview
(MINI) (14), the Present State Examination (PSE)
(15) and others.
Screening for PPD is currently recommended in
Australia and the USA, but not the UK (16). In the
UK, the policy has shifted towards opportunistic
case finding for PPD. The National Institute for
Health and Clinical Excellence guideline ÔAntena-
tal and postnatal mental healthÕ (17) recommends
the use of the two Whooley questions (18) by all
healthcare professionals at each appointment from
first detection of pregnancy to the postnatal period,
despite a lack of evidence of effectiveness in this
context. These are ÔDuring the past month, have you
often been bothered by feeling down, depressed or
hopeless?Õ and ÔDuring the past month, have you
often been bothered by having little interest or
pleasure in doing things?Õ.
Three other groups have surveyed validation
studies of the EPDS (10, 19, 20). Boyd et al. (10)
conducted a review of screening instruments for
PPD which included a literature review up to
December 2004. This paper focused on comparing
the psychometric data of eight self-report measures
including the EPDS. This was not a systematic
review, and data are quoted for 11 studies as
opposed to 37 studies included in this paper.
Gaynes et al. (19) addressed three key questions in
their paper, including assessment of the accuracy
of various screening tools in detecting APD and
PPD. They identified 23 studies for all screening
tools, of which only eight specifically addressed the
accuracy of the EPDS. The results for the English-
speaking studies only were analysed in detail.
Eberhard-Gran et al. (20) systematically reviewed
validation studies of the EPDS up to 2000. They
concluded that the EPDS had limited utility
because of its high level of heterogeneity. They
only included papers until October 2000. There-
fore, a systematic review of studies comparing the
EPDS with a diagnostic Ôgold standardÕ for APD
and PPD, which includes more recently published
studies, is desirable. This systematic review used a
comprehensive search strategy, which generated a
larger group of relevant studies than the previous
reviews. This area is still relevant and important
because of the differences in approach to screening
in different countries and because of the variation
in administration across different parts of the UK.
Aims of the study
The aim of this study was to determine, by system-
atically reviewing the literature, whether the
Edinburgh Postnatal Depression Scale compares
favourably with a structured clinical interview for
the detection of antepartum depression and post-
partum depression across a variety of settings and in
different languages of administration.
Material and methods
Data search
A computerised literature search was undertaken
using webspirs software in the following databases:
Medline, Embase, Cinahl, PsychInfo and SIGLE.
The time frame of the search was from 1987 (when
the EPDS was published) to July 2008. A broad
search strategy was employed first using the terms
ÔEPDSÕ (or ÔEPDSÕ) and ÔvalidationÕ as search
terms. A focused search strategy including more
precise descriptions such as Ôscreening toolÕ and
Ôlikelihood ratioÕ was used later (Appendix 1). In
addition, the bibliographies of retrieved papers
were inspected by the authors and experts in the
field were contacted for unpublished findings.
Inclusion exclusion criteria
Studies were included: i) if the population of interest
was antepartum or postpartum women (up to
1 year postpartum), ii) if the EPDS was used as a
Systematic review of studies validating the EPDS
351
Page 2
screening instrument, iii) if the EPDS was compared
with a gold standard, defined as a structured or
semistructured clinical interview to diagnose
depression and iv) if data on sensitivity, specificity,
positive likelihood ratio (LR +ve) and negative
likelihood ratio (LR )ve) for relevant cut-off points
could be extracted or sufficient data were available
to enable these parameters to be calculated.
Studies were excluded: i) if the population was
£1 week postpartum (to avoid the period when
symptoms of the baby blues may be confused with
emerging PPD) or ii) if there was a delay of 24 h
between administration of the EPDS and gold
standard (when information was available) to
avoid measurement of changing symptoms over
time or iii) if the postpartum population was
restricted to a subgroup, e.g. to mothers with
learning disability. Any disagreements were
resolved by discussion between the two reviewers
or by a third investigator if required.
A number of studies were reported in more than
one publication. Every effort was made to ensure
that only one paper per study was included in the
review.
Data extraction
All abstracts were read by two reviewers indepen-
dently to identify relevant papers. Papers deemed
relevant or of uncertain relevance, were obtained
and read in full. Data extraction was performed
independently by two researchers for each study
using a standardised data extraction form (see
Appendix 2). A third reviewer was used in the few
cases when the original reviewers could not agree.
Data quality
The quality of the selected studies was assessed
using a grading system based on the York CRD
system (21). Each study was assigned a grade
according to blinding of the assessors, comparison
with a reference standard for diagnosis and the
population spectrum.
Meta-analysis of included studies was considered
inappropriate because of the likely heterogeneity of
the included studies and based on the previous
report by Eberhard Gran et al. (20).
Data analysis
We tabulated extracted data, plotted the results from
studies reporting findings for major and combined
depression on a scatterplot of true-positive rate (sen-
sitivity) against false-positive rate (1 ) specificity),
and calculated relevant summary statistics.
Results
The combined broad and focused searches yielded
274 studies for further inspection. The focused
search strategy yielded a number of papers not
generated by the broad strategy. Of these 274
studies, there were 135 which were not relevant; 20
which did not compare the EPDS to a Ôgold
standardÕ; 18 which did not use a structured semi-
structured interview as a diagnostic reference
standard; 17 which did not use the EPDS; 17
which were of an inappropriate study type; 13
which reported insufficient data for calculations
(this included papers which did not present figures
for at least one of the EPDS cut-off point of
interest, as described below); six which used an
inappropriate population sample, e.g. men
included or breast-feeding mothers only; five
which had a delay between EPDS and interview;
five in which data were repeated from the same
sample as in other papers; and one in which the
mothers were assessed at <1 week postpartum.
Therefore, a total of 37 studies met the inclusion
criteria. A full list of excluded studies with reasons
for exclusion is available from the corresponding
author on request.
Figure 1 shows the flow of
studies through the review.
Edinburgh Postnatal Depression Scale cut-off
points at 9 10 and 12 13 for PPD were used. As
described in the Introduction, these are the widely
accepted cut-off points to indicate ÔpossibleÕ and
ÔprobableÕ depression respectively. For studies in
antepartum populations, a cut-off point of 14 15
was presented. This is the value found to be most
Titles and abstracts screened
(n = 274)
Not relevant (n = 135)
Full articles screened (n = 139)
Excluded studies (n = 102):
No gold standard used (n = 20)
No semi/structured interview (n = 18)
EPDS not used (n = 17)
Inappropriate study type (n = 17)
Insufficient data (n = 13)
Delay between assessments (n = 5)
Inappropriate population (n = 6)
Sample data in >1 paper (n = 5)
Assessment at < 1week (n = 1)
Included studies (n = 37)
Fig. 1. Flow of studies through review process.
Gibson et al.
352
Page 3
predictive of probable depression during preg-
nancy (12). Many papers provided data for multi-
ple cut-off points.
Study sample setting selection
The size of the study samples varied between 23
and 876 (details in attached file to the online
version of this article). Of the 37 studies included,
three validated the EPDS for use in an antepartum
population (12, 23, 24) and the remaining 34 used a
postpartum population. The majority of samples
were drawn from the community (recruited at
antenatal or postnatal clinics). However, some
samples contained a higher proportion of women
who were depressed or women identified as likely
to be depressed (11, 25–29).
Diagnostic criteria
Twenty-two studies (22, 24, 27–46) validated the
EPDS against the Diagnostic and Statistical
Manual of Mental Disorders (47–49) criteria for
depression. Six studies (11–13, 26, 50, 51) used the
Research Diagnostic Criteria (RDC) (52) as the
gold standard. The International Classification of
Diseases (53, 54) was used in six studies (23, 55–
59), and two studies (60, 61) used the PSE (15) as
the reference standard. One study (62) did not state
which criteria were used.
Most studies included assessment for both major
and minor depressive disorders in the definition of
PPD (11, 12, 22–24, 28–32, 34–39, 41, 50, 51, 55–
62) whilst other studies used a stricter definition of
major depression only (25, 27, 33, 40, 42, 43). All
three studies validating the EPDS in the antenatal
population included minor and major depression
in the definition of APD and used a higher cut-off
point to screen for illness (12, 22, 23).
Diagnostic instruments
Eleven different diagnostic instruments were used
across studies. The SCID (13) was used in 10 studies
(24, 27, 29, 31, 32, 37, 39, 42, 44, 45); the MINI (14)
in five studies (22, 30, 34, 35, 41); GoldbergÕs
Standardised Psychiatric Interview (SPI) (63) in
four studies (11, 12, 25, 28); the Composite Inter-
national Diagnostic Interview (CIDI) (64) in three
studies (56–58) and the PSE (15) was used in a
further four studies (26, 56, 60, 61). The Clinical
Interview Schedule (CIS) (65) was used in three
studies (23, 59, 62); the Montgomery–Asberg
Depression Rating Scale (MADRS) (66), which is
an observer rating scale, was used in two studies
(38, 43); the Schedule of Affective Disorders and
Schizophrenia (SADS) (67), the Psychiatric Assess-
ment Schedule (PAS) (68), the Diagnostic Interview
Schedule (DIS) (69), and the PRIME-MD (70) were
used in one study each (33, 36 and 51). One study
(40) used a Ôstructured clinical interview based on
the criteria defined in the current fourth edition of
the DSM-IVÕ and another (46) used a Ôsemistruc-
tured interviewÕ but neither named the interview.
Quality assessment
Table 1 summarises the results of the quality
assessment. Study quality was generally reasonable
with 12 of 37 studies achieving a grade A. This
denoted that an appropriate study design was used
in a representative population and that the com-
parison of the EPDS result to the Ôgold standardÕ
was blind. Of the remaining 25 studies, 24 achieved
a grade B because of one or two of the following
characteristics: lack of uncertain blinding, narrow
population sample and differential use of the
reference standard. For one study (59), there was
insufficient data available to grade its quality.
Sensitivity
1
and specificity
2
There was a large variation in estimates of sensi-
tivity and specificity between studies (see
Tables2
5). Using an EPDS score cut-off point of 9 10, the
sensitivity for detecting postnatal depression
ranged from 59 to 100% and the specificity
ranged from 44 to 97%. Using an EPDS cut-off
point of 12 13, the sensitivity for postnatal depres-
sion ranged from 34 to 100% and the specificity
Table 1. Table to show quality grading system for included studies
Grade Test accuracy
Number
of studies
A High quality studies with a blind
comparison of test to reference
standard in an appropriate
population spectrum
12
B Any one or two of the following: 24
Narrow population spectrum
Differential use of reference standard
Reference standard not blind
Case control study design
C Any three or more of the above 0
D Expert opinion 0
Other Insufficient data available to grade study 1
Modified from Sackett et al.
1
Sensitivity is the proportion of those with the disease who test
positive.
2
Specificity is the proportion of those without the disease who
have negative test results.
Systematic review of studies validating the EPDS
353
Page 4
from 49 to 100%. For the three studies validating
the EPDS in an antenatal sample, the cut off point
of 14 15 was selected. The sensitivity for major
depression ranged from 57 to 100% and the
specificity from 93 to 99%.
Positive
3
and negative predictive values
4
The positive predictive values (PPV) of the EPDS
for detecting depression (major, or major and
minor combined) at a cut-off score of 9 10 ranged
from 9 to 64% for major depression and from 23 to
93% for combined depression. At a cut-off score
on the EPDS of 12 13, the PPV was 17–100% for
major depression and 30–100% for combined
major and minor depression. If used in an antena-
tal population to detect APD, a cut-off of 14 15
was used and the PPV for major depression ranged
from 60 to 80% (two studies) and from 80 to 90%
(three studies) for major and minor depression
combined.
The range of negative predictive values (NPV) of
the EPDS for detecting depression (major, or
major and minor combined) at a cut-off score of
9 10 was 82–100% for both major and combined
depression. For a cut-off score of 12 13 on the
EPDS, the NPV for major depression was 47–
100% and for combined major and minor depres-
sion was 84–100%. The median values and ranges
for the PPV and NPV from all postpartum studies
are reported in
Table 6.
In antepartum populations, the range of PPV
results was 60–88% for major depression and 80–
81% for combined major and minor depression.
The NPV results ranged from 97 to 100% for
major depression and from 94 to 95% in combined
depression. Median values could not be calculated
for the antepartum studies as there were data for
only two studies within each category.
Likelihood ratios
5
Estimates of the positive likelihood ratios for the
presence of any postnatal depression (i.e. combined
minor and major) ranged from 1.81 to 88 using an
EPDS cut-off of 9 10, and from 0 to ¥ using a cut-
off point of 12 13. Estimates of the negative
likelihood ratio ranged from 0 to 0.48 at a cut-off
point of 9 10 and from 0 to 1 at a cut-off point of
12 13.
In the group of antepartum women, estimates of
the positive likelihood ratios for the presence of
any APD ranged from 22 to 29 and the negative
likelihood ratio ranged from 0.35 to 0.44.
Scatterplots
Scatterplots of the rate of true positives (sensitiv-
ity) against the rate of false positives (1 ) specific-
ity) were drawn for studies validating the EPDS in
detecting major depression and combined major
and minor depression in the postpartum period
(
Figs 2 and 3). Scatterplots for the validity of the
EPDS in detecting minor PPD or APD were not
constructed because of the small number of studies
in each category (two and three respectively). The
plots show some scattering of the results, reflecting
the level of heterogeneity in performance of the
EPDS in different populations and at different
times in the postnatal period.
Language of administration
There were six studies (12, 24, 25, 28, 38, 44)
using the English language version of the EPDS
alone in postpartum populations. There were 25
using alternative languages and two studies
which administered the EPDS in English and
one or more other languages (33, 58). In addi-
tion, one paper administered the EPDS in
English and Yoruba but did not present results
Table 2. Summary of studies validating the EPDS for detection of minor depression in the postpartum period
Author, country
No.
women
Language
and postpartum
timing of EPDS
administration
Diagnostic
instrument
Diagnostic
criteria
Sensitivity
% at EPDS
cut-off point
of 9 10
Specificity
% at EPDS
cut-off point
of 9 10 PPV % NPV %
Positive
likelihood ratio
at EPDS cut-off
point of 9 10
Negative
likelihood
ratio at EPDS
cut-off
point of 9 10
Benvenuti, Italy (34) 67 Italian EPDS at 8–12 weeks MINI DSM-IIIR 83 90 60 N A 7.9 0.2
Murray, UK (28) 646 English EPDS at 6 weeks SPI DSM-III 82 82 17 N A 4.6 0.22
PPV, positive predictive value; NPV, negative predictive value.
3
Positive predictive value is the proportion of people scoring
positive in a test who actually have the disorder.
4
Negative predictive value is the proportion of people scoring
negative in a test who actually do not have the disorder.
5
Likelihood ratios describe how many times more (or less)
likely a person with disease is to receive a particular test result
(positive or negative) than a person without disease.
Gibson et al.
354
Page 5
Table 3. Summary of studies validating the EPDS for detection of major depression in the postpartum period
Author, country
No.
women
Language and
postpartum timing
of EPDS
administration
Diagnostic
instrument
Diagnostic
criteria
Sensitivity (%) at
different EPDS
cut-off points
Specificity (%) at
different EPDS
cut-off points
PPV (%) at
different EPDS
cut-off points
NPV (%) at
different EPDS
cut-off points
Positive Likelihood
Ratio at different
EPDS cut-off
points
Negative Likelihood
Ratio at different
EPDS cut-off points
9 10 12 13 9 10 12 13 9 10 12 13 9 10 12 13 9 10 12 13 9 10 12 13
Adewuya,
Nigeria (29)
876 English Yoruba EPDS at
6 weeks postnatal infant
clinic
SCID-NP
DSM-IIIR
N A 100 N A98N A 100 N A91N A50N A0
Barnett,
Australia (33)
105 AC*
98 A*
113 V*
English, Arabic and
Vietnamese EPDS at
6 weeks
DIS
DSM-IIIR
AC86
A78
V 100
AC57
A56
V 100
AC84
A80
V69
AC94
A91
V89
AC27
A29
V13
AC40
A39
V29
N AN A AC5.4
A 3.9
V 3.2
AC9.4
A6
V9
AC0.17
A 0.28
V0
AC0.46
A 0.49
V0
Beck, USA (24) 150 English EPDS at at
2–12 weeks
SCID
DSM-IV
N A78N A9964938296N A78N A 0.22
Benvenuti,
Italy (34)
67 Italian EPDS at
8–12 weeks
MINI
DSM-IIIR
N A56N A9960N A91N AN A56N A 0.4
Berle,
Norway (35)
100 Norwegian EPDS at
6–12 weeks
MINI
DSM-IV
96 56 62 89 48 65 N AN A 2.53 5.09 0.06 0.49
Boyce,
Australia (25)
103 English EPDS at
mean of 12 weeks
SPI
RDC
100 100 89 96 47 69 100 100 9.1 25 0 0
Cox, UK (11) 84 English EPDS at
6 weeks
SPI
RDC
N A96N A70N A56N A98N A 3.2 N A 0.06
Eberhard-Gran,
Norway (20)
310 Norwegian EPDS at
3 and 6 weeks
PRIME-MD
DSM-IV
N A44N A96N
AN AN AN AN A 10.43 N A 0.58
Garcia Esteve,
Spain (37)
334 Spanish EPDS at
6 weeks
SCID
DSM-IV
100 86 89 95 N A46N A 100 9.1 17.2 0 0.15
Kadir, Malaysia
(62)
52 Malay EPDS
at 4–12 weeks
CIS 100 75 83 94 33 50 N AN A 5.9 12.5 0 0.27
Lawrie, South
Africa (38)
102 English EPDS with
translator at 6 weeks
MADRS
DSM-IV
100 88 51 72 15 21 100 99 2.04 3.15 0 0.17
Murray, UK (28) 646 English EPDS at
6 weeks
SPI
DSM-III
96 81 82 96 N A43N AN A 5.3 18.8 0.05 0.2
Muzik, Austria
(27)
49 Community sample
2–3 months
SCID
DSM-IIIR
78 67 75 95 44 75 94 92 3.12 13.4 0.29 0.35
Regmi, Nepal (40) 100 Nepalese EPDS at
2–3 months
Structured clinical
interview based
on DSM-IV
criteria, DSM-IV
N A 100 N A93N A42N A 100 N A 14.29 N A0
Vega-Dienstmaier,
Peru (42)
321 Spanish EPDS within
1st postpartum year
SCID
DSM-IV
90 90 44 72 9 17 99 99 1.61 3.22 0.24 0.15
Wickberg, Sweden
(43)
128 Swedish EPDS at
2–3 months
MADRS
DSM-IIIR
N A85N A63N A64N A47N A 2.3 N A 0.24
AC, Anglo-Celtic; A, Arabic; V, Vietnamese; PPV, positive predictive value; NPV, negative predictive value.
Systematic review of studies validating the EPDS
355
Page 6
Table 4. Summary of studies validating the EPDS for detection of combined minor and major depression in the postpartum period
Author, country No. women
Language and
postpartum timing
of EPDS administration
Diagnostic
instrument
Diagnostic
criteria
Sensitivity (%) at
different EPDS
cut-off points
Specificity (%) at
different EPDS
cut-off points
PPV (%) at
different EPDS
cut-off points
NPV (%) at
different EPDS
cut-off points
Positive Likelihood
Ratio at different
EPDS cut-off
points
Negative Likeli-
hood Ratio at
different EPDS
cut-off points
9 10 12 13 9 10 12 13 9 10 12 13 9 10 12 13 9 10 12 13 9 10 12 13
Adewuya, Nigeria (29) 876 English Yoruba EPDS at
6 weeks
SCID-NP
DSM-IIIR
88 100 99 49 93 100 97 92 88 1.96 0.12 0
Agoub, Morocco (30) 144 Arabic EPDS at 15–20 days MINI
DSM-IV
N A92N A966586N AN AN A23 N A 0.08
Ascaso, Spain (31) 334 Spanish EPDS at 6 weeks SCID
DSM-IV
89 62 71 92 N AN AN AN A 3.01 7.64 0.16 0.41
Aydin, Turkey (32) 341 Turkish EPDS in first
postpartum year
SCID
DSM-IV
96 76 47 72 23 30 99 95 1.81 2.72 0.09 0.33
Beck, USA (24) 150 English EPDS at 2–12 weeks SCID
DSM-IV
59 N A86N A 64 93 82 96 4.21 N A 0.48 N
A
Berle, Norway (35) 100 Norwegian EPDS at
6–12 weeks
MINI
DSM-IV
90 95 71 95 N A87N AN A 3.1 9.81 0.14 0.54
Carpiniello, Italy (60) 61 Italian EPDS at 4–6 weeks PSE
PSE-ID-Catego
100 67 83 100 50 100 100 95 5.88 ¥ 0 0.33
Clarke, Canada (44) 103 English EPDS up to 1 year
postpartum
SCID
DSM-IV
N A81N A88N A56N A96N A 6.75 N A 0.22
Cox, UK (11) 84 English EPDS at 6 weeks SPI
RDC
89 N A51N A58N A93N A 1.82 N A 0.22 N A
Eberhard-Gran, Norway (20) 310 Norwegian EPDS at 3 and
6 weeks
PRIME-MD
DSM-IV
67 78 97 98 N AN AN AN A 23 38.9 0.34 0.23
Felice, Malta (23) 229 Maltese EPDS used at
8–10 weeks
CIS-R
ICD-10
83 78 92 98 43 56 98 96 10.04 39 0.18 0.22
Garcia Esteve, Spain(37) 334 Spanish EPDS at 6 weeks SCID
DSM-IV
89 62 93 98 56 46 99 100 12.7 31 0.12 0.39
Gausia, Bangla-desh (45) 126 Bangla EPDS at 6–8 weeks SCID
DSM-IV
89 67 80 93 31 50 99 97 4.4 10.3 0.14 0.35
Ghubash, UAE (61) 95 Arabic EPDS at 1 week PSE
ICD-10
91 73 84 90 44 50 99 96 5.69 7.3 0.11 0.3
Guedeney, France (26) 87 French EPDS at 4 months
postpartum
PSE
RDC
80 60 92 97 64 78 96 93 10 20 0.22 0.41
Jadresic, Chile (50) 108 Spanish EPDS at 2–3 months PAS
RDC
100 55 80 94 37 50 100 95 5 9.12 0 0.48
Kadir, Malaysia (62) 52 Malay EPDS at 4–12 weeks CIS 73 55 90 100 67 50 93 98 7.3 ¥ 0 0.45
Lawrie, South Africa (38) 102 English EPDS with translator
at 6 weeks
MADRS
DSM-IV
84 76 57 82 39 58 92 91 1.95 4.18 0.28 0.29
Lee, Hong Kong (39) 145 Chinese EPDS used at
6 weeks
Chinese SCID-NP
DSM-IV
82 41 86 95 44 N A97N
A 5.86 8.2 0.21 0.62
Leverton, UK (55) 207* English EPDS at 3 months PSE
ICD-10
90 70 84 93 23 33 99 98 5.6 10 0.12 0.32
Mahmud, Malaysia (56) 64 Malay EPDS at 4–12 weeks CIDI
ICD-10
100 78 93 98 69 88 100 96 13.7 43.2 0 0.23
Gibson et al.
356
Page 7
Table 4. (Continued)
Author, country No. women
Language and
postpartum timing
of EPDS administration
Diagnostic
instrument
Diagnostic
criteria
Sensitivity (%) at
different EPDS
cut-off points
Specificity (%) at
different EPDS
cut-off points
PPV (%) at
different EPDS
cut-off points
NPV (%) at
different EPDS
cut-off points
Positive Likelihood
Ratio at different
EPDS cut-off
points
Negative
Likelihood Ratio
at different EPDS
cut-off points
9 10 12 13 9 10 12 13 9 10 12 13 9 10 12 13 9 10 12 13 9 10 12 13
Murray, UK (28) 646 English EPDS at 6 weeks
postnatal
SPI
DSM-III
89 68 82 96 39 67 N AN A 5.0 15.7 0.1 0.34
Pitanupong, Thailand (46) 615 Thai EPDS at 6–8 weeks ÔSemistructured
interviewÕ
DSM-IV
60 34 90 97 42 59 95 92 6 11.3 0.44 0.68
Rushidi, Malaysia (59) 54 Malay EPDS at 4–12 weeks CIS
ICD-10
77 62 90 100 71 100 93 89 7.7 0 0.26 0.38
Teng, Taiwan (41) 531 Chinese EPDS at 3 days
(n = 328) or 6 weeks
(n = 203) postpartum
MINI
DSM-IV
N A96N A85N A46N A99N A 6.4 N
A 0.05
Uwakwe, Nigeria (57) 225 Igbo EPDS at 7 days CIDI
ICD-10
75 50 97 98 75 80 97 84 25 25 0.26 0.51
Werrett, UK (58) 23 English and Punjabi EPDS
at 5–8 weeks
CIDI
ICD-10
E 100
P71
E71
P71
E63
P69
E81
P94
E54
P50
E63
P83
N AN A E 2.67
P 2.28
E 5.71
P 11.33
E0
P 0.42
E 0.35
P 0.31
Yoshida, Japan (51) 186 Japanese EPDS at
1 month
SADS
RDC
J82
UK N A
J55
UK N A
J95
UK N A
J98
UK 100
N AN AN AN A J 16.4
UK N A
J 27.5
UK N A
J 0.19
PUK N A
J 0.46
UK 1
PPV, positive predictive value; NPV, negative predictive value; E, English; P, Punjabi; J, sample of Japanese women in Japan; UK, sample of Japanese women living in UK.
*Includes 1 case of mania, 1 case of Ôanxiety neurosisÕ and 1 case of paranoid psychosis.
Sensitivity and specificity reported in reverse order in paper.
Systematic review of studies validating the EPDS
357
Page 8
separately for the two languages (29). The three
papers which used two or more different lan-
guage versions of the EPDS were not included in
the examination.
The authors considered the hypothesis that the
language of administration of the EPDS may have
contributed to the heterogeneity of results. We
therefore examined the results from English-speak-
ing populations separately from non-English-
speaking groups. At a diagnostic threshold of
9 10 on the EPDS, there was no appreciable
difference between the two groups for specificity
(English language EPDS 51–89%; non-English
language EPDS 44–97%). The sensitivity range
was larger for the non-English papers (English
language EPDS 82–100%; non-English language
EPDS 60–100%). At a diagnostic threshold of
12 13, both the sensitivity and specificity ranges
showed a more marked difference depending upon
language of administration. Sensitivity values for
English language EPDS studies ranged from 76 to
100% and for non-English language EPDS studies
from 49 to 100%. Specificity values for English
language EPDS papers ranged from 70 to 99%,
whilst for non-English language EPDS studies the
range was from 34 to 100%. It was not possible to
consider the difference between countries as very
few countries produced more than one study.
For the antepartum studies, only one was
conducted in English (12) and its results did not
differ appreciably from the other two studies.
Discussion
In conclusion, 37 validation studies of the EPDS
had been published up to the end of July 2008,
which met our inclusion and exclusion criteria.
However, the degree of heterogeneity among
studies was such that it was not possible to form
distinct groups for further subanalysis. There were
significant differences in study design, population
sampled, the timing of testing, language version of
the EPDS used and diagnostic criteria. As a result,
each subgroup of interest that we considered was
itself too heterogeneous to analyse statistically.
Were we to do so, we were concerned that the
results would not be valid or useful. The degree of
heterogeneity may also suggest that these studies
have not provided evidence for the EPDS as a valid
screening tool across different cultural groups.
There was a wide range of values for sensitivity
and specificity at all cut-off points. There are
various explanations to be considered. Both meth-
ods and populations varied greatly between the
studies. Samples were drawn from rural and urban
areas, from affluent and poor women, and from
Table 5. Summary of studies validating the EPDS for detection of depression in the antepartum period
Author, country
No.
women
Language and
antepartum
timing of EPDS
administration
Diagnostic
Criteria
Diagnostic
Instrument
EPDS
cut-off point
Sensitivity (%) Specificity (%) PPV (%) NPV (%)
Positive likeli-
hood ratio
Negative likeli-
hood ratio
Major Comb Major Comb Major Comb Major Comb Major Comb Major Comb
Adewuya, Nigeria (22) 182 Yoruba EPDS at
>32 weeks gestation
MINI
DSM-IV
14–15 78 N A99N A88N A97N A78N A 0.22 N A
Felice, Malta (23) 229 Maltese EPDS at
36–40 weeks gestation
CIS-R
ICD-10
14–15 73 66 93 97 N A81N A 94 10 22 0.29 0.35
Murray, UK (12) 100 English EPDS at
28–34 weeks gestation
SPI
RDC
14–15 100 57 96 98 60 80 100 95 25 29 0 0.44
Comb, combined major and minor depression; PPV, positive predictive value; NPV, negative predictive value.
Gibson et al.
358
Page 9
countries with different cultural attitudes to feeling
and expressing distress. Different diagnostic inter-
views and criteria were used, and screening was
performed at various times in the antepar-
tum postpartum period. Another important con-
sideration is that the screening and diagnostic
instruments used in the studies have been devel-
oped in the West. Although an international
qualitative study of the experience of women in
pregnancy and the postpartum suggests that there
is a universal experience of morbid unhappiness
with similar signs and symptoms to APD and PPD
(8), the understanding of the nature of this state
and its aetiology differ according to the cultural
setting. It is possible that instruments devised to
detect the Western concept of depression do not
accurately detect the presence of significant unhap-
piness in other cultural settings. If so, this would
contribute to the heterogeneity of the results.
The degree of variation found in the PPV and
NPV results is likely to be explained by several
factors. First, the prevalence of major and or
minor depression varied across studies. As the
prevalence of the disorder increases, so too does
the PPV. Therefore, the studies in which women
with PND were over-represented showed higher
PPV values. The PPV also varies according to the
EPDS cut-off used. Barnett et al. (33) found that a
higher cut-off score (14 15) was recommended to
identify PND in a Vietnamese-speaking sample of
mothers in Australia, compared with their English-
and Arabic-speaking counterparts where a cut-off
score of 9 10 was suggested. This may imply that
different cut-off scores should be used in different
cultural groups and could contribute to the heter-
ogeneity found in these studies.
It has been suggested that, for use in clinical
practice, Ôa positive likelihood ratio of greater than
10 may provide convincing diagnostic evidence,
whilst those greater than 5 give strong diagnos-
tic evidenceÕ (71). The median positive likelihood
ratios (Table 6) show that, at a cut-off point of
9 10, for major PPD, performance is modest.
However, when the cut-off point is increased to
12 13, the ÔaverageÕ study suggests that the EPDS
provides Ôconvincing diagnostic evidenceÕ for the
presence of major depression. When the EPDS is
used to screen for women with any type of PPD
(i.e. combined minor and major), at a cut-off point
of 12 13, the average study suggests that the EPDS
provides Ôconvincing diagnostic evidenceÕ, while at
a cut-off point of 9 10 the average study suggests
that the EPDS provides only Ôstrong diagnostic
evidenceÕ.
Regarding the negative likelihood ratio, it has
been suggested that Ôa negative likelihood ratio of
<0.1 may provide convincing diagnostic evidence
Table 6. Summary statistics for performance of EPDS in detecting postpartum depression at different cut-off points
PPV
Median (range)
NPV
Median (range)
LR+
Median (range)
LR)
Median (range)
Minor depression EPDS cut-off 9 10 N AN AN AN A
Major depression EPDS cut-off 9 10 33 (9–64) 97 (82–100) 3.9 (1.61–9.1) 0.05 (0–0.29)
Major depression EPDS cut-off 12 13 48 (17–100) 98.5 (47–100) 10.47 (2.7–78) 0.22 (0–0.58)
Combined depression EPDS cut-off 9 10 50 (23–93) 97 (82–100) 5.88 (1.81–88) 0.15 (0–0.48)
Combined depression EPDS cut-off 12 13 61 (30–100) 96 (84–100) 10.3 (0–¥) 0.35 (0–1)
PPV, positive predictive value; NPV, negative predictive value; LR+, positive likelihood ratio; LR), negative likelihood ratio.
Median values could not be calculated for the antenatal group as there were only data for 2 studies within each category.
0
0.2
0.4
0.6
0.8
1
1.2
0 0.1 0.2 0.3 0.4 0.5 0.6
1-Specificity
Sensitivity
Cut-off 12–13
Cut-off 9 –10
Fig. 2. Scatterplot of sensitivity vs. 1 ) specificity for detecting
major depression in each study.
0
0.2
0.4
0.6
0.8
1
1.2
0 0.1 0.2 0.3 0.4 0.5 0.6
1-Specificity
Sensitivity
Cut-off 12–13
Cut-off 9 –10
Fig. 3. Scatterplot of sensitivity vs. 1 ) specificity for detecting
major minor depression in each study.
Systematic review of studies validating the EPDS
359
Page 10
and below 0.2 gives strong diagnostic evidenceÕ
(71). Again, the range of likelihood ratios through-
out the studies is wide. The median negative
likelihood ratios (Table 6) show that at the
higher EPDS cut-off point of 12 13, the test
performs poorly in screening out those women
without PPD, both with regard to major depres-
sion alone and combined minor major depression.
However, at the lower cut-off point of 9 10 the
EPDS performs considerably better and the aver-
age study suggest that it provides Ôstrong evidenceÕ
(combined depression) or Ôconvincing evidenceÕ
(major depression) by which to rule out the
diagnosis. As with any screening tool, in selecting
a cut-off point, there is a trade-off between the
accuracy of ruling out a diagnosis and the accuracy
of ruling in a diagnosis. To determine the most
appropriate EPDS cut-off, it is therefore necessary
to determine the needs of the specific population,
the resources available and the consequences both
of missing diagnoses and of over-diagnosis.
In terms of the results for likelihood ratios, it has
been suggested that for use in clinical practice
Ôa positive likelihood ratio of greater than 10
may provide convincing diagnostic evidence, whilst
those greater than 5 give strong diagnostic
evidenceÕ (71). There are wide ranges of results for
the likelihood ratios, but the median positive
likelihood ratios (Table 6) show that at a cut-off
point of 9 10 there is little evidence to suggest that
the EPDS can accurately identify women with
major PPD. However, when the cut-off point is
increased to 12 13 there is Ôconvincing diagnostic
evidenceÕ for the test. When the EPDS is used to
screen for women with any type of PPD (i.e.
combined minor and major depression) over half of
the studies at cut-off points of both 9 10 and 12 13
show Ôconvincing diagnostic evidenceÕ. This
suggests that the EPDS can be an effective screening
tool for detecting combined depression in post-
partum women at a cut-off point of 9 10, but that
its accuracy increases in both major and combined
PPD if the cut-off point is increased to 12 13.
Regarding the negative likelihood ratio, it has
been suggested that Ôa negative likelihood ratio of
<0.1 may provide convincing diagnostic evidence
and below 0.2 gives strong diagnostic evidenceÕ
(71). Again, the range of likelihood ratios through-
out the studies is wide. The median negative
likelihood ratios (Table 6) show that at the higher
EPDS cut-off point of 12 13, the test performs
poorly in screening out those women without PPD,
both when considering major depression alone, or
minor and major depression combined. However,
at the lower cut-off point of 9 10 the EPDS
performs considerably better and in over half of
the studies presented it provides Ôstrong diagnostic
evidenceÕ (combined depression) or Ôconvincing
diagnostic evidenceÕ (major depression).
Thus, in over half of the studies the EPDS could
convincingly rule out the presence of PPD in non-
depressed women when a cut-off point of 9 10 is
used. However, if the cut-off point is increased to
12 13 the EPDS no longer provides either ÔstrongÕ
or ÔconvincingÕ evidence for excluding the disorder.
It is therefore necessary to determine, according
to the characteristics of the population, whether it
is more important to use the EPDS to rule in cases
of PPD, or rule out women who do not have the
disorder. This will determine the most useful EPDS
cut-off point to select.
Whilst the included studies scored well on the
quality rating scale (all achieving a grade A or B)
there were some general points of weakness. The
method of selection of study samples was not
always clearly described and in some cases was
Ôad hocÕ which may bias the results towards
including more motivated, such as psychologically
vulnerable women. In several cases, it was unclear
if the comparison of the EPDS results to the
reference standard was blind (and in some there
was clearly no blinding) which again could bias the
results. Four of the studies (28, 38, 55, 58) were
conducted on a narrowly defined population,
e.g. women who had experienced obstetric compli-
cations, and the results are not likely to be
generalisable to a standard community setting.
Studies were excluded if there was a delay of
more than 24 h between the administration of the
EPDS and the diagnostic interview. This was to
avoid the possibility of measuring different symp-
tom profiles in the same individual, given the
potential for fluctuations in the mental state in the
postpartum period. We examined the five studies
excluded for this reason. There was a delay of up to
8 weeks in these studies. We found that there was
no appreciable difference in any of the results.
However, because of the small number of studies
involved, we could not conclude that the EPDS
remains a valid screening tool for PND when a
time delay occurs between screening and the
diagnostic interview.
In addition, the wide range of values for both
sensitivity and specificity across studies make
interpretation of the utility of this screening
instrument difficult. The heterogeneity in results
may suggest that the accuracy of the EPDS varies
depending on the clinical setting, country and
language of administration. This may be compli-
cated by the use of diagnostic interviews and
instruments devised to detect depression as it is
understood and expressed in Western societies.
Gibson et al.
360
Page 11
The examination of the language of administra-
tion of the EPDS found a larger difference in the
sensitivity range at a cut-off point of 12 13,
implying that the tool may be more sensitive
when administered in the English language. This
may suggest that the EPDS performs best at a
higher cut-off point and for women who are
familiar with expressing their distress in English,
and that the screening and diagnostic tools are
geared towards identifying more severe illness in
English-speaking populations. There may be sys-
tematic differences in the foreign language versions
of the EPDS and the way it is applied between
countries. It is possible also that these findings
result from the differences in study design and
population sampled.
At present, when the EPDS is used in the
general population it will generate a substantial
proportion of false positives, which is costly to
service providers in terms of further assessment. In
addition, it will miss a considerable number of
cases. However, the majority of screening tools
suffer from the same difficulties. It should be
noted that such tools are designed to indicate the
possibility of illness. A clinical assessment is
required to make a definitive diagnosis. It is also
the case that clinical assessment carries a higher
cost. The utility of the EPDS rests in its free
availability, ease of administration and general
acceptability to women, if given sympathetically.
Therefore, if the above caveats are observed, this
remains a useful tool in the field of perinatal
mental health.
Strengths
This systematic review used both a broad and
focused search strategy. It was able to include
studies published between 2000 and 2008 thus
making the review more up to date than the existing
Eberhard-Gran review. It additionally considered
studies examining an antepartum population.
Limitations
This systematic review was not able to answer the
question of whether or not the EPDS is an accurate
screening tool for APD and PPD. The heteroge-
neity of included studies prevented progression to
meta-analysis and therefore statistical comparison
of the EPDS across settings was not possible.
Future research
A point which emerges from this, and other,
studies (19, 20) is the high level of heterogeneity.
Therefore, future quantitative and qualitative
research should focus on identifying the underlying
causes of this. Finally, there are only three valida-
tion studies in antenatal populations and therefore
further studies in this group are important.
Acknowledgements
We thank the authors who provided further information about
their studies. We also thank Maria Quigley for commenting on
an earlier draft of this paper. This paper reports on an
independent study, which is part-funded by the Policy
Research Programme in the Department of Health. The
views expressed are not necessarily those of the Department.
Declaration of interest
Jonathan Price received a research grant from Servier R&D
Limited for the development of a questionnaire to measure
antidepressant side-effects.
References
1. Scottish Intercollegiate Guidelines Network. Management
of postnatal depression and puerperal psychosis. A na-
tional clinical guideline, guideline number 60. Edinburgh:
SIGN Executive, 2002.
2. OÕhara MW, Swain AM. Rates and risk of postpartum
depression a meta-analysis. Int Rev Psychiatry
1996;8:37–54.
3. Boath EH, Pryce AJ, Cox JL. Postnatal depression: the
impact on the family. J Reprod Infant Psychol 1998;
16:199–203.
4. Lovestone S, Kumar R. Postnatal psychiatric illness: the
impact on partners. Br J Psychiatry 1993;163:210–216.
5. Murray L, Cooper PJ. Postpartum depression and child
development. Psychol Med 1997;27:253–260.
6. Caplan HL, Cogill SR, Alexandra H, Robson KM, Katz R,
Kumar R. Maternal depression and the emotional devel-
opment of the child. Br J Psychiatry 1989;154:818–822.
7. Murray L, Sinclair D, Cooper P, Ducournau P, Turner P,
Stein A. The socio-emotional development of 5-year-old
children of postnatally depressed mothers. J Child Psychol
Psychiatry 1999;40:1259–1271.
8. Evans J , Heron J, Francomb H, Oke S, Golding J. Cohort
study of depressed mood during pregnancy and after
childbirth. BMJ 2001;323:257–260.
9. Green JM. Postnatal depression or perinatal dysphoria?
Findings from a longitudinal community-based study
using the Edinburgh Postnatal Depression Scale. J Reprod
Infant Psychol 1998;
16:143–155.
10. Boyd RC, Le HN, Somberg R. Review of screening instru-
ments for postpartum depression. Arch Womens Ment
Health 2005;8:141–153.
11. Cox JL, Holden JM, Sagovsky R. Detection of postnatal
depression. Development of the 10-item Edinburgh Post-
natal Depression Scale. Br J Psychiatry 1987;150:782–786.
12. Murray D, Cox J. Screening for depression during preg-
nancy with the Edinburgh Depression Scale (EPDS).
J Reprod Infant Psychol 1990;8:99–107.
13. Spitzer RL, Williams JB, Gibbon M, First MD. The Struc-
tured Clinical Interview for DSM-III-R (SCID): History,
rationale, description. Arch Gen Psychiatry 1992;49:624–
629.
Systematic review of studies validating the EPDS
361
Page 12
14. Sheehan DV, Lecrubier Y, Sheehan KH et al. The Mini
International Neuropsychiatric Interview (MINI); the
development and validation of a structured diagnostic
psychiatric interview for DSM-IV and ICD-10. J Clin
Psychiatry 1998; 59(suppl. 20):22–33; quiz 34–57. Review.
15. Wing JK, Stuart E. The PSE-ID-Catego System supple-
mentary manual. London: MRC Social Unit, 1978.
16. Postnatal Depression Screening. National Screening Com-
mittee UK, 2006 [Cited 16 December 2007]. Available at:
http://www.library.nhs.uk/screening/ViewRe-
source.aspx?resID=60978.
17. Antenatal and Postnatal Mental Health. Clinical man-
agement and service guidance. Clinical Guideline 45.
London:National Institute for Clinical Excellence (NICE),
2007
18. Whooley MA, Avins AL, Miranda J, Browner WS. Case-
finding instruments for depression. Two questions are as
good as many. J Gen Intern Med 1997;12:439–445.
19. Gaynes BN, Gavin N, Meltzer-Brody S et al. Perinatal
depression: prevalence, screening accuracy, and screening
outcomes [Cited 20 March 2008]. Available at: http://
www.ahrq.gov/clinic/epcsums/peridepsum2.htm.
20. Eberhard-Gran M, Eskild A, Tambs K, Opjordsmoen S,
Samuelson SO. Review of validation studies of the Edin-
burgh Postnatal Depression Scale. Acta Psychiatr Scand
2001;104:243–249.
21. Khan KS, ter Riet G, Popay J, Nixon J, Kleijnen J. In: Khan
KS, ter Riet G,
Glanville J, Snowden AJ, Kleijnen J, eds.
Understanding Systematic Reviews on Effectiveness. 2nd
edn. Centre for Reviews and Dissemination Report
Number 4 2001. Available at: http://www.york.ac.uk/inst/
crd/report4.htm (accessed 20 February 2009).
22. Adewuya AO, Ola BA, Dada AO, Fasoto OO. Validation of
the Edinburgh Postnatal Depression Scale as a screening
tool for depression in late pregnancy among Nigerian
women. J Psychosom Obstet Gynecol 2006;2:267–272.
23. Felice E, Saliba J, Grech V, Cox J. Validation of the Mal-
tese version of the Edinburgh Postnatal Depression Scale.
Arch Womens Ment Health 2006;9:75–80.
24. Beck CT, Gable RK. Comparative analysis of the perfor-
mance of the Postpartum Depression Screening Scale with
two other depression instruments. Nurs Res 2001;50:242–
250.
25. Boyce P, Stubbs J, Todd A. The Edinburgh Postnatal
Depression Scale: validation for an Austrian sample. Aust
N Z J Psychiatry 1993;27:472–476.
26. Guedeney N, Fermanian J. Validation of the French version
of the Edinburgh Postnatal Depression Scale (EPDS): new
results about use and psychometric properties. Eur Psy-
chiatry 1998;13:83–89.
27. Muzik M, Klier CM, Rosenblum KL, Holzinger A, Umek W,
Katschnig H. Are commonly used self-report inventories
suitable for screening postpartum depression and anxiety
disorders? Acta Psychiatr Scand 2000;102:71–73.
28. Murray L, Carothers AD. The validation of the Edinburgh
Postnatal Depression Scale on a community sample. Br J
Psychiatry 1990;157:288–290.
29. Adewuya AO, Eegunranti AB, Lawal AM. Prevalence of
postnatal depression in Western Nigerian Women: a con-
trolled study. Int J Psychiatry Clin Pract 2005;9:60–64.
30. Agoub M, Moussaoui D,
Battas O. Prevalence of post-
partum depression in a Moroccan sample. Arch Womens
Ment Health 2005;8:37–43.
31. Ascaso Terren C, Garcia Esteve L, Navarro P, Aguado J,
Ojuel J, Jesus Tarragona M. Prevalence of postpartum
depression in Spanish mothers: a comparison between
structured clinical interview and the Edinburgh post-
partum depression scale. Med Clin (Barc) 2003;120:326–
329.
32. Aydin N, Indani T, Yigit A, Hodoglugil NNS. Validation of
the Turkish version of the Edinburgh Postnatal Depression
Scale among women within their first postpartum year. Soc
Psychiatry Psychiatr Epidemiol 2004;39:483–486.
33. Barnett B, Matthey S, Gyaneshwar R. Screening for post-
natal depression in women of non-English speaking
background. Arch Womens Ment Health 1999;2:67–74.
34. Benvenuti P, Maurizio F, Niccolai C, Valoriani V, Cox JL.
The Edinburgh Postnatal Depression Scale: validation for
an Italian sample. J Affect Disord 1999;53:137–141.
35. Berle JO, Aarre TF, Mykletum A, Dahl AA, Holsten F.
Screening for postnatal depression. Validation of the
Norwegian version of the Edinburgh Postnatal Depression
Scale, and assessment of risk factors for postnatal
depression. J Affect Disord 2003;76:151–156.
36. Eberhard-Gran M, Eskild A, Tambs K, Schei B, Opjordsmoen
S, Samuelson SO . The Edinburgh Postnatal Depression
Scale: validation in a Norwegian community sample. Nord
J Psychiatry 2001;
55:113–117.
37. Garcia-Esteve L, Ascaso C, Ojuel J, Navarro P. Validation
of the Edinburgh Postnatal Depression Scale (EPDS) in
Spanish mothers. J Affect Disord 2003;75:71–76.
38. Lawrie TA, Hofmeyr GJ, De Jager M, Berk M. Validation
of the Edinburgh Postnatal Depression Scale in a cohort
of South African women. S Afr Med J 1998;86:1340–
1344.
39. Lee DT, Yip SK, Chiu HF. Detecting postnatal depression in
Chinese women. Validation of the Chinese version of the
Edinburgh Postnatal Depression Scale. Br J Psychiatry
1998;172:433–437.
40. Regmi S, Sligl W, Carter D, Grut W, Seear M. A controlled
study of postpartum depression among Nepalese women:
validation of the Edinburgh Postnatal Depression Scale in
Kathmandu. Trop Med Int Health 2002;7:378–382.
41. Teng HW, Hsu CS, Shih SM, Lu ML, Pan JJ, Shen WW.
Screening postpartum depression with the Taiwanese ver-
sion of the Edinburgh Postnatal Depression Scale. Compr
Psychiatry 2005;46:261–265.
42. Vega-Dienstmaier JM, Mazzotti Suarez G, Campos Sanchez
M. Validation of a Spanish version of the Edinburgh
Postnatal Depression Scale. Actas Esp Psiquiatr 2002;
30:106–111.
43. Wickberg B, Hwang CP. The Edinburgh Postnatal
Depression Scale: validation on a Swedish community
sample. Acta Psychiatr Scand 1996;94:181–184.
44. Clarke PJ. Validation of two postpartum depression
screening scales with a sample of First Nations and Metis
women. Can J Nurs Res 2008;40:112–125.
45. Gausia K, Fisher C, Algin S, Oosthuizen J. Validation of the
Bangla version of the Edinburgh Postnatal Depression
Scale for a Bangladeshi sample. J Reprod Infant Psychol
2007;25:308–315.
46. Pitanupong J, Liabsuetrakul T , Vittayanont A. Validation of
the Thai Edinburgh Postnatal Depression Scale for
screening postpartum depression. Psych Res 2007;149:253–
259.
47. American Psychiatric Association. Diagnostic and statisti-
cal manual of mental disorders, 3rd edn. Washington, DC:
American Psychiatric Press, 1980.
48. American Psychiatric Association. Diagnostic and statisti-
cal manual of mental disorders, revised 3rd edn. Wash-
ington, DC: American Psychiatric Press, 1987.
49. American Psychiatric Association. Diagnostic and statisti-
cal manual of mental disorders, 4th edn. Washington, DC:
American Psychiatric Press, 1994.
Gibson et al.
362
Page 13
50. Jadresic E, Araya R, Jara C. Validation of the Edinburgh
Postnatal Depression Scale (EPDS) in Chilean post-
partum women. J Psychosom Obstet Gynecol 1995;
16:187–191.
51. Yoshida K, Yamashita H, Ueda M, Tashiro N. Postnatal
depression in Japanese mothers and the reconsideration of
ÔSatogaeri bunbenÕ. Ped Int 2001;43:189–193.
52. Spitzer RL, Endicott J, Robins E. Research Diagnostic
Criteria Instrument no. 58. New York: New York State
Psychiatric Institute, 1975.
53. World Health Organisation. Mental disorders, glossary
and guide to their classification in accordance with the 8th
REVISION of the international classification of diseases
(ICD-8). Geneva: WHO, 1974.
54. World Health Organisation. The ICD-10 classification of
mental and behavioural disorders. Clinical descriptions
and diagnostic guidelines. Geneva: WHO, 1992.
55. Leverton TJ, Elliott SA. Is the EPDS a magic wand? 1. A
comparison of the Edinburgh Postnatal Depression Scale
and health visitor report as predictors of diagnosis on the
present state examination. J Reprod Infant Psychol
2000;18:279–296.
56. Mahmud WMRW, Awang A, Mohamed MN. 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. Malay J Med Sci 2003;10:71–
75.
57. Ukwakwe R. Affective (depressive) mornidity in puerperal
Nigerian women: validation of the Edinburgh Postnatal
Depression Scale. Acta Psychiatr Scand 2003;107:251–259.
58. Werrett J, Clifford C. Validation of the Punjabi version of
the Edinburgh postnatal depression scale (EPDS). Int J
Nurs Stud 2006;43:227–236.
59. Rushidi WMWM, Azidah AK, Shaiful B ahari I , Janil MY.
Validation of the Malay version of the Edinburgh Post-
natal Depression Scale (EPDS). Malaysian Journal of
Psychiatry 2002;10:44–49.
60. Carapinello B, Pariante C, Serri F, Costa G, Carta MG.
Validation of the Edinburgh Postnatal Depression Scale in
Italy. J Psychom Obstet Gynecol 1997;18:280–285.
61. Ghubash R, Abou-Saleh MT, Daradekh TK. The validity of
the Arabic Edinburgh Postnatal Depression Scale. Soc
Psychiatr Epidemiol 1997;32:474–476.
62. Kadir AA, Nordin R, Shaiful BI, Mohd JY , Wan Mohd
RWM. Validation of the Malay version of Edinburgh
Postnatal Depression Scale for Postnatal Women in Ke-
lantan, Malaysia. IMJ 2005;12:105–109.
63. Goldberg D. The detection of psychiatric illness by ques-
tionnaire. Maudsley monograph 21. Oxford: Oxford Uni-
versity Press, 1972.
64. World Health Organisation. The composite international
diagnostic interview version 2.0. Geneva: WHO, 1997.
65. Lewis G, Pelosi AJ. Manual if the revised clinical interview
schedule (CIS-R). London: MRC Institute of Psychiatry,
1990.
66. Montgomery SA, Asberg M. A new depression scale designed
to be sensitive to change. Br J Psychiatry 1984;144:35–47.
67. Spitzer RL, Endicott J. Schedule for affective disorders and
schizophrenia, 3rd edn. New York: New York State Psy-
chiatric Institute, 1979.
68. Dean C, Surtees P, Sashidharan S. Comparison of research
diagnostic systems in an Edinburgh community sample. Br
J Psychiatry 1983;142:247–256.
69. Robins L, Helzer J, Cottler L, Goldring E
. NIMH diag-
nostic interview schedule, version 3, revised. Bethesda,
MA: NIH, 1989.
70. Spitzer RL, Williams JB, Kroenke K et al. Utility of a new
procedure for diagnosing mental disorders in primary care.
The PRIME-MD 1000 study. JAMA 1994;272:1749–1756.
71. Egger M, Davey Smith G, Altman D. Systematic reviews in
health care. London: BMJ Publishing Group, 2001.
Appendix 1
Focused search strategy
Medline search: (((Puerper* or Pos?tnatal or Post?partum or Ante?natal or Ante?partum or Pregnan* or Pre?natal) near5 (depress*))
or (explode ÔDepression-PostpartumÕ all subheadings in MIME,MJME)) and ((EPDS or Edinburgh Postnatal Depression Scale) or
((diagnos* or screen*) near (tool* or questionnaire* or scale* or inventor*)) or (depress* near (scale* or questionnaire* or inventor*))
or (explode ÔQuestionnaires-Õ all subheadings in MIME,MJME) or (explode ÔSeverity-of-Illness-IndexÕ all subheadings in
MIME,MJME) or (ÔDiagnostic-and-Statistical-Manual-of-Mental-DisordersÕ all subheadings in MIME,MJME)) and (Diagnos*
interview or Clinical interview* or semi?structured interview*) and ((explode ÔSensitivity-and-SpecificityÕ all subheadings in
MIME,MJME) or (explode ÔLikelihood-FunctionsÕ all subheadings in MIME,MJME) or (Valid* or Specificity or Sensitivity or
predict* value* or Ppv or Npv or Likelihood ratio*)).
PsychInfo search: (((Puerper* or Pos?tnatal or Post?partum or Ante?natal or Ante?partum or Pregnan* or Pre?natal) near5
(depress*)) or (explode ÔPostpartum-DepressionÕ in MJ,MN)) and ((EPDS or Edinburgh Postnatal Depression Scale) or ((diagnos* or
screen*) near (tool* or questionnaire* or scale* or inventor*)) or (depress* near (scale* or questionnaire* or inventor*)) or (explode
ÔPersonality-MeasuresÕ in MJ,MN) or (explode ÔQuestionnaires-Õ in MJ,MN) or (explode ÔDiagnostic-and-Statistical-ManualÕ in
MJ,MN)) and ((Diagnos* interview or Clinical interview* or semi?structured interview*) or (explode ÔPsychodiagnostic-Interview Õ in
MJ,MN) or (explode ÔInterview-SchedulesÕ in MJ,MN)) and ((Valid* or Specificity or Sensitivity or predict* value* or Ppv or Npv or
Likelihood ratio*) or (ÔStatistical-ValidityÕ in MJ,MN) or (ÔPredictability-MeasurementÕ in MJ,MN) or (explode ÔProbability-Õ in
MJ,MN) or (ÔConsistency-MeasurementÕ in MJ,MN) or (ÔTest-ReliabilityÕ in MJ,MN)).
Embase search: ((explode Ôpuerperal-depressionÕ all subheadings in DEM,DER,DRM,DRR) or ((Puerper* or Pos?tnatal or
Post?partum or Ante?natal or Ante?partum or Pregnan* or Pre?natal) near5 (depress*))) and ((EPDS or Edinburgh Postnatal
Depression Scale) or ((diagnos* or screen*) near (tool* or questionnaire* or scale* or inventor*)) or (depress* near (scale* or
questionnaire* or inventor*)) or (explode Ôpsychologic-testÕ all subheadings in DEM,DER,DRM,DRR) or (explode Ôquestionnaire-
Õ all
subheadings in DEM,DER,DRM,DRR)) and ((Diagnos* interview or Clinical interview* or semi?structured interview*) or ((explode
Ôclinical-examinationÕ all subheadings in DEM,DER,DRM,DRR) and (explode Ôinterview-Õ all subheadings in
DEM,DER,DRM,DRR))) and ((Valid* or Specificity or Sensitivity or predict* value* or Ppv or Npv or Likelihood ratio*) or
(explode Ôsensitivity-and-specificityÕ all subheadings in DEM,DER,DRM,DRR)).
Systematic review of studies validating the EPDS
363
Page 14
Gibson et al.
364
Page 15
    • "In this study, the LLV-EPDS had high sensitivity (86.7 %) and specificity (91.5 %), even though it didn't meet most of our recommended steps for culturally sensitive translation processes [31]. This suggests that these highly educated participants having greater emotional literacy [3, 23] and familiarity with test-taking [7], but does not provide evidence that this LLV-EPDS will be useful for the majority who have not had opportunities for education and social participation. Further areas of inconsistency and suboptimal practice appeared to have occurred during the process of formal validation against a diagnostic interview (Fig. 1, criteria: 22–28). "
    [Show abstract] [Hide abstract] ABSTRACT: Background The Edinburgh Postnatal Depression Scale (EPDS), originally developed in Britain, is one of the most widely used screening instruments for assessing symptoms of the Perinatal Common Mental Disorders (PCMDs) of depression and anxiety. However, its potential to detect PCMDs in culturally diverse low- and lower-middle income countries (LALMICs) is unclear. This systematic review aimed to appraise formally validated local language versions of the EPDS from these resource-constrained settings. Methods Following the PRISMA protocol, we searched MEDLINE-OVID, CINAHL-Plus and PUBMED to identify studies reporting translation, cultural adaptation and formal validation of the EPDS to detect PCMDs among women in LALMICs. The quality of the studies meeting inclusion criteria was assessed using standard criteria and a new process-based criteria; which was developed specifically for this study. Results We identified 1281 records among which 16 met inclusion criteria; three further papers were identified by hand-searching reference lists. The publications reported findings from 12 LALMICs in14 native languages. Most of these local language versions of the EPDS (LLV-EPDS) had lower precision for identifying true cases of PCMDs among women in the general perinatal population compared to the original English version. Only one study met all criteria for culturally sensitive translation, the others had not established the comprehensibility of the local version amongst representative groups of women in pre-testing. Many studies tested the LLV-EPDS only amongst convenience samples recruited at single health facilities. Diagnostic interviews for confirmation of mental disorders could have been influenced by the mental health professionals’ lack of blinding to the initial screening results. Additionally, even when diagnostic-interviews were carried out in the local language, questions might not have been understood as most studies followed standard diagnostic protocol which had not been culturally adapted. Conclusions Most of the LLV-EPDS from non-English speaking low- and middle-income-countries did not meet all criteria for formal validation of a screening instrument. Psychometric properties of LLV-EPDS could be enhanced by adopting the new process-based criteria for translation, adaptation and validation. Electronic supplementary material The online version of this article (doi:10.1186/s12884-016-0859-2) contains supplementary material, which is available to authorized users.
    Full-text · Article · Dec 2016 · BMC Pregnancy and Childbirth
    • "The EPDS, a widely used and sensitive measure of postnatal depression (Cox et al., 1987), consists of 10 items rated from 0 to 3, with higher scores indicative of more depressive symptoms. A continuous total sum score was used in the lack of definitive cut-point for clinical depression in pregnant women (Gibson, McKenzie-McHarg, Shakespeare, Price, & Gray, 2009). The State Anxiety measure of The State and Trait Anxiety Inventory (Spielberger, Gorsuch, Lushene, Vagg, & Jacobs, 1983), a reliable and valid measure of anxiety in both clinical and general population screening, was used to measure anxiety symptoms 3 months postpartum. "
    [Show abstract] [Hide abstract] ABSTRACT: Postnatal mother-infant bonding refers to the early emotional bond between mothers and infants. Although some factors, such as maternal mental health, especially postnatal depression, have been considered in relation to mother-infant bonding, few studies have investigated the role of infant temperament traits in early bonding. In this study, the effects of maternal postnatal depressive and anxiety symptoms and infant temperament traits on mother-infant bonding were examined using both mother and father reports of infant temperament. Data for this study came from the first phase of the FinnBrain Birth Cohort Study (n = 102, father reports n = 62). After controlling for maternal symptoms of depression and anxiety, mother-reported infant positive emotionality, measured by infant smiling was related to better mother-infant bonding. In contrast, infant negative emotionality, measured by infant distress to limitations was related to lower quality of bonding. In regards to father-report infant temperament, only infant distress to limitations (i.e., frustration/anger) was associated with lower quality of mother-infant bonding. These findings underline the importance of infant temperament as one factor contributing to early parent-infant relationships, and counseling parents in understanding and caring for infants with different temperament traits.
    No preview · Article · May 2016 · Infant Behavior and Development
    • "The scale includes ten questions about symptoms of depression and the total score ranges from 0 to 30. Several cut off scores have been used to indicate depression , but a cut-off of 15 or higher has been suggested for antepartum depression depression [29]. Previous studies in South Africa used a score of 13 or above, a cut-off with a sensitivity of 0.69 and specificity of 0.78 for the detection of antenatal depression [30, 31]. "
    [Show abstract] [Hide abstract] ABSTRACT: Maternal depression, including antepartum and postpartum depression, is a neglected public health issue with potentially far-reaching effects on maternal and child health. We aimed to measure the burden of antepartum depression and identify risk factors among women in a peri-urban community in Swaziland. We conducted a cross-sectional study within the context of a community outreach peer support project involving "Mentor Mothers". We used of the Edinburgh Postnatal Depression Scale (EPDS) to screen women for depression during the third trimester of pregnancy, using a cut-off score of ≥13 to indicate depression. We also collected demographic and socioeconomic factors, and assessed the association of these factors with EPDS score using logistic regression models. A total of 1038 pregnant women were screened over a period of 9 months. Almost a quarter (22.7 %) had EPDS scores ≥13 and 41.2 % were HIV positive. A fifth, 17.5 % were teenagers and 73.7 % were unemployed. Depression was not associated with HIV status, age or employment status. However, women with multiple socioeconomic stressors were found to be more likely to score highly on the EPDS. Depression was common among pregnant women in the peri-urban areas of Swaziland. Screening for depression using the EPDS is feasible and can be included in the community health worker standard tool box as a way to improve early detection of depression and to highlight the importance of maternal mental health as a core public health concern.
    Full-text · Article · Mar 2016 · Journal of Community Health
Show more

Similar publications

Discover more