Predictive validation study of the Edinburgh Postnatal Depression
Scale in the first week after delivery and risk analysis for
Renaud Jardri⁎, Jerome Pelta, Michel Maron, Pierre Thomas, Pierre Delion,
Xavier Codaccioni, Michel Goudemand
Hôpital Michel Fontan, Centre Hospitalier Universitaire de Lille 59 037, Lille cedex, France
Received 16 October 2005; received in revised form 16 March 2006; accepted 17 March 2006
Available online 27 April 2006
Background: Postnatal depression is a major public health problem. The aim of this study is to validate the use of the Edinburgh
Postnatal Depression Scale (EPDS) in the early postpartum, and to identify the markers for risk of postnatal depression.
Methods: 815 women filled out an EPDS and general information questionnaire between the third and the fifth day postpartum.
The women with an EPDS score of N8 and a randomized control group from those with scores of b8 were contacted 8 weeks
postpartum. 363 women therefore had a structured diagnostic interview by telephone at 8 weeks postpartum (MINI-DSM-IV),
without knowledge of their EPDS scores, to screen for a major or minor depressive episode.
Results: The sensitivity of EPDS was measured as 0.82 [0.78–0.86], with a positivity threshold of 9.5/30. For an estimated
prevalence for all depressive episodes of 16.1%, the positive predictive value of EPDS was measured as 42.8% [39.1–46.5%].
Multivariate risk analysis using logistical regression identified the following as risk markers for postnatal depression: previous
history of depression (postnatal or other), unemployment, premature delivery or stopping breast-feeding in the first month for non-
Conclusion: The use of EPDS between the third and fifth day postpartum is valid. An EPDS score of N10 should be completed by a
clinical assessment and suitable management. The risk markers identified here are clinical indices that can be used for first-line
early screening by non-psychiatric health workers.
© 2006 Elsevier B.V. All rights reserved.
Keywords: Postnatal depression; EPDS; Postpartum; Predictive validity; MINI; Risk factor
Postnatal depression (PND) is a frequent psychiatric
disorder whose prevalence is evaluated as 12.8%
(O'Hara and Swain, 1996). It is currently the most
frequent complication following giving birth. The
development of medicine during the 20th century has
produced a large reduction in obstetrical morbidity and
mortality, but the frequency of PND has remained stable
over this period. PND has a three-fold impact (Cox and
on the cognitive and psychoaffective development of the
Journal of Affective Disorders 93 (2006) 169–176
⁎Corresponding author. Tel.: +33 3 20 44 67 47; fax: +33 3 20 44 49
E-mail address: email@example.com (R. Jardri).
0165-0327/$ - see front matter © 2006 Elsevier B.V. All rights reserved.
baby. Apart from its high prevalence, PND causes many
residual symptoms in young mothers. In fact, 60% of
clinically depressed 1 year after delivery (MacMahon et
al., 2005). The risk of depression is also increased in the
partner (Lovestine and Kumar, 1993). Mother–baby
interactions are less frequent, and their inter-dependence
and reciprocity are less marked (Campbell et al., 1995).
Babies of depressed mothers are also more likely to have
insecure attachment styles (Murray et al., 1999) and a
delay in cognitive development (Grace et al., 2003).
These risks are increased if the father also has a psycho-
pathologicaldisorder (Marmorstein etal.,2004).Inview
to develop strategies for early screening of PND.
Identification of at-risk women before delivery is an
not benefit from medical follow up, will be missed.
Moreover, the peak prevalence of PND is between 4 and
10 weeks postpartum (Cooper and Murray, 1995), but
many patients are lost to follow-up after leaving the ma-
ternity unit, this being compounded by the gradual re-
Assessment of the women after leaving the maternity
unit, therefore, encountersthe same difficultiesasduring
the antenatal period. Immediate postpartum screening in
the Maternity Unit allows the whole of the exposed
population to be assessed.
The Edinburgh Postnatal Depression Scale (EPDS), a
questionnaire with 10 items, filled out by the patient, was
developed to help screen for PND at 6 weeks postpartum
in several languages (see review, Jardri, 2004). Early use
is possible, since EPDS measurement immediately
postpartum has been shown to be independent of the
et al., 1997). Also, the accuracy over time (Hannah et al.,
1992; Dennis, 2004; Teissedre and Chabrol, 2004), and
the internal consistency of EPDS (Chabrol and Tesseidre,
2004), have already been evaluated using two measure-
ments at 1 and 6 weeks postpartum. We believe that this
psychometric tool fulfils the requirementsfor use asa test
for early screening of PND in the maternity unit. How-
ever, the studies mentioned above did not use a standard-
ized diagnostic interview.
We believe that a study is required to measure the
predictability of a postnatal depressive episode by EPDS
between the third and fifth days postpartum. The
secondary objective would be to determine the medical,
social and demographic markers, which are significantly
associated with the risk of PND in the study population.
This type of risk table for PND would be a useful com-
plement for use alongside EPDS. We feel that risk mark
ers specific to our population in the north of France
should be identified, in view of its importance (1.4% of
all births in France) and the large percentage of women
coming from lower social groups.
This naturalistic study (using standard clinical prac-
tice conditions) took place in the Jeanne de Flandre
Maternity Unit (University Hospital Centre of Lille,
France) during two periods of 2 months each, in order to
correct for possible seasonal variations (Hiltunen et al.,
2004): December 2003–January 2004 (minimal sun)
and July–August 2004 (maximal sun).
The inclusion criterion was that of giving birth in the
concerned received an information and consent form on
the second day of their long-stay in the Postnatal De-
partment. The exclusion criteria were refusal to parti-
schizophrenia, according to the DSM-IV diagnostic
criteria (A.P.A., 1996), noted during a MINI test at
generally already benefited from psychiatric treatment,
offer to depressed patients, including those with other
Between the third (D3) and fifth days (D5) postpar-
tum, the patients included in the study filled out the
EPDS and a general information questionnaire. The data
were: age, parity, employment and social status, whether
they had their own accommodation, their past medical
history, a past history of PND, or of another depressive
episode. The medical data were noted according to the
International Classification of Diseases ICD-10 (W.H.
O., 1992). A second series of data about how the preg-
together with the patient's consent form from the
computerized medical records. These data were checked
against the paper medical records. These data included
the presence of an antenatal complication, a complica-
tion during delivery or postpartum, the type of delivery,
prematurity or the need for neonatal resuscitation, and
the type of feeding.
170R. Jardri et al. / Journal of Affective Disorders 93 (2006) 169–176
The EPDS filled out between D3 and D5 was the
French version, validated at 6 weeks postpartum, with a
positivity threshold of 11.5/30 (Guedeney and Ferma-
nian, 1998). The questionnaire was collected in a sealed
envelope upon discharge from the department. All of the
studies validating EPDS found in the literature (Jardri,
2004), found positivity thresholds of more than 8.5/30.
Also, the preliminary studies of the use of EPDS in early
30 (Hannah et al., 1992; Dennis, 2004). In view of these
findings, two groups were made: the women with EPDS
scores of N8 and a control group with the same number
of patients, who were chosen randomly from all of the
since the aim of our study was to find a threshold to
predict possible depression, and this score gave us some
leeway so that as few depressed women as possible
would be missed. The women from these two groups
were contacted 8 weeks postpartum. Two doctors (JR,
PJ) telephoned the women from the two groups at their
homes without knowledge of their EPDS scores.
The following general information was obtained at
8 weeks postpartum: the health status of the baby (as it
was noted in the baby's healthcheck booklet by the
paediatrician or the family doctor) and the mother, if
breast-feeding had been stopped during the first month
for non-medical reasons, or if anti-depressants were
being taken for more than 3 weeks. A standardized
diagnostic evaluation was made during this interview,
using the French version of the Mini International
depression (Rapaport et al., 2002) which is a frequent
form of postpartum depression, particularly with less-
marked somatic symptoms (Paykel, 2002). From a
“criteriological” point of view, these minor episodes
were defined by the presence of between two and four
criteria for a major depressive episode, lasting for more
than 15 days (APA, 1996; Berle et al., 2003). The use of
MINI-DSM-IV by telephone has been validated
(Duburcq et al., 1999). The inter-judge fidelity coeffi-
cient for the two participating psychiatrists using this
questionnaire was evaluated on a preliminary sample
using Cohen's Kappa coefficient as 0.8.
The results of the diagnostic and therapeutic mea-
surements performed in our department are briefly pre-
sently here. All of the women with an EPDS score of
N8 were contacted by telephone and assessed using
MINI-DSM-IV. The patients with a positive diagnosis of
depression were given an appointment for a perinatal
consultation during the telephone interview. We did not
take account of any health problems that the baby may
have had, but they were treated by the paediatrician and
noted in the baby's healthcheck booklet.
Statistical analysis was performed using the Statisti-
cal Analysis System v.8 software. The follow-up and the
representativity of the study sample were performed
using the exact Fischer and Chi squared (χ2) tests for the
categorial variables, and the Student's t-test for the
(two-tailed). Single-factor variance analysis (ANOVA)
with an independent factor, the postpartum mood status
of the patients (not depressed, mild depression, major
depression), was performed using the EPDS scores.
When significant differences were found between the
patients in the different mood-status groups, contrast
analysis was performed using Tukey's test. The thres-
hold for positivity of EPDS, to allow the best balance
between sensitivity and specificity, was calculated using
Receiver Operation Characteristics analysis, a compar-
ison being made with the reference test, MINI-DSM-IV.
95% confidence interval (95% CI). Risk analysis was
sion analysis in case of any link between some of the
factors liable to cause confusion.
3.1. General data
992 women gave birth in the study centre during the
inclusion period. Their mean age was 28.8 years old
(SD=5.6), with a range of 16–44 years old. 44.8% of
them were primiparous and 3.6% had multiple preg-
nancy. 68.9% had normal delivery and 7.7% delivered
prematurely. Out of a total of 992 deliveries, 815 women
responded to the inclusion criteria and were accepted for
participation in the study. 57 records were later excluded
because of missing data about the main variable of
interest (EPDS). 427 women were telephoned at
8 weeks postpartum, that is 227 women with an EPDS
score of N8, and the control group of 200 women with
an EPDS score of b8 (see Methods). It was not possible
to contact 60 patients by telephone (a mean of 4
attempts). 367 had a standardized diagnostic interview.
4 women were excluded because schizophrenia was
diagnosed during the MINI-DSM-IV assessment. The
final analysis included 363 women (usable data, see
171R. Jardri et al. / Journal of Affective Disorders 93 (2006) 169–176
3.2. Representivity of the population
There was no significant difference in the medical,
social and demographic variables between the complete
and incomplete medical records for EPDS, nor between
the randomized controls and the other patients with a
score of b8 (χ2test). The social data of the patients not
included (n=177) differed from that of the study
sample: isolation (pb10−4), unemployment (pb10−4)
and no personal home (pb0.01). The patients, whom we
were able to recontact at 8 weeks postpartum (n=60),
also showed a significant difference for isolation
(pb0.05) and unemployment (pb0.05) (see Fig. 1).
3.3. Distribution of EPDS scores
The EPDS scores ranged from 0 to 27/30. The mean
EPDS score of those included in the study was 6.8±4.5.
The distribution of EPDS scores between D3 and D5,
variance uniformity (Levene's test), variance analysis
(ANOVA) and post-hoc analysis showed that the EPDS
scores of the women with a diagnosis of a major de-
a minor depressive episode, and they in turn were
depressed [F(2.363)=58.53, pb0.001)].
3.4. Calculation of the positivity score for EPDS
ROC analysis (see Fig. 3) identified a positivity
threshold of 9.5 as the best rate between the sensitivity
and specificity of EPDS for screening for major and
minor depression. The sensitivity of EPDS was 0.82
area under the curve was measured as 0.79 [0.74–0.84].
For a threshold of 11.5/30 (Guedeney and Fermanian,
1998), the sensitivity of early EPDS was 0.70 [0.65–
0.75] and its specificity was 0.74 [0.70–0.78] for major
10.5/30 if only major depressive episodes are included,
Fig. 1. A flow diagram of the women's outcomes in the study.
172R. Jardri et al. / Journal of Affective Disorders 93 (2006) 169–176
with a sensitivity of 0.84 [0.80–0.88] and a specificityof
0.71 [0.66–0.76]. All of the values are shown in Table 1.
3.5. Prevalence of postnatal depression and predictive
values for EPDS
The prevalence of a major depressive episode at
8 weeks postpartum, according to the DSM-IV criteria,
was 8.7%, 95% CI [6.6–10.8] (n=61), and for minor
depression 7.4%, 95% CI [4.7–10.1] (n=51). For major
and minor depressive episodes, the prevalence of
postnatal depression at 8 weeks postpartum was
estimated as 16.1% [13.4–18.8] (n=112). It should be
noted that no seasonal effect was found as shown by the
absence of a significant difference between the preva-
lence measurements from the two study periods. With a
prevalence of postnatal depression of 16.1%, the
positive predictive value (PPV) of EPDS was measured
as 42.8% [39.1–46.5%] and the negative predictive
value (NPV) of EPDS as 95.2% [93.6–96.8%].
3.6. Analysis of the risk markers
The first step of the logistic regression was univariate
analysis of all the medical, social and demographic data.
Seven factors were found to be linked to depression at
8 weeks postpartum (major or minor depressive syn-
drome according to MINI). In the second step, these
factors were included successively in the logistic regres-
sion multivariate analysis. Only 5 of these factors were
identified as risk markers for postnatal depression:
unemployment (adjusted odds ratio (aOR)=2.8, 95%
confidence interval (95% CI) [1.1–4.9]), a positive
history of postnatal depression (aOR=4.3, 95% CI
[1.7–10.9]), a positive history of depression (aOR=4.4,
95% CI [2.2–9.0]), a preterm delivery of b37 weeks
(aOR=4.5, 95% CI [1.4–14.6]) and stopping breast
Fig 3. EPDS 3–5 days postpartum R.O.C. curve (n=363).
Results of R.O.C. analysis for EPDS in the first week postpartum,
validated against the DSM-IV depression criteria (n=363; 1/2 95% CI:
semi 95% confidence interval)
Threshold score Sensitivity 1/2 95% CI Specificity 1/2 95% CI
Fig. 2. EPDS score 3–5 days postpartum according to the depressive
state. ND: not depressed; aD: attenuated depression; MD: major
Logistic regression model of the most significant risk factors of
postnatal depression (n=363)
Predictors for major and minor depressive disorderaOR CI
Positive history of postnatal depression
Positive history of depression
Preterm delivery b37 weeks
Stopping breast-feeding during 1st month (non-
173R. Jardri et al. / Journal of Affective Disorders 93 (2006) 169–176
feeding during first month for non-medical reasons
(aOR=5.4, 95% CI [1.4–20.3]) (see Table 2). For the
other 2 markers, the study numbers were too small to
judge if they were linked with postnatal depression:
multiple pregnancies (aOR=3.3, 95% CI [0.6–16.0])
and postpartum complications (aOR=2.3, 95% CI [0.8–
6.3]). The explained variance for the 5 risk markers
identified was estimated as 72%.
This study confirms that it is valid to use EPDS
between D3 and D5 postpartum with a positivity
threshold lowered to 9.5/30, as opposed to its classical
use at 6 weeks postpartum. This result is compatible with
the data in the literature (e.g. Jardri, 2004). Hannah et al.
noted a significant correlation between the mother's
both measured using EPDS (r=0.60, pb0.001, n=217).
more chance of having a score of N9/30 at 6 weeks
postpartum (Hannah et al., 1992). Tesseidre and Chabrol
also assessed the fidelity with time of EPDS, using a
larger number of women (r=0.59, pb0.0001, n=1154).
They measured EPDS twice and classified it into 4
categories [0/1–9/10–12/N13], on the third day postpar-
tum and between 4 and 6 weeks postpartum (Teissedre
and Chabrol, 2004). They suggested that the detection
score should be reduced from 11.5/30 to 10.5/30, when it
is used early on, at 3 days postpartum. Finally, Dennis
found the EPDS scores to be stable with time, between
4 weeks (r=0.72, pb0.001, n=535) and the third at
8weeks (r=0.65, pb0.001,n=498). Shediscoveredthat
the women who obtained a score of N9/30 in the first
week postpartum had a 30.3 times greater risk of having
depressive symptoms at 4 weeks postpartum (95%
CI=17.5–42.3), and a 19.1 greater risk at 8 weeks
postpartum (95% CI=11.0–32.9). The depressive symp-
toms included in the studies by Hannah, Tesseidre and
Dennis, however, have not been assessed using a
standardized tool for diagnosing depression. Since the
diagnosis of PND was confirmed in our study at 8 weeks
were tested by these two psychometric tests (EPDS and
MINI-DSM-IV) and since a distinction was made
between major and minor depression, this allows us to
validate early use of EPDS and shows that the detection
threshold needs to be reduced with respect to its classical
use at 6 weeks postpartum.
For this threshold of 9.5/30, the NPVis high (95.2%),
important for a screening tool. On the other hand, the
PPV is low (42.8%). The fact that depressive episodes
were diagnosed using the DSM-IV criteria may explain
these values, since they are less sensitive to atypical
postnatal depression. We tried to correct this effect by
taking into account minor depressive episodes, never-
theless defined from a strictly “criteriological” point of
view. It should be noted that the positive EPDS scores
should always be followed by a complementary assess-
ment by an experienced clinician, in order to judge if
postnatal depression is indeed present. We consider that
it is important to remember that the instructions given to
the patients for filling out the questionnaire made it clear
that the answers should correspond with the 7 days pre-
ceding the assessment. In our study, this means that the
EPDS score obtained between D3 and D5 postpartum
contain information about the pre-, peri- and the imme-
diate postpartum. In the light of our results, our hypo-
thesis is that the EPDS score could therefore, at the time
of childbirth, be a measurement of the mother's own
perinatal affective state, which would be predictive of a
depressive state at 8 weeks postpartum. To avoid an ex-
cessively rigidinterpretationoftheEPDSscore, itwould
seem to be necessary to use an uncertainty interval (grey
area for scores close to the EPDS threshold), or to repeat
the scores close to the threshold. Matthey proposed the
use ofthe ReliableChange Index (RCI), inorder to seeif
a change between two measurements was significant. At
6 weeks postpartum,the RCI was calculated to be within
four points of the EPDS score (Matthey, 2004). The re-
sults of our study could be repeated by measuring EPDS
again, in order to calculate the RCI value in the early
a “naturalistic” approach, was based on the idea that
EPDS could be used in practice as a mass screening tool
for postpartum depression in the maternity unit by non-
psychiatric staff. With this in mind, we did not look for
any complex psychopathological indices, such as a
personality disorder, which could confuse non-specia-
lists. The 60 patients lost to follow up after discharge
underline the difficulty in following up women with a
low socio-economic level or who are isolated, even
though they are a population with a high frequency of
one or more of the risk markers for postnatal depression
(unemploymentb0.05, aOR=2.8 [1.1–4.9)]. The prev-
alence of postpartum depression is certainly under-
estimated in this sample. Authors like Wickberg et al.
noted that if PND was not identified using EPDS during
the screening phase, there was a risk that it would not be
found later on either (Wickberg and Hwang, 1996). We
would add that if depression is not screened for in the
Maternity Unit, irrespective of the screening programme
174R. Jardri et al. / Journal of Affective Disorders 93 (2006) 169–176
women who were lost to follow up. These results show
that these patients can be offered suitable management
and follow up by home care services (general practi-
tioners and nursery nurses from maternal and child
The results of logistic regression are compatible with
the data in the literature (Warner et al., 1996). These
markers are weakly linked with postnatal depression,
with adjusted OR ranging from 2.8 to 4.5. The antenatal
mood (Green, 1998) or the presence of a personality
disorder were not assessed according to the study's
protocol. However, the presence of a psychological
distress during pregnancy has previously been identified
as a risk marker for postnatal depression (Nielsen
Forman et al., 2000). We consider that it is important to
include this risk analysis in this validation study, since
clinical screening by EPDS can not be dissociated from
the medical history data if a usable protocol is to be
defined. The risk markers identified in this study
constitute key items for clinical screening by midwives,
nursery nurses, paediatricians and obstetricians. Socio-
economic difficulties, a previous history of mood
disorders and mother–baby interaction difficulties
(early separation due to prematurity or stopping
breast-feeding in the first month for non-medical
reasons), are indices which limit bias in filling out the
EPDS patient questionnaire. These results should be
confirmed and completed. With this in view, we are
currently assessing the quality of clinical screening by
midwives in the Maternity Unit, before and after
training to help them identify these risk markers.
EPDS is a simple, quick-to-use tool for early screen-
ing of postnatal depression in the maternity unit by using
a positivity threshold of 9.5/30 between D3 and D5
postpartum. To optimize this screening strategy, the psy-
chiatrists need to be available to provide adequate treat-
ment when necessary. Finally, the use of an integrated
medical, psychiatric and social network at the patients'
home would allow treatment and tertiary prevention to
be continued by encouraging interaction between the
parents and the baby early on.
American Psychiatric Association, 1994. Diagnostic Statistical
Manual of Mental Disorders, 4th edn. DSM-IV. American
Psychiatric Press, Washington D.C.
Austin, M.P., Lumley, J., 2003. Antenatal screening for postnatal
depression: a systematic review. Acta Psychiatr. Scand. 107,
Berle, J.O., Aarre, T.F., Mykletun, A., Dahl, A.A., Hiosten, F., 2003.
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. 76, 151–156.
Campbell, S., Cohn, J., Meyers, T., 1995. Depression in first-time
mothers. Mother infant interaction and depression chronicity. Dev.
Psychol. 31, 349–357.
Cooper, P., Murray, L., 1995. Course and recurrence of postnatal
depression. Evidence for the specificity of the diagnostic concept.
Br. J. Psychiatr. 166, 191–195.
Chabrol, H., Tesseidre, F., 2004. A study of the Edinburgh Postnatal
Depression Scale (EPDS) on 859 mothers: detection of mothers at
risk for postpartum depression. Encéphale 30, 376–381.
Cox, J.L., Holden, J.M., 2003. Perinatal Mental Health: A guide to the
Edinburgh Postnatal Depression Scale. Gaskell, Glasgow.
Cox, J.L., Holden, J.M., Sagovsky, R., 1987. Detection of postnatal
depression. Development of the 10 item Edinburgh Postnatal
Depression Scale. Br. J. Psychiatr. 150, 782–786.
Dennis, C.L., 2004. Can we identify mothers at risk for postpartum
Postnatal Depression Scale? J. Affect. Disord. 78, 163–169.
Duburcq, A., Blin, P., Charpak, Y., Blachier, C., Allicar, M.P.,
Bouhassira, M., Hergueta, T., Lecrubier, Y., 1999. Use of a
epidemiologic study: a posteriori interval validation. Rev. Épidé-
miol. Santé Publique 47, 455–463.
Grace, S., Evindar, A., Stewart, D., 2003. The effect of postpartum
depression on child cognitive development and behaviour: a
review and critical analysis of the literature. Arch. Women Ment.
Health 6, 263–274.
Green, J.M., 1998. Postnatal depression or perinatal dysphoria?
Findings from a longitudinal community-based study using the
EPDS. J. Reprod. Infant Psychol. 16, 143–155.
Guedeney, N., Fermanian, J., 1998. Validation study of the French
version of the Edinburgh Postnatal Depression Scale (EPDS): new
results about use and psychometric properties. Eur. Psychiatr. 13,
Hannah, P., Adams, D., Lee, A., Glover, V., Sandler, M., 1992. Links
between early postpartum mood and postnatal depression. Br.
J. Psychiatr. 160, 777–780.
Hiltunen, P., Jokelainen, J., Ebeling, H., Szajnberg, N., Moilanen, I.,
2004. Seasonal variation in postnatal depression. J. Affect. Disord.
Jardri, R., 2004. Postnataldepression screening: a qualitative review of
the validation studies of the Edinburgh Postnatal Depression Scale.
Devenir 16, 245–262.
Lovestine, S., Kumar, R., 1993. Postnatal psychiatric illness: the
impact on partners. Br. J. Psychiatr. 163, 210–216.
MacMahon, C., Barnett, B., Kowalenko, N., Tennant, C., 2005.
past and current relationships, defence styles and the mediating role
of insecure attachment style. J. Affect. Disord. 84, 15–24.
Marmorstein, N., Malone, S., Iacono, W., 2004. Psychiatric disorders
among offspring of depressed mothers: association with paternal
psychopathology. Am. J. Psychiatry 161, 1588–1594.
Matthey, S., 2004. Calculating clinically significant change in
postnatal depression studies using the EPDS. J. Affect. Disord.
175R. Jardri et al. / Journal of Affective Disorders 93 (2006) 169–176
Murray, L., Sinclair, D., Cooper, P., Ducournau, P., Turner, P., Stein,
A., 1999. The socio-emotional development of 5-year-old children
of postnatally depressed mothers. J. Child Psychol. Psychiatry 40,
Nielsen Forman, D., Videbech, P., Hedegaard, M., Dalby Salvig, J.,
Secher, N.J., 2000. Postpartum depression: identification of
women at risk. Br. J. Obst.Gynaecol. 107, 1210–1217.
O'Hara, M.W., Swain, A.M., 1996. Rates and risk of postpartum
depression—a meta-analysis. Int. Rev. Psychiatr. 8, 37–54.
Paykel, E.S., 2002. Mood disorders: review of current diagnostic
systems. Psychopathology 35, 94–99.
Kupfer, D.J., Frank, E., Plewes, J.M., Tollefson, G.D., Rush, A.J.,
2002. Descriptive analysis of minor depression. Am. J. Psychiatr.
Sheehan, D.V., Lecrubier, Y., Sheehan, K., Amerim, P., Janars, J.,
Weiller, E., Hergueta, T., Baker, R., Dunbar, G.C., 1998. The Mini
International Neuropsychiatric Interview (M.I.N.I.): the develop-
mentandvalidationofa structureddiagnostic psychiatricinterview
for DSM-IVand ICD-10. J. Clin. Psychiatr. 59 (suppl. 20), 22–33.
Sutter, A.L., Leroy, V., Dallay, D., Verdoux, H., Bourgeois, M., 1997.
postpartum blues and mild depressive symptomatology at day's
three to five after delivery. J.Affect. Disord. 44, 1–4.
Teissedre, F., Chabrol, H., 2004. Detecting women at risk for postnatal
depression using the EPDS at 2 to 3 days postpartum. Can.
J. Psychiatr. 49, 51–54.
Warner, R., Appleby, L., Whitton, A., Faragher, B., 1996. Demo-
graphic and Obstetric risk factors for postnatal psychiatric
morbidity. Br. J. Psychiatr. 168, 607–611.
Wickberg, B., Hwang, C., 1996. The Edinburgh Postnatal Depression
Scale: validation on a Swedish community sample. Acta Psychiatr.
Scand. 94, 181–184.
World Health Organization, 1992. International Classification of
Diseases, 10th edn. W.H.O., Geneva.
176R. Jardri et al. / Journal of Affective Disorders 93 (2006) 169–176