the course of postpartum depression: a review of ......the children.7,8 indeed, mothers suffering...
TRANSCRIPT
REVIEW
The Course of Postpartum Depression: A Review ofLongitudinal Studies
From tLeuvenogy, U
Origin29 Marevised
Corresof LeuvEmail:
questioHarvarby clickwebsitedirectly
© 2014
DOI: 1
Harvar
Nicole Vliegen, PhD,* Sara Casalin, PhD,* and Patrick Luyten, PhD
Learning Objectives: After participating in this educational activity, the physician should be better able to1. Identify the risk factors associated with persistence of postpartum depression.2. Evaluate the limitations of the literature.3. Determine the implications of the findings on women with postpartum depression and their children.This article aims to critically review studies published between 1985 and 2012 concerning the course of postpartumdepression (PPD), as well as factors implicated in PPD with a chronic course. We provide a systematic, qualitative re-view of studies on the course of PPD, following PRISMA guidelines. The results show that although the majority ofwomen recover from PPD, it becomes chronic in a relatively large subgroup of women. Several studies have identifiedrisk factors predicting a chronic course of PPD. This review also emphasizes and discusses important conceptual andmethodological limitations in existing research, which preclude drawing strong conclusions. Finally, the implicationsof these findings and suggestions for future research and clinical intervention are outlined.
Keywords: chronic depression, course and severity, longitudinal studies, postpartum depression
Alarge body of research has documented the associa-tion between maternal depression and adverseoutcomes in offspring (for a review, see Downey
& Coyne1 and Weissman et al.2). Several studies suggestthat these adverse effects are especially pronounced inmothers who suffer from chronic depression.3–5 Further-more, maternal chronic depression may be especially harm-ful in infancy since children are exposed to maternaldisengagement and negative affect during their earliestmonths of life, a sensitive developmental period; this ex-perience carries the risk of impairing their cognitive andverbal abilities and school readiness, and of leading to psy-chological problems.6 It has also been shown that the sever-ity and chronicity of maternal depression, rather than itsdiagnosis per se, are related to negative outcomes inthe children.7,8 Indeed, mothers suffering from chronic
he Faculty of Psychology and Educational Services, University of; Research Department of Clinical, Educational and Health Psychol-niversity College London (Dr. Luyten).
al manuscript received 3 February 2013; revised manuscript receivedy 2013, accepted for publication subject to revision 24 July 2013;manuscripts received 15 September 2013 and 13 October 2013.
pondence: Sara Casalin, PhD, Department of Psychology, Universityen, Tiensestraat 102 bus 3720, 3000 Leuven, [email protected]
Harvard Review of Psychiatry offers CME for readers who completens about featured articles. Questions can be accessed from thed Review of Psychiatry website (www.harvardreviewofpsychiatry.org)ing the CME tab. Please read the featured article and then log into thefor this educational offering. If you are already online, click here to goto the CME page for further information.
President and Fellows of Harvard College
0.1097/HRP.0000000000000013
d Review of Psychiatry
Copyright @ 2014 President and Fellows of Harvard Colle
depression, compared to those with intermittent, episodic,or transient depression, have been shown to be less posi-tive, sensitive, and engaged with their infants.4,6,9 It is clearthat more knowledge about the course of postpartum de-pression is urgently needed. Mothers’ depression after thebirth of their children may have significant, prolonged neg-ative consequences not just for the mothers themselves butalso for the children.
Although pregnancy is typically associated with ex-pectations of new motherhood as a happy time, for asubstantial number of mothers the postpartum period isprimarily associated with feelings of sadness, fear, andloneliness. Between 50% and 80% of new mothers experi-ence an episode of the “baby blues,” characterized by mildbut unmistakable sadness, exhaustion, unstable mood, andfear.10 For mothers with full-blown postpartum depres-sion (PPD), however, this period is even more difficult. IntheDiagnostic and Statistical Manual of Mental Disorders(DSM-IV),11 PPD is defined as a current, or the most re-cent, episode of major depressive disorder (MDD) or of bi-polar I or bipolar II disorder, if the episode has an onsetwithin four weeks postpartum. Based on a meta-analysisof 59 studies using both self-report questionnaires anddiagnostic interviews, O’Hara and Swain12 estimated theprevalence of PPD to be 13%. They also found that theprevalence of PPD was slightly, although significantly,higher when assessed with self-report questionnaires (14%)compared to diagnostic interviews (12%). A more recentreview by Gavin and colleagues13 excluded studies using
* The first two authors contributed equally to this article.
www.harvardreviewofpsychiatry.org 1
ge. Unauthorized reproduction of this article is prohibited.
N. Vliegen et al.
only self-report measures of PPD and found the prevalenceof major depression during the first three months postpar-tum to be 7.1%. During this early postpartum period,mothers are at heightened risk for depression,14,15 andthe number of depressive episodes can be twice as high asduring other periods in a woman’s life.16 During the firstthree months postpartum, parents are also at increased riskof hospital admission for mental disorders compared tononparents of the same age,17 with the highest risk of admis-sion between 10 and 19 days postpartum.17 This pattern,however, changes with increasing age: the risk of admissionwas found to be higher among nonparents, compared toage-matched parents, starting at about 30 years of age.
Because PPD has significant consequences for the baby,the depressed mother, and the early relationship betweenmother and child, knowledge about prolonged changes inthe mental health of mothers with PPD may not only im-prove our understanding of the course of PPD but also in-form prevention and intervention strategies.18
Hence, this article systematically reviews longitudinalstudies that have examined the course of PPD, followingthe guidelines of the PRISMA statement.19 It addresses thefollowing questions: (1) How does PPD evolve beyond thepostpartum period? Has empirical research identified dif-ferent trajectories in processes of recovery? (2) Does the lit-erature identify factors influencing the course of PPD?Which factors have been studied, and what are the con-clusions that can be drawn from this body of literature? Thediscussion of the main results is followed by a broad dis-cussion of the limitations of existing research that affect theinterpretability and generalizability of research findings.Issues discussed are the conceptualization of PPD and its se-verity and chronicity; postpartum onset of depression versusprevious vulnerability; the instruments used to assess PPDand its course; the samples investigated; and the role oftreatment.
METHODSFor this review, empirical studies published in peer-reviewedjournals in English between 1 January 1985 and 1 August2012 were retrieved using several search engines (PubMed,PsycINFO, Lime, Google Scholar, and LibriSource) and thefollowing combinations of subject headings: longitudinalresearch, follow-up, postpartum, postnatal, depression, de-pressive symptoms, depressive symptomatology, course,and chronic depression. The data search revealed 64 records.Studies were then screened according to five inclusion/exclusion criteria: (1) The study investigated PPD withinthe first three months after giving birth, since diagnosticcriteria specify that the onset of PPD should be within fourweeks postpartum and since PPD most frequently occurs inthe first three months postpartum.20 Studies that assessedPPD later than three months after childbirth were includedonly if they focused on PPD during the first three monthsafter birth—for example, by assessing symptoms at some
2 www.harvardreviewofpsychiatry.org
Copyright @ 2014 President and Fellows of Harvard College. U
point during the first three months postpartum using a psy-chiatric interview (see below). (2) The study had a pro-spective, longitudinal design that included at least a secondmeasure of depressive symptomatology beyond the post-partum period (e.g., at 3 months). (3) The study presentedquantitative data for the PPD group (e.g., mean scores,correlations, prevalences) on clearly defined time pointsfrom childbirth. (4) Studies that did not differentiate betweenmothers with and without PPD were not considered fur-ther for review. (5) Studies on postpartum psychosis orbipolar affective disorders were excluded.
After two of the authors (NV and SC) screened for theseinclusion and exclusion criteria, 23 longitudinal studieswere identified for this review. A detailed list of studiesexcluded from the review can be obtained from the firstauthor. Of the 23 included studies, 18 were conducted incommunity samples, and 5 used clinical samples (for anoverview, see Tables 1 & 2). Sociodemographic or other riskfactors for PPDwere investigated in 15 of the reviewed studies.
RESULTS I: LONGITUDINAL COURSE OF PPDResearch on the course of PPD has examined (1) the sever-ity of depression (mean depression rates on a depressionquestionnaire) over time, (2) the prevalence of depressionafter the postpartum period, or (3) different subgroups ofwomen with PPD. In the following sections, we presentthe results of the 23 studies that we reviewed.
Severity of Depressive SymptomatologySeven studies—four conducted in community samples23,26,28,29
and three in clinical samples37,39,42 (see Table 3)—investigatedmean depression levels in the postnatal period. All the studiesfound decreasing depression scores over time. Follow-up inthese studies ranged from 1 to 3.5 years. Two studies foundthat there was a statistically significant decrease in depres-sion between initial assessment and follow-up,29,42 whereasfour studies described a decrease but did not calculate sta-tistical significance.26,28,38,39 Holden and colleagues23 founda slight but statistically nonsignificant decrease in the levelof depression in a sample of mothers who did not receivetreatment for PPD, compared to a significant decrease in asample who were treated.
In four studies, the mean level of depression decreasedto below the cutoff value for depression defined in eachstudy;23,26,28,38 and in the other three studies, the levelof depression at follow-up remained above the cut-off.23,29,42,43 In the study by Holden and colleagues,23 de-pression levels decreased below the cutoff for the treatedmothers but not for the control sample (see Table 3). In thestudy by Milgrom and Beatrice,39 a cutoff score for theHamilton Depression Rating Scale (HDRS) was not reported.
In summary, these studies suggest that the severity ofPPD decreases over time, consistent with the 1991 resultsof Cooper and colleagues,44 who stated that a majority ofdepressive episodes after childbirth resolve within three to
Volume 22 • Number 1 • January/February 2014
nauthorized reproduction of this article is prohibited.
Table1
Reviewed
Long
itudina
lStudies
onPo
stpa
rtum
Dep
ressionin
Com
mun
itySamples
Partic
ipan
tsAssessm
enttim
epostpartum
Mea
suresofmaternal
symptomatology
Types
of
results
a
Wrate
etal.(1985),2
1
follo
w-upofCoxet
al.(1982)22
30PPD
(rec
ruite
dfrom
aco
mmunity
sample,n=105):
11PPD
16postnatal
symptoms
T1:with
in1wee
k
T2:3–
5months
T3:3ye
ars
Atalltim
epoints:SP
I(2),(3)
Holden
etal.(1989)23
50PP
D(recruite
dfrom
aco
mmun
itysample,
n=73
4):
26PPD
in8wee
ksofco
unselin
gprovided
byhea
lthvisitors
24PPD
controls(noco
unselin
g)
T1:6&
12wee
ks
T2:25wee
ks
Atalltim
epoints:EP
DS(cutoff≥12),SP
I(1),(2)
Cam
pbell&
Coh
n(199
7),24ba
sed
onCam
pbelle
tal.(199
5)4
130subjects(rec
ruite
dfrom
aco
mmunity
sample,n=2760):
67PPD
63nondep
ressed
mothers
T1:2months
T2:4months
T3:6months
T4:9months
T5:12months
T6:18months
T7:24months
Atalltim
epoints:CES
-D,RDC,SA
DS
(2),(3),
(4)
Viin
amäkiet
al.(1997)25
39with
“minorpsych
iatric
disorder”(rec
ruite
dfrom
aco
mmunity
sample,
n=139;maternity
centerforroutin
ehea
lthch
ecku
p)
T1:4–
8wee
ks
T2:2ye
ars
Atalltim
epoints:GHQ,ZSD
S(2),(3),
(4)
NICHD
EarlyChild
Care
Resea
rchNetwork
(1999)6
552PPD
(rec
ruite
dfrom
aco
mmunity
sample,
n=1215;U.S.hospita
ls)
T1:1mth
T2:6months
T3:15months
T4:24months
T5:36months
Atalltim
epoints:CES-D
(cutoff>16)
(2),(3)
Edhborg
etal.(2000)26
22PPD
(rec
ruite
dfrom
aco
mmunity
sample,n=326)
T1:2months
T2:1ye
ar
Atalltim
epoints:EP
DS(cutoff≥12)
AtT1:ICQ
(both
paren
ts)
AtT2:EM
Q/EFQ
(both
paren
ts)
(1),(2),
(4)
Zelko
witz
&Mile
t(2001)27
20PPD-affe
cted
mothers(from
sample
ofwomen
with
postpartum
psych
iatric
disorder,n=48)
50co
ntrols
T1:2months
T2:6months
Atalltim
epoints:EP
DS(cutoff≥10),LIFE
,PSI,
SCID
,SC
L-90-R
(2),(4)
Contin
ued
onnextpage
Course of Postpartum Depression
Harvard Review of Psychiatry www.harvardreviewofpsychiatry.org 3
Copyright @ 2014 President and Fellows of Harvard College. Unauthorized reproduction of this article is prohibited.
Table1
Con
tinue
d
Partic
ipan
tsAssessm
enttim
epostpartum
Mea
suresofmaternal
symptomatology
Types
of
results
a
Bee
ghly
etal.(2002)28
48PPD
(CES
-D>16;recruite
dfrom
aco
mmunity
cohort,n=106)
57nondep
ressed
mothers
(CES
-D=2–
12)
T1:2months
T2:3months
T3:6months
T4:12months
Atalltim
epoints:CES
-D(1),(2),
(3),(4)
Cam
pbelle
tal.(2004),9
based
on
NICHD
(1999)6
385
PPD
(rec
ruite
dfrom
aco
mmunity
sample,n=1077;U.S.hospita
ls)
T1:1month
T2:6months
T3:15months
T4:24months
T5:36months
Atalltim
epoints:CES
-D(cutoff≥16)
(2),(3),
(4)
Horowitz
&Goodman
(2004)29
62PPD
(rec
ruite
dfrom
aco
mmunity
sample,n=563)
T1:2–
4wee
ks
T2:4–
8wee
ks
T3:10–
14wee
ks
T4:14–
18wee
ks
T5:2ye
ars
AtT1:EP
DS(cutoff≥10),PSI
Atother
timepoints:BDI-II(cutoff≥14)
(1),(2),
(4)
Seim
yret
al.(2004)30
54
PPD
(rec
ruite
dfrom
aco
mmunity
sample,n=272)
T1:2months
T2:1ye
ar
Atalltim
epoints:EP
DS(cutoff≥9/10)
(2),(4)
Cam
pbelletal.(200
7),31follo
w-up
ofNICHD
(199
9)6
215PPD
(recruite
dfrom
aco
mmunity
sample,n=1261)
T1–
T5
T6:4.5
years
T7:7ye
ars
Atalltim
epoints:CES
-D(cutoff≥16)
(3)
Ashman
etal.(2008)3
159PPD
(rec
ruite
dfrom
hospita
ls,
clinics,new
spap
ers,etc.)
T1:14months
T2:24months
T3:3.5
years
T4:4.5
years
T5:6.5
years
Atalltim
epoints:C
ES-D
(cutoff>16),LIFE,SCID
Family
contextual
risk
factors:co
mposite
scores
ofDAS,
FES,
LES,
PSI,SS
Q
(3),(4)
Klie
ret
al.(2008)32
14PP
D(recruitedfro
macommun
itysample,
n=10
5;Vienn
a)T1:at
birth
T2:6months
T3:18months
AtT1:structuredpsych
iatric
interview,EP
I(risk
index
,crea
tedbysumming7risk
variab
les)
AtT2&
T3:EP
DS(cutoff≥10/11)
(4)
Blabe
yet
al.(20
09)33
104
PPD
(rec
ruite
dfrom
aco
mmunity
sample,n=444;Alaska)
T1:3.6
months
T2:2ye
arslater
AtT
1:PR
AMS(tw
oDSM
-IVde
pressive
symptom
s&risk
factorsforch
ronicde
pression
)
AtT2:CUBS
(2),(4)
Contin
ued
onnextpage
N. Vliegen et al.
4 www.harvardreviewofpsychiatry.org Volume 22 • Number 1 • January/February 2014
Copyright @ 2014 President and Fellows of Harvard College. Unauthorized reproduction of this article is prohibited.
Table1
Con
tinue
d
Partic
ipan
tsAssessm
enttim
epostpartum
Mea
suresofmaternal
symptomatology
Types
of
results
a
Cam
pbellet
al.(2009),3
4based
onNICHD
(1999)6
270PPD
(rec
ruite
dfrom
aco
mmunity
sample,n=1357)
T1–
T7:at
1month
and
then
everyyear
T8:9ye
ars
T9:11
years
T10:12ye
ars
Atalltim
epoints:CES-D
(cutoff≥16)
(3)
Howellet
al.(2009)35
215PPD
(rec
ruite
dfrom
aco
mmunity
sample,
n=563;En
glish-orSp
anish-spea
king
patientsin
anurban
U.S.tertiary
care
acad
emic
med
ical
center)
T1:2wee
ks
T2:6months
AtT1:trigge
rs&
buffe
rsofdep
ression
Atalltim
epoints:PHQ-2
(2),(3),
(4)
Glavinet
al.(2010)36
186PPD
(rec
ruite
dfrom
aco
mmunity
sample,n=2247):
128in
supportiveco
unselin
g
58in
trea
tmen
t-as-usual
condition
T1:6wee
ks
T2:3months
T3:6months
T4:12months
Atalltim
epoints:EP
DS(cutoff≥10)
(2)
aFo
urtypes
ofresults
aregive
n:(1)m
eandep
ressionleve
ls,(2)p
reva
lence
ofd
epressionafterthepostpartum
period,(3)subgroupsofd
epressed
mothers,an
d(4)influ
encing
factorswith
regard
toch
ronicity.
BDI,Bec
kDep
ressionInventory;B
DI-II,
Bec
kDep
ressionInventory,S
econ
dEd
ition;C
ES-D
,Cen
terforEp
idem
iologica
lStudiesDep
ressionSc
ale;
CUBS,
Child
hoodUnde
rstandingBeh
aviors
Survey;DAS,
Dya
dicAdjustmen
tSc
ale;
EMQ/EFQ
,Ex
perience
ofMotherhood/Fatherho
odQuestio
nnaire;EP
DS,
Edinbu
rghPostnatal
Dep
ressionSc
ale;EP
I,Ex
perience
ofPregn
ancy
Inter-
view
;FE
S,Fa
mily
Environmen
tSca
le;G
HQ,G
eneral
Hea
lthQuestio
nnaire;ICQ,InfantC
harac
teristic
Que
stionna
ire;
LES,
Life
Experience
sSu
rvey;L
IFE,
Longitudinal
Interval
Follo
w-upEval-
uation;PHQ-2,PatientHea
lthQuestio
nnaire;PPD,motherswith
postpartum
dep
ression;PRAMS,
Pregn
ancy
RiskAssessm
entMonito
ringSy
stem
;PSI,Pa
rentin
gStress
Index
;RDC,Resea
rch
Diagn
osticCriteria;SA
DS-L,Sc
hed
uleforA
ffectiveDisordersan
dSc
hizop
hrenia–Life
time;SC
ID,Structured
Clin
icalInterviewforD
SM-III-R;SCL-90-R,Sym
ptom
Chec
klist9
0–R;SPI,Stan
dard-
ized
Psych
iatric
Interview;SSQ,So
cial
Suppo
rtQue
stionnaire;ZSD
S,ZungSelf-RatingDep
ressionScale.
Course of Postpartum Depression
Harvard Review of Psychiatry www.harvardreviewofpsychiatry.org 5
Copyright @ 2014 President and Fellows of Harvard College. Unauthorized reproduction of this article is prohibited.
Table2
Reviewed
Long
itudina
lStudies
onPo
stpa
rtum
Dep
ressionin
Clin
ical
Samples
Partic
ipan
tsAssessm
enttim
epostpartum
Mea
suresofmaternal
symptomatology
Types
of
results
a
Buist(199
8)37(T1)
Buist&Janson
(200
1)38(T2)
45PPD
inpatients(23mothersreported
ahistory
ofch
ildhoodsexu
alab
use)
T1:when
child
was,onav
erag
e,3.9
monthsold
(betwee
n2.2
and
6.9
monthsold)
T2:when
child
was,onav
erag
e,36.8
monthsold
(betwee
n31an
d39monthsold)
AtT
1:fullpsychiatricinterview,B
DI
(cutoff≥
23),DAS,HDRS(cutoff
≥15
),MMI,NPI,P
BI,PS
E,PS
I,SSQ,STA
I
AtT2:BDI,DAS,
HDRS,
MMI,
PSE
,PSI,SS
Q,STAI
(1),(2),
(4)
Milg
rom
&Bea
trice
(2003)39
41PPD
inpatients
47nondep
ressed
mothers
T1:3months
T2:24months
Atalltim
epo
ints:B
DI(cu
toff≥17
),EP
DS
(cutoff>
12),HDRS
(1),(2)
McM
ahonet
al.
(2005)40
62PPD
recruite
dfrom
100mothers
admitted
toaparen
t-crafthospita
lfor
awee
k
T1:4months
T2:12months
Atalltim
epo
ints:A
SQ,C
ES-D
(cutoff>
16),
CIDI,DAS,DSQ
,PBI
(2),(4)
Cornish
etal.(20
08),4
1
follo
w-upof
McM
ahon
etal.(20
05)40
77PPD
recruite
dfrom
111primiparous
mothersin
a5-day
residen
tialprogram
forparen
tingproblems
T1:4months
T2:12months
T3:15months
AtT1&
T2:CID
I
Atalltim
epoints:CES-D
(cufoff≥16)
(2),(3),
(4)
Vliegenetal.(20
10)42
Vlie
genet
al.(2013)43
41PPD
inpatients
T1:d
uringpo
stpa
rtum
hospita
lization
T2:onav
erag
e,3.5
yearsafterT1
AtT1&
T2:BDI-II,
DEQ
AtT2:lifehistory
calendar
(1),(2),
(3),(4)
aFo
urtypes
ofresults
aregive
n:(1)m
eandep
ressionleve
ls,(2)p
reva
lence
ofd
epressionafterthepostpartum
period,(3)subgroupsofd
epressed
mothers,an
d(4)influ
encing
factorswith
regard
toch
ronicity.
ASQ
,Atta
chmen
tStyleQue
stionnaire
;BDI,Bec
kDep
ressionInventory;BDI-II,
BeckDep
ressionInventory,Se
cond
Edition;CES
-D,Cen
terforEp
idem
iologicalStudiesDep
ressionScale;
CID
I,Composite
International
Diagn
ostic
Interview;D
AS,
Dyadic
Adjustmen
tScale;D
EQ,D
epressiveExpe
rien
cesQuestio
nnaire;D
SQ,Defen
seStyleQuestio
nnaire;E
PDS,
Edinbu
rghPostnatal
De-
pressionSc
ale;
HDRS,
Ham
iltonDep
ressionRatingSc
ale;
MMI,MonashMother-Infan
tInteractionSc
ale;
NPI,Neo
natal
Perce
ptionInve
ntory
II;PBI,Paren
talB
ond
ingInstrumen
t;PPD,p
ostpa
r-tum
dep
ression;PSE,Paren
tingSe
lf-Effic
acySc
ale;
PSI,Paren
tingStress
Index
;SS
Q,So
cial
Suppo
rtQuestio
nnaire;STAI,State-TraitAnxietyInve
ntory.
N. Vliegen et al.
6 www.harvardreviewofpsychiatry.org Volume 22 • Number 1 • January/February 2014
Copyright @ 2014 President and Fellows of Harvard College. Unauthorized reproduction of this article is prohibited.
Table3
MeanLevelsof
Postpa
rtum
Dep
ressiveSymptom
atolog
yat
Initial
Assessm
enta
ndFo
llow-upPo
intsin
Com
mun
itySamples
andClin
ical
Samples
Meanlevelsofdep
ressivesymptomology
Other
conclusions
Com
mun
itysamples
Holden
etal.
(1989)23,a
Trea
tedmothers:
EPSD
mea
n=16at
T1;EP
SDmea
n=10.5
atT2(noSD
give
n)(p
<.001)
SPIsymptomsmea
n=25.5
atT1;SP
Isymptomsmea
n=14at
T2(noSD
give
n)
(p<.001)
Sign
ifica
ntch
ange
fortrea
tedmothers;
EPSD
mea
ndep
ression
leve
ldec
reased
toleve
lbelow
thecu
toffscore
(=12)
Controlgroup:
EPSD
mea
n=15.5
atT1;EP
SDmea
n=12at
T2(N
S)
SPIsymptomsmea
n=24at
T1;SP
Isymptomsmea
n=23at
T2(N
S)
Small/n
onsign
ificantch
ange
fornontreateddep
ressed
mothers;
EPSD
mea
ndep
ressionleve
lremained
above
thecu
toff
score
(=12)
Edhborg
etal.
(2000)26
EPSD
mea
n=15.5
(SD
=3.4,range
=13–
24)at
T1;EP
SDmea
n=7.6
(SD
=4.6)at
T2
EPSD
mea
ndep
ressionleve
ldec
reased
from
above
tobelow
the
cutoffscore
(=12)
Bee
ghly
etal.
(2002)28
CES
-Dmea
n=20.4
(SD
=5.2)at
T1;CES
-Dmea
n=13.1
(SD
=8.2)at
T3;CES
-Dmea
n=11
.4(SD
=8.2)at
T4
CES
-Dmea
ndep
ressionleve
ldec
reased
from
above
tobelow
thecu
toffscore
(=16)
Horowitz
&Goodman
(2004)29,a
EPSD
mea
n=14.12(SD
=7.34)a
tT2;E
PSD
mea
n=10.75(SD
=6.84)a
tT3;E
PSD
mea
n=10.65(SD
=6.87)at
T4;EP
SDmea
n=11
.57(SD
=9.49)at
T5
Sign
ifica
ntdec
reasebetwee
n4–
8wee
ks&
10–
12wee
ks(pairw
iseco
mparisons)
EPSD
mea
ndep
ressionleve
ldec
reased
butrem
ained
above
the
cutoffscore
(=10)
Clin
ical
samples
Buist(1998)37
(T1)
Buist&
Janson
(2001)38(T2)
Dep
ressed
motherswith
history
ofsexu
alab
use:
HDRSmea
n=18.7
(SD
=4)at
T1;HDRSmea
n=9.4
(SD
=6.7)at
T2
BDImea
n=36(SD
=9.3)at
T1;BDImea
n=13.2
(SD
=9.6)at
T2
SASmea
n=63.3
(SD
=11
.4)at
T1;SA
Smea
n=47.9
(SD
=16.1)at
T2
Dec
reased
tounder
thecu
toffscore
(15forHDRS;
23forBDI);
smallerdec
reasein
BDIdep
ressionleve
lin
PPD
mothers
Dep
ressed
motherswith
outhistory
ofsexu
alab
use
(n=22):
HDRSmea
n=18.7
(SD
=3.5)at
T1;HDRSmea
n=3.8
(SD
=3.8)at
T2
BDImea
n=30.1
(SD
=9.3)at
T1;BDImea
n=7.8
(SD
=6)at
T2
SASmea
n=55.7
(SD
=15.3)at
T1;SA
Smea
n=37.0
(SD
=13.1)at
T2
BDImea
ndep
ressionleve
ldec
reased
inPPD
mothersfrom
above
tobelow
thecu
toffscore
(15forHDRS;
23forBDI)
Milg
rom
&Bea
trice
(2003)39
Dep
ressed
sample:HDRSmea
n=13.3
(SD
=7.5)at
T1;HDRSmea
n=6.8
(SD
=6.8)at
T2
Nondep
ressed
sample:HDRSmea
n=3.2
(SD
=2.9)at
T1;HDRS
mea
n=2.6
(SD
=2.7)at
T2
HDRSmea
ndep
ressionleve
ldec
reased
Vlie
genet
al.
(2010)42,a
Vlie
genet
al.
(2013)43,a
BDImea
n=24.06(SD
=13.30)at
T1;BDImea
n=14.28(SD
=13.75)
atT2(t=4.63;p<.001)
BDI-IImea
ndep
ressionleve
ldec
reased
sign
ifica
ntly
from
highly
above
toslightly
above
thecu
toffscore
(=13)
aStudiesthat
calculatedthech
ange
inmea
ndep
ressionleve
love
rtim
e.
BDI,Bec
kDep
ressionInventory;B
DI-II,
BeckDep
ressionInventory,S
econd
Edition;C
ES-D
,Cen
terfor
Epidem
iologicalStudies
Dep
ressionScale;EP
SD,Edinb
urgh
Postna
talD
epressionScale;HDRS,
Ham
ilton
DepressionRatingScale;NS,no
nsignifican
t;PP
D,p
ostpartum
depression
;SAS,Sp
ielbergerAnxietyScale,statesubscale;SD,stand
arddeviation;
SPI,Stan
dardized
Psychiatric
Interview.
Course of Postpartum Depression
Harvard Review of Psychiatry www.harvardreviewofpsychiatry.org 7
Copyright @ 2014 President and Fellows of Harvard College. Unauthorized reproduction of this article is prohibited.
Table 4
Prevalence of Postpartum Depression Mothers That Meet Criteria for a Later Episode of Depression in CommunitySamples and Clinical Samples
3–4 months 6 months 9 months 12 months 18 months 24 months 3–3.5 years
Community samples
Wrate et al. (1985)21 62%
Holden et al. (1989)23 62%
Campbell & Cohn (1997)24 48% 30% 25% 24% 18% 13%
Viinamäki et al. (1997)25 31%
NICHD (1999)6 17%
Edhborg et al. (2000)26 6%
Zelkowitz & Milet (2001)27 62%
Beeghly et al. (2002)28 31% 35% 31%
Campbell et al. (2004)9 23%
Horowitz&Goodman (2004)29 27% 31%
Seimyr et al. (2004)30 39%
Blabey et al. (2009)33 46%
Howell et al. (2009)35 36%
Glavin et al. (2010)36 31.7% 27.8% 20.2%
Clinical samples
Buist (1998),37 Buist & Janson(2001)38
58%
Milgrom & Beatrice (2003)39 25%
McMahon et al. (2005)40 60%
Cornish et al. (2008)41 49%
Vliegen et al. (2010)42 39%
N. Vliegen et al.
six months. Depression levels do not always decrease to be-low the cutoff, however, and the decrease is not always sta-tistically significant. In addition, the standard deviationsare relatively high, indicating that change in depressionlevels within a sample is highly variable, suggesting that de-pressed mothers cannot be considered a homogeneousgroup (as discussed later in this review).
Prevalence of Depression After the Postpartum PeriodAll but three studies3,4,32 report prevalences at differentfollow-up time points, with various intervals (see Table 4).
Studies in community samples suggest that 27% to 48%(median = 31.3%) of mothers with PPD still met criteria formajor depression at 3 to 4 months postpartum (n = 4 stud-ies); 30% to 62% (median = 35.5%) at 6 months (n = 6studies); 25% to 31% (median = 28%) at 9 months (n =2 studies); 6% to 39% (median = 23.1%) at 12 months(n = 4 studies); 18% at 18 months (n = 1 study); 13% to46% (median = 31%) at 2 years (n = 4 studies); and 17%
8 www.harvardreviewofpsychiatry.org
Copyright @ 2014 President and Fellows of Harvard College. U
to 62% (median = 38.5%) at 3 years (n = 2 studies). Over-all, these findings suggest that at any time point between4 months and 3 years postpartum, about 30% of mothersdiagnosed with PPD still meet criteria for depression.
In clinical samples, as might be expected, these figuresare even higher: 49% to 60% of mothers with PPD continueto satisfy criteria for major depression at 12 to 15 monthspostpartum (n = 2 studies); 25% do so at 24 months postpar-tum (n = 1 study); and 39% to 58% of mothers still havemajor depression at 3 to 3.5 years postpartum (n = 2 studies).
In sum, these results indicate that about 30% of PPD-affected mothers in community samples and about 50%in clinical samples remain depressed throughout and be-yond the first postnatal year.
Subgroups of Depressed MothersTwelve studies (ten conducted in community samples andtwo in clinical samples) focused on subgroups of depressedmothers (see Table 5).
Volume 22 • Number 1 • January/February 2014
nauthorized reproduction of this article is prohibited.
Table5
Subg
roup
sof
PPD
Mothe
rsTh
atMet
CriteriaforaLaterEp
isod
eof
Dep
ressionin
Com
mun
itySamples
andClin
ical
Samples
Results
Com
mun
itysamples
Wrate
etal.(1985),2
1follo
w-upofCoxet
al.(1982)22
Seve
rity
ofdep
ressionat
baseline:
11(12%)seve
rely
dep
ressed
mothers(PPD)
16(17%)mild
lydep
ressed
mothers(PS)
Courseofdep
ression:
12(40%,a11
%b)c
hronically
dep
ressed
:n=2seve
rely
dep
ressed
(PPD);n=10mild
lydep
ressed
(PS)
15(16%
b)late
dep
ressed
:suffe
red≥
1dep
ressiveep
isodes
afterpostpartum
period
49(54%
b)nev
erdep
ressed
Cam
pbell&
Cohn(1997),2
4based
on
Cam
pbellet
al.(1995)4
Courseofdep
ression:
20(30%)ch
ronically
dep
ressed
(SADS≥
18)
27(41%)women
with
subclinical
symptoms(SADS≤
16)
19(29%)dep
ressionin
remission(SADS≤
14)
Viin
amäkiet
al.(1997)25
12(31%
,a9%
b)chron
icallyde
pressed:ZSD
Smean=41
.9(SD=5.2)atT1
;ZSD
Smean=44
.6(SD=9.5)atT2
27(69%,a19%
b)remitted
:ZSD
Smea
n=40.4
(SD
=5.2)at
T1;ZSD
Smea
n=34.4
(SD
=6.5)at
T2
100(72%
b)nev
erdep
ressed
:ZSD
Smea
n=32.9
(SD
=4.9)at
T1;ZSD
Smea
n=30.8
(SD
=6.2)at
T2
NICHD
EarlyChild
CareResea
rchNetwork
(1999)6
92(17%,a7.6%
b)ch
ronically
dep
ressed
:CES
-D>16,4or5tim
es
663(54.5%
b)nev
erdep
ressed
:CES
-D<16at
alltim
epoints
460(37.9%
b)sometim
esdep
ressed
:CES
-D>16,1to
3tim
es
Bee
ghly
etal.(2002)28
22(46%
a)ch
ronically
dep
ressed
:CES
-Dmea
n>12at
4assessmen
ttim
es
26(54%
a)remissiongroup
57(controlsample):CES
-Dmea
n≤
6.2
(SD
=5.0)at
4assessmen
ttim
es
Cam
pbellet
al.(2004),9
based
on
NICHD
(1999)6
Seve
rity
andco
urseofdep
ression(fu
rther
distin
ctionwith
inthe“sometim
esdep
ressed
”sample
ofthe
NICDH
study):
198(51%,a18.4%
b)remitted
mothers:CES
-D≥
16at
1,6,or15months
87(23%,a8.1%
b)ch
ronically
dep
ressed
:CES
-D≥
16,at
least3tim
epoints
100(26%,a9.3%
b)interm
itten
tlydep
ressed
:CES
-D≥16ataminim
um
of2
timepoints,sep
arated
bya
periodwith
CES
-D<16
Non-PPD
mothers:
594(55.1%
b)nev
erdep
ressed
:CES
-D<16at
alltim
epoints
98(9.1%
b)late
dep
ressed
:CES
-D≥
16at
24or36months
Contin
ued
onnextpage
Course of Postpartum Depression
Harvard Review of Psychiatry www.harvardreviewofpsychiatry.org 9
Copyright @ 2014 President and Fellows of Harvard College. Unauthorized reproduction of this article is prohibited.
Table5
Con
tinue
d
Results
Com
mun
itySa
mples
Cam
pbellet
al.(2007),3
1based
on
NICHD
(1999)6
Severity
&co
urseofdep
ression(fu
rther
distin
ctionwith
inthe“sometim
esdep
ressed
”sample
ofthe
NICHD
study):
31(14%,a2.5%
b)high–
chronic
mea
nCES
-Dfrom
32.8
(T1)to
27(T7)
73(34%,a6.2%
b)moderate–
increa
sing:
mea
nCES
-D=14.9
(T1);mea
nCES
-D=25.8
(T7)
71(33%,a5.6%
b)high–
dec
reasing:
mea
nCES
-Dfrom
25.5
(T1)to
13.2
(T7)
40(19%,a3.6%
b)interm
itten
t:marke
dva
riationsin
mea
nCES
-Dove
rtim
e
577(45.6%
b)low–stable:CES
-Dmea
nfrom
6.0
(T1)to
4.2
(T7)
469(36.4%
b)moderate–
stab
le(subclinical):CES
-Dmea
nfrom
12.2
(T1)to
10.5
(T7)
Ashman
etal.(2008)3
11(8%
a)ch
ronic
seve
redep
ression:av
erag
eCES
-Dscoresin
theclinical
range
&thehighestfreq
uen
cyofmajordep
ressiondiagn
osesat
alltim
epoints
40(30%
a)d
ecreasingdep
ressivesymptomsove
rtim
e:from
highto
moderate;from
majortosubthreshold
dep
ressiondiagn
oses
82(62%
a)stab
lemild
dep
ression,av
erag
eCES
-Dscoresbelow
clinical
range
:co
nsisten
tlylow
leve
lof
dep
ressiondiagn
osesthrough
outthestudy
Cam
pbellet
al.(2009),3
4based
on
NICHD
(1999)6
Severity
&co
urseofdep
ression(fu
rther
distin
ctionwith
inthe“sometim
esdep
ressed
”sample
ofthe
NICHD
study):
66(24%,a4.7%
b)ch
ronically
dep
ressed
:CES
-D≥
16at
alltim
epoints
59(22%
,a5.1%
b )early
anddecreasing:C
ES-D
≥16
mostofthe
time,with
adecreasetosubclinicallevelat24mon
ths
145(54%
,a10
.9%
b)m
oderatelyelevated
:CES-D
≥16
,with
ade
crease
tosubc
linicallevelsat6&15
mon
ths
416(30.8%
b)stab
lesubclinical:CES
-Dalway
s<16,but>nev
erdep
ressed
671(48.5%
b)nev
erdep
ressed
:CES
-D<16at
alltim
epoints
Howellet
al.(2009)35
PPD
sample:
78(36%,a14%
b)still
dep
ressed
atT2
137(64%,a24%
b)remissiongroup
Non-PPD
sample:
54(10%
b)late
onset(dep
ressed
atT2)
294(52%
b)nev
erdep
ressed
Clin
ical
samples
Cornishet
al.(2008)41(T3ofthesamesample
asdescribed
by
McM
ahonet
al.(2005))4
038(49%,a34%
b)ch
ronically
dep
ressed
:CES
-Dmea
n=18.0
(SD
=10.6)at
T1;CES
-Dmea
n=22.3
(SD
=10.1)at
T2;CES
-Dmea
n=22.4
(SD
=10)at
T3
39(51%,a35%
b)b
rief
dep
ression:C
ES-D
mea
n=11
.0(SD
=6.2)a
tT1;C
ES-D
mea
n=7.3
(SD
=4.7)a
tT2;CES
-Dmea
n=6.9
(SD
=4.4)at
T3
35(31%
b)n
on-dep
ressed
:CES
-Dmea
n=6.4
(SD=4.1)atT
1;C
ES-D
mea
n=5.4
(SD
=3.4)atT
2;C
ES-D
mea
n=6.3
(SD
=4.0)at
T3
Contin
ued
onnextpage
N. Vliegen et al.
10 www.harvardreviewofpsychiatry.org Volume 22 • Number 1 • January/February 2014
Copyright @ 2014 President and Fellows of Harvard College. Unauthorized reproduction of this article is prohibited.
Table5
Con
tinue
d
Results
Clin
ical
samples
Vlie
genet
al.(2010)42
Vlie
genet
al.(2013)43
16(39%
a)ch
ronically
dep
ressed
:BDImea
n=29.76(SD
=23.64)at
T1;BDImea
n=28.81
(SD
=11
.43)at
T2
25(61%
a)remitted
:BDImea
n=20.77(SD
=15.54)at
T1;BDImea
n=5.60(SD
=4.43)at
T2
aThepercentageiscalculatedonPPD
sample
only.
bThe
percentage
iscalculatedontotalsample
(norm
alco
ntrolsincluded
).BDI,BeckDep
ressionInve
ntory;C
ES-D
,Cen
terforEp
idem
iologica
lStudiesDep
ressionSc
ale;
PPD,p
ostpa
rtum
dep
ression;P
S,postna
talsym
ptoms;PSI,P
aren
tingStress
Index
;SADS,
Sched
ule
forAffe
ctiveDisordersan
dSchizop
hrenia;SD
,stan
darddeviatio
n;ZSD
S,ZungSelf-RatingDep
ressionScale.
Course of Postpartum Depression
Harvard Review of Psychiatry
Copyright @ 2014 President and Fellows of Harvard Colle
Five of the studies specifically focused on the course ofthe depression. All five studies described two PPD groups.Howell and colleagues35 found a remission group reportingdepressive symptoms at baseline but not at 6 months, andan always-depressed group reporting depressive symp-toms both at baseline and at 6 months. A one-year28 anda two-year25 follow-up study also found improved andchronically depressed mothers. In a 15-month follow-upstudy of a clinical sample, Cornish and colleagues41 identi-fied mothers who were depressed for a short period—thatis, for no longer than one year postpartum—and thosewho were chronically depressed—that is, still depressed at15 months postpartum. In a sample of inpatient motherswith PPD, Vliegen and colleagues42,43 found remitted andchronic groups of mothers, with the latter being moder-ately to severely depressed at follow-up 3.5 years afterbaseline. Taking these results as a whole, all five studiesfound a group of mothers whose depression remitted, com-prising 51% to 64% of the sample of initially depressedmothers, and a second group who were chronically de-pressed, representing 36% to 49% of the sample.
Seven other studies distinguished subgroups on the ba-sis of both the course and severity of PPD. An early studyby Wrate and colleagues21 followed up a sample firstdescribed by Cox and colleagues.22 Initially, Cox andcolleagues22 delineated two groups of depressed mothersin this sample, one severely depressed and one with mildpostnatal symptoms. Of the total sample of 27 depressedmothers, 44% had subsequent depressive episodes. Wrateand colleagues21 found, quite surprisingly, that the groupwith mild postnatal symptoms had a higher risk (62%) ofdeveloping chronic depression three years later than theseverely depressed group (18%). Other studies includedseverity of depression but took it into account only atthe follow-up points. Campbell and colleagues,4,24 in atwo-year follow-up study, found three subgroups: remit-ted (29%), subclinical (41%), and chronic (30%). At twomonths postpartum, these three subgroups were equallysymptomatic and differed significantly from controls, butnot from each other, on total symptom ratings. By fourmonths, the subclinical and chronically depressed subgroupsshowed a significant and substantial decrease in symptomsrelative to their initial high levels. Only the remitted sub-group, however, had dropped to the level of the control groupon overall symptom ratings. Furthermore, only the chronicsubgroup continued to show statistically elevated levels ofdepression throughout the follow-up period, despite somevariability in symptom severity from one assessment tothe next.
In a subsequent study, Campbell and colleagues9 re-analyzed findings from the National Institute of ChildHealth and Human Development study.6 Based on as-sessments three years postpartum, they found five sub-groups, of which only two fell within the rubric of PPD:(1) early depressed (i.e., until 6 or 15 months; 18.4%
www.harvardreviewofpsychiatry.org 11
ge. Unauthorized reproduction of this article is prohibited.
N. Vliegen et al.
of the total sample, or 69% of those with PPD), and(2) chronically depressed (i.e., for a period of 3 years;8.1% of the total, or 30% of those with PPD). The othersubgroups were (3) never depressed (55.1% of the total),(4) late depressed (i.e., depressed since two or three yearsafter childbirth; 9.1% of the total), and (5) intermittentlydepressed (i.e., experienced a minimum of two episodesof depression, but not necessarily in the postpartum period;9.3% of the total).
At seven-year follow-up of the same sample, Campbelland colleagues31 found six subgroups of depressedmothers, three of which can be considered to have PPD:(1) moderate–increasing (6.2% of the total communitysample, or 34% of those with PPD), (2) high–decreasing(i.e., partially remitted; 5.6% of the total, or 33% of thosewith PPD), and (3) high–chronic (i.e., remaining highly de-pressed; 2.5% of the total, or 14% of those with PPD).The three other trajectories that they found were (4) low–stable (i.e., very low symptomatology; 45.6% of the total),(5) moderate–stable, or subclinically depressed (36.4% ofthe total), and (6) intermittent (i.e., marked variations ofdepressed versus nondepressed status, but not necessarilyPPD; 3.6% of the total). At 12-year follow-up, Campbelland colleagues34 found evidence for five subgroups, threewith PPD: (1) early and decreasing (i.e., partial remissionsince 2 years postpartum; 5.1% of the total sample, or22% of those with PPD), (2) moderately elevated (i.e.,early partial remission; 10.9% of the total, or 54%of those with PPD), and (3) chronically depressed (4.7%of the total, or 24% of those with PPD). Others were(4) never depressed (48.5% of the total) and (5) stable sub-clinically depressed (30.8% of the total).
Ashman and colleagues3 identified three subgroups ofdepressed mothers in a follow-up study from childbirth toage 7: (1) decreasing depressive symptoms (30%), (2) sta-ble mild depression (i.e., consistent low level of depression;62%), and (3) chronic severe depression (8%).
To summarize, despite variability in the number ofsubgroups found by different studies, all have consistentlyreported a chronically depressed group—that is, PPDmothers who continue to show clinically elevated depres-sive symptomatology, with estimates of prevalence varyingbetween 23%9 and 49%41 (median = 38%). Most studies(with the exception of Ashman et al.3 and Wrate et al.21)also identified a remitted group—that is, mothers who ex-perience an acute major depression during the first threemonths postpartum but no longer show elevated symp-toms during follow-up (from 6 months to 3.5 years). Twostudies that followed up PPD mothers over a longer period(6.5 years3 and 7 to 12 years31,34), however, did not finda fully remitted group. More specifically, a decreasing de-pression group was identified, indicating a decrease fromhigh levels of depression at the baseline measurementto persisting subclinical depression at the final follow-uppoint.3,31,34 Hence, it seems that when the depressive
12 www.harvardreviewofpsychiatry.org
Copyright @ 2014 President and Fellows of Harvard College. U
symptoms of mothers with PDD are examined for a longerfollow-up period, all mothers experience a relapse of depres-sive symptoms, although not always to a clinical level. Also,a subgroup of PPD mothers were found who showed stablesubclinical depressive symptoms throughout the study pe-riod.24,31,34 Although the authors did not specifically definethis last form as chronic, we highlight—consistent with theterminology in the PPD literature (see discussion)—that itis, in a strict sense, a chronic form of PPD. Furthermore, adistinction between high–chronic depression and so-calledmoderate–increasing depression has been made,31 with thelatter group comprising mothers who were moderately (sub-clinically) depressed during the postpartum period andshowed increasing levels of depressive symptoms that haverisen to a clinical level seven years postpartum.
Other studies have found subsamples of mothers whocould not be considered chronically depressed or remitted.These mothers are labeled as showing mild21 and stablemild3 depression.
RESULTS II: FACTORS PREDICTING THE COURSEOF PPDA considerable body of research has examined predictorsof the onset of PPD, eliciting eight factors that clearly pre-dict PPD: prenatal depression, a history of previous depres-sion, a lack of social support, life stress, child-care stress,“baby blues,” marital dissatisfaction, and prenatal anxiety(for a meta-analysis, see Beck).45 However, no one hasreviewed prospective studies that explore the risk factorsassociated with the persistence of PPD. Seventeen of the23 studies included in the present review investigatedsociodemographic or other risk factors, which we will dis-cuss in more detail below.
Sociodemographic Differences Between SubgroupsIn studies of the course of PPD, sociodemographic dif-ferences between depressed and nondepressed mothers orbetween different subgroups of depressed mothers areoften controlled for in order to minimize confoundinginfluences. In total, we identified 17 publications meet-ing the inclusion criteria of this review that examinedsociodemographic factors (see Table 6).
Of the 17 studies examining the factors that influencethe course of PPD, 4 studies either did not investigatesociodemographic differences between chronic and non-chronic PPD mothers8,21,23,25,36,37,39,42 or used matchedsamples.24
AGE Out of nine studies that investigated the impactof age, only three reported an association between thechild’s or mother’s age and chronic course of PPD.Ashman and colleagues3 found that mothers whose de-pressive symptoms declined over time had younger chil-dren than mothers with subclinical depression, although nodifference was found in the chronic depressive group.
Volume 22 • Number 1 • January/February 2014
nauthorized reproduction of this article is prohibited.
Table6
Influ
encing
Factorswith
Regardto
Chron
icity
inPP
DMothe
rsTh
atMeetC
riteriaforaLaterEp
isod
eof
Dep
ressionin
Com
mun
itySamples
and
Clin
ical
Samples
Results
relatin
gto
dem
ograp
hic
variab
les
Results
relatin
gto
other
risk
factors
Com
mun
itysamples
Cam
pbell&
Cohn(1997)24
Nosociodem
ograp
hic
factors
wereinve
stigated
(bec
ause
ofmatch
edsamples)
Sign
ifica
ntdifferen
ceswerefoundwith
rega
rdto
spousalsupport
(F(3,84)=5.38;p<.002)
Nosign
ifica
ntdifferen
ceswerefoundwith
rega
rdto
familial
or
personal
history
ofdep
ression
Viin
amäkiet
al.(1997)25
Sign
ifica
ntdifferen
ceswerefoundwith
rega
rdto
poorfin
ancial
situation(p
<.05)
Sign
ifica
ntdifferen
ceswerefoundwith
rega
rdto
(1)inad
equatesocial
support
(p<.0001);(2)subjectivemen
talproblemsbefore
&duringpregn
ancy
(p<.001),&
espec
ially
2ye
arslater(p
<.0001);(3)deterioratedrelatio
nship
with
partner
before
&duringpregn
ancy
(p<.05),&
espec
ially
2ye
arslater
(p<.0001);(4)stressfullifeev
entsafterdelivery(p
<.0001)&
(5)dep
ressive
symptomsin
postpartum
period(p
<.0001)
Highbaselinedep
ressionscoresco
mbined
with
adeterioratin
gpartner
relatio
nship
predictedch
ronicity
ofmen
talhea
lthproblems(lo
gistic
regressionan
alysis)
Edhborg
etal.(2000)26
Nodifferen
ceswerefound
Sign
ifica
ntdifferen
ceswerefoundwith
rega
rdto
stress
about
motherhood1ye
arpostpartum
(F(1,90)=12.30;p<.0006)
Zelko
witz
&Mile
t(2001)27
Sign
ificant
differenc
eswerefoun
dwith
regard
to(1)m
aterna
ledu
catio
n(t=−2
.1;p
<.05)
&(2)o
ccup
ationa
lstatus(χ2=10
.3;p
<.05)
Sign
ifica
ntdifferen
ceswerefoundwith
rega
rdto
history
ofmen
tal
hea
lthproblems(in
partic
ular,majordep
ression)
Bee
ghly
etal.(2002)28
Nosign
ifica
ntdifferen
ceswerefound(dueto
strict
inclusioncrite
ria)
Cam
pbell
etal.
(2004),9
based
onNICHD
(1999)6
Sign
ificant
differenc
eswerefoun
dwith
regard
to(1)levelof
educ
ation(F(2,1
212)
=38
.77;
p<.001
),(2)incom
e-to-needs
ratios(F(2,12
10)=
41.09;p<.001
)&(3)a
partn
erpresent(χ2
(2,121
5)=57
.30;p<.001
)
Sign
ifica
ntdifferen
ceswerefoundwith
rega
rdto
perce
ived
social
support(p
<.001)
Horowitz
&Goodman
(2004)29
Parity
&inco
mewereco
rrelated
with
dep
ressionat
T5;maternal
agewas
not
Dep
ressionhistory
(β=.423;p<.0001),lower
social
support(β
=.216;
p=.035)&
higher
paren
taldistress(β
=−.221;p=.036)were
sign
ifica
ntpredictors
ofdep
ressionat
T5
Seim
yret
al.(2004)30
Sign
ifica
ntdifferen
ceswerefoundwith
rega
rdto
finan
cial
worries(p
<.0001)
Nosign
ifica
ntdifferen
ceswerefoundwith
rega
rdto
(1)losses
&strains,
(2)hea
lthproblems&
(3)partner
support
Ashman
etal.(2008)3
Sign
ifica
ntdifferen
ceswerefoundwith
rega
rdto
ageofthech
ildren
Contextual
risk
factors
wereassociated
with
maternal
dep
ression
(r=.46,p<.001)&
werestab
leove
rtim
e
Contin
ued
onnextpage
Course of Postpartum Depression
Harvard Review of Psychiatry www.harvardreviewofpsychiatry.org 13
Copyright @ 2014 President and Fellows of Harvard College. Unauthorized reproduction of this article is prohibited.
Table6
Con
tinue
d
Results
relatin
gto
dem
ograp
hic
variab
les
Results
relatin
gto
other
risk
factors
Com
mun
itysamples
Klie
ret
al.(2008)32
Nosign
ifica
ntdifferen
ceswerefound
Dep
ressivesymptomsat
T2predictedT3dep
ressivesymptomsev
enwhen
controllingfortheco
ntributio
nofmaternal
dep
ressionat
birth
(r=.44;p<.001)
Separated
psych
osocial
risk
factors
wereunrelatedto
thech
ronicity
ofPPD
Cumulativ
epsych
osocial
risk
(riskindex
)has
amoderatinginflu
ence
onthe
stab
ility
ofdep
ressivesymptomatology
:
Women
with
≥2outof7risk
factors
atbirth
weremore
likelyto
hav
estab
ledep
ressivesymptomatology
across
theinfant’s
first18monthsoflife
,from
T1to
T2
(r=0.36;p=.05)&
from
T2to
T3(r=0.55;p=.005)
Intercorrelationsam
ongthedep
ressionmea
suresat
all3tim
epointswerenot
sign
ifica
ntforwomen
with
fewer
than
2risk
factors
Blabey
etal.(2009)33
Nosign
ifica
ntdifferen
ceswerefound
Aco
ntrollingpartner
was
sign
ifica
ntly
associated
with
persisten
tPPD
(OR=6.9,
95%
CI=1.5–3
1.8;p=.014)
Howellet
al.(2009)35
Proportionally
fewer
chronically
dep
ressed
mothers
werewhite
(OR=2.90,95%
CI=1.33–
6.33)
Chronically
dep
ressed
mothersweremore
likelyto
hav
eahistory
ofdep
ression
(OR=4.24,95%
CI=1.93–9
.32)than
other
subgroups
Chronically
dep
ressed
&remitted
mothersreported
thesameleve
lsofphysical
symptomsat
T1(M
diff=−0
.31,NS),butch
ronically
dep
ressed
mothersshowed
asm
allerreductionin
leve
lsofphysical
symptoms(M
diff=−0
.78;p<.05)
Self-effic
acyincrea
sedfrom
T1to
T2forremitted
mothers(F(1,136)=60.20;
p<.01)&
remained
unch
ange
din
thech
ronic
sample
(F(1,77)=0.43,NS)
Clin
ical
samples
Buist(1998)37
Nosociodem
ograp
hic
factors
wereinve
stigated
PPD
motherswith
ahistory
ofch
ildhoodsexu
alab
use
improve
dless
compared
tonon-abusedPPD
mothers(p
<.05forboth
HDRS&
SAS)
McM
ahonet
al.(2005)40
Beingnon–Englishspea
kingwas
sign
ifica
ntly
associated
with
depression
atT2
(Waldχ2
=5.04
;df
=1;
p=.02;
OR=1.13
)
Thefollo
wingfactorsweresign
ificantlyassociated
with
depression
atT2
:(1)dep
ressionscoreatT1
(Waldχ2
=13
.59;df=1;p=.00;OR=1.35
),(2)low
materna
lcare
during
child
hood
(Waldχ2
=7.12
;df
=1;
p=.01;
OR=1.10
),(3)an
xietyov
errelatio
nships
(Waldχ2
=6.80
;df=
1;p=.01;
OR=1.18
),(4)immaturede
fensestyle
(Waldχ2
=7.38
;df=
1;p=.01;
OR=1.05
)&(5)low
maritalsatisfactionatT1
(Wald
χ2=5.04
;df=
1;p=.02;
OR=1.05
)
Cornishet
al.(2008)41
Sign
ificant
differenc
eswerefoun
dwith
regard
toma-
ternalage(F(2,1
09)=
2.50
;p<.10)
Noother
risk
factors
wereinve
stigated
Contin
ued
onnextpage
N. Vliegen et al.
14 www.harvardreviewofpsychiatry.org Volume 22 • Number 1 • January/February 2
Copyright @ 2014 President and Fellows of Harvard College. Unauthorized reproduction of this article is prohibited.
014
Table6
Con
tinue
d
Results
relatin
gto
dem
ograp
hic
variab
les
Results
relatin
gto
other
risk
factors
Clin
ical
samples
Vlie
genet
al.(2010)42
Vlie
genet
al.(2013)43
Sign
ifica
ntdifferen
ceswerefoundwith
rega
rdto
finan
cial
prob
lems(69%
vs.3
1%:χ
2=4.39
;p<.05)
Current
depressedmothe
rsrepo
rted(1)moreillne
ssof
closerelatives
(100
%versus
46%:χ2
=5.21
;p<.05),(2)more-fre
quen
tmov
esfro
mon
ereside
nceto
anothe
r(92%
vs.7
3%:χ
2=2.55
(NStre
nd))&(3)m
orefreq
uent
leavingthefather
ofthech
ild(46%
versus
35%:χ
2=.49(NStren
d))
Conce
rningpersonality
factors:self-criticism
,butnotdep
enden
cy,assessed
atT1
predictedboth
dep
ressiondiagn
osis&
leve
lsofdep
ressionat
follo
w-up
HDRS,
Ham
iltonDep
ressionRatingSc
ale;
HMRA,H
ierarchical
Multip
leReg
ressionAnalysis;N
S,nonsign
ifica
nt;P
PD,p
ostpa
rtum
dep
ression;S
AS,
Spielberge
rAnxietySc
ale,
statesubscale.
Course of Postpartum Depression
Harvard Review of Psychiatry
Copyright @ 2014 President and Fellows of Harvard Colle
Cornish and colleagues41 showed that chronically depressedmothers were slightly younger (p < .10) than mothers whohad never been depressed. Campbell and colleagues9 like-wise found that intermittently and chronically depressedmothers were significantly younger than never-depressedand remitted-depressed mothers. By comparison, seven stud-ies found no influence of the mother’s age on developmentof chronic PPD.26–29,32,33,35
SOCIOECONOMIC STATUS Of the nine studies investigatingvariables related to educational, occupational, and maritalstatus, only three found that they affected the chroniccourse of PPD. The National Institute of Child Healthand Human Development study6,9 found that chronicallydepressed women had a lower level of education and alower income-to-needs ratio than the remitted group. Atone month postpartum, mothers classified as chronicallydepressed were more likely to have no partner and no plansto go out to work. Zelkowitz and Milet27 found chroni-cally depressed mothers to be lower in occupational and ed-ucational status than the group who had remitted (p < .05).Furthermore, Horowitz and Goodman29 found a negativecorrelation between income and maternal depression scoresat four to fivemonths postpartum (p < .001). By comparison,six studies found no influence of socioeconomic status on de-velopment of chronic PPD.3,30,32,33,35,41
PARITY Of the six studies that investigated the mother’sparity, none found that it had influenced the course ofPPD.26,27,29,32,33,35
ETHNICITY Five studies investigated the possible influence ofethnicity. Howell and colleagues35 found that chronicallydepressed mothers were more frequently nonwhite thanmothers in their other three subgroups (never depressed,in remission, late onset). McMahon and colleagues40 foundthat chronically depressed mothers were more often non–English speaking. By comparison, three studies found noinfluence of ethnicity on the course of PPD.3,27,33
INFANT GENDER Of the five studies investigated the impact ofthe child’s gender, none found that it influenced the chroniccourse of PPD.26–28,35,41
Risk FactorsA number of factors in the postpartum period have beenfound to predict a chronic course of depression, and someof these factors are also known risk factors for PPD. Thesefactors can be divided into four categories, as follows:
PARTNER RELATIONSHIP AND SOCIAL SUPPORT Three studies foundthat chronically depressedmothers reported significantly less sat-isfaction with spousal support than mothers who were subclini-cally depressed and thosewhose PPD remitted.24,25,40 Two otherstudies took into account more specific aspects of the partner
www.harvardreviewofpsychiatry.org 15
ge. Unauthorized reproduction of this article is prohibited.
N. Vliegen et al.
relationship: Blabey and colleagues33 found that a controlling,threatening partner was significantly associated with persistentPPD, and Vliegen and colleagues42,43 found a (nonsignificant)trend for chronically depressed mothers to leave the fathers oftheir children more frequently. In addition, two studies foundthat chronically depressed mothers reported receiving less socialsupport thanmotherswhose depression remitted.9,29 By compar-ison, two studies did not find that social support differentiatedbetween chronic and remitted PPD mothers.30,35
HISTORY OF MENTAL HEALTH PROBLEMS OR MORE SPECIFIC CHILD-
HOOD FEATURES Three studies found that mothers withchronic PPD were more likely to have a history of depres-sion29,35 or mental health problems,25 though two studiesfound that a history of depression24 or mental healthproblems27 did not differentiate between subgroups withdifferent courses of PPD. Buist and colleagues37,38 foundthat PPD mothers with a history of childhood sexual abusefailed to improve as much as mothers who had not beenabused. McMahon and colleagues40 reported that a historyof low maternal care during childhood was significantlyassociated with chronicity of PPD.
CONTEXTUAL RISK FACTORS A chronic course of PDD wasfound to be associated with parental stress in two studies26,29
and with financial worries in three.25,26,42,43 Howell andcolleagues35 found that chronically depressed mothers showeda smaller reduction in postpartum physical symptoms overtime compared to the remitted group. This study also showedthat the incidence of health problems in the infants, such ascolic and illness, did not differ between remitted and chronicPPD mothers. Vliegen and colleagues42,43 found that chroni-cally depressed mothers reported significantly more illnessamong close relatives and a (nonsignificant) trend of more-frequent moves from one residence to another. In two stud-ies, a composite measure of stress was used. Ashman andcolleagues3 calculated a Contextual Risk Composite Mea-sure that included stress in five domains (life events, maritaldissatisfaction, parenting, family conflict, and lack of socialsupport). A chronic course of depression was associated withhigher levels of these contextual risks. Although the studydid not examine whether contextual risk predicted achronic course of PPD, these factors were stable over time,suggesting that the subgroup of mothers with chronic, severedepression experienced repeated exposure to stressful lifecircumstances. Klier and colleagues32 computed a MultipleRisk Index including seven variables (household crowding,maternal age, problems during pregnancy, weight gain ofless than 9 kg during pregnancy, dissatisfaction with socialsupport, financial resources, and the child being unwanted).Women with two or more risk factors at birth were likelyto have stable depressive symptomatology from birth to6 months postpartum, and from 6 to 18 months; by compar-ison, the correlations among the depression measures atthese three time points were not significant for women withfewer than two risk factors.
16 www.harvardreviewofpsychiatry.org
Copyright @ 2014 President and Fellows of Harvard College. U
PERSONALITY FACTORS Problems managing role demands,35
anxiety over relationships and an immature defense style,40
and high standards and excessive self-criticism42,43 havebeen found to put mothers at increased risk of developingchronic PPD.
DISCUSSION AND CONCLUSIONSThe general findings of this review of 23 longitudinal stud-ies that followed up mothers after a diagnosis of PPD canbe summarized as follows. First, the reviewed studies foundthat the severity of the depression (mean level of depres-sion) in mothers with PPD decreased at different timepoints in the postpartum period, but not always to a nonde-pressed level (below the cutoff), and that the decrease isnot always statistically significant. Second, with regard toprevalence, studies indicate that an estimated 30% of PPD-affected mothers in community samples and 50% in clinicalsamples continue to havemajor depression during their child’sfirst year of life and even beyond. Third, 12 publications dis-tinguishing between subgroups of mothers with PPD consis-tently identified (1) a chronically depressed group of motherswho continued to experience clinically elevated depressivesymptoms and (2) a remitted group of mothers who experi-enced acute major PPD but were no longer depressed atfollow-up.When severity of depressionwas also taken into ac-count, a more differentiated picture of chronic depressionemerged, with subgroups of mothers showing (1) chronic ma-jor depression, (2) stable minor depression, or (3) recurrentmajor depression without full recovery between episodes.Taken together, these studies show that mothers with PPDcannot be considered a homogeneous group. A considerableproportion of mothers who experience PPD return to the levelof depression of normal controls at four months postpartumand remain in remission throughout a long follow-up period,indicating that for these mothers, depression constitutes atime-limited problem. For a large group of mothers (38%),however, PPD is the prelude to the development of a chronicdepressive disorder. Alternatively, as discussed in more detailbelow, chronic PPD may perhaps simply be the continuationof chronic depression or dysthymia that is already present.Finally, although a minority of the studies found evidencefor sociodemographic differences (younger age of mother;lower maternal income and occupational and educationalstatus; minority/nonwhite ethnicity), results regarding thefour domains of risk factors seem to be more equivocal.More specifically, lower quality of partner relationship (withthe exception of two of nine studies that did not find socialsupport to have an influence), a history of depression, sexualabuse, or lower-quality maternal care, higher parental stressand contextual risk factors, and personality-related vulnera-bility have been consistently found to predict a chroniccourse of PPD. The incidence of infant colic and infant illnessdid not seem to differ between mothers whose PPD remittedand those whose PPD became chronic. In this regard, the
Volume 22 • Number 1 • January/February 2014
nauthorized reproduction of this article is prohibited.
Course of Postpartum Depression
composite measures used in two studies3,32 are of major in-terest. Both measures, which included five and sevendomains of risk, respectively, revealed a chronic course of de-pression to be significantly associated with higher levels ofcontextual risk. These risk factors were stable over time,suggesting repeated or chronic exposure to stressful lifecircumstances.
LimitationsThe results of this review are subject to various limitations.No “gold standard” is available for defining and measuringPPD and its course. As a consequence, studies have useddifferent criteria to define PPD and different instrumentsto measure it at diverging time points. The large differencesbetween these studies therefore need to be discussed care-fully and critically. Of special note in this context are(1) the conceptualization of PPD and its severity and chro-nicity, (2) postpartum onset versus previous vulnerability todepression, (3) the instruments used to assess PPD and itscourse, (4) the samples investigated, and (5) the role oftreatment. Finally, we propose guidelines for designingstudies whose results would be easier to compare, whichcould lead, in turn, to a more consistent body of knowledgein this domain.
ConceptualizationThe researchwe have reviewed concerning the course of PPDis hampered by the use of differing definitions that often donot take into account that depression is a heterogeneousdisorder that ranges from mild, transitory disturbances ofmood to severe, persistent depressed mood associated withsevere social and occupational impairments.7 Hence, aclear definition of key concepts is needed that considersthe severity and course of the depression as distinct and im-portant factors, particularly since studies have often con-founded them.
With regard to the severity of depression, DSM-IV dis-tinguishes between mild, moderate, and severe depression.11
Earlier classification systems based on the Research Diagnos-tic Criteria46 distinguished between definite major depres-sion (depressed mood and five symptoms), probable majordepression (depressed mood and four symptoms), and minordepression (depressed mood and three symptoms).24
In terms of the course of depression, DSM-IV describeschronic depression as depressive symptoms persisting forat least two years, and distinguishes between differenttypes of chronic depression on the basis of the severity ofsymptoms:47 (1) chronic major depression, referring to full-blown clinical depression lasting for a minimum of twoyears; (2) recurrent major depression without complete in-terepisode recovery, in which full clinical symptoms (i.e.,meeting full DSM-IV criteria) and subthreshold symptomsalternate over a period of at least two years; and (3) dysthymia,referring to a condition characterized by depressed moodthat persists for at least two years (which is conceptualized as
Harvard Review of Psychiatry
Copyright @ 2014 President and Fellows of Harvard Colle
a mild form of chronic depression). The Research DiagnosticCriteria diagnosis of intermittent depression is similar todysthymia as defined in DSM-IV.48 Dysthymia in combinationwith a major depressive episode is termed double depression.47
In the reviewed studies, operationalizations other thanthose described in the DSM-IV or Research Diagnostic Cri-teria are often used. In some studies, chronic depression isdefined in terms of long-lasting depressive symptoms thatare present from one follow-up period to the next, butnot necessarily over a two-year period.27,32 However, in-vestigation of the course of PPD by the use of differentfollow-up points makes it difficult to distinguish betweenchronic depression and recurrent, or episodic, depressionas a discrete acute episode of depression after a symptom-free period.7 Furthermore, it is hard to detect whether de-pression is actually chronic, with depressive symptomscontinuing from one follow-up period to the next, whenonly a fewmeasurements are made at points over an extendedperiod of time. If the follow-up points are spread out, it maynot be possible to identify shorter-term changes in thecourse of the PPD. In this regard, some authors stress theimportance of assessing depression in the interval betweenthe study’s established time points for measurement.3,47
PPD Onset Versus Previous VulnerabilityMost of the studies examining the course of PPD start tolook at depressive symptomatology from childbirth on,partly guided by the postpartum-onset specifier of fourweeks postpartum described in the DSM-IV-TR. In thestudies that we reviewed, none examined whether PPDamong the sample could actually be an extension of de-pression with onset during or before pregnancy. The timeof onset of “postpartum” depression has, in fact, been fre-quently debated in the literature.49,50 Some authors evenquestion whether PPD should be considered as a specificentity.51 These authors argue that being pregnant and giv-ing birth to a child are major stressors that may lead to anew, or even the first, episode of depression in womenwho are vulnerable for this disorder. Of course, even thenthe term postpartum depression seems justified, as depres-sion in early motherhood is characterized by many specificfeatures such as being preoccupied with worries aboutthe baby’s well-being and about one’s own abilities asa mother. Jones and Cantwell52 similarly recommend thecontinued use of a pregnancy and postpartum onset speci-fier in each category of mood disorder in DSM-5.53 Never-theless, studies suggesting that 11.5%50 and even up to50%49 of depressed women report depression onset beforeor during pregnancy point to the importance of assessingdepression, as well as the vulnerability to depression, be-fore the postpartum period. Congruent with these views,DSM-5 has extended the PPD specifier to include depres-sive symptoms occurring during pregnancy.
Only five studies covered in the present review investi-gated a history of depression among participants (as
www.harvardreviewofpsychiatry.org 17
ge. Unauthorized reproduction of this article is prohibited.
N. Vliegen et al.
discussed above), and three of them found such a history tobe a possible risk factor for chronic PPD.25,29,35 Conse-quently, we have only limited knowledge about how PPDmight be a recurrence of a preexisting disorder in a subsetof women. Furthermore, none of the studies specifically in-vestigated prenatal onset of depression as a possible factorheightening the risk of a chronic course of PPD. Stowe andcolleagues50 therefore correctly argue that the heterogene-ity of previous depression histories in samples of chroni-cally depressed PPD women may explain the limitedsuccess of existing efforts to identify the risk factors forchronic PPD. This situation illustrates a disadvantage of adisorder-centered approach to the study of depression—an approach that tends to focus on the onset and courseof specific disorders (such as PPD).54 By contrast, a devel-opmental, person-centered approach—one that followswomen, over time, during the transition to motherhood—may be more appropriate and may yield important newinsights into the nature of PPD.55
The current review therefore cannot draw conclusionsregarding the association between the onset (prenatal orpostpartum) of depression and its course, although thefew existing studies suggest that, at least in a subsampleof women, chronic PPD may be an extension of previousepisodes of depression and thus probably reflects a nonspe-cific vulnerability to depression.
Instruments UsedDifferent self-report instruments have been used to assessPPD symptoms, whereas various clinical interviews havebeen used to diagnose PPD as a syndrome. Studies suggestthat these various instruments and clinical assessmentsmay result in considerable differences between findings,depending on the instruments/assessments and cutoffs thatare used.
QUESTIONNAIRES In empirical research, questionnaires arefrequently used to assess the occurrence and severity ofPPD. These include the HDRS,56 Beck Depression Inven-tory (BDI, BDI-II),57,58 General Health Questionnaire,59
Center for Epidemiological Studies Depression Scale,60
Comprehensive Psychiatric Rating Scale,61 EdinburghPostnatal Depression Scale (EPDS),62 Postpartum Depres-sion Screening Scale,63 and Patient Health Questionnaire.64
These measures are also often used to define clinical de-pression using a cutoff score. In studies using the Centerfor Epidemiological Studies Depression Scale, for instance,individuals scoring 16 or higher are commonly categorizedas clinically depressed,3,9,28,31,34,41,60 and different cutoffscores are often used to distinguish between different levelsof severity. According to the scoring guidelines for the BDI,for example, a score of 11 is the lowest score for mild tomoderate depression, whereas a score of 19 indicates themoderate-to-severe range.65 In the BDI-II, total scoresranging from 0 to 13 represent “minimal” depression, 14
18 www.harvardreviewofpsychiatry.org
Copyright @ 2014 President and Fellows of Harvard College. U
to 19 “mild,” 20 to 28 “moderate,” and 29 to 63 “severe”depression.58 EPDS scores exceeding a threshold score of10 (within family medical practices when used for screen-ing purposes) and 12 (within research studies) indicate agreater likelihood of depression.62
Potential methodological problems that are associatedwith the use of cutoff scores66 are the instruments’ lowersensitivity (i.e., the number of women who are correctly di-agnosed as depressed) and specificity (i.e., the number ofwomen who are correctly diagnosed as not depressed).First, in several self-report instruments that were not specif-ically developed for assessing PPD, high scores (e.g., on theBDI) may reflect fatigue, low levels of bodily satisfaction,and loss of libido, which are all considered normal in thepostpartum period. Self-report measures such as the EPDSwere developed specifically for use within a postnatalpopulation and take into account that these symptomsare common in the postpartum period. However, a meta-analysis reports that, whereas the specificities of the BDI,EPDS, and Postpartum Depression Screening Scale areall relatively high and similar, the sensitivity of theseinstruments varies considerably, with the latter two beingmore sensitive.67
Second, different cutoff scores have been used in differ-ent studies, making it difficult to compare results acrossstudies. In defining clinical thresholds, studies have usedBDI scores of �12,68 �17,39 and �23,38 and BDI-II scoresof �13,42,43 and �14.29 For the HDRS, Buist and Janson38
used a cutoff of �15. For the EPDS, studies have used cut-off points of �9,30,68 �10,27,29,36 >12,8,18,39�12,23,26,69 and�13.70 See also in this context Matthey and colleagues’ re-view71 of several validation studies using the EPDS;62,72–74
in discussing the use of “validated” versus “unvalidated”cutoff scores, the authors suggest that using validated cut-off scores of 13 for major depression and 10 for minor de-pression represents an optimal balance between sensitivityand specificity.
A final concern with regard to cutoff values relates tocultural differences. The prevalence of PPD,68 the expres-sion of depressive symptoms,75 and the sensitivity andspecificity of measures such as the EPDS71,76 vary amongdifferent cultures. Consequently, the cutoff values validatedin particular English-speaking populations cannot neces-sarily be generalized to other countries or cultures.
CLINICAL INTERVIEWS Numerous standardized clinical inter-views have been used in studies examining the course ofPPD: Goldberg’s Standardized Psychiatric Interview,77 PresentState Examination,78 Montgomery–Asberg Depression RatingScale,79 Schedule for Affective Disorders and Schizophrenia–Lifetime,16 Diagnostic Interview Schedule III–Revised,80
Structured Clinical Interview for DSM-III-R and DSM-IV,81
and Composite International Diagnostic Interview.82
Although interviews may be more reliable than ques-tionnaires, self-report measures are easier and less costly
Volume 22 • Number 1 • January/February 2014
nauthorized reproduction of this article is prohibited.
Course of Postpartum Depression
to administer and do not require trained interviewers,which may explain their greater popularity, particularlyin studies with large samples. Affonso and colleagues68
state that questionnaire data should be used to analyzewomen’s experiences of depressive symptoms but not todiagnose clinical depression, for which clinical interviewsare necessary.
Another limitation of the reviewed studies is that theyvaried in sampling procedures, leading to samples beingrecruited from different populations of depressed indi-viduals, including hospital patients, psychiatric clinic out-patients, clients of social service agencies, and people inthe community. Some results are based on very smallsamples,27,32 making it hard to generalize findings. Further-more, the results from studies using community samplescannot simply be generalized to clinical populations. Giventhat few studies to date used clinical samples (n = 5) andthat some studies included mothers with heterogeneousdiagnoses,27 it is difficult to draw conclusions about thespecific postpartum depressed group.
The Role of TreatmentThe heterogeneity of PPD makes it difficult to assess treat-ment. In two of the studies with nonclinical samples,23,36
and in all five studies with clinical samples,37–39,42,43 themothers received some kind of support or treatment, butit is difficult to determine whether recovery occurred be-cause of treatment or other factors.
Two studies examined the effect of supportive counsel-ing on the course of PPD in nonclinical samples andshowed that the prevalence of depression at one-year fol-low up was lower in participants who received the inter-vention than in those who did not (15% vs. 32%;36
31% vs. 62%23).
RECOMMENDATIONS FOR FUTURE RESEARCH ANDCLINICAL INTERVENTIONSIn this section, we outline a number of recommendations forfuture research and intervention based on the above review.
First, future research should replicate published findingsusing larger samples, in a variety of clinical contexts, withan exclusive focus on mothers with PPD.
Second, the considerable heterogeneity in the conceptu-alization of PPD should be addressed. In order to obtain asystematic body of research on PPD and its course, studiesshould use more common, internationally agreed criteria; vali-dated cutoff scores for instruments; more consistent and paral-lel procedures; and comparable measurement time points.With regard to this last category, future studies of PPD shoulddefine clear time points for assessing patients, assess depres-sive symptoms between these measurement points, and col-lect data on subclinical depressive symptomatology. Thenew definitions and criteria in DSM-5 represent an impor-tant step in this direction.
Harvard Review of Psychiatry
Copyright @ 2014 President and Fellows of Harvard Colle
Third, congruent with a person-centered approach—andfollowing the recommendations of several researchers inthe field83 who advocate a move beyond thinking of PPDas applying solely to the postpartum period—future re-search should investigate mothers’ history of depression,as well as possible risk and protective factors, before deliv-ery. A possible method for taking into account the onsetand timing of depressive symptoms, treatment, and lifeevents is the life history calendar84,85 method, which isdesigned to collect detailed, individual-level event timingand sequencing data within an interview framework. Withthis method, it is easy to obtain information on mothers’history of depressive episodes, including any current de-pression (onset, timing, and duration), on important lifeevents, and on treatments received (see Vliegen et al.).42,43
In this regard, we recommended the use of compositemeasures that take into account several domains of con-textual risk.3,32
Fourth, this review stresses the need for culturally sensi-tive studies investigating different racial and ethnic groups,and the need to consider other demographic variables suchas socioeconomic status.13
Finally, studies should take into account individual per-sonality factors that may influence the treatment-seekingpatterns of mothers with PPD. We need more systematic,theory-driven research concerning the psychosocial42,55
and biological factors54 that may function, either sepa-rately or together, as either risk or protective factors inthe development of PPD.
With regard to clinical implications, professionals shouldbe aware that the onset of PPD is often before the post-partum period and that PPD can persist far beyond the firstyear after childbirth. Clinicians who treat PPD mothersneed to be well informed about the prognosis and the differ-ent possible trajectories of the depression. Whereas forsome mothers recover from their depressive symptomsfairly quickly, other mothers’ experience of PPD signalsthe start of what proves to be a chronic disease, with con-siderable implications for the mother, child, and broaderfamily. Consequently, programs specifically designed forPPD mothers need to include follow-up assessments inorder to identify risk factors for the depression becomingchronic. Furthermore, clinicians need to be aware thatmothers’ previous episodes of depression and that possiblecontextual factors heighten vulnerability for a chroniccourse of depression.
In sum, families with mothers suffering from PPD needthe engagement of clinicians who are sensitive to the signsthat the depression may become chronic. Moreover, giventhat parental depression has a detrimental (and poten-tially long-term) influence on the development of children,parents need ongoing contact with clinicians who aremindful of the parent-child interaction as well as of eachchild’s need for developmental support during early child-hood and beyond.
www.harvardreviewofpsychiatry.org 19
ge. Unauthorized reproduction of this article is prohibited.
N. Vliegen et al.
Declaration of interest: The authors report no conflicts ofinterest. The authors alone are responsible for the contentand writing of the article.
REFERENCES1. Downey G, Coyne J. Children of depressed parents: an
integrative review. Psychol Bull 1990;108:50–76.2. Weissman M, Wickramaratne P, Nomura Y, Warner V,
Pilowksy D, Verdeli H. Offspring of depressed parents:20 years later. Am J Psychiatry 2006;163:1001–8.
3. Ashman SB, Dawson G, Panagiotides H. Trajectories of ma-ternal depression over 7 years: relationswith child psychophysiol-ogy and behavior and role of contextual risks. Dev Psychopathol2008;20:55–77.
4. Campbell SB, Cohn JF, Meyers T. Depression in first timemothers: mother-infant interaction and depression chronicity.Dev Psychol 1995;31:349–57.
5. Moehler E, Kagan J, Parzer P, et al. Childhood behaviouralinhibition and maternal symptoms of depression. Psycho-pathology 2007;40:446–52.
6. Chronicity of maternal depressive symptoms, maternalsensitivity, and child functioning at 36 months. NICHDEarly Child Care Research Network. Dev Psychol 1999;35:1297–310.
7. Brennan PA, Hammen C, Andersen MJ, Bor W, Najman JM,Williams GM. Chronicity, severity, and timing of maternaldepressive symptoms: relationships with child outcomes atage 5. Dev Psychol 2000;36:759–66.
8. Halligan SL, Murray L, Martins C, Cooper PJ. Maternaldepression and psychiatric outcomes in adolescent off-spring: a 13-year longitudinal study. J Affect Disord 2007;97:145–54.
9. Campbell SB, Brownell CA, Hungerford A, Spieker SI, MohanR, Blessing JS. The course of maternal depressive symptomsand maternal sensitivity as predictors of attachment securityat 36 months. Dev Psychopathol 2004;16:231–52.
10. Seifer R, Dickstein S. Parental mental illness and infantdevelopment. New York, London: Guilford, 2000.
11. American Psychiatric Association. Diagnostic and statisticalmanual of mental disorders. 4th ed., text rev. Washington, DC:American Psychiatric Press, 2000.
12. O’Hara MW, Swain AM. Rates and risk of postpartumdepression—ameta-analysis. Int Rev Psychiatry 1996;8:37–55.
13. Gavin NI, Gaynes BN, Lohr KN, Meltzer-Brody S,Gartlehner G, Swinson T. Perinatal depression: a systematicreview of prevalence and incidence. Obstet Gynecol 2005;106:1071–83.
14. Cox JL, Murray D, Chapman G. A controlled study of theonset, duration and prevalence of postnatal depression. Br JPsychiatry 1993;163:27–31.
15. Vesga-Lopez O, Blanco C, Keyes K, Olfson M, Grant BF,Hasin DS. Psychiatric disorders in pregnant and post-partum women in the United States. Arch Gen Psychiatry2008;65:805–15.
16. O’Hara MW, Neunaber DJ, Zekoski EM. Prospective studyof postpartum depression: prevalence, course, and predictivefactors. J Abnorm Psychol 1984;93:158–71.
17. Munk-Olsen T, Laursen TM, Pedersen CB, Mors O, MortensenPB. New parents and mental disorders. A population-basedregister study. JAMA 2006;296:2582–9.
18. Verkerk GJM, Pop VJM, Van Son MJM, Van Heck GL.Prediction of depression in the postpartum period: a lon-gitudinal follow-up study in high-risk and low-risk women. JAffect Disord 2003;77:159–66.
20 www.harvardreviewofpsychiatry.org
Copyright @ 2014 President and Fellows of Harvard College. U
19. Moher D, Liberati A, Tetzlaff J, Altman DG; PRISMA
Group. Preferred reporting items for systematic reviews and
meta-analyses: the PRISMA statement. PLoS Med 2009;6:
e1000097.20. Beck CT. Postpartum depression: a metasynthesis. Qual
Health Res 2002;12:453–72.21. Wrate RM, Rooney AC, Thomas PF, Cox JL. Postnatal
depression and child development. A three-year follow-upstudy. Br J Psychiatry 1985;146:622–7.
22. Cox JL, Connor Y, Kendell RE. Prospective study ofthe psychiatric disorders of childbirth. Br J Psychiatry1982;140:111–7.
23. Holden JM, Sagovsky R, Cox JL. Counselling in ageneral practice setting: controlled study of health visitorintervention in treatment of postnatal depression. BMJ1989;298:223–6.
24. Campbell S, Cohn J. The timing and chronicity of post-partum depression: implications for infant development. In:Cooper P, Murray L, eds. Postpartum depression and childdevelopment. New York: Guilford, 1997:165–97.
25. Viinamaki H, Niskanen L, Pesonen P, Saarikoski S. Evolutionof postpartum mental health. J Psychosom Obstet Gynaecol1997;18:213–9.
26. EdhborgM, Seimyr L, LundhW,Widstrom AM. Fussy child—difficult parenthood? Comparisons between families witha ‘depressed’ mother and non-depressed mother 2 monthspostpartum. J Reprod Infant Psychol 2000;18:225–38.
27. Zelkowitz P, Milet TH. The course of postpartum psychiatricdisorders in women and their partners. J Nerv Ment Dis2001;189:575–82.
28. Beeghly M, Weinberg MK, Olson KL, Kernan H, Riley J,Tronick EZ. Stability and change in level of maternal de-pressive symptomatology during the first postpartum year. JAffect Disord 2002;71:169–80.
29. Horowitz JA, Goodman J. A longitudinal study of maternalpostpartum depression symptoms. Res Theory Nurs Pract2004;18:149–63.
30. Seimyr L, Edhborg M, Lundh W, Sjogren B. In the shadow ofmaternal depressed mood: experiences of parenthood duringthe first year after childbirth. J Psychosom Obstet Gynaecol2004;25:23–34.
31. Campbell S, Matestic P, von Stauffenberg C, Mohan R,Kirchner T. Trajectories of maternal depressive symptoms,maternal sensitivity, and children’s functioning at schoolentry. Dev Psychol 2007;5:1202–15.
32. Klier CM, Rosenblum KL, Zeller M, Steinhardt K, BergemannN, Muzik M. A multirisk approach to predicting chronic-ity of postpartum depression symptoms. Depress Anxiety2008;25:718–24.
33. Blabey MH, Locke ER, Goldsmith YW, Perham-Hester KA.Experience of a controlling or threatening partner amongmothers with persistent symptoms of depression. Am JObstet Gynecol 2009;201:173 e1–9.
34. Campbell SB, Morgan-Lopez AA, Cox MJ, McLoyd VC. Alatent class analysis of maternal depressive symptoms over12 years and offspring adjustment in adolescence. J AbnormPsychol 2009;118:479–93.
35. Howell EA, Mora PA, DiBonaventura MD, Leventhal H.Modifiable factors associated with changes in postpartumdepressive symptoms. Arch Womens Ment Health 2009;12:113–20.
36. Glavin K, Smith L, Sorum R, Ellefsen B. Redesigned communitypostpartum care to prevent and treat postpartum depressionin women—a one-year follow-up study. J Clin Nurs 2010;19:3051–62.
Volume 22 • Number 1 • January/February 2014
nauthorized reproduction of this article is prohibited.
Course of Postpartum Depression
37. Buist A. Childhood abuse, parenting and postpartum depres-sion. Aust N Z J Psychiatry 1998;32:479–87.
38. Buist A, Janson H. Childhood sexual abuse, parenting andpostpartum depression—a 3-year follow-up study. ChildAbuse Negl 2001;25:909–21.
39. Milgrom J, Beatrice G. Coping with the stress of motherhood:cognitive and defence style of woman with postnal depres-sion. Stress Health 2003;19:281–7.
40. McMahon C, Barnett B, Kowlenko N, Tennant C. Psycho-logical factors associated with persistent postnatal depression:past and current relationships, defence styles and the mediatingrole of an insecure attachment style. J Affect Disord 2005;84:15–24.
41. Cornish AM, McMahon C, Ungerer JA. Postnatal depressionand the quality of mother-infant interactions during thesecond year of life. Aust J Psychol 2008;60:142–51.
42. Vliegen N, Luyten P, Besser A, Casalin S, Kempke S, Tang E.Stability and change in levels of depression and personality:a follow-up study of postpartum depressed mothers hospitalizedin a mother-infant unit. J Nerv Ment Dis 2010;198:45–51.
43. Vliegen N, Casalin S, Luyten P, et al. Hospitalization-based treatment for postpartum depressed mothers and theirbabies: rationale, principles, and preliminary follow-up data.Psychiatry 2013;76:150–68.
44. Cooper PJ, Murray L, Stein A. Postnatal depression. In: SevaA, ed. European handbook of psychiatry and mental health.Barcelona: Anthropos, 1991:1255– 62.
45. Beck CT. Predictors of postpartum depression: an update.Nurs Res 2001;50:275–85.
46. Spitzer RL, Endicott J, Robins E. Research diagnosticcriteria: rationale and reliability. ArchGen Psychiatry 1978;36:773–82.
47. McCullough J, J.P., Klein DN, Borian FE, et al. Group com-parisons of DSM-IV subtypes of chronic depression: validity ofthe distinctions, part 2. J Abnorm Psychol 2003;112:614–22.
48. Frankel KA, Harmon RJ. Depressed mothers: they don’talways look as bad as they feel. J Am Acad Child AdolescPsychiatry 1996;35:289–98.
49. Yonkers KA, Ramin SM, Rush AJ, et al. Onset and persistenceof postpartum depression in an inner-city maternal healthclinic system. Am J Psychiatry 2001;158:1856–63.
50. Stowe ZN, Hostetter AL, Newport DJ. The onset of postpartumdepression: implications for clinical screening in obstetrical andprimary care. Am J Obstet Gynecol 2005;192:522–6.
51. Riecher-Rossler A, Hofecker Fallapour M. Postpartum depres-sion: dowe still need this diagnostic term? Acta Psychiatr Scand2003;108(suppl 418):51–6.
52. Jones I, Cantwell R; Nosology Working Group, RoyalCollege of Psychiatrists, Perinatal Section. The classificationof perinatal mood disorders—suggestions for DSMV andICD11. Arch Womens Ment Health 2010;13:33–6.
53. American Psychiatric Association. Diagnostic and statis-tical manual of mental disorders. 5th ed. Arlington, VA: Amer-ican Psychiatric Press, 2013.
54. Luyten P, Blatt SJ. Integrating theory-driven and empirically-derivedmodels of personality development and psychopathology: a pro-posal for DSM V. Clin Psychol Rev 2011;31:52–68.
55. Besser A, Vliegen N, Luyten P, Blatt SJ. Systematic empiricalinvestigation of vulnerability to postpartum depression froma psychodynamic perspective—commentary on issues raisedby Blum (2007). Psychoanal Psychol 2008;25:392–410.
56. Hamilton M. A rating scale for depression. J NeurolNeurosurg Psychiatry 1960;23:56–62.
57. Beck A, Ward C, Mendelson M, Mock J, Erbaugh J. Aninventory of depression. Arch Gen Psychiatry 1961;4:561–71.
Harvard Review of Psychiatry
Copyright @ 2014 President and Fellows of Harvard Colle
58. Beck AT, Steer RA, Brown GK. Beck depression inventorymanual. San Antonio, TX: Psychological, 1996.
59. Goldberg D. The detection of psychiatry illness by ques-tionnaire. London: Oxford University Press, 1972.
60. Radloff LS. The CES-D scale: a self-report depression scalefor research in the general population. Appl Psychol Meas1977;1:385–401.
61. Asberg M, Perris C, Schalling D, Sedvall G. The CPRS—development and applications of a psychiatric rating scale.Acta Psychiatr Scand 1978;(suppl 271):5–27.
62. Cox JL, Holden JM, Sagovsky R. Detection of postnataldepression. Development of the 10-item Edinburgh Post-natal Depression Scale. Br J Psychiatry 1987;150:782–6.
63. Beck CT, Gable RK. Postpartum depression screening scale(PDSS). Los Angeles, CA: Western Psychological Services, 2002.
64. Kroenke K, Spitzer RL, Williams JB. The patient healthquestionnaire-2: validity of a two-item depression screener.Med Care 2003;41:1284–92.
65. Beck AT, Steer RA, Garbin MG. Psychometric propertiesof the Beck Depression Inventory: twenty-five years ofevaluation. Clin Psychol Rev 1988;8:77–100.
66. Stewart DE, Robertson E, Dennis CL, Grace SL, Wallington TA.Postpartum depression: literature review of risk factors and inter-ventions. Toronto: UniversityHealthNetworkWomen’s HealthProgram, 2003. http://www.who.int/mental_health/prevention/suicide/lit_review_postpartum_depression.pdf
67. Gaynes BN, Gavin N, Meltzer-Brody S, et al. Perinataldepression: prevalence, screening accuracy, and screeningoutcomes. Evid Rep Technol Assess (Summ) 2005;(119):1–8.
68. Affonso DD, De AK, Horowitz JA, Mayberry LJ. Aninternational study exploring levels of postpartum de-pressive symptomatology. J Psychosom Res 2000;49:207–16.
69. de Tychey C, Spitz E, Briancon S, et al. Pre- and postnataldepression and coping: a comparative approach. J AffectDisord 2005;85:323–6.
70. Luoma I, Tamminen T, Kaukonen P, et al. Longitudinal studyof maternal depressive symptoms and child well-being. J AmAcad Child Adolesc Psychiatry 2001;40:1367–74.
71. Matthey S, Henshaw C, Elliot S, Barnett B. Variability in useof cut off scores and formats on the Edinburgh PostnatalDepression Scale: implications for clinical and researchpractice. Arch Womens Ment Health 2006;9:309–15.
72. Harris B, Huckle P, Thomas R, Johns S, Fung H. The useof rating scales to identify post-natal depression. Br JPsychiatry 1989;154:813–7.
73. Boyce P, Stubbs J, Todd A. The Edinburgh Postnatal Depres-sion Scale: validation for an Australian sample. Aust N Z JPsychiatry 1993;27:472–6.
74. Murray L, Carothers A. The validation of the EdinburghPostnatal Depression Scale on a community sample. Br JPsychiatry 1990;157:288–90.
75. Oates MR, Cox JL, Neema S, et al. Postnatal depressionacross countries and cultures: a qualitative study. Br JPsychiatry 2004;46:10–6.
76. Adouard F, Glangeaud-Freudenthal NM, Golse B. Validationof the Edinburgh Postnatal Depression Scale (EPDS) in asample of women with high-risk pregnancies in France.Arch Womens Ment Health 2005;8:89–95.
77. Goldberg D, Cooper B, Eastwood MR, Kedward HB,Shepherd M. A standardized psychiatric interview for use incommunity surveys. Br J Prev Soc Med 1970;24:18–23.
78. Wing JK, Cooper JE, Sartorius N. The measurement andclassification of psychiatric symptoms. London: CambridgeUniversity Press, 1974.
79. Montgomery SA, Asberg M. A new depression scale designed tobe sensitive to change. Br J Psychiatry 1979;134:382–9.
www.harvardreviewofpsychiatry.org 21
ge. Unauthorized reproduction of this article is prohibited.
N. Vliegen et al.
80. Robins L, Helzer J, Cottler L, Goldring E. Diagnostic InterviewSchedule (version III, revised). St. Louis, MO: WashingtonUniversity, 1988.
81. Spitzer R, Williams J, Gibbon M, First M. Structured clinicalinterview for DSM-III-R (with psychotic screen). New York:New York State Psychiatric Institute, Biometric ResearchDepartment, 1989.
82. World Health Organization. Composite International Diag-nostic Interview (CIDI). Geneva: WHO, 1997.
22 www.harvardreviewofpsychiatry.org
Copyright @ 2014 President and Fellows of Harvard College. U
83. Austin MP. Antenatal screening and early intervention for“perinatal” distress, depression and anxiety: where to fromhere? Arch Womens Ment Health 2004;7:1–6.
84. Caspi A, Moffitt T, Thornton A, et al. The life historycalendar: a research and clinical assessment method forcollecting retrospective event-history data. Int J MethodsPsychiatr Res 1996;6:101–14.
85. Axinn W, Pearce L, Ghimire D. Innovations in life historycalendar applications. Soc Sci Res 1999;2:243–64.
Volume 22 • Number 1 • January/February 2014
nauthorized reproduction of this article is prohibited.