Article Analysis 1

REVIEW ARTICLE

Effectiveness of self-help psychological interventions for treating and preventing postpartum depression: a meta-analysis

Ping-Zhen Lin1 & Jiao-Mei Xue2 & Bei Yang1 & Meng Li1 & Feng-Lin Cao1

Received: 19 October 2017 /Accepted: 21 March 2018 /Published online: 4 April 2018 # Springer-Verlag GmbH Austria, part of Springer Nature 2018

Abstract Previous studies have reported different effect sizes for self-help interventions designed to reduce postpartum depression symp- toms; therefore, a comprehensive quantitative review of the research was required. A meta-analysis was conducted to examine the effectiveness of self-help interventions designed to treat and prevent postpartum depression, and identified nine relevant randomized controlled trials. Differences in depressive symptoms between self-help interventions and control conditions, chang- es in depressive symptoms following self-help interventions, and differences in postintervention recovery and improvement rates between self-help interventions and control conditions were assessed in separate analyses. In treatment trials, depression scores continued to decrease from baseline to posttreatment and follow-up assessment in treatment subgroups. Changes in treatment subgroups’ depression scores from baseline to postintervention assessment were greater relative to those observed in prevention subgroups. Self-help interventions produced larger overall effects on postpartum depression, relative to those observed in control conditions, in posttreatment (Hedges’ g = 0.51) and follow-up (Hedges’ g = 0.32) assessments; and self-help interventions were significantly more effective, relative to control conditions, in promoting recovery from postpartum depression. Effectiveness in preventing depression did not differ significantly between self-help interventions and control conditions.The findings suggested that self-help interventions designed to treat postpartum depression reduced levels of depressive symptoms effectively and decreased the risk of postpartum depression.

Keywords Self-help . Depression . Postpartum . Meta-analysis . Randomized controlled trial

Introduction

Postpartum depression refers to major and minor depressive episodes that occur within 12 months of parturition (Gavin et al. 2005). It is a common mental illness in women, with high prevalence rates ranging from 6.5 to 12.9% (Gavin et al. 2005). Postpartum depression exerts negative effects including in- creased obstetric complications in depressed mothers (Leung & Kaplan 2009), compromised mother-child relationships, and disturbed neurobiological, social, and cognitive develop- ment in infants (Brummelte & Galea 2016). Consequently, the development and validation of effective interventions and pre- ventative programs are crucial.

Despite observation of positive effects and improvements in depressive symptomatology with traditional treatments, such as antidepressant medication and individual psychother- apy, research has shown that low participation rates (5–14%) (Andersson et al. 2003; Andersson et al. 2006; Kelly et al. 2001) limit these interventions (Bijl & Ravelli, 2000).

* Feng-Lin Cao caofenglin2008@126.com

Ping-Zhen Lin lpz385675123@163.com

Jiao-Mei Xue xuejiaomeiyan@163.com

Bei Yang yangbeilucky@126.com

Meng Li gflemon11@163.com

1 School of Nursing, Shandong University, No. 44 Wenhua Xi Road, Jinan 250012, Shandong, People’s Republic of China

2 Society and Law School, Shandong Women’s University, Jinan, Shandong, People’s Republic of China

Archives of Women's Mental Health (2018) 21:491–503 https://doi.org/10.1007/s00737-018-0835-0

Antidepressant medication is associated with adverse preg- nancy and neonatal outcomes. Antidepressants could exert adverse effects during pregnancy and breastfeeding on both mother and child, such as preterm birth, reduction in maternal weight gain and infant birth weight, and poor neonatal out- comes (Wisner et al. 2009). Moreover, there are other disad- vantages, such as time inflexibility, the need for health profes- sionals’ participation, and unaffordability for many women (Cowpertwait & Clarke 2013; Roness et al. 2005). Therefore, self-help interventions, which involve little burden on clinical resources, few time constraints, and accessibility for a broad range of users (Beatty & Binnion 2016), have attracted increased attention.

Self-help interventions refer to psychological interventions that patients largely complete independently at home, in ac- cordance with a standardized protocol, using written material (e.g., books, booklets), CD-ROMs, DVDs, computerized soft- ware packages, and websites (Cuijpers & Schuurmans 2007). Cowpertwait & Clarke (2013) conducted a systematic review examining web-based psychological interventions for depres- sion and showed such interventions exerted a statistically sig- nificant and moderately large effect, relative to that observed in the control condition (placebo or treatment-as-usual or waitlist), and led to a significant reduction in depressive symp- toms. Moreover, another systematic review that evaluated the clinical effectiveness of cognitive behavioral therapy (CBT)- based, guided self-help interventions for anxiety and depres- sive disorders demonstrated the effectiveness of guided self- help at posttreatment assessment; however, this effectiveness was considerably diminished at follow-up and in clinically representative samples (Coull & Morris 2011). Furthermore, self-help interventions may be more effective for treatment rather than prevention of depression in the general population (Gellatly et al. 2007).

The effectiveness of a wide range of self-help interventions for postpartum depression has been demonstrated in random- ized controlled trials (RCTs). For example, behavioral activa- tion (BA) is an effective method for reducing depressive symptoms when compared to treatment as usual (TAU) (O’Mahen, 2013, O’Mahen et al. 2014). Symptoms of post- partum depression decreased more for participants in the CBT group compared to those participants in the waitlist control group (Pugh et al. 2016). In addition, self-help exercise inter- vention was more effective than wellness/support contact con- trol condition in alleviating symptoms of postpartum depres- sion (Lewis et al. 2014). However, some studies failed to demonstrate that self-help interventions were superior to usual care in the treatment or prevention of postpartum depression (Costa et al. 2009; Daley et al. 2015a; Mohammadi et al. 2015). Moreover, with the exception of a study conducted by Milgrom et al. (2016), research has consistently shown that intervention effects were not sustained during follow-up pe- riods of various durations (Costa et al. 2009; Daley et al.

2015a; Mohammadi et al. 2015; O’Mahen et al. 2014). Most studies reported that depressive symptoms decreased signifi- cantly from baseline to postintervention assessment (Costa et al. 2009; Daley et al. 2015a; King 2009; O’Mahen et al. 2014; O’Mahen et al. 2014; Pugh et al. 2016), while some others reported that symptoms decreased significantly from postintervention to follow-up assessment (O’Mahen et al. 2014; Pugh et al. 2016). In the only published RCT to com- pare the effectiveness of an Internet-based self-help interven- tion with a face-to-face control intervention in the prevention of postpartum depression, King (2009) demonstrated that re- duction of depression from pretest to posttest of the Internet- based program was equivalent to that of the face-to-face intervention.

Given that different effect sizes have been observed for various types of self-help interventions designed to reduce postpartum depression symptoms, comprehensive and quan- titative reviews of potential outcomes are required. Therefore, the primary aim of the current meta-analysis was to assess the effectiveness of a range of self-help interventions in the treat- ment and prevention of postpartum depression observed in RCTs.

We compared depressive symptom severity between a self- help intervention condition and a control condition and exam- ined the overall effect of these interventions on depressive symptoms over time. In addition, the difference in postinter- vention recovery and improvement rates between the two con- ditions was also examined.

Materials and methods

Data sources and search strategy

We conducted a meta-analysis in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (2015) statement (Moher et al. 2015) to examine the impact of self-help psychological interventions on postpartum depression. Systematic literature searches of the PubMed, Scopus, Web of Science, Cochrane Central Register of Controlled Trials, MEDLINE databases, and ProQuest Dissertation and Theses were performed to identify relevant RCTs. There were no re- strictions regarding publication date. Searches were last updated in September 31, 2017. Searches were conducted within the domains of title, abstract, and keywords. A string was used within each database: (Bself-help^ OR BInternet^ OR Bweb^ OR Bonline^ OR Bcomputer^ OR Bhome^ OR Btelephone^) AND (Bpostpartum depress*^ OR Bpostnatal depress*^ OR Bmaternal depress*^) AND (Btreatment^ OR Btherapy^ OR Bintervention^ OR Btrial^ OR Banalysis^). In addition, we per- formed manual searches of the reference lists in the identified articles and publications that cited them.

492 P.-Z. Lin et al.

Literature selection and data extraction

The eligibility criteria for study selection were as follows: Study design: Studies were eligible for inclusion if they

reported RCTs that examined self-help interventions for at least 4 weeks that involved one of the following four factors: routine clinical care/treatment-as-usual (TAU), receive no treatment, join a waiting list (WL), or participate in a face- to-face psychological intervention. Where studies reported multiple self-help intervention groups with a control group, they were recoded as multiple subgroups, and effect sizes were calculated between each intervention group and control group separately.

Participants: Included studies were based solely on women during pregnancy or the postpartum period (defined as the 12 months following the birth of a child), regardless of gender, race, nationality, or depressive status.

Interventions: Self-help interventions, as a type of psycho- logical interventions (interventions which could exert influ- ence on the mental activity, personality characteristics, or psy- chological problems of objects under the guidance of psycho- logical theory), are defined as an individual’s access to self- help materials (e.g., written material and websites) in the treat- ment of mood disorders. Maximum professional input for self- help interventions equals to half of that for the lower range in conventional therapy, in accordance with the recommenda- tions of a recent systematic review examining self-help inter- ventions for depression (Gellatly et al. 2007).

Outcomes: Measurement and reporting of postpartum de- pression scores using a validated self-report or clinician- administered measure at postintervention and follow-up as- sessment. The manuscript reported that sufficient outcome data for the calculation of effect sizes was included. When studies did not report sufficient outcome data for the compu- tation of effect sizes, authors were contacted to request addi- tional data, and this information was used to calculate effect sizes when provided.

Studies in which all female participants were classified as depressed using diagnostic criteria or symptom severity at baseline were eligible for inclusion in the meta-analysis of studies involving treatment; and studies in which not all par- ticipants fulfilled the criteria for a depressive episode or ex- hibited clinically significant depressive symptoms at baseline were eligible for inclusion in the meta-analysis of studies in- volving prevention subgroups.

Search results were not limited with respect to language, but all of the identified articles were in English. The exclusion criteria were as follows: (1) absence of a control group, (2) examination of methods other than self-help interventions, (3) irretrievable intervention results, and (4) total overlap of sam- ple and results reported in a different publication.

The data extracted from all studies were as follows: First Author, publication date, age (mean), education (%), race (%),

employment (%), marital status (%), months since postpar- tum, depression at entry, intervention type (exercise vs. stress management program vs. CBT vs. BA), control type (TAU vs. WL vs. face-to-face psychological intervention), outcome measure (the Edinburgh Postnatal Depression Scale, EPDS vs. the Beck Depression Inventory-II, BDI-II), intervention approaches (Internet-based vs. telephone-based), intervention duration (months or weeks), follow-up duration (months or weeks), and the mean and standard deviation from each study’s depressive symptomatology scale scores.

The systematic search and data extraction were performed independently by two of the researchers. The risk of bias was assessed using the Cochrane Risk of Bias Tool (Peters et al. 2015), and a third researcher checked for consistency of bias. Divergence of opinions was resolved through consultation.

Statistical analysis

Statistical analyses were performed using Stata Version 14.0. Pooled mean effect sizes were calculated (Hedges’ g) with 95% confidence intervals (CIs), using random effects models based on the assumption that included studies represented the true distribution of intervention data, to compare outcomes between self-help interventions and control conditions and assess changes in depressive symptoms for self-help interven- tions. Hedges’ g values of 0.2, 0.5, and 0.8 represent small, moderate, and large effect sizes, respectively (Cohen 1988). Odds ratios (ORs) were calculated to compare postinterven- tion recovery and improvement rates between the intervention and control conditions.

Q and I2 statistics were used to assess heterogeneity, and P values of < .100 for the Q statistic indicated high heterogene- ity levels. For the I2 statistic, scores of 25, 50, and 75% indi- cated low, moderate, and high heterogeneity levels, respec- tively (Higgins et al. 2003). As fewer than 10 studies were included in the review, we did not perform a meta-regression.

Subgroups classified according to intervention purpose, control type, intervention type, and depression measure were analyzed. Except for intervention purpose and intervention type, all the other subgroup analyses did not show significant results, which would not be displayed in the result.

In each of these analyses, outliers were identified using the sample-adjusted meta-analytic deviance (SAMD) statistic (Huffcutt & Arthur 1995). If the SAMD value was ≥ 2.58 and the scree plot suggested that the SAMD did not differ from the overall distribution, the study was retained.

Publication bias was assessed via visual examination of funnel plots, Duval and Tweedie’s (2000) trim-and-fill proce- dure, and classic fail-safe N values (Rosenthal 1979). According to Rosenthal’s (1991) recommendation, a value of 5K + 10, where K is the number of observed studies, was used as the cutoff point for an unlikely number of nonsignif- icant studies. Given the small number of studies in the meta-

Effectiveness of self-help psychological interventions for treating and preventing postpartum depression: a... 493

analysis, these tests demonstrated low sensitivity in detecting publication bias.

Assessment of study quality

The methodological quality of each study was assessed via the Delphi List Criteria (Verhagen et al. 2017). The criteria in- clude nine standard items for RCTs. Two of the nine list criteria (i.e., BWas the intervention provider blinded?^ and BWas the participant blinded?^) were not considered, as it is difficult to conduct self-help intervention trials in which par- ticipants and the intervention provider are blind to the inter- vention (Daley et al. 2015b). Therefore, we evaluated the de- sign of each study using seven criteria. One point was assigned for each criterion, providing a maximum score of 7.

Results

Study characteristics

Figure 1 shows the literature search flow chart. Of the 1921 potentially eligible articles identified via the electronic search, nine met all of the review criteria and were included in the analysis (Costa et al. 2009; Daley et al. 2015a; King 2009; Lewis et al. 2014; Milgrom et al. 2016; Mohammadi et al. 2015; O’Mahen et al. 2014; O’Mahen et al. 2014; Pugh et al. 2016) after removal of duplicates, title and abstract screening, and full-text article assessment. Most studies were journal articles, with only one doctoral thesis included (King 2009).

Study quality and sensitivity analysis

Table 1 shows the results regarding the methodological quality of the included studies. Eight studies were awarded 6 points, which indicated high methodological quality. The remaining study (King 2009) was awarded 4 points. All studies involved random sampling, specified eligibility criteria, and presented point estimates and measures of variability for the primary outcome measures.

Elimination of the data from the study that was awarded 4 points resulted in a reduction in heterogeneity in the overall depressive symptomatology analysis, from I2 = 63.2% to I2 = 57.9%. In addition, the effect sizes for the psychological interventions were maintained (Hedges’ g = 0.42; 95% CI 0.20 to 0.64) following the exclusion of this study, and the reduction in effect sizes did not reach statistical significance (P = .098). This study (King 2009) was remained in the following analysis.

Study designs and participants

The characteristics of each study are presented in Table 2. All interventions were administered after parturition, with the ex- ception of that in one study, which was implemented between 26 and 32 weeks of pregnancy (Mohammadi et al. 2015). Intervention durations ranged from 4 weeks to 6 months, and the follow-up period ranged from 3 weeks to 6 months, with the exception that three studies did not have follow-up. The study conducted by Mohammadi et al. (2015) involved two self-help intervention groups and a control group, which were recoded as two subgroups: Mohammadi 2015a (antenatal exercise intervention versus TAU) and Mohammadi 2015b

(antenatal and postnatal exercise intervention versus TAU). Effect sizes were calculated between each intervention group and the control group separately.

Six and four of the intervention subgroups received Internet- and telephone-based interventions, respectively. Control groups included TAU, WL, and face-to-face psycho- logical intervention. Intervention subgroups involved five ex- ercise subgroups, one stress management program, two CBT subgroups, and two BA subgroups. The studies included six treatment subgroups and four prevention subgroups.

Sample sizes ranged from 17 to 165. The total numbers of participants in self-help intervention and control groups were 513 and 438, respectively. Seven studies used EPDS, and two used BDI-II.

Meta-analysis findings

Postintervention postpartum depressive symptoms following self-help interventions

The pooled effect size for eight subgroups was − 1.08 (95% CI − 1.61 to − 0.55, Fig. 2), suggesting that depressive symptoms in the self-help intervention groups decreased following the interventions, with large effect sizes. In addition, the effect sizes were significantly heterogeneous (χ2 = 76.19, P < .001), indicating that potential moderators existed. One study (O’Mahen et al. 2014) reported SAMD values exceed- ing 2.58. Visual inspection of the scree plot of rank-ordered SAMD scores indicated that the value for this study was con- sistent with the overall distribution of SAMD scores. Therefore, this study was excluded from further analyses.

Postintervention effect sizes were − 0.60 (95% CI − 1.16 to − 0.05, P = .001, I2 = 80.7%) for exercise; − 0.84 (95% CI − 1.55 to − 0.14) for the stress management program; − 1.88 (95% CI: − 2.29 to − 1.46, P = .153, I2 = 51.0%) for BA; and − 1.50 (95% CI − 2.18 to − 0.82) for CBT.

The difference in intervention purposes between subgroups was significant (P = .013), and Hedges’ g value for the treat- ment subgroups (− 1.51; 95% CI − 1.90 to − 1.12, P = .004, I2 = 73.6%) was higher relative to that observed for the

494 P.-Z. Lin et al.

prevention subgroups (− 0.29; 95% CI − 0.70 to 0.12, P = .151, I2 = 47.1%). Significant heterogeneity was observed for the treatment and exercise subgroups.

In the treatment studies, the fail-safe N was 654, which exceeded the tolerance level for an unlikely number of non- significant studies (35). In the treatment and prevention stud- ies, the funnel plot and trim-and-fill procedures indicated that there was no publication bias (Fig. 3).

Comparison of postpartum depressive symptoms between postintervention and follow-up assessments for self-help interventions

The pooled effect size for six subgroups was − 0.32 (95% CI − 0.52 to − 0.12, Fig. 4), indicating that depression symptoms in the self-help intervention groups decreased between post- intervention and follow-up assessments, with small-to-

3

Full-text articles assessed for eligibility

(n=36) Full-text articles excluded:

No RCT (n=5)

Study design (n=1)

Population (n=1)

Non-self-help (n=11)

Insufficient data (n=9) Studies included in

treatment

meta-analysis

(n=6)

Studies included in

prevention

meta-analysis

(n=3)

Records after duplicates removed

(n=991)

Records identified though database searching

＆ reference review

Title and abstract screening

(n=79)

In cl

u d

ed

E li

gi b

il it

y S

cr ee

n in

g Id

en ti

fi ca

ti on

S

cr ee

n in

g

(n=1921)

Fig. 1 Study selection flow diagram

Table 1 Methodological quality of included studies

Quality characteristics Costa 2009

Daley 2015a

King 2009

Lewis 2014

Milgrom 2016

Mohammadi 2015

O’Mahen 2013

O’Mahen 2014

Pugh 2016

1. Was a method of randomization performed? Y Y Y Y Y Y Y Y Y

2. Was the treatment allocation concealed? Y Y – Y Y Y Y Y Y

3. Were the groups similar at baseline regarding the most important prognostic indicators?

Y Y Y Y Y Y Y Y Y

4. Were the eligibility criteria specified? Y Y Y Y Y Y Y Y Y

5. Was the outcome assessor blinded? – N – Y – N – – –

6. Were point estimates and measures of variability presented for the primary outcome measures?

Y Y Y Y Y Y Y Y Y

7. Did the analysis include an intention-to-treat analysis? Y Y N N Y Y Y Y Y

Total 6 6 4 6 6 6 6 6 6

Effectiveness of self-help psychological interventions for treating and preventing postpartum depression: a... 495

Ta b le 2

S el ec te d ch ar ac te ri st ic s fo r in cl u d ed

st u d ie s

S tu dy , y ea r

C o u n tr y N o f

in te rv en ti o n/

co n tr o l at

b as el in e

A g e

(m ea n ) E d uc at io n

(% )

R ac e

(% )

E m p lo y

(% )

M ar it al

st at u s

(% )

M o n th s si n ce

p o st p ar tu m

D ep re ss io n at en tr y

In te rv en ti o n /c o n tr o l

ty p e

O ut co m e

m ea su re

In te rv en ti on

ap p ro ac h es

In te rv en ti o n

d u ra ti o n

F o ll o w -u p

d u ra ti o n

C os ta et

al ., 2 00 9

C an ad a

4 6 /4 2

3 3 .5 0

– 0. 78

– –

4– 38

w ee ks

E P D S ≥ 10

E x er ci se /T A U

E P D S

T el ep h o n e

3 m o n th s

3 m o n th s

D al ey

et al ., 20 1 5 a

U K

4 7 /4 7

3 1 .7 0

– 0 .5 7

0 .5 4

0 .8 9

6 m o n th s

E P D S > 1 2

E x er ci se /T A U

E P D S

In te rn et

6 m o n th s

6 m o n th s

K in g , 2 0 0 9

A m er ic a

1 7 /2 1

3 2 .5 0

0 .9 8

0 .6 1

– 0 .6 3

12 m o n th s

W it h o u t st an d ar d

S tr es s m an ag em

en t/

fa ce -t o- fa ce

tr ea tm

en t

B D I- II

In te rn et

4 w ee ks

–

L ew

is et

al ., 2 0 1 4

A m er ic a

6 6 /6 4

3 1 .6 9

– 0. 80

0. 06

0. 83

8 w ee ks

H ad

a hi st or y of

de pr es si on

or a m at er na l fa m il y

h is to ry

o f d ep re ss io n , w er e no t

cu rr en tl y de pr es se d

E x er ci se /

T A U

E P D S

T el ep h o n e

6 m o n th s

–

M il gr om

et al ., 20 1 6

A u st ra li a

2 1 /2 2

3 1 .7 0

1 .0 0

– –

0 .8 6

12 m o n th s

E P D S = 11 – 23

C B T /T A U

B D I- II

In te rn et

6 w ee ks

3 w ee ks

M oh am

m ad i, 2 0 1 5a

Ir an

43 /4 2

25 .2 0

0. 57

– 0. 02

0. 31

26 w ee k s pr e-

2 m o n th s p o st

E P D S < 1 5

E x er ci se /T A U

E P D S

T el ep h o n e

10 -1 3 w ee ks

2 m o n th s

M oh am

m ad i, 2 0 1 5b

Ir an

42 /4 2

25 .4 0

0. 57

– 0. 06

0. 33

26 w ee k s pr e-

2 m o n th s p o st

E P D S < 1 5

E xe rc is e/

T A U

E P D S

T el ep h o n e

18 -2 1 w ee ks

2 m o n th s

O ’M

ah en , 2 0 1 3

U K

1 8 1 /1 6 2

3 2 .3 0

0 .9 8

– 0 .2 6

0 .9 2

12 m o n th s

E P D S > 1 2

B A /

T A U

E P D S

In te rn et

11 w ee k s

–

O ’M

ah en , 2 0 1 3

U K

4 1 /4 2

– 0 .9 8

0 .9 3

0 .1 5

0 .9 3

12 m o n th s

E P D S > 1 2

B A /

T A U

E P D S

In te rn et

12 w ee k s

6 m o n th s

P ug h et

al ., 2 0 16

C an ad a

2 5 /2 5

– 0. 92

0. 92

– 0 .9 6

12 m o n th s

E P D S ≥ 10

C B T /

W L

E P D S

In te rn et

7 w ee k s

1 m o n th s

N ot e. ed u ca ti o n (%

): th e p er ce n ta g e o f h ig h sc h o o l o r hi g h er ; ra ce

(% ): th e p er ce n ta ge

o f C au ca si an ; em

p lo y (%

): p er ce n ta g e o f em

p lo y ed ; m ar it al

st at u s (%

): th e p er ce n ta g e o f m ar ri ed /c o h ab it in g;

M o h am

m ad i (2 0 1 5 )a : an te n at al ex er ci se

in te rv en ti o n v er su s tr ea tm

en t- as -u su al ; M o h am

m ad i (2 0 1 5 )b : an te na ta l an d po st na ta l ex er ci se

in te rv en ti on

v er su s tr ea tm

en t- as -u su al ; T A U = tr ea tm

en t as

u su al ,

W L = w ai ti n g li st , B D I- II = B ec k D ep re ss io n In v en to ry -I I, E P D S = E d in b u rg h P o st n at al D ep re ss io n S ca le , C B T = co g n it iv e b eh av io ra l th er ap y, B A = b eh av io ra l ac ti v at io n

496 P.-Z. Lin et al.

moderate effect sizes. There was no significant heterogeneity observed for effect sizes (χ2 = 4.96, P = .421), and no studies reported SAMD values exceeding 2.58 (range − 1.71 to 1.56).

Effect sizes for depression symptoms between follow-up and postintervention assessments were − 0.21 (95% CI − 0.44 to 0.02, P = .923, I2 = 0.0%) for exercise, − 0.54 (95% CI − 1.03 to − 0.06) for BA, and − 0.93 (95% CI − 1.64 to − 0.21) for CBT.

The difference in intervention purposes between subgroups was nonsignificant (P = .510), with Hedges’ g values of − 0.38 (95% CI − 0.70 to − 0.07, P = .200, I2 = 35.4%) in treat- ment subgroups and − 0.25 (95% CI − 0.58 to 0.07, P = .790, I2 = 0.0%) in prevention subgroups. No significant heteroge- neity was observed for the subgroups.

In the treatment studies, the fail-safe N was 12, which did not exceed the tolerance level for an unlikely number of non- significant studies (30). The funnel plot and trim-and-fill pro- cedures suggested that there was no publication bias for treat- ment or prevention studies (Fig. 3).

Comparison of postintervention postpartum depressive symptoms between intervention and control conditions

A forest plot presenting the pooled between-group effect sizes for postintervention depression scores is presented in Fig. 5. The pooled postintervention between-group effect size for nine subgroups was 0.32 (95% CI 0.09 to 0.56, P = .004), indicating that self-help interventions

produced stronger overall effects relative to those of control interventions, which were small to moderate. No studies reported SAMD values exceeding 2.58 (range − 2.22 to 2.24).

Postintervention between-group effect sizes for depressive symptoms were 0.20 (95% CI − 0.06 to 0.46, P = .113, I2 = 46.4%) for exercise, − 0.35 (95% CI − 0.99 to 0.30) for stress management programs, 0.56 (95% CI 0.37 to 0.76, P = .696, I2 = 0.0%) for BA, and 1.06 (95% CI 0.42 to 1.70) for CBT.

The difference in intervention purposes between subgroups was nonsignificant (P = .063), and Hedges’ g for treatment subgroups (0.51; 95% CI 0.27 to 0.75, P = .146, I2 = 41.3%) was higher relative to that observed for prevention subgroups (0.05; 95% CI − 0.35 to 0.45, P = .032, I2 = 66.0%). Significant heterogeneity was observed in the prevention subgroups.

In the treatment studies, the fail-safe N was 64, which exceeded the tolerance level for an unlikely number of non- significant studies (35). In addition, the funnel plot was slight- ly asymmetric (Fig. 3), and trim-and-fill procedures suggested one missing study with a value to the left of the mean. The overall effect size after trim-and-fill correction was 0.45 (95% CI 0.19 to 0.70). This adjusted value suggested that if the included studies reflected publication bias, it occurred in the direction of overestimation of true effect sizes for the inter- ventions. In the prevention studies, the funnel plot showed no publication bias (Fig. 3), and trim-and-fill procedures sug- gested no missing studies.

Fig. 2 Forest plot displaying effect sizes of depression scores at post-intervention versus at baseline in self-help intervention group. Note. Mohammadi (2015)a: antenatal exercise intervention versus treatment-as- usual, Mohammadi (2015)b: antenatal and postnatal exercise intervention versus treatment-as- usual.

Effectiveness of self-help psychological interventions for treating and preventing postpartum depression: a... 497

Fig. 3 Funnel plots to assess for publication bias by relating effect sizes of the studies to standard errors

498 P.-Z. Lin et al.

Comparison of postpartum depressive symptoms between interventions and control conditions at follow-up

The pooled between-group effect size at follow-up in six sub- groups was 0.19 (95% CI − 0.05 to 0.43, see Fig. 6),

indicating that self-help interventions produced larger overall effects relative to those of control conditions, which were small to moderate. In addition, the effect sizes were signifi- cantly heterogeneous (χ2 = 7.24, P = .203). No studies report- ed SAMD values exceeding 2.58 (range − 1.79 to 1.61).

Fig. 4 Effect sizes of depression scores at follow-up versus at post- intervention in self-help intervention group. Note. Mohammadi (2015)a: antenatal exercise intervention versus treatment-as-usual, Mohammadi (2015)b: antenatal and postnatal exercise intervention versus treatment-as-usual.

Fig. 5 Forest plot displaying effect sizes of depression scores at post-intervention comparing self- help interventions with control conditions. Note. Mohammadi (2015)a: antenatal exercise intervention versus treatment-as- usual, Mohammadi (2015)b: antenatal and postnatal exercise intervention versus treatment-as- usual.

Effectiveness of self-help psychological interventions for treating and preventing postpartum depression: a... 499

Between-group effect sizes for depressive symptoms at follow-up were 0.03 (95% CI − 0.20 to 0.26, P = .981, I2 = 0.0%) for exercise; 0.48 (95% CI − 0.04 to 1.00) for BA; and 0.83 (95% CI 0.20 to 1.45) for CBT.

Treatment trials showed a trend toward larger overall ef- fects for self-help treatment, relative to those in control con- ditions, with a Hedges’ g value of 0.32 (95% CI − 0.01 to 0.65, P = .170, I2 = 40.3%). In prevention studies, the Hedges’ g value was − 0.02 (95% CI − 0.34 to 0.31, P = 0.100, I2 = 0.0%), which was nonsignificant. Subgroup anal- yses of intervention purposes showed no difference between subgroups (P = .240). No heterogeneity was observed for the subgroups.

In the treatment studies, the funnel plot showed no publication bias (Fig. 3); and trim-and-fill procedures sug- gested no missing studies. In the prevention studies, the funnel plot showed no publication bias (Fig. 3), and trim- and-fill procedures suggested that the overall effect size did not change with one missing study (Hedges’ g = − 0.02, 95% CI − 0.28 to 0.25).

Postintervention recovery and improvement rates

A forest plot presenting the pooled between-group effect sizes for postintervention recovery and improvement rates is shown in Fig. 7. Participants in seven subgroups (n = 372) recovered or showed significant improvement following self-help interventions.

One treatment study (Milgrom et al. 2016) reported SAMD values exceeding 2.58 and was excluded from subsequent analyses. Following exclusion of this outlier, the average ef- fect size was 2.50 (95% CI 1.76 to 3.55, P = .772, I2 = 0.0%), indicating that recovery rates following self-help interventions were significantly higher relative to those observed in control conditions. No significant heterogeneity was observed for the treatment studies. The fail-safe N value was 35, which exceeded the tolerance value of 30. While the funnel plot was asymmetric (Fig. 3), trim-and-fill procedures showed two missing studies with values to the left of the mean. The overall effect size following trim-and-fill correction was 2.31 (95% CI 1.53 to 3.08). This adjusted value suggested that if the included studies reflected publication bias, it occurred in the direction of overestimation of the true effect size for the interventions.

One of the prevention subgroups (Mohammadi 2015b) re- ported SAMD values exceeding 2.58 and was excluded from subsequent analyses. Excluding this outlier, the effect size for the subgroups was 1.13 (95% CI: 0.31 to 4.08), indicating that improvements in depression symptoms did not differ between the self-help prevention and control conditions in nonde- pressed women.

Discussion

This meta-analysis summarized the effects of self-help inter- ventions in the prevention and treatment of postpartum

Fig. 6 Forest plot displaying effect sizes of depression scores at follow-up comparing self-help interventions with control conditions. Note. Mohammadi (2015)a: antenatal exercise intervention versus treatment-as- usual, Mohammadi (2015)b: antenatal and postnatal exercise intervention versus treatment-as- usual.

500 P.-Z. Lin et al.

depression. The meta-analysis showed that self-help interven- tions constituted an effective method of treatment for postpar- tum depression. Specifically, (1) depression symptoms contin- ued to decrease from baseline to posttreatment and follow-up assessments in treatment subgroups. Changes in depression scores from baseline to postintervention assessment in treat- ment subgroups were greater relative to those observed in prevention subgroups. (2) Posttreatment and follow-up de- pression symptoms showed greater reductions following self-help interventions, relative to those observed in control conditions; and (3) self-help interventions were significantly more effective in aiding recovery from postpartum depression, relative to control conditions. However, their effectiveness in improvement of depression symptoms did not differ signifi- cantly from that of control conditions in prevention studies.

Depression scores continued to decrease from baseline to postintervention and follow-up assessment in treatment sub- groups, which verified the effectiveness of self-help treatment for postpartum depressive symptoms. In treatment studies, self-help interventions produced larger overall effects on post- partum depressive symptoms relative to those observed for control conditions (Hedges’ g = 0.51). In addition, effect sizes observed at follow-up assessment (Hedges’ g = 0.32) were consistent with those observed in previously published re- views examining treatments for perinatal depression, which reported an average posttreatment effect size of 0.57 for post- partum depressive symptoms (Sockol et al. 2011). Moreover, the effectiveness of self-help interventions in aiding recovery from postpartum depressive symptoms was significantly

greater relative to that observed for control conditions. As stated by Cuijpers et al. (2010), guided self-help and face-to- face treatments for depression exhibited comparable benefits at follow-up of up to 1 year. This pattern of results suggested that, similar to face-to-face treatments, self-help interventions were more effective for postpartum depressive symptoms rel- ative to TAU or WLs, and these benefits could be maintained in the long term.

Treatment effects were stronger, relative to prevention ef- fects, with respect to symptomatic improvement of postpar- tum depression from baseline to postintervention assessment. Similar results were observed in another meta-analysis exam- ining self-help intervention for depression in the general pop- ulation (Gellatly et al. 2007). Moreover, no advantages of self- help interventions for postpartum depression were observed in prevention trials. Furthermore, it is possible that depression symptom severity could have influenced effect sizes. In other words, self-help interventions could have produced a signifi- cant effect size in the treatment of women who met the criteria for depression or displayed depression scores above the cutoff point for clinically significant depressive symptoms at base- line. However, the effects of the interventions could have been limited for women with low levels of depressive symptoms (Bortolotti et al. 2008; Cuijpers, Smit, & Van 2007a).

Self-help CBTand BA produced stronger overall posttreat- ment effects on postpartum depression relative to those ob- served in control conditions. Moreover, these interventions could provide steady symptomatic improvements in postpar- tum depression from baseline to follow-up assessment. The

Fig. 7 Forest plot of the odds ratio of patients recovered or improved comparing self-help psychological interventions with control conditions. Note. Mohammadi (2015)a: antenatal exercise intervention versus treatment-as-usual.

Effectiveness of self-help psychological interventions for treating and preventing postpartum depression: a... 501

results showed that self-help CBTand BAwere more effective for postpartum depression, relative to control conditions. A systematic review of the effectiveness of CBT in treating and preventing perinatal depression reported an overall effect size of 0.65 (95% CI 0.54 to 0.76, P < .001), indicating that participants who received CBT exhibited significantly greater reductions in depressive symptoms relative to those observed in control conditions (Sockol 2015). Another meta-analysis examining BA for depression in adults showed that BA was as effective as cognitive therapy and other psychological treat- ments for depression, with a large effect size of 0.87 (Cuijpers, Van & Warmerdam 2007b). There is strong evidence indicat- ing that dysfunctional patterns of cognition constitute a key risk factor for emotional distress (Beck & Haigh 2014). In addition, CBTwas an effective treatment, as it helped patients to identify, evaluate, challenge, and modify dysfunctional be- liefs. The current meta-analyses showed that while relatively few studies had examined each condition, the overall numbers of studies and participants provided sufficient power for the identification of differences in effectiveness between treatments.

The meta-analysis was subject to several limitations. For example, the sample size was small; therefore, the results should be interpreted with caution. In addition, the body of existing research examining this topic is rel- atively small, and all of the studies included in the review focused on the effects of self-help interventions for post- partum depression. Therefore, the homogeneity of the sample limited generalization of the results and exposed the shortage in existing research. Another limitation that should be noted is the existence of publication bias in the included studies. Trim-and-fill procedures suggested that the publications showed significant positive intervention effects. In addition, following correction for publication bias, the overall effects of the interventions remained sig- nificant, but null findings from well-designed studies are required to enhance understanding of these interventions.

In summary, self-help interventions were more effec- tive, relative to TAU and WLs, and as effective as face- to-face psychological interventions in treating postpar- tum depression. Considering the advantage of conve- nience, self-help interventions, such as self-help, CBT, and BA, have the potential to be effective therapy methods for the treatment of postpartum depression.

Funding information This work was financially supported by grants from the Science and Technology Major Project of Shandong Province (grant number: 2015ZDXX0801A01).

Compliance with ethical standards

Conflict of interest The authors declare that there is no conflict of interest.

References

Andersson L, Sundström-Poromaa I, Bixo M, Wulff M, Bondestam, ÅStrö M (2003) Point prevalence of psychiatric disorders during the second trimester of pregnancy: a population-based study. Am J Obstet Gynecol 189:148–154. https://doi.org/10.1067/mob.2003. 336

Andersson L, Sundström-Poromaa I, Wulff M, Åström M, Bixo M (2006) Depression and anxiety during pregnancy and six months postpar- tum: a follow-up study. Acta Obstet Gynecol Scand 85:937–944. https://doi.org/10.1080/00016340600697652

Beatty L, Binnion C (2016) A systematic review of predictors of, and reasons for, adherence to online psychological interventions. Int J Behav Med 23:776–794. https://doi.org/10.1007/s12529-016-9556-9

Beck AT, Haigh EAP (2014) Advances in cognitive theory and therapy: the generic cognitive model. ANNU REV CLIN PSYCHO 10:1–24. https://doi.org/10.1146/annurev-clinpsy-032813-153734

Bijl RV, Ravelli A (2000) Psychiatric morbidity, service use, and need for care in the general population: results of The Netherlands Mental Health Survey and Incidence Study. Am J Public Health 90:602– 607. https://doi.org/10.2105/ajph.90.4.602

Bortolotti B, Menchetti M, Bellini F, Montaguti MB, Berardi D (2008) Psychological interventions for major depression in primary care: a meta-analytic review of randomized controlled trials. Gen Hosp Psychiatry 30:293–302. https://doi.org/10.1016/j.genhosppsych. 2008.04.001

Brummelte S, Galea LAM (2016) Postpartum depression: etiology, treat- ment and consequences for maternal care. Horm Behav 77:153– 166. https://doi.org/10.1016/j.yhbeh.2015.08.008

Cohen J (1988) Statistical power analysis for the behavioral sciences. J Am Stat Assoc 2:19–74

Costa D, Lowensteyn I, Abrahamowicz M et al (2009) A randomized clinical trial of exercise to alleviate postpartum depressed mood. J Psychosom Obstet Gynaecol 30:191–200. https://doi.org/10.1080/ 01674820903212136

Coull G, Morris PG (2011) The clinical effectiveness of CBT-based guid- ed self-help interventions for anxiety and depressive disorders: a systematic review. Psychol Med 41:2239–2252. https://doi.org/10. 1017/s0033291711000900

Cowpertwait L, Clarke D (2013) Effectiveness of web-based psycholog- ical interventions for depression: a meta-analysis. Int J Ment Health Addict 11:247–268. https://doi.org/10.1007/s11469-012-9416-z

Cuijpers P, Schuurmans J (2007) Self-help interventions for anxiety dis- orders: an overview. Curr Psychiatry Rep 9:284–290. https://doi. org/10.1007/s11920-007-0034-6

Cuijpers P, Smit F, Van SA (2007a) Psychological treatments of sub- threshold depression: a meta-analytic review. Acta Psychiatr Scand 115:434–441. https://doi.org/10.1111/j.1600-0447.2007.00998.x

Cuijpers P, Van SA, Warmerdam L (2007b) Behavioral activation treat- ments of depression: a meta-analysis. Clin Psychol Rev 27:318– 326. https://doi.org/10.1016/j.cpr.2006.11.001

Cuijpers P, Donker T, Van SA, Li J, Andersson G (2010) Is guided self- help as effective as face-to-face psychotherapy for depression and anxiety disorders? A systematic review and meta-analysis of com- parative outcome studies. Psychol Med 40:1943–1957

Daley AJ, Blamey RV, Jolly K, Roalfe AK, Turner KM, Coleman S, McGuinness M, Jones I, Sharp DJ, MacArthur C (2015a) A prag- matic randomized controlled trial to evaluate the effectiveness of a facilitated exercise intervention as a treatment for postnatal depres- sion: the PAM-PeRS trial. Psychol Med 45:2413–2425. https://doi. org/10.1017/S0033291715000409

Daley AJ, Foster L, Long G, Palmer C, Robinson O, Walmsley H, Ward R (2015b) The effectiveness of exercise for the prevention and

502 P.-Z. Lin et al.

treatment of antenatal depression: systematic review with meta-anal- ysis. BJOG 122:57–62. https://doi.org/10.1111/1471-0528.12909

Duval S, Tweedie R (2000) Trim and fill: a simple funnel-plot-based method of testing and adjusting for publication bias in meta-analy- sis. Biometrics 56:455–463. https://doi.org/10.1111/j.0006-341X. 2000.00455.x

Gavin NI, Gaynes BN, Lohr KN, Meltzer-Brody S, Gartlehner G, Swinson T (2005) Perinatal depression: a systematic review of prev- alence and incidence. Obstet Gynecol 106:1071–1083. https://doi. org/10.1097/01.aog.0000183597.31630.db

Gellatly J, Bower P, Hennessy SUE, Richards D, Gilbody S, Lovell K (2007) What makes self help interventions effective in the manage- ment of depressive symptoms ? Meta-analysis and meta-regression. Psychol Med 37:1217–1228. https://doi.org/10.1017/ S0033291707000062

Higgins JP, Thompson SG, Deeks JJ, Altman DG (2003) Measuring inconsistency in meta-analyses. Br Med J 327:557–560

Huffcutt AI, Arthur W (1995) Development of a new outlier statistic for meta-analytic data. J Appl Psychol 80:327–334

Kelly RH, Russo J, Katon W (2001) Somatic complaints among pregnant women cared for in obstetrics: normal pregnancy or depressive and anxiety symptom amplification revisited? Gen Hosp Psychiatry 23: 107–113. https://doi.org/10.1016/s0163-8343(01)00129-3

King E (2009) The effectiveness of an internet-based stress management program in the prevention of postpartum stress, anxiety and depres- sion for new. Walden University, Mothers

Leung BMY, Kaplan BJ (2009) Perinatal depression: prevalence, risks, and the nutrition link—a review of the literature. J Am Diet Assoc 109:1566–1575. https://doi.org/10.1016/j.jada.2009.06.368

Lewis BA, Gjerdingen DK, Avery MD, Sirard JR, Guo H, Schuver K, Marcus BH (2014) A randomized trial examining a physical activity intervention for the prevention of postpartum depression: the healthy mom trial. Ment Health Phys Act 7:42–49. https://doi.org/10.1016/j. mhpa.2013.11.002

Milgrom J, Danaher BG, Gemmill AW, Holt C, Holt CJ, Seeley JR, Tyler MS, Ross J, Ericksen J (2016) Internet cognitive behavioral therapy for women with postnatal depression: a randomized controlled trial of MumMoodBooster. J Med Internet Res 18(54):e54. https://doi. org/10.2196/jmir.4993

Mohammadi F, Malakooti J, Babapoor J, Mohammad-Alizadeh- Charandabi S (2015) The effect of a home-based exercise interven- tion on postnatal depression and fatigue: a randomized controlled trial. Int J Nurs Pract 21:478–485. https://doi.org/10.1111/ijn.12259

Moher D, Shamseer L, Clarke M et al (2015) Preferred reporting items for systematic review and meta-analysis protocols (PRISMA-P) 2015 statement. Syst Rev 4(1). https://doi.org/10.1186/2046-4053-4-1

O'Mahen HA, Woodford J, McGinley J, Warren FC, Richards DA, Lynch TR, Taylor RS (2013) Internet-based behavioral activation– treatment for postnatal depression (Netmums): a randomized con- trolled trial. J Affect Disord 150: 814-22. https://doi.org/10.1016/j. jad.2013.03.005

O’Mahen HA, Richards DA, Woodford J, Wilkinson E, McGinley J, Taylor RS, Warren FC (2014) Netmums: a phase II randomized controlled trial of a guided Internet behavioural activation treatment for postpartum depression. Psychol Med 44:1675–1689. https://doi. org/10.1017/S0033291713002092

Peters J, Booth A, Peters R (2015) Potential for specific dihydropyridine calcium channel blockers to have a positive impact on cognitive function in humans: a systematic review. Ther Adv Chronic Dis 6: 160–169. https://doi.org/10.1177/2040622315582353

Pugh NE, Hadjistavropoulos HD, Dirkse D (2016) A randomised con- trolled trial of therapist-assisted, Internet-delivered cognitive behav- ior therapy for women with maternal depression. PLoS One 11:1– 14. https://doi.org/10.1371/journal.pone.0149186

Roness A, Mykletun A, Dahl AA (2005) Help-seeking behaviour in patients with anxiety disorder and depression. Acta Psychiatr Scand 111:51–58

Rosenthal R (1979) The file drawer problem and tolerance for null results. Psychol Bull 86:638–641

Rosenthal R (1991) Meta-analytic procedures for social research. Sage, Newbury Park, CA

Sockol LE (2015) A systematic review of the efficacy of cognitive be- havioral therapy for treating and preventing perinatal depression. J Affect Disord 177:7–21. https://doi.org/10.1016/j.jad.2015.01.052

Sockol LE, Epperson CN, Barber JP (2011) A meta-analysis of treat- ments for perinatal depression. Clin Psychol Rev 31:839–849. https://doi.org/10.1016/j.cpr.2011.03.009

Verhagen AP, de Vet HCW, de Bie RA, Kessels AGH, Boers M, Bouter LM, Knipschild PG (2017) The Delphi list. J Clin Epidemiol 51: 1235–1241. https://doi.org/10.1016/S0895-4356(98)00131-0

Wisner KL, Sit DKYY, Hanusa BH et al (2009) Major depression and antidepressant treatment: impact on pregnancy and neonatal out- comes. Am J Psychiatry 166:557–566. https://doi.org/10.1176/ appi.ajp.2008.08081170

Effectiveness of self-help psychological interventions for treating and preventing postpartum depression: a... 503

Archives of Women's Mental Health is a copyright of Springer, 2018. All Rights Reserved.