Journal of Midwifery & Women’s Health www.jmwh.

org
Review

Risk Factors for Postpartum Depression: An Umbrella Review

Bridget F. Hutchens1 , CNM, PhD, RN , Joan Kearney2,3 , PhD, PMHCNS, APRN-BC

Introduction: A deeper understanding of risk factors for postpartum depression (PPD) is essential to better target prevention and screening. An
umbrella review was conducted to summarize and synthesize previously published systematic reviews and meta-analyses.
Methods: Eight databases were searched in October of 2016, including PubMed, CINAHL, MEDLINE, PsycINFO, Embase, SCOPUS, PsycEX-
TRA, and Cochrane. Studies were included if they were reviews examining one or more risk factors for PPD and published between 1996 and
2016. The final sample included 21 articles, which varied in numerous ways, including the scope of risk factors explored and statistical methods.
Results: Because of methodological variations between reviews, standardized statistical aggregation was not possible. From this body of literature,
25 statistically significant risk factors emerged with 2 additional risk factors presenting inconclusive findings. The most common risk factors
identified were high life stress, lack of social support, current or past abuse, prenatal depression, and marital or partner dissatisfaction. The 2
strongest risk factors for PPD were prenatal depression and current abuse.
Discussion: Because untreated PPD leaves women and their children vulnerable to numerous negative short-term and long-term outcomes, a
better understanding of PPD risk factors serves to improve maternal and child outcomes by allowing health care providers to better anticipate the
needs of affected women.
J Midwifery Womens Health 2020;00:1–13 
c 2020 by the American College of Nurse-Midwives.

Keywords: depression, mothers, postpartum depression, review, risk factors

INTRODUCTION diagnoses, but it does not specifically include PPD as an inde-

Postpartum depression (PPD) has been described as “the thief pendent diagnosis.6 Rather, it lists major depressive disorder
that steals motherhood” by depriving women of the antic- with the specifier “with peripartum onset” defined as depres-
ipated joy of a new infant.1 According to the US National sive symptoms occurring during pregnancy and up to 4 weeks
Library of Medicine, PPD is moderate to severe depression after birth. This definition differs from what is typically used
with onset during the first year after birth and often occurring in clinical and research settings, which identify PPD as occur-
within the first 3 months.2 An estimated 13% to 19% of child- ring within the first 12 months after birth.2 However, the def-
bearing women will experience PPD,3 making it one of the inition in the DSM-V is helpful in that it clarifies that PPD is a
most common morbidities related to childbirth.4 With close type of major depressive disorder with unique timing. Symp-
to 4 million births each year in the United States,5 that trans- toms of PPD include extreme sadness, diminished pleasure,
lates to anywhere from 520,000 to 760,000 women affected by low energy, thoughts of death, and others.6 Extreme cases of
PPD annually. Given the prevalence of PPD, it is important PPD can lead to maternal suicide.7
that professionals have a thorough understanding of this con- Moreover, PPD’s unique timing has a secondary impact
dition’s risk factors to better target prevention and screening. on the infant through maternal impairment. For example, ma-
This article aims to deepen our understanding of risk factors ternal depression can lead to impaired maternal-child inter-
for PPD using an umbrella review to summarize previously action and disrupted attachment.8,9 Children of women with
published systematic reviews and meta-analyses. Umbrella re- PPD have been found to have sleeping and eating problems
views are review articles that synthesize previously published as well as behavioral difficulties.10 Long-term outcomes for
review articles. these children as adults can include developmental delays11
The Diagnostic and Statistical Manual of Mental Disor- and depression.12 Studies on adverse childhood experiences
ders, fifth edition (DSM-V), sets the standards for psychiatric (ACEs) show that negative experiences during childhood in-
crease the risk for poor health and psychosocial outcomes
in adulthood.13,14 The original study in this well-recognized
1
Hahn School of Nursing and Health Science and Beyster body of work included 7 categories of ACEs: exposure to
Institute for Nursing Research, University of San Diego, San abuse, intimate partner violence, parental mental illness, and
Diego, California
2 other toxic stressors. The authors found a graded relation-
Yale School of Nursing, Yale University, New Haven,
Connecticut ship between the number of ACEs and numerous poor health
3
Yale Child Study Center, Yale University, New Haven, and psychosocial outcomes including adult obesity, heart dis-
Connecticut ease, and cancer.13 These findings support the concept that
Correspondence untreated PPD, as a type of parental mental illness, can in-
Bridget F. Hutchens teract with complex biological and psychosocial factors lead-
Email: bhutchens@sandiego.edu ing to a host of poor outcomes extending beyond child-
ORCID hood. The results of the ACEs studies emphasize the im-
Bridget F. Hutchens https://orcid.org/0000-0001-8341-015X portance of screening for and treating PPD to minimize

1526-9523/09/$36.00 doi:10.1111/jmwh.13067 
c 2020 by the American College of Nurse-Midwives 1
 ✦ Postpartum depression (PPD) is moderate to severe depression affecting an estimated 13% to 19% of childbearing women.
✦ A summary review identified 25 statistically significant risk factors for PPD, including high life stress, lack of social support,
current or past abuse, prenatal depression, and marital or partner dissatisfaction.
✦ The 2 strongest risk factors were prenatal depression and current abuse.
✦ Health care providers can use the summary of risk factors for PPD to target prevention and screening strategies.

poor long-term health outcomes in the children of depressed this article, we selected the term umbrella review. Accord-
women. ing to the methods outlined by Aromataris and colleagues, an
There are valid and reliable treatments for PPD includ- umbrella review samples from other systematic reviews and
ing psychotherapy and antidepressant medication.10 Despite meta-analyses to synthesize a more comprehensive summary
this, significant barriers persist that prevent women from ob- of the topic under study.30 Umbrella reviews must have a pri-
taining care and effective treatments. Such barriers include ori objectives and inclusion criteria, a clear and transparent
stigma, health care providers not screening for PPD, and a search strategy, and assessment of methodological quality, and
lack of skilled health care providers with special training to they must present a clear summary of evidence in the reported
treat this condition.15 It is estimated that only 15% of women findings. This method provides a wide lens to appraise the
with PPD symptoms seek professional care, leaving 85% of state of the science and enables the exploration of convergent
affected women untreated.16,17 Of those women who screen and divergent research findings in the literature. Umbrella re-
positive for PPD, it is estimated that only 22% follow up with views provide clinicians a succinct view on the topic of inter-
professional mental health services.18 This leaves a critical gap est, helping to establish a broad foundation for evidence-based
between those women who suffer from PPD symptoms and practice.
those who receive treatment. The objective of this review was to summarize the sys-
A recent cost analysis conducted by the Mathematica tematic reviews and meta-analyses focused on risk factors for
Policy Research organization examined the cost of perinatal PPD, which began in 1996 and extended to current work. The
mood and anxiety disorders, including PPD. They found that year 1996 was chosen as the lower limit because 2 seminal re-
it cost $32,000 for each mother-child dyad that was affected view articles were published that year, which captured the re-
but untreated. In 2017, this cost the United States an estimated search on risk factors for PPD up to 1996.31,32 The aim was to
$14.2 billion.19 Sixty percent of these costs were attributed to capture the contributions of existing systematic reviews and
maternal outcomes and the remaining 40% to child outcomes. meta-analyses to create a more in-depth understanding of risk
Given the overall burden of PPD and the barriers to treat- factors for PPD. The objective and aim set the foundation for
ment, efforts to address PPD are reflected in the Healthy Peo- the search methods, subsequent sample, and findings.
ple 2020 priority of decreasing the number of women who
experience symptoms of PPD.20 National attention to PPD is
also reflected in the US Preventive Services Task Force empha- Search Methods
sis on the importance of screening for major depressive dis- The literature search was conducted by the primary author
order at least once during the perinatal period.21 To address of this article between October 5 and October 16, 2016,
this priority, there must be a comprehensive understanding and included 8 databases: PubMed, CINAHL, MEDLINE,
of PPD risk and vulnerability factors and those groups who PsycINFO, Embase, SCOPUS, PsycEXTRA, and Cochrane.
may be more likely to suffer from the disorder. This is chal- The primary search term included postpartum, postnatal,
lenging given the large number and wide variety of scientific puerperal, or perinatal, which was joined with depression
articles that explore risk factors for PPD. Topics range from or postpartum depression. The next key terms added in-
mode of birth22–26 to breastfeeding27–29 and numerous oth- cluded risk factors, antecedents, or predictors and finally
ers. As this array of individual articles is difficult to report systematic review, meta-analysis, or synthesis. In addition to
on succinctly, many have been summarized in systematic re- this systematic literature search, the author conducted a hand
views and meta-analyses. This article used the umbrella re- search abstracting review articles from a previous literature
view method30 to synthesize these review articles focused on search on risk factors for PPD.
risk factors for PPD.

METHODS Inclusion and Exclusion Criteria

As the number of systematic reviews grows, it is a logical next Articles were included if they were meta-analyses or system-
step to synthesize the findings of these reviews to compare atic reviews that used PPD as a dependent variable and risk
and contrast findings. This type of review article has been factors as independent variables. In keeping with the review
referred to by a variety of names, including a review of re- objectives, the only participants qualified for inclusion were
views, an overview of reviews, a summary of systematic re- women in the postpartum period. English-language articles
views, a synthesis of reviews, and an umbrella review. In published in the United States or other developed nations were

2 Volume 00, No. 0, xxxx 2020

included if they were published in or after 1996. Articles that In 1996, 2 articles broadly examined risk factors for
focused solely on low- to middle-income countries were ex- PPD.31,32 They both found statistically significant relation-
cluded to maximize comparative validity. Articles in which ships between PPD and the following variables: prenatal de-
the population focus was limited or too narrowly focused (eg, pression, life stress, social support, and history of depression.
parents with a newborn in the neonatal intensive care unit or Beck also found childcare stress, prenatal anxiety, and ma-
adolescent participants only) were also excluded. ternity blues to have statistically significant correlations with
PPD. O’Hara and Swain identified the same relationship with
respect to income, woman’s occupation, pregnancy and birth
Search Outcomes complications, marital satisfaction, neuroticism, and nega-
tive cognitive attribution style. Following the publication of
The initial search yielded 621 articles plus 6 articles from hand
these 2 seminal reviews, Wilson and colleagues also used a
searching. However, title screening, abstract, and full-text re-
broad perspective to examine risk factors for PPD but limited
view greatly reduced the final number of articles used in this
their scope to antenatal factors.38 These authors’ findings sup-
review. Within the original collection, 167 duplicates were
ported the results reported by Beck31 and O’Hara and Swain,32
eliminated, and 428 articles were removed after title and ab-
identifying the same 4 common variables: prenatal depres-
stract review, yielding 32 articles. A full-text review of these
sion, life stress, social support, and history of depression. Wil-
32 articles was conducted, and a final 21 articles met criteria
son added both marital satisfaction (supporting O’Hara and
for inclusion in this umbrella review (Figure 1).
Swain’s findings) and the unique variable of abuse. Beck up-
dated her findings in 2001, adding a few key statistically sig-
nificant risk factors including income, marital satisfaction, in-
Quality and Bias Appraisal
fant temperament, self-esteem, and unplanned or unwanted
First, the studies were reviewed for inclusion status. Then, pregnancy.35 Robertson and colleagues added to this litera-
each of the 21 articles was again reviewed by both authors for ture by exploring a broad spectrum of risk factors for PPD and
quality and bias using the guidelines set forth by the assess- supported previous findings: prenatal depression, life stress,
ment of multiple systematic reviews (AMSTAR) tool.33 There social support, prenatal anxiety, income, pregnancy and birth
are 11 criteria in the AMSTAR tool, and the presence of each factors, and neuroticism.37 From these early studies up until
criterion was assigned one point out of a possible total score 2015, there were no further systematic reviews conducted to
of 11. Examples of AMSTAR criteria include “Was an ‘a pri- explore risk factors from a broad perspective. Rather, subse-
ori’ design provided?” and “Was there duplicate study selec- quent reviews centered on specific risk factors.
tion and data extractions?” AMSTAR does not provide a cut- Of the studies that examined specific risk factors, 6 dif-
off for an acceptable minimum score and therefore was used ferent studies focused on current or past abuse.40,41,43,50,52,53
solely to measure quality and not for exclusion of articles. Two studies focused on sleep quality.44,47 The remaining stud-
For quality assessment, the 2 authors independently ies all had unique and specific foci: infant temperament,42
scored each study and then compared the results for each of preeclampsia and/or HELLP,45 obesity,48 negative breast-
the 11 criteria of AMSTAR for all 21 articles included in the feeding experiences,46 body dissatisfaction,51 and partner
review. When disagreement was found, the 2 authors confer- support.49 The review by Ross and Dennis also focused on
enced to consensus. Overall, initial interrater agreement was substance abuse and chronic disease in addition to abuse
88.7%. Cohen’s kappa score was calculated and found to be history.50
0.799 (95% CI, 0.725-0.872), which is considered a moderate Across the 21 studies in this review, 25 statistically sig-
to strong level of agreement.34 After analysis of the 21 articles nificant risk factors were identified for PPD. In addition to
was complete, a data display matrix was created and is sum- these 25 risk factors, 2 variables yielded inconclusive findings:
marized in Table 1.33 a history of child abuse and preeclampsia and/or HELLP (Ta-
ble 2). The most common risk factors identified were life stress
and social support, each appearing 7 times.31,32,35–39 Abuse,
RESULTS including current, past, and child sexual abuse, also appeared
The 21 articles reviewed varied in numerous ways. Seven ex- 7 times.38,40,41,43,50,52,53 Prenatal depression was found to be a
amined a broad variety of risk factors,31,32,35–39 whereas the risk factor in 6 of the systematic reviews.31,32,35–38 Marital or
remaining 14 articles focused on specific risk factors.40–53 partner satisfaction appeared 5 times.32,35,38,39,49 Of the most
The number of published studies synthesized in each re- frequently appearing risk factors, 2 of the strongest predic-
view ranged from as few as 6 articles45,53 to more than 200 tors of PPD were prenatal depression (r = 0.49)31 and current
articles.36,39 The total number of articles in the final sample abuse with an odds ratio of 3 or higher.36,53
of 21 systematic reviews and meta-analyses entries was 1143 Statistical analyses varied widely among the
without removing duplicates. The mean number of articles 21 studies. Twelve of the studies calculated pooled
per entry across the 21 entries was 54. statistics,31,32,35,37,41–44,48,49,52,53 whereas the remaining 9
A final quality score was tabulated based on the 11 AM- did not.35,36,38,39,45–47,50,51 Of the studies with pooled cal-
STAR criteria. Quality scores ranged from 1 out of 11 (9.1%)40 culations, the statistical analyses varied widely. Robertson
to a score of 10.5 out of 11 (95.4%).41 The average quality score and colleagues37 used Cohen’s d to report the standardized
across the 21 review articles was 7.1 out of 11 (64.5%), which difference between means, whereas O’Hara and Swain32 used
is considered overall moderate to low quality. Quality scores delta to report the difference between means divided by the
are listed in Table 1. pooled standard deviation. Five reviews with pooled statistics

Journal of Midwifery & Women’s Health r www.jmwh.org 3

4
Table 1. Overview of Review Articles on Risk Factors for PPD
Number of
First Author, Year Databases Searched
AMSTAR Quality Review Type Tool Used to Assess Quality (Date Range)
a
Score of Studies Number of Studies Objective Overall Sample Outcome
Antoniou Systematic review of bibliography 1 To explore the relationship between domestic Domestic violence during pregnancy is an
2008 Not reported (1996-2007) violence and PPD important risk factor for PPD
1/11 (9.1%) 33 Perinatal women
Beck Meta-analysis 4 To explore the relationship between infant Effect size (r)b : 0.31 (weighted by sample size;
1996b A quality scoring instrument was (1974-1993) temperament and PPD 95% CI, 0.261-0.369)
9/11 (81.8%) designed specifically for PPD 17 Women with PPD Indicates a significant moderate correlation
research between PPD and infant temperament
Beck Meta-analysis 4 To explore broad risk factors for PPD Effect size (r)b :
1996a Scoring system developed by Beck in (1974-1993) Women with PPD Prenatal depression: 0.49
8.5/11 (77.3%) 1995 with interrater check 44 Childcare stress: 0.48
Life stress: 0.40
Social support: 0.37-0.39
Prenatal anxiety: 0.30-0.36
Maternity blues: 0.35-0.37
Marital satisfaction 0.29-0.37
Previous depression: 0.27-0.29
Beck Meta-analysis 9 To explore broad risk factors for PPD Effect size (r)b :
2001 Scoring system developed by Beck in (1990-1999) Women with PPD Prenatal depression: 0.44-0.46
7.5/11 (68.2%) 1995 84 Self-esteem: 0.45-0.47
Childcare stress: 0.45-0.46
Prenatal anxiety: 0.41-0.45
Life stress: 0.38-0.40
Social support: 0.36-0.41
Marital relationship: 0.38-0.39
Previous depression: 0.38-0.39
Infant temperament: 0.33-0.34
Maternity blues: 0.25-0.31
Marital status: 0.21-0.35
Socioeconomic status: 0.19-0.22
Unplanned or unwanted Pregnancy:
0.14-0.17
(Continued)

Volume 00, No. 0, xxxx 2020

 Table 1. Overview of Review Articles on Risk Factors for PPD
Number of
First Author, Year Databases Searched
AMSTAR Quality Review Type Tool Used to Assess Quality (Date Range)
Scorea of Studies Number of Studies Objective Overall Sample Outcome
Beydoun Systematic review and meta-analysis 1 To explore the relationship between intimate 7 of 37 articles addressed intimate partner
2012 Not reported (1980-2010) partner violence relationship and major violence and PPD
7.5/11 (68.2%) 37 depressive disorder, depressive symptoms, Combined risk ratio for PPD: 1.43 (95% CI,
and PPD 1.22-1.67)
Adult women with intimate partner violence
Bhati Systematic review 7 Sleep disturbances and PPD Consistent strong relationship between sleep

Journal of Midwifery & Women’s Health r www.jmwh.org

2015 GRADE (1990-2014) Perinatal women disturbances and PPD
5.5/11 (50%) 13
Delahaije Systematic review 2 To explore the relationship between PE/HELLP Positive association between PE/HELLP and
2013 New castle-Ottawa Quality (Inception of and PPD or postpartum anxiety PPD, but not statistically significant in 50%
8/11 (72.7%) Assessment database until Postpartum women who suffered PE/HELLP of the studies
March 2013)
6
Dias Systematic review 3 To explore the relationship between Association between breastfeeding and PPD
2015 Not reported (1980-2013) breastfeeding relationship and PPD found in 18 articles with negative
8/11 (72.7%) 48 Perinatal women breastfeeding experiences preceding PPD
Howard Systematic review and meta-analysis 17 To explore the relationship between domestic Cross sectional studies pooled estimates: 3- to
2013 Critical appraisal skills programme (Inception of violence relationship and perinatal mental 5-fold increased unadjusted odds of having
10.5/11 (95.4%) database until disorders experienced domestic violence if prenatal or
March 21, 2011) PPD
67 Longitudinal studies pooled estimates: 3-fold
increased unadjusted odds of PPD if
domestic violence experienced during
pregnancy
Lawson Systematic review 6 To explore the relationship between sleep and Modest support for participant-reported sleep
2015 Effective public health practice (Inception of postpartum mental disorders disturbances and PPD
8.5/11 (77.3%) project quality assessment tool for database until July Postpartum women
quantitative studies 30, 2013; updated
in June 2014)
31
(Continued)

5
6
Table 1. Overview of Review Articles on Risk Factors for PPD
Number of
First Author, Year Databases Searched
AMSTAR Quality Review Type Tool Used to Assess Quality (Date Range)
a
Score of Studies Number of Studies Objective Overall Sample Outcome
Molyneaux Systematic review and meta-analysis 7 To explore the relationship between obesity and Obese vs normal weight:
2014 Author adapted quality assessment (Inception of perinatal mental disorders OR, 1.30 (95% CI, 1.20-1.42)
10/11 (90.9%) tool database until Perinatal women Overweight vs normal weight:
January 7, 2013) OR, 1.09 (95% CI, 1.05-1.13)
23 Obese vs overweight:
OR, 1.20 (95% CI, 1.13-1.27)
Norhayati Systematic review 5 To explore broad risk factor for PPD Significant contributors to PPD:
2015 Not reported (2005-2014) Adult postpartum women Antenatal depression
4.5/11 (40.9%) 202 Antenatal anxiety
Previous psychiatric illness
Poor marital relationship
Stressful life events
Negative attitude toward pregnancy
Lack of social support
O’Hara Meta-analysis Not reported To explore prevalence and predictors of PPD Effect size (Cohen’s d)b :
1996 Not reported (Not reported) Women with PPD Prenatal depression: 0.75
4/11 (36.4%) Not reported Life events: 0.60
Social support: −0.63
Personal or family psychopathology: 0.57
Family income: −0.141
Woman’s occupation: −0.146
Pregnancy and birth complication: 0.26
Marital satisfaction: −0.13
Neuroticism: 0.39
Negative cognitive style: 0.24
Pilkington Systematic review and meta-analysis 5 To explore partner factors related to perinatal Effect size (r)b :
2015 Quality assessed by (Inception of depression and anxiety Communication: −0.23
9/11 (81.8%) author-determined standards database until Women with perinatal anxiety or depression Conflict: 0.28
March 2014) Emotional support: −0.22
120 Instrumental support: −0.19
Relationship satisfaction: −0.33
(Continued)

Volume 00, No. 0, xxxx 2020

 Table 1. Overview of Review Articles on Risk Factors for PPD
Number of
First Author, Year Databases Searched
AMSTAR Quality Review Type Tool Used to Assess Quality (Date Range)
Scorea of Studies Number of Studies Objective Overall Sample Outcome
Robertson Systematic review 19 To explore broad antepartum risk factors for Effect size (Cohen’s d)b :
2004 Quality assessed by authors (1990-2002) PPD Prenatal depression: 0.75
4.5/11 (40.9%) determined standards Not reported Perinatal women Prenatal anxiety: 0.68
Life events: 0.61

Journal of Midwifery & Women’s Health r www.jmwh.org

Social support: −0.64
History of depression: 0.58
Neuroticism: 0.39
Socioeconomic status: −0.14
Pregnancy and birth factors: 0.26
Ross Systematic review 5 To explore the relationship between substance Substance use, current abuse, and past abuse
2009 Not reported (Inception of use/chronic illness and PPD consistently associated with increased risk of
6.5/11 (59.1%) database until Perinatal women who used substances, had PPD, but not chronic illness
August 1, 2008) current or past experience of abuse, or had
17 chronic illness
Silveira Systematic review 1 To explore the relationship between body Majority of studies found body dissatisfaction
2015 Not reported (1994 to May image and perinatal depression is weakly but consistently associated with
5.5/11 (50%) 2014) Perinatal women prenatal and PPD
19
Wilson Systematic review 5 To explore the relationship between antenatal Risk factors for PPD:
1996 Canadian Task Force on the Periodic (1980-1993) psychosocial risk factors and adverse Poor marital adjustment
6.5/11 (59.1%) Health Examination (modified) 118 postpartum family outcomes, including PPD Recent life stressors
Perinatal women Antepartum depression
Lack of social support
Abuse of the woman
History of psychiatric disorder in the woman
(Continued)

7
8
Table 1. Overview of Review Articles on Risk Factors for PPD
Number of
First Author, Year Databases Searched
AMSTAR Quality Review Type Tool Used to Assess Quality (Date Range)
Scorea of Studies Number of Studies Objective Overall Sample Outcome
Wosu Epidemiologic review, 7 To explore the relationship between a history of Association of child sexual abuse to PPD
2015 Newcastle-Ottawa Scale (Inception of child sexual abuse and perinatal depression Pooled unadjusted OR, 1.82 (95% CI,
8/11 (72.7%) database until Perinatal women 0.92-3.60)
August 2014) Pooled adjusted OR, 1.20 (95% CI, 0.81-1.76).
14 Findings on the postpartum period were
inconsistent
Wu Meta-analysis 4 To explore the relationship between violence Association of violence to PPD: positive
2012 Newcastle-Ottawa Scale for (Inception of and PPD correlation
9.5/11 (86.4%) nonrandomized studies database until July Women with PPD OR, 3.47 (95% CI, 2.13-5.64)
1, 2011)
6
Yim Systematic review 2 To explore the relationship between biological Biological predictors of PPD:
2015 Not reported (2000-2013) and psychosocial risk factors and PPD Hypothalamic-pituitary-adrenal
7/11 (63.6%) 214 Postpartum women dysregulation
Inflammatory processes
Genetic vulnerabilities
Psychosocial predictors of PPD:
Severe life events
Chronic strain
Relationship quality
Support from partner
Support from mother
Abbreviations: AMSTAR, assessment of multiple systematic reviews; GRADE, Grading, Recommendation, Assessment, Development, and Evaluation; OR, odds ratio; PE, preeclampsia; PPD, postpartum depression.
a
From Shea et al (2007).33 There are 11 criteria in the AMSTAR tool; each criterion is assigned one point out of a possible total score of 11. The scores have been converted to percents.
b
Effect size (r and Cohen’s d): small (0.2), medium (0.5), large (0.8).62

Volume 00, No. 0, xxxx 2020

 Figure 1. Flow Diagram Summarizing Study Selection
Abbreviations: DV, dependent variable; IV, independent variable; PPD, postpartum depression.

used r to report correlations.31,35,42,44,49 Four studies with ratios. Consensus on uniform definitions and methods would
pooled statistics reported their findings in terms of odds allow for valid comparison moving forward.
ratios to report the odds that the disorder will occur,41,48,52,53 Another challenge is being able to identify whether the
and one review reported its findings in terms of relative risk listed associated attribute is a true risk factor or a characteris-
to report the likelihood of the disorder occurring.43 tic of those who are suffering from depression. For example,
Bhati and Richards44 define sleep disturbance as “a combina-
DISCUSSION tion of sleep deprivation and sleep fragmentation.” However,
sleep disturbance is also listed as a symptom of depression.6
The purpose of this umbrella review was to provide an
Therefore, it could be argued that depression is causing the
overview and synthesis of the systematic reviews and meta-
sleep disturbance rather than the sleep disturbance being a
analyses that focused on risk factors for PPD over the span
risk factor for depression. This chicken-or-egg dilemma ap-
of 20 years. Although pertinent findings were extracted from
plies to other identified risk factors such as self-esteem, mari-
this collection of published studies, the breadth and variety
tal dissatisfaction, hypothalamic-pituitary-adrenal dysregula-
of risk factors explored made it difficult to concisely sum-
tion, and inflammatory process.
marize the literature. Within this challenge, operational def-
In synthesizing these systematic reviews and meta-
initions of risk factors varied across studies even when dis-
analyses, an interesting trend emerged regarding the timing of
cussing similar or identical topics. For example, Antoniou and
publications. In 1996, there was a surge in reviews exploring
colleagues40 referred to their independent variable as “domes-
this topic, with 4 articles published that year. In the ensuing
tic violence,” whereas Beydoun and colleagues43 referred to a
15 years, only 4 reviews exploring risk factors for PPD were
similar variable as “intimate partner violence.” Another ex-
published. This was followed by a second surge in published
ample is that Beck31 referred to one of the independent vari-
reviews, so that from 2013 to 2016 there were 13 systematic
ables as “life stress,” whereas O’Hara and Swain32 referred to a
reviews and meta-analyses published, with 8 of them appear-
similar variable as “life events.” Additionally, the variety of sta-
ing in the literature in 2015. This pattern may be related to
tistical methods used restricted the ability to compare results
the initial saturation in 1996, perhaps dissuading researchers
across articles. For example, it is challenging to draw overall
from covering this topic again until recently when the research
conclusions when comparing correlations or means with odds
from 1996 needed updating.

Journal of Midwifery & Women’s Health r www.jmwh.org 9

 Table 2. Frequency of Risk Factors in the 21 Reviews Included in This Umbrella Review

Molyneaux (2014)

Pilkington (2015)
Robertson (2004)

Norhayati (2015)
Antoniou (2008)

Delahaije (2013)
Beydoun (2012)

Howard (2013)

Lawson (2015)

Silveira (2015)
O’Hara (1996)
Wilson (1996)
Beck (1996b)

Beck (1996a)

Wosu (2015)
Bhati (2015)
Beck (2001)

Ross (2009)

Dias (2015)

Yim (2015)
Wu (2012)
Risk Factor
Prenatal depression r r r r r r
Childcare stress r r
Life stress r r r r r r r
Social support r r r r r r r
Prenatal anxiety r r r
Maternity blues r r
History of depression r r r r
Family income r r r
Woman’s occupation r
Pregnancy and birth complication r r
Marital satisfaction r r r r r
Neuroticism r r
Negative cognitive attributional style r
Current abuse or violence r r r r r r
Infant temperament r r
Self-esteem r
Unplanned or unwanted pregnancy r r
Substance abuse r
Preeclampsia or HELLP M
Obesity r
Sleep disturbances r r
Breastfeeding r
Body image dissatisfaction r
Child sexual abuse M
Hypothalamic-pituitary-adrenal r
dysregulation
Inflammatory process r
Genetic vulnerability r
r
Symbols: , statistically significant risk factor; M, mixed results, inconclusive.

Another interesting pattern emerged with respect to the of pooling statistics has become more sophisticated over the
variety of statistical analyses employed by the studies in this past 20 years, which renders application of the current method
review. Of the studies published before 2006, 5 of the 6 stud- to earlier studies unlikely.
ies all included pooled statistical analysis (83%). Of those pub-
lished after 2006, only 7 out of 15 articles used pooled statistics
Limitations and Opportunities
(47%), and by 2015 only 3 of 8 articles used this method (37%).
Clearly, the more recent systematic reviews were less likely to According to the AMSTAR criteria, there was a wide range in
used pooled statistics contrasted with earlier research. This quality across the 21 articles with a mean score of 7.1 out of
may be explained in part by the sheer amount of research that 11. Scores ranged from 1 out of 11 (9.1%) in the review article
is available on this topic that grows with each passing year. As by Antoniou, Vivilaki, and Daglas40 to a score of 10.5 out of
the body of literature grows, so does the heterogeneity among 11 (90.9%) in the review article by Howard and colleagues.41
studies, which may make it more difficult to pool statistics. For This indicates an overall moderate to low quality across all
example, Yim and colleagues39 included 214 articles in their reviews and provides an opportunity to develop high qual-
review, making it exceedingly difficult to calculate any pooled ity systematic reviews and meta-analyses on the topic. Exam-
statistics because of the inherent variation across this volume ples of AMSTAR criteria that were frequently missing from
of studies. It may also be explained by the fact that the method articles include the status of publication (ie, grey literature)

10 Volume 00, No. 0, xxxx 2020

used as an inclusion criterion and assessment of publication and contributes to the development of targeted prevention
bias.33 and screening strategies for the most vulnerable women. Us-
There were several additional limitations. The first limi- ing such targeted strategies may significantly affect treatment
tation is that the literature search and data extractions were success, including increasing identification and improving
conducted only by the primary author, which may have in- treatment rates in high-risk women. Because untreated PPD
troduced selection bias. Also, the varying statistical methods leaves women and their children vulnerable to numerous neg-
limited the synthesis of data to counting how frequently a ative short-term and long-term outcomes, better understand-
risk factor appeared in the review articles. This does not re- ing of PPD risk factors serves to improve maternal and child
flect the overall strength of risk factors or alert practitioners outcomes and limit the damage done by “the thief that steals
to what are the strongest risk factors to look for in individ- motherhood.”1 Systematic and adequate attention to PPD and
ual patients. Clear opportunities for further research include its risk factors can bolster maternal-child outcomes and en-
methodological analysis of what is now a large body of hetero- sure more opportunities for women and children to thrive.
geneous studies on the topic conducted over a period of years.
Because the more recent reviews often lack pooled statistical CONFLICT OF INTEREST
calculations, it may be helpful to have current work employ
The authors have no conflicts of interest to disclose.
pooled effect sizes or odds ratios. Another opportunity relates
to examination and analysis of interactions among risk factors
themselves. Large-scale studies that employ multilevel model- ACKNOWLEDGMENTS
ing techniques would address this. The authors would like to acknowledge the Robert Wood
Johnson Foundation Future of Nursing Scholars Program and
Yale School of Nursing for funding this research through the
Clinical and Policy Implications support of Bridget F. Hutchens’s doctoral studies.
Clinically, this umbrella review provides health care providers
with a summary of risk factors for PPD that can be used to REFERENCES
target screening and prevention strategies. However, the list
is extensive and does not represent what puts women at the 1.Beck CT. Postpartum depression. Stopping the thief that steals moth-
highest risk for PPD. Also, it is important to keep in mind that erhood. AWHONN Lifelines. 1999;3(4):41-44.
patients may be hesitant to disclose certain risk factors, such as 2.Postpartum depression. MedlinePlus website. https://medlineplus.
gov/ency/article/007215.htm. Updated 2016. Accessed January 11,
prenatal depression or a history of abuse. To address this and 2017.
other concerns, the implementation of universal screening for 3.O’Hara MW, McCabe JE. Postpartum depression: current status and
PPD as outlined by numerous professional organizations4,21,54 future directions. Annu Rev Clin Psychol. 2013;9:379-407.
would standardize our approach, clarify and strengthen prac- 4.American College of Obstetricians and Gynecologists Committee on
tice parameters, and place the responsibility on the health Obstetric Practice. Committee opinion no. 630: Screening for perinatal
care provider to gather these critical data without undue bur- depression. Obstet Gynecol. 2015;125(5):1268-1271.
5.Centers for Disease Control and Prevention. Births and natality. Cen-
den on vulnerable women. There are several validated screen-
ters for Disease Control and Prevention website. http://www.cdc.gov/
ing tools for PPD, including the Edinburgh Postnatal De- nchs/fastats/births.htm. Updated 2016. Accessed May 26, 2016.
pression Screen,55 Beck’s Postpartum Depression Screening 6.American Psychiatric Association. Diagnostic and Statistical Man-
Scale,56 the Patient-Health Questionnaire-9,57 and the abbre- ual of Mental Disorders. 5th ed. Arlington, VA: American Psychiatric
viated Patient-Health Questionnaire-2.58 Publishing; 2013.
Because PPD has been shown to have long-term impli- 7.Kendall-Tackett KA. Depression in New Mothers: Causes, Conse-
cations for woman and child, this umbrella review has im- quences, and Treatment Alternatives. 2nd ed. New York: Routledge;
2010.
plications for health policy. Creating policies to address PPD
8.Stein A, Pearson RM, Goodman SH, et al. Effects of perinatal mental
is an upstream intervention preventing future complications disorders on the fetus and child. Lancet. 2014;384(9956):1800-1819.
and societal costs. In recent years, there has been more leg- 9.Martins C, Gaffan EA. Effects of early maternal depression on patterns
islative attention given to PPD, such as the passing in 2016 of infant-mother attachment: a meta-analytic investigation. J Child
of the Bringing Postpartum Depression Out of the Shadows Psychol Psychiatry. 2000;41(6):737-746.
Act59 and 2 key pieces of PPD legislation passed in California 10.National Institute of Mental Health. Postpartum depression facts. Na-
in 2018.60 Also in 2019, the US Preventive Services Task Force tional Institute of Health website. http://www.nimh.nih.gov/health/
publications/postpartum-depression-facts/index.shtml. Updated
published a grade B recommendation that women at higher 2016. Accessed December 15, 2016.
risk for PPD should receive preventive counseling.61 This um- 11.Beck CT. The effects of postpartum depression on maternal-infant in-
brella review in part serves to support that recommendation, teraction: a meta-analysis. Nurs Res. 1995;44(5):298-304.
as it can further alert health care providers to which women 12.Murray L, Arteche A, Fearon P, Halligan S, Goodyer I, Cooper P. Ma-
may be at increased risk. ternal postnatal depression and the development of depression in off-
spring up to 16 years of age. J Am Acad Child Adolesc Psychiatry.
2011;50(5):460-470.
CONCLUSION 13.Felitti VJ, Anda RF, Nordenberg D, et al. Relationship of childhood
abuse and household dysfunction to many of the leading causes of
There has been extensive research on risk factors for PPD over
death in adults: the Adverse Childhood Experiences (ACE) study. Am
the span of 20 years. Although not exhaustive, this umbrella J Prev Med. 1998;14(4):245-258.
review serves to summarize this work. More in-depth under- 14.Anda RF, Felitti VJ, Bremner JD, et al. The enduring effects of
standing of risk factors for PPD informs health care providers abuse and related adverse experiences in childhood. A convergence of

Journal of Midwifery & Women’s Health r www.jmwh.org 11

 evidence from neurobiology and epidemiology. Eur Arch Psychiatry 34.McHugh ML. Interrater reliability: the kappa statistic. Biochem Med
Clin Neurosci. 2006;256(3):174-186. (Zagreb). 2012;22(3):276-282.
15.Byatt N, Biebel K, Friedman L, Debordes-Jackson G, Ziedonis D, Pbert 35.Beck CT. Predictors of postpartum depression: an update. Nurs Res.
L. Patient’s views on depression care in obstetric settings: how do they 2001;50(5):275-285.
compare to the views of perinatal health care professionals? Gen Hosp 36.Norhayati MN, Nik Hazlina NH, Asrenee AR, Wan Emilin WM. Mag-
Psychiatry. 2013;35(6):598-604. nitude and risk factors for postpartum symptoms: a literature review.
16.Center on the Developing Child at Harvard University. Maternal J Affective Disord. 2015;175:34-52.
Depression Can Undermine the Development of Young Children. 37.Robertson E, Grace S, Wallington T, Stewart DE. Antenatal risk factors
Working paper no. 8. Cambridge, MA: Center on the Develop- for postpartum depression: a synthesis of recent literature. Gen Hosp
ing Child at Harvard University; 2009. http://developingchild. Psychiatry. 2004;26(4):289-295.
harvard.edu/wp-content/uploads/2009/05/Maternal-Depression- 38.Wilson LM, Reid AJ, Midmer DK, Biringer A, Carroll JC, Stewart DE.
Can-Undermine-Development.pdf. Accessed January 18, 2017. Antenatal psychosocial risk factors associated with adverse postpar-
17.Vesga-López O, Blanco C, Keyes K, Olfson M, Grant BF, Hasin tum family outcomes. CMAJ. 1996;154(6):785-799.
DS. Psychiatric disorders in pregnant and postpartum women in the 39.Yim IS, Tanner Stapleton LR, Guardino CM, Hahn-Holbrook J,
United States. Arch Gen Psychiatry. 2008;65(7):805-815. Dunkel Schetter C. Biological and psychosocial predictors of postpar-
18.Byatt N, Levin LL, Ziedonis D, Moore Simas TA, Allison J. Enhancing tum depression: systematic review and call for integration. Annu Rev
participation in depression care in outpatient perinatal care settings: a Clin Psychol. 2015;11:99-137.
systematic review. Obstet Gynecol. 2015;126(5):1048-1058. 40.Antoniou E, Vivilaki V, Daglas M. Correlation of domestic violence
19.Luca DL, Garlow N, Staatz C, Margiotta C, Zivin K. Societal during pregnancy with postnatal depression: systematic review of bib-
Costs of Untreated Perinatal Mood and Anxiety Disorders in the liography. Health Science Journal. 2008;2(1):15-19.
United States. Cambridge, MA: Mathematica Policy Research; 2019. 41.Howard LM, Oram S, Galley H, Trevillion K, Feder G. Domestic vio-
https://www.mathematica-mpr.com/our-publications-and-findings/ lence and perinatal mental disorders: A systematic review and meta-
publications/societal-costs-of-untreated-perinatal-mood-and-anxie analysis. PLoS Med. 2013;10(5):e1001452.
ty-disorders-in-the-united-states. Accessed May 10, 2018. 42.Beck CT. A meta-analysis of the relationship between postpartum de-
20.Office of Disease Prevention and Health Promotion. Healthy People pression and infant temperament. Nurs Res. 1996;45(4):225-230.
2020: Maternal, infant, and child health. Healthy People 2020 website. 43.Beydoun HA, Beydoun MA, Kaufman JS, Lo B, Zonderman AB.
https://www.healthypeople.gov/2020/topics-objectives/topic/mater Intimate partner violence against adult women and its association
nal-infant-and-child-health/objectives. Updated 2017. Accessed with major depressive disorder, depressive symptoms and postpar-
January 18, 2017. tum depression: a systematic review and meta-analysis. Soc Sci Med.
21.Siu AL, Bibbins-Domingo K, Grossman DC, et al; US Preventive Ser- 2012;75(6):959-975.
vices Task Force. Screening for depression in adults: US Preventive 44.Bhati S, Richards K. A systematic review of the relationship between
Services Task Force recommendation statement. JAMA. 2016;315(4): postpartum sleep disturbance and postpartum depression. J Obstet
380-387. Gynecol Neonatal Nurs. 2015;44(3):350-357.
22.Hanlon AJ, Beckmann MM. Mode of birth and early postnatal psycho- 45.Delahaije DH, Dirksen CD, Peeters LL, Smits LJ. Anxiety and depres-
logical morbidity. Aust N Z J Obstet Gynaecol. 2015;55(6):578-583. sion following preeclampsia or hemolysis, elevated liver enzymes, and
23.Bland M. The influence of birth experience on postpartum depression. low platelets syndrome. A systematic review. Acta Obstet Gynecol
Midwifery Today Int Midwife. 2009;(89):20-1, 65-6. Scand. 2013;92(7):746-761.
24.Goker A, Yanikkerem E, Demet MM, Dikayak S, Yildirim Y, Koyuncu 46.Dias CC, Figueiredo B. Breastfeeding and depression: a systematic re-
FM. Postpartum depression: is mode of delivery a risk factor? ISRN view of the literature. J Affect Disord. 2015;171:142-154.
Obstet Gynecol. 2012;2012:616759. 47.Lawson A, Murphy KE, Sloan E, Uleryk E, Dalfen A. The relationship
25.Adams S, Eberhard-Gran M, Sandvik ÅR, Eskild A. Mode of delivery between sleep and postpartum mental disorders: a systematic review.
and postpartum emotional distress: a cohort study of 55 814 women. J Affect Disord. 2015;176:65-77.
BJOG. 2012;119(3):298-305. 48.Molyneaux E, Poston L, Ashurst-Williams S, Howard LM. Obe-
26.Sword W, Landy CK, Thabane L, et al. Is mode of delivery associated sity and mental disorders during pregnancy and postpartum: a sys-
with postpartum depression at 6 weeks: a prospective cohort study. tematic review and meta-analysis. Obstet Gynecol. 2014;123(4):857-
BJOG. 2011;118(8):966-977. 867.
27.Brown A, Rance J, Bennett P. Understanding the relationship between 49.Pilkington PD, Milne LC, Cairns KE, Lewis J, Whelan TA. Modifi-
breastfeeding and postnatal depression: the role of pain and physical able partner factors associated with perinatal depression and anxiety:
difficulties. J Adv Nurs. 2016;72(2):273-282. a systematic review and meta-analysis. J Affect Disord. 2015;178:165-
28.Borra C, Iacovou M, Sevilla A. New evidence on breastfeeding and 180.
postpartum depression: the importance of understanding women’s in- 50.Ross LE, Dennis CL. The prevalence of postpartum depression among
tentions. Matern Child Health J. 2015;19(4):897-907. women with substance use, an abuse history, or chronic illness: a sys-
29.Beyca H, Mutlu H, Özdamar Ö. Effect of breastfeeding on depression tematic review. J Womens Health (Larchmt). 2009;18(4):475-486.
levels of mothers, weight and height measurements of babies in the 51.Silveira ML, Ertel KA, Dole N, Chasan-Taber L. The role of body image
postpartum first two months [in Turkish]. Zeynep Kamil Tıp B ülteni. in prenatal and postpartum depression: a critical review of the litera-
2014;45(3):124-130. ture. Arch Womens Ment Health. 2015;18(3):409-421.
30.Aromataris E, Fernandez R, Godfrey CM, Holly C, Khalil H, Tung- 52.Wosu AC, Gelaye B, Williams MA. History of childhood sexual
punkom P. Summarizing systematic reviews: methodological devel- abuse and risk of prenatal and postpartum depression or depressive
opment, conduct and reporting of an umbrella review approach. Int J symptoms: an epidemiologic review. Arch Womens Ment Health.
Evid Based Healthc. 2015;13(3):132-140. 2015;18(5):659-671.
31.Beck CT. A meta-analysis of predictors of postpartum depression. 53.Wu Q, Chen HL, Xu XJ. Violence as a risk factor for postpartum de-
Nurs Res. 1996;45(5):297-303. pression in mothers: a meta-analysis. Arch Womens Ment Health.
32.O’Hara MW, Swain AM. Rates and risk of postpartum depression - a 2012;15(2):107-114.
meta-analysis. Int Rev Psychiatry. 1996;8(1):37-54. 54.Earls MF; American Academy of Pediatrics Committee on Psychoso-
33.Shea BJ, Grimshaw JM, Wells GA, et al. Development of AMSTAR: a cial Aspects of Child and Family Health. Incorporating recognition
measurement tool to assess the methodological quality of systematic and management of perinatal and postpartum depression into pedi-
reviews. BMC Med Res Methodol. 2007;7:10. atric practice. Pediatrics. 2010;126(5):1032-1039.

12 Volume 00, No. 0, xxxx 2020

55.Cox JL, Holden JM, Sagovsky R. Detection of postnatal depression: de- https://www.perinatalmentalhealthcare.com/news/bringing-postpart
velopment of the 10-item Edinburgh Postnatal Depression Scale. Br J um-depression-out-of-the-shadows-act. Published July 20, 2017.
Psychiatry. 1987;150(6):782-786. Accessed May 10, 2019.
56.Beck CT, Gable RK. Postpartum Depression Screening Scale: develop- 60.2018 CA legislation. 2020 Mom website. https://www.2020mom.org/
ment and psychometric testing. Nurs Res. 2000;49(5):272-282. 2018-ca-legislation. Updated 2018. Accessed May 10, 2019.
57.Kroenke K, Spitzer RL, Williams JB. The PHQ-9: validity of a brief 61.US Preventive Services Task Force; Curry SJ, Krist AH, Owens DK,
depression severity measure. J Gen Intern Med. 2001;16(9):606-613. et al. Interventions to prevent perinatal depression: US Preventive Ser-
58.Löwe B, Kroenke K, Gräfe K. Detecting and monitoring depres- vices Task Force recommendation statement. JAMA. 2019;321(6):580–
sion with a two-item questionnaire (PHQ-2). J Psychosom Res. 587.
2005;58(2):163-171. 62.Polit DF, Beck CT. Nursing Research: Generating and Assessing
59.Bringing Postpartum Depression Out of the Shadows Act [blog Evidence for Nursing Practice. 10th ed. Philadelphia, PA: Wolters
post]. Perinatal Mental Health Care: A Psychology Clinic website. Kluwer; 2017.

Journal of Midwifery & Women’s Health r www.jmwh.org 13