Preventive Factors of Postpartum Depression in Adult Mothers: A Systematic Review of Randomized Controlled Trials


 Background: Postpartum Depression (PPD) is defined as major depressive disorder with peripartum onset. It is the most common mental illness that occurs after childbirth, affecting not only mother but also newborn. Considering the great impact PPD has, it is relevant to investigate the effectiveness of preventive interventions.Objective: This review investigates published literature of the last decade (2009-2019) regarding the topic of preventive factors of PPD in adult mothers.Methods: An electronic search was conducted by using PubMed, psycINFO and Google Scholar. The search term included a combination of postpartum depression, postnatal depression, predictor, prevention and risk factor. Inclusion criteria were: 1) randomized controlled trials (RCT) published in English or German language that deal with 2) adult pregnant or adult mothers (>18 years). Extraction of articles by two independent raters using predefined data fields, including study quality indicators.Results: The electronic search resulted in a total of 905 articles, whereby only 22 articles were relevant for this synthesis. The interventions were divided into five major topics, these being psychological, psychosocial, nutritional, medical, and exercise interventions. Significant preventive effects concerning PPD in adult mothers were found in psychological and psychosocial interventions. Nutritional, medical and exercise interventions did not have a significant preventive impact.Conclusion: Psychological and psychosocial interventions seem to be effective in the prevention of PPD. Consequently, preventive programs should include such components. Implications for future research might include the comparison of already existing prevention programs for PPD and the observation of their long-term effects.

interventions, prenatal antenatal and postnatal education programs and infant massage training were effective; as well as psychological interventions including psycho-educational and cognitive behavioral therapy and group interpersonal therapy.
Werner et al. [19] conducted a literature review regarding biological, psychological, and psychosocial preventive interventions for PPD. They analyzed 45 published, English-language RCTs. Out of those, 20 showed clear positive effects of an intervention. Among biological studies, the intake of calcium, sertraline and selenium had successful effects. Out of the psychological interventions interpersonal therapy, individual cognitive behavioral therapy (rather than group-based interventions) and postnatal psychological debrie ng found positive effects. Out of the psychosocial interventions only antenatal and postnatal classes were effective.
It is important to foster the research on the prevention of PPD, since it is a disease with serious impact on mother and child. The aim of this review is to provide a current overview on published RCTs investigating preventive factors for PPD. Different interventions designed to protect high risk as well as low risk pregnant and postpartum adult women in culturally diverse samples are investigated.

Methods
The systematic review was conducted in accordance with the Prisma guideline [22]. The search was applied to PubMed, psycINFO and Google Scholar. All articles published between the years 2009 to 2019 were considered. The following search term was used to search all databases: ((("postpartum depression" OR "postnatal depression") AND "predictor" [Title/Abstract]) OR ("postpartum depression" OR "postnatal depression") AND "prevention" [Title/Abstract]) OR ("postpartum depression" OR "postnatal depression") AND "risk factor" [Title/Abstract]).
Primary outcome measures were depressive symptoms and major or minor depression with peripartum onset. No restrictions regarding types of intervention were imposed. The inclusion criteria were as following: (1) already published studies in English or German language; (2) studies involving human subjects; and (3) adult pregnant or adult mothers (>18 years). At the second step research designs other than RCT, such as non-comparative studies, qualitative studies, literature reviews, systematic reviews and meta-analyses, were excluded. Including only randomized studies ensured a reduction of bias and the integration of studies of a high-quality design.
Eligibility assessment was performed in an unblinded standardized manner by two independent raters (NM, KF). Disagreements were resolved by consensus between the two review authors (NM, KF); if no agreement could be reached, a third author was consulted (TO). All titles and abstracts were reviewed by NM to assess their eligibility. The eligibility of the included studies was checked by KF. Afterwards, the full texts from the eligible studies which report on risk and preventive factors of PPD were analyzed by NM and checked by KF.
The quality assessment tool Qualsyst [23] was used to evaluate the quality of the included studies. It includes 14 items, for example "Question/objective su ciently described?" and "Study design evident and appropriate?" [23] (the remaining items are detailed in Table 1). Each of the items is scored on a 3point Likert scale ("yes" = 2, "partial" = 1, "no" = 0), not applicable criteria are marked "n/a" [23]. For each paper a Qualsyst sum score and a percentage score was calculated. Criteria marked with "n/a" were excluded from both calculations. The quality classi cation was based on the percentage score: poor quality (score<50%), fair quality (score=50-69%), good quality (score=70-79%) and strong quality (score>80%) [21].
The following information was extracted from the included studies: author(s), year of publication, study title, study design, sample characteristics (mean age, inclusion/exclusion criteria, sample size), type of intervention (including duration and frequency), control intervention, protective factors for PPD, outcome measures and ndings. The structured approach commonly known by the acronym PICOS ("the patient population or the disease being addressed (P), the interventions or exposure (I), the comparator group (C), the outcome or endpoint (O), and the study design chosen (S)" [24]) was used for the evaluation of results.

Search Results
A total of 905 records were identi ed in the databases. After removing duplicates, 884 records were screened for relevance in protective factors for PPD based on titles and abstracts. 751 records were excluded based on not tting content and 11 records were not available in the English or German language. No unpublished relevant studies were obtained. The remaining 153 records were screened for full text of which 106 records had a different study design than RCT, 9 records included participants under the age of 18, 8 records were study protocols, therefore not yet completed, and 8 records were thematically un tting. This resulted in 22 records that met the inclusion criteria and were included in the systematic review (see Figure 1).

Methods
All studies nally selected for the review were RCTs published in English. The duration of the interventions varied strongly, with the shortest one being only one single session during which a drug was administered [39] and the longest being 12 months with a follow-up at 27 months [5].

Participants
The 22 included studies involved 10448 participants in total. Study sample sizes ranged from 27 to 2,241 participants. The main inclusion criteria were ful lling certain age limits, if pregnant within determined weeks of gestation, having a history of depression/scoring a set baseline score on a depression instrument or having no depression at all.

Interventions
In the reviewed studies, psychological (n=7642), psychosocial (n=1318), nutritional (n=185), medical (n=330) and exercise-based (n=849) interventions were investigated. Table 2 provides an overview about the interventions and their effect. Out of the 22 studies 15 turned out to have a signi cant impact on the prevention of postpartum depression. 7 studies had no overall signi cant impact. In the following section the reviewed studies are reported in detail.

Psychological interventions
Lara et al. [33] conducted the rst randomized control depression prevention trial on high-risk pregnant Mexican women. 138 adult pregnant women with ≤ 26 weeks gestation and a baseline depression score on CES-D > 16 or a self-reported history of depression, took part in an 8-week prospective. The psychoeducational intervention included a manual that covered the following three components: educational (information on normal pregnancy and the postpartum period), psychological (aimed at reducing depression trough e.g. increasing positive thinking) and group component (atmosphere of support of facilitator). The control group received usual postpartum care. Primary outcome were major depression and depressive symptoms, anxiety symptoms as well as perceived intervention impact. There were signi cantly fewer new depression cases in the intervention group (p<0.05). Repeated-measures analysis of variance found neither signi cant effects of treatment (p=0.54) nor signi cant effects within group (time effect) (p=0.00) and in the interaction of time by condition (p=0.02). Main effect for treatment and for initial CES-D level was signi cant (p=0.54), but interaction between initial level of symptoms and treatment was not signi cant. Within groups signi cant effects of time (p=0.00), interaction between time by initial level of depression (p=0.00), time by treatment (p=0.02), and time by treatment by initial level of depression (p=0.04) were found.
Brugha et al. [4] tested in a cluster RCT, whether receiving care from a specialized health visitor is effective in preventing depression in low-risk women. 2241 adult pregnant women at low risk (EPDS score < 12) received either psychologically orientated sessions based on cognitive behavioral therapy (CBT) or person-centered principles for 8 weeks or usual postpartum care. Primary outcome was depressive symptoms. At 6 months postpartum more women in the control group scored ⩾12 on the EPDS than in the intervention group (p=0.016). The covariates living alone, having a history of postnatal depression, having experienced an adverse life event and the 6-week EPDS score were signi cant predictors of PPD at 6 months (p=0.031).
Mao et al. [7] studied the impact of an emotional self-management training program on postnatal depression symptomology. 240 primiparous pregnant females enrolled in the RCT. The emotional selfmanagement group training program, which is based on CBT with elements of Chinese culture of delivery included 4 weekly group sessions and 1 individual counseling session, whereas the control group received usual antenatal care (four 90-minute childbirth education sessions). Primary outcome was depressive symptoms. Results show signi cantly lower mean score of PHQ-9 at 36 weeks antenatal in the intervention group compared to the control group (p<0.01) and signi cantly lower mean score of EPDS at 6 weeks postnatal in the intervention group than in the control group (p<0.05). This resulted in more people having PPD within the control group (p<0.05).
In Kozinszky et al.'s [32] study the impact of a preventive group intervention for PPD was investigated. All pregnant women living in the catchment area were approached for the RCT, resulting in 1719 participants. Psychoeducation and psychotherapy for PPD were held in group sessions, including elements of interpersonal psychotherapy (IPT) and CBT. The control group received treatment as usual but in group meetings, also 4 times. Primary outcome was depressive symptoms. There were signi cant lower LQ scores (p<0.001) and PPD prevalences (p<0.01) 6 weeks postpartum in the intervention group than the control group.
Moshki et al. [26] assessed the effectiveness of application of health locus of control (HLC) for prevention of PPD. In a pre-post experimental design with a control group, 230 pregnant females with 28-30 weeks gestation took part. The intervention combined information on anatomical and physiological changes, nutrition, common complications during pregnancy, mental health and communication skills, familiarity with stages of pregnancy, delivery and pain reduction methods, postpartum health, emotions and attitudes of women. This was performed with particular emphasis on components of the HLC, including internal HLC, strong other HLC and random HLC, lasting in total 36 hours. The control group received usual postpartum care. Primary outcomes were health behaviors according to participants' beliefs and depressive symptoms. This led to a signi cantly reduced chance HLC (p=0.003) and a signi cantly increased internal HLC (p=0.03) immediately after the intervention. Moreover, a signi cant difference between the two groups in reduced PPD outcomes one month after intervention (p=0.001) and a signi cant correlation between internal belief (p<0.05) and chance belief (p<0.01) with EPDS was found.
Hiscock et al. [6] evaluated in a RCT a program for preventing infant sleep and cry problems and postnatal depression. 781 infants born at 32 weeks or later were recruited with their families to participate. The intervention supplied various informations on topics like infant sleep, cry behavior and settling techniques. Families in the control group received usual childcare. Primary outcome was depressive symptoms and sleep quality and quantity. Between the two caregiver groups there were no differences in depression symptoms at 4 months (p=0.07), but a signi cant greater reduction of depressive symptoms in the intervention caregivers at 6 months (p=0.03) was found.
Nugent et al. [27] tested the e cacy of the Newborn Behavioral Observations (NBO) in lowering postnatal depression. 104 primiparous mothers were eligible for the participation in this RCT. The inclusion criteria were raising the baby with the baby's father, having a vaginal delivery between 36 and 42 weeks of gestation and an Apgar score above 7. The intervention included routine care plus NBO within two days post-delivery (in hospital) and one-month postpartum (during home visit), meaning 18 NBO of behavioral spectrum of infants along four dimensions (autonomic, motor, state organization and attentionalinteractional). The control group received, next to the usual postpartum care, a short home visit to carry out the EPDS. Primary outcome was depressive symptoms. The intervention group had signi cantly fewer mothers with elevated depression scores compared to the control group (p=0.05).
Maimburg et al. [36] studied the in uence of a structured general antenatal program on PPD. 1062 nulliparous adult pregnant females between 10 + 0 to 21 + 6 days of gestation with a singleton pregnancy took part in the parallel RCT. The antenatal education program was conducted between the 30th -35th weeks of gestation. It provided material on the content of the birth module, the newborn module and the parent module with information about PPD, lasting 9 hours in total. The control group received usual antenatal care. Primary outcome were depressive symptoms and identifying risk factors for PPD. There were no signi cant differences in EPDS scores at 6 weeks postpartum between the two groups (OR= 0.89, 95% CI: 0.57-1.40).
Dimidjian et al. [30] evaluated treatment acceptability and e cacy of mindfulness-based cognitive therapy adapted for perinatal depression (MBCT-PD). 86 adult pregnant females with >32 weeks gestation that met criteria for prior major depressive disorder (but not in the last two months) participated in the pilot randomized clinical trial. The MBCT-PD consisted of eight sessions and additionally at-home practice for 6 days each week between Sessions 1 and 7. The intervention lasted 42 days in total. They were compared to the control group, which received usual antenatal care. Primary outcomes were baseline diagnostic status, treatment acceptability, women's satisfaction with study condition, depression relapse/recurrence and depression symptom severity. The intervention group had signi cantly improved depressive outcomes compared to the control group, including signi cantly lower rates of depressive relapse/recurrence (p=0.005) in the postpartum period and fewer depressive symptoms throughout the trial (p=0.002).
Werner et al. [41] examined the effectiveness of Practical Resources for Effective Postpartum Parenting (PREPP) to treat women at risk for PPD. In a RCT 54 pregnant females in 2nd or 3rd trimester of their pregnancy and at high-risk (score of > 24 on Predictive Index of Postnatal Depression), aged 18-45 years took part. The intervention included 3 PREPP sessions in which ve speci c infant behavioral (feeding, day-night rhythm, carrying duration, swaddling) and mindfulness techniques were covered. Moreover, the participants received supportive psychological interviewing (using motivational interviewing techniques) and psychoeducation conducted by a psychologist. Participants in the control group received enhanced treatment as usual, they met on two occasions with a psychologist and were provided with information about PPD and referred to treatment if necessary. Primary outcome were depressive symptoms, symptoms of anxiety and infant fuss/cry episodes. This showed signi cantly decreased depressive symptoms in the intervention group in contrast to the control group (p=0.01) at 36-38 weeks gestation and 6 weeks postpartum and signi cantly more depressive symptoms at 6 weeks postpartum in control group in contrast to intervention group (p=0.02).
Tsai et al. [25] evaluated the link between food shortages, depression and the role of social support. For this population-based, cluster-randomized trial all pregnant females living in the study catchment area who were willing to participate were recruited, resulting in 958 subjects. The home visiting intervention offered problem-solving and cognitive-behavioral techniques on topics like HIV, tuberculosis, malnutrition and alcohol use in 4 visits and a 36-month follow-up interview. The control group received standard clinic care which consisted of primary health care, tuberculosis and HIV testing, antiretroviral therapy, antenatal and postnatal care. Primary outcomes were depression symptom severity, number of days of hunger in the past week and emotional and instrumental support. This trial showed a strong signi cant association between lagged food insu ciency and depression symptom severity (β=0.70; 95% CI, 0.46-0.94), no statistically signi cant association with food insu ciency and depression symptom severity among women with a level of instrumental support greater than the median (β=0.12; 95% CI, −0.35 to 0.59), but a signi cant association between food insu ciency and depression among women with low levels of instrumental support (β=0.79; 95% CI, 0.51-1.07).
Duncan et al. [31] conducted a mindfulness-based childbirth preparation course. In the RCT 29 primiparous pregnant females in the last part of their singleton pregnancy with no extensive prior experience with meditation or yoga practice were suitable. The intervention group participated in the program called Mind in Labor (MIL): Working with Pain in Childbirth [31], which is a 2.5-day mindfulnessbased childbirth preparation course based on Mindfulness-Based Childbirth and Parenting (MBCP). The control group received usual antenatal care and a list of approved, comparable childbirth courses, but with no mind-body focus. Primary outcomes were childbirth self-e cacy, maladaptive pain appraisal, perceived pain in labor, use of pain medication in labor, using medical record review, birth satisfaction and depression. Results for perinatal and PPD symptoms showed a signi cant interaction between group and time (p=0.04), the estimated treatment effect was -3.34 (80% CI [-5.22, -1.28]).

Psychosocial interventions
Barnes et al. [28] evaluated in a cluster-randomized study if a home-visiting support program is effective in preventing maternal depression. 250 adult mothers with social disadvantage (Social Disadvantage Screening Index score >9), infant birthweight ≥2500g and infant being under 5 days in special care baby unit took part. The home visits included different activities like providing advice or assistance with childcare and varied individually in length and frequency. The matched control group receiving no support and those who declined home-start-support also receiving no support were compared to the intervention group. Primary outcomes were major or minor depression and depressive symptoms. Depression at 2 months was the only signi cant predictor for having depression at 12 months. Social support at 2 months led to signi cantly lower depression scores (p<0.0001).
In a cross-sectional RCT, Dennis [29] evaluated whether peer support is effective in the prevention of PPD. 701 adult mothers with a EPDS baseline score < 9 participated. In contrast to the control group, which just received usual postpartum care, the intervention group also received telephone-based peer support, which provided emotional, informational and appraisal support. Primary outcome was maternal perceptions of peer support, which was evaluated by using PSEI at 12 weeks postpartum. The items in Maternal perceptions of perceived bene ts subscale the perspective Stress and Coping in the domain Depression showed moderate to strong correlations (Life is more enjoyable (r=0.41), Less depressed (r=0.62) [29]), resulting in signi cantly lower EPDS scores at 12 weeks in the intervention group (p<0.001).
Dugravier et al. [5] tested in a prospective RCT the impact of a home-visiting program conducted by psychologists on postnatal depression. 367 primiparous pregnant females with 12-27 weeks gestation who are under 26 years old and ful ll at least one of the following characteristics: "1) having less than twelve years of education, 2) planning to bring up their child without the child's father, 3) having low income" [5] took part. The CAPEDP (Compétences parentales et Attachement dans la Petite Enfance: Diminution des risques lies aux troubles de santé mentale et Promotion de la résilience -Parental Skills and Attachment in Early Childhood: reducing mental health risks and promoting resilience [5]) included 14 intensive multifocal home visits done by psychologists who promoted mental and physical health and provided support. It lasted for 12 months, with a 27-month follow-up. The control group received usual postpartum care. Primary outcome was depressive symptoms. The home-visiting program had no overall impact on postnatal depression symptomatology, however post-hoc analysis revealed that the intervention group had signi cantly lower EPDS scores in the following subgroups: women with lower depressive scores at inclusion (p = 0.05), women who were planning on cohabiting with the child's father (p = 0.04) and well-educated women (p = 0.05).
Lewis et al. [35] examined the relation between employment status and PPD among high-risk women. This was a post hoc analysis from a previously conducted RCT, which was also included in this review and is described in more detail below (see [34]). 124 healthy women (six participants were lost to followup), who had a (maternal family) history of depression aged between 18-42 years have been examined. They lled out a demographic questionnaire, completed a telephone interview at the beginning of the study on race/ethnicity, age, body mass index and use of antidepressants, and a telephone interview after 7 months on employment and breastfeeding status and history of depression. Lewis et al. found evidence for a signi cant association between employment status and higher depression symptomatology (p=0.010), especially employed women at 7 months postpartum had signi cantly fewer depressive symptoms than non-employed women. However, there were no signi cant differences between women who worked full-time versus part-time on depression symptoms (r=-0.022, ns.). They also found a signi cant correlation between being single and higher depression symptomatology at 7 months postpartum (p=0.001).

Nutritional interventions
Mozurkewich et al. [37] conducted a double-blind RCT to investigate whether different types of sh oil supplementation lowers depressive symptoms during pregnancy or postpartum. 126 adult pregnant women with 12-20 weeks gestation and at risk for PPD took part. They either received 6 capsules of EPArich (1060 mg EPA plus 274 mg DHA) sh oil supplementation or DHA-rich (900 mg DHA plus 180 mg EPA) sh oil supplements or soy oil placebo twice daily for 3 months. No differences between the groups in any depression outcomes at any of the three time points after supplementation were found (signi cance=0.051). but DHA levels did predict BDI score at 34-36 weeks (p<0.05).
Vaz et al. [38] evaluated the e cacy of sh oil supplementation on the prevention of PPD. 32 pregnant females who were at risk for PPD (past history of depression or baseline EPDS score ≥9), between 5-13 weeks of gestation at the time of enrolment and aged 20-40 years were eligible to participate in the randomized, placebo-controlled, double blind trial. They either received 6 capsules per day of sh oil (1.08 g of EPA and 0.72 g of DHA) or soybean oil placebo starting at week 22-24 of gestation for 16 consecutive weeks. Primary outcomes were prevalence of EPDS, length of gestation and birth weight.
There were no signi cant differences between intervention and control groups in EPDS scores (T0 p=0.809, T1 p=0.299, T2 p=0.630, T3 p=1.000), but women in the intervention group with a history of depression scored signi cantly lower on the EPDS from the second to the third trimester in the ITT analyses compared to the control group (p=0.038).
Young et al. [40] analyzed the maternal placentophagy's effect on postpartum mood, energy, and bonding. 27 adult pregnant females who had previously decided they would ingest their placenta after birth took part in the randomized, double-blind, placebo-controlled pilot study. Participants consumed their processed, encapsulated placenta: two 550 mg capsules 3 times daily for the rst 4 days of supplementation, decreased the dosage of 2 capsules twice daily on days 5 through 12 of supplementation, and reduced the dosage to 2 capsules once daily through completion. The control group received a similarly prepared placebo with the same intake times. Both groups had 4 meetings with researchers. Primary outcomes were basic demographic information, various psychometric, social and lifestyle variables, depression symptoms, salvia sample, length and weight of the neonate. No signi cant main effects were found (p=0.016).

Medical interventions
Xu et al. [39] evaluated in a randomized, double-blind, placebo-controlled design the preventive effect of ketamine on PPD. 330 women who were scheduled to undergo caesarean section with spinal anesthesia received either intravenous low-dose ketamine (0.25 mg/kg diluted to 10 mL with 0.9% saline) or placebo (10 mL of 0.9% saline) within a short time after giving birth. Primary outcome was depression. There were no signi cant differences found between the two groups in the prevalence of PPD at 3 days (p=0.965) and 6 weeks (p=0.900) after delivery.

Exercise
Lewis et al. [34] examined in a RCT the effectiveness of a telephone-based exercise program in preventing PPD among high-risk women. 130 adult mothers with a (maternal family) history of depression who are currently not exercising were recruited for The Healthy Mom Study. The intervention group took part in a telephone-based exercise intervention including telephone counseling sessions, motivational print materials and completion of exercise logs. The wellness/support contact control condition provided a counselor who only gave support for general issues related to health and wellness and no information regarding exercise, also lasting 6 months. Primary outcome was depression. There were no signi cant differences in PHQ-9 (p=0.09), PSS (p=0.87), and PSQI (p=0.995) scores between participants in the two groups. The three baseline measures were all correlated with one another: PHQ-9 scores were signi cantly correlated with the PSS scores (r=0.43, p<0.0001) and PSQI scores (r= 0.41, p<0.0001).
Songøygard et al. [8] investigated whether exercising during pregnancy is effective in preventing postnatal depression. They conducted a RCT with 719 adult pregnant women. The intervention consisted of a 12-week exercise program exercise groups once weekly and a 45-min home exercise program several times a week, starting from week 20 of pregnancy. The control group received usual antenatal care.
Primary outcome was depressive symptoms. There were no signi cant differences in depressive symptoms between the two groups (p=0.35), but women in the intervention group who did not engage in physical activity regularly before pregnancy had a reduced risk of postnatal depression (p<0.05).

Discussion
PPD is one of the most frequent mental diseases of the perinatal period [1] and the importance of prevention of PPD has been highlighted frequently [11,19,21]. Therefore, the aim of this current review was to provide an overview of current literature on preventive interventions for PPD in adult mothers. In our systematic review we have investigated 22 studies in total. Out of these, 12 studies included psychological interventions, four studies included psychosocial interventions, three studies included nutritional interventions, one study performed a medical intervention and two studies included exercise programs. 15 studies had a signi cant effect in preventing PPD. All psychological interventions investigated showed a signi cant preventive effect concerning PPD, except for one [36]. The psychological interventions which included psychoeducation, elements of CBT and IPT, prevention programs, home visits and a focus on mindfulness had a signi cant impact in the reduction of PPD. The study of Maimburg et al. [36] carried out an antenatal education program with information about birth and PPD. In comparison to the other studies investigated the intervention was carried out relatively late in pregnancy (30-35 weeks of gestation), which may be a possible explanation for lack of signi cant preventive effect.
All psychosocial interventions showed signi cant effects, they contained home-visiting and peer support. Nutritional interventions did not show any signi cant effects. In the studies of Mozurkewich et al. [37] and Vaz et al. [38] Omega 3 supplementation was given to pregnant women. Both studies did not show signi cant results regarding the prevention of PPD, which leads to the assumption that Omega 3 is not effective in that case. Young et al. [40] investigated the effects of placentophagy. No signi cant results were found, this might be due to the small sample size (n=27). The medical intervention showed no signi cant effect, this could be caused by the fact that in the study of Xu et al. [39] a relatively low dose of ketamine (0.25 mg/kg) was used. Exercise programs did not result in an overall lower prevalence of PPD, but only partially in one study [8]. One difference was, that Lewis et al. [34] examined women postpartum (on average 5.7 weeks postpartum), whereas the intervention of Songøygard et al. [8] took place during pregnancy. Songøygard et al. found that a subgroup of women in the intervention group who did not exercise regularly prior to pregnancy but then exercised during pregnancy had a reduced risk of PPD, indicating that exercise programs during pregnancy can be effective among certain participants for preventing PPD.
These results can be compared with the ndings of previous reviews. Sangsawang et al. [21] found effective results for psychological and psychosocial interventions, whereas Werner et al. [19] found both signi cant and non-signi cant effects for psychological and psychosocial interventions. In line with those ndings, we found psychological and psychosocial interventions to be effective in the prevention of PPD. Werner et al. showed that biological interventions, namely with anti-depressants and nutrients provided the most evidence, but interventions using thyroxine, Omega 3 and progestogen showed negative results. In our study no comparable intervention was successful. A factor that could be responsible for this is that these groups had the lowest total number of participants (nutritional n=185, medical n=330). Parallels can be drawn for Omega 3, since both studies in our review investigating the effect of Omega 3 on PPD in high risk women led to negative results, as well as the three negative results The majority of the studies did not put any emphasis on reaching over or under certain cut-offs for baseline depression scores, nor was there a special focus on ethnic minority populations. Due to the fact, that the studies were conducted globally, a wide variability of various populations in different cultures and countries was ensured. The ndings may be generalized and implemented in the general adult female population of childbearing age.
Most of the studies (n=19) classi ed as strong quality, none had poor quality. The items 5 (random allocation), 6 and 7 (blinding of investigator and subjects) and 9 (appropriate sample size) were the ones where reporting was most lacking. Therefore, more attention should be drawn to proper randomization techniques including the description of the method used, if applicable reporting of blinding of investigators and subjects, if not, clearly saying why not applicable and having a large enough sample size based on sample size calculations and mentioning of power/effect size.
It is important to foster the research regarding the topic of preventive interventions for PPD since there are ambivalent ndings. Implications for future research might include the comparison of already existing prevention programs for PPD and the observation of their long-term effects.

Limitations
Our study has several limitations. The search term was strictly limited to a combination of postpartum or postnatal depression, predictor, prevention and risk factor. This excludes some studies that may be important for this work in advance. For example, studies that focus on the Baby Blues were neglected, which might be interesting since there is a link between Baby Blues and PPD [10]. Furthermore, only three databases were used to conduct the electronic search, meaning there was no greater variability in the outcome sources. Another limitation is that the screening of PPD symptoms varied widely between the different studies. Although at the beginning we determined the primary outcomes with depressive symptoms and major or minor depression with peripartum onset, it would have been more consistently and comparable if we had decided on a reliable screening instrument, like the EPDS, to measure PPD [42].

Consent for publication
Not applicable.

Availability of data and materials
All data generated or analyzed during this study are included in this published article and its supplementary information les.

Competing interests
The authors declare that they have no competing interests.    Figure 1 Flow diagram of the selection process

Supplementary Files
This is a list of supplementary les associated with this preprint. Click to download. CoverLetterNM30.5.docx