Prenatal and Postpartum Maternal Psychological Distress and Infant Development: A Systematic Review

Child Psychiatry Hum Dev (2012) 43:683 714 DOI 10.1007/s10578-012-0291-4 REVIEW PAPER Prenatal and Postpartum Maternal Psychological Distress and Infant Development: A Systematic Review Dawn Kingston Suzanne Tough Heather Whitfield Published online: 10 March 2012 Ó Springer Science+Business Media, LLC 2012 Abstract Infant development plays a foundational role in optimal child development and health. Some studies have demonstrated an association between maternal psychological distress and infant outcomes, although the main emphasis has been on postpartum depression and infant-maternal attachment. Prevention and early intervention strategies would benefit from an understanding of the influence of both prenatal and postpartum maternal distress on a broader spectrum of infant developmental outcomes. We conducted a systematic review of studies assessing the effect of prenatal and postpartum maternal psychological distress on five aspects of infant development: global; cognitive; behavioral; socio-emotional; and psychomotor. These findings suggest that prenatal distress can have an adverse effect on cognitive, behavioral, and psychomotor development, and that postpartum distress contributes to cognitive and socio-emotional development. Keywords Infant development Maternal psychological distress Systematic review Maternal psychosocial care Background Healthy child development has been viewed as a necessary foundation for reducing health and social inequities across the life course [1, 2]. Early years programming has been an important strategy for the prevention of developmental problems, largely influenced by an increasing understanding of environmental influences on the neuroplasticity of the young D. Kingston (&) Faculty of Nursing, Rm 5-258 Edmonton Clinic Health Academy, University of Alberta, 11405-87th Avenue, Edmonton, AB T6G 1C9, Canada e-mail: Dawn.Kingston@ualberta.ca S. Tough Faculty of Paediatrics and Community Health Sciences, Centre for Child, Family, and Community Research, University of Calgary, Calgary, AB, Canada H. Whitfield McMaster University, Hamilton, ON, Canada

684 Child Psychiatry Hum Dev (2012) 43:683 714 brain [2]. From an economic standpoint, investment in the early years in the form of quality education, child development, and parenting programs has shown greater return than investments allocated post-kindergarten [2]. Yet, based on the Early Development Index (EDI), 25 30% of children in Canada enter school with some form of physical, socioemotional, or cognitive-language delay, and Canadian trends reveal an increase in developmental vulnerability in several provinces over the past decade [2, 3]. Advocates suggest that the decline in healthy child development at a time when there has been increased attention in this area implies that greater investment in early years services is required [2]. However, it also compels us to consider earlier influences in a child s life that have not been sufficiently addressed to date. Tandem to the movement to develop and enhance universal services to support healthy early years development has been a growing interest in the long-term effects of risks that occur during pregnancy and the postpartum period on child development. Although maternal health may represent a key point of early, upstream intervention, the evidence surrounding early life factors has not been translated into prevention/intervention strategies or policy. The recently released Marmot Review, Fair Society, Healthy Lives, proposed a second revolution in the early years to increase the support of parents starting in pregnancy and continuing through primary school [4]. This report recommended giving priority to pre- and post-natal interventions that reduce adverse outcomes of pregnancy and infancy (p. 16) [4]. An earlier report by the World Health Organization s Commission of Social Determinants stated, Implementing a more comprehensive approach to early life includes comprehensive support to and care of mother before, during, and after pregnancy including interventions that help to address prenatal and postnatal maternal mental health problems (p. 53) [5]. These documents formally acknowledge the important influence of maternal health during the prenatal and postpartum periods in child health and development. Given that maternal psychological distress (e.g., stress, anxiety, depression) in pregnancy is common [6, 7], and a substantial proportion of women who experience distress in pregnancy or during the postpartum period continue to have symptoms into their child s early years [8 10], maternal psychological distress represents a prevalent, enduring, and modifiable influence that may significantly impact fetal and child development. Interest in the effect of maternal psychological distress on infant outcomes and its underlying mechanisms has surged over the past two decades. While much of the early research in this field focused on studying the impact of postpartum depression (PPD) on outcomes such as maternal-infant interaction and infant temperament, more recent studies have explored the effects of different forms of maternal psychological distress and their timing on a broader array of infant outcomes, including infant development. Some key findings of this body of evidence suggest that: (a) developmental delay in infancy is associated with delay in later childhood stages [11]; (b) predictors of neurodevelopmental delay can be detected during the first 10 months of life [12, 13]; (c) infant developmental delay and its causes are amenable to early intervention targeted at the infant and its family [14, 15]; and (d) intervention aimed at reducing maternal psychological distress can lower the risk of adverse infant developmental outcomes [16]. Very few reviews have synthesized evidence relating prenatal and postpartum maternal psychological distress to infant development. Of these, few are methodologically rigorous, systematic reviews; most address a single form of maternal distress during the prenatal or postpartum period; and, the majority focus on older children. Given the importance of the role of infant development in future child development and health, a substantive review of the impact and magnitude of effect of maternal prenatal and postpartum psychological

Child Psychiatry Hum Dev (2012) 43:683 714 685 distress on infant development would inform prevention, early intervention, and policy strategies for reducing the risks of developmental delay that occur in a child s earliest environment. Objectives The objectives of this systematic review were to: (a) assess the relationship between prenatal and postnatal maternal psychological distress and infant development from birth to 12 months; (b) estimate the magnitude of effect of the relationship between various forms of maternal psychological distress and infant developmental outcomes; (c) describe the quality of the evidence for the relationship between maternal psychological distress and infant development; (d) identify gaps in the existing evidence; (e) describe the implications of the review findings; and (f) formulate research, clinical, and policy related recommendations. Methods Inclusion and Exclusion Criteria and Definitions Studies were included in this review if the: (a) exposure was any form of maternal psychological distress (e.g., anxiety, depression, stress, psychological distress) occurring during pregnancy or the postpartum period (i.e., 1 year following birth); (b) outcome was a measure of child development that was assessed from birth up to and including 12 months and included global indices of development, behavior, cognitive development, socio socio-emotional development, and psychomotor development; (c) study recruited women and children from developed countries; (d) study was published in English; and, (e) study was a primary study that was published between 1990 and 2010. Studies were excluded from this review if: (a) maternal distress during the prenatal or postpartum period was part of a composite variable that extended beyond 1 year postpartum; (b) the exposure was a pharmacologic treatment for maternal distress; or, (c) the study did not have a comparison group. The categorization of infant outcomes (e.g., as cognitive or behavioral) was based on the investigators own descriptions. A global index of development combined a number of developmental components (e.g., socio-emotional, behavioral, cognitive development) into a single index (e.g., Child Behavior Checklist Total Score; Bayley Scales of Infant Development). Trimesters of pregnancy were categorized as first (0 13 weeks gestation), second ([13 26 weeks), and third ([26 40? weeks). Effect sizes were based on Cohen s guidelines (trivial when r \ 0.10; small when d = 0.20 or r = 0.10 0.30; medium when d = 0.50 or r = 0.30 0.50; and, large when d = 0.80 or r C 0.50) [17]. Odds ratios (ORs) of \1.7 were defined as small, [1.7 2.5 were medium, and [2.5 were large [17]. Search Strategy, Title and Abstract Review, Critical Appraisal, Data Extraction, Analysis The search strategy was developed in consultation with a university-based librarian. Five electronic databases were searched, including Embase, CINAHL, Eric, PsycInfo, and Medline. Reference lists were reviewed and key journals were hand-searched. The search

686 Child Psychiatry Hum Dev (2012) 43:683 714 encompassed the period from January 1, 1990, to August 10, 2010. The detailed search strategy is available from the authors. The titles and/or abstracts of each article were reviewed independently by 2 individuals based on the a priori inclusion and exclusion criteria. Disagreements related to inclusion or exclusion were resolved by discussion and consensus. In cases where decisions could not be reached based on title or abstract review, the full-text version of the article was retrieved and reviewed. A modified version of the critical appraisal form for observational studies developed by the Scottish Intercollegiate Guideline Network (http://www.sign.ac.uk/methodology/checklists.html) was used to assess the quality of each article. Each study was appraised based on its study design, potential for selection bias, confounders, withdrawals and dropouts, follow-up, blinding, and measurement of exposures and outcomes. The quality of the articles was assessed by 2 independent reviewers with experience in critical appraisal. Disagreements were resolved by consensus. Data were extracted by one reviewer using a standardized data extraction form that was developed for this review. Studies were also reviewed for the potential to conduct a meta-analysis. Results: Overview The search strategy yielded a total of 17,792 studies with 18 studies having infant outcomes (Fig. 1). These studies recruited participants from a variety of different countries (e.g., New Zealand, Finland, Netherlands, US, Australia, Sweden, Israel). The majority of these studies were published in the past decade (n = 16) with over 60% of these being published in the past 5 years (2005 2010). All but 2 studies were longitudinal and the majority (n = 12) represented community-based samples. Among these 18 studies, most evaluated the effect of maternal distress on infant cognitive (n = 7) or psychomotor (n = 7) development, with fewer assessing the effects on global indices of infant development (n = 2), behavior (n = 2), and socio-emotional development (n = 4). There was a fairly even distribution of the number of studies that assessed prenatal (n = 7) and postpartum maternal distress (n = 8), with 3 studies addressing both. We examined each infant outcome for the potential to conduct a meta-analysis. The diversity of the outcomes (e.g., dichotomous and continuous measures) precluded metaanalysis. We also reviewed the studies for the potential to calculate mean differences where outcomes were dichotomized, but insufficient data were available. As such, a thorough qualitative analysis was conducted. We included all studies in our qualitative analysis because few studies met the inclusion criteria (n = 18), particularly across the various forms of infant development; in addition, their inclusion enabled us to qualitatively describe the heterogeneity that existed between weak and moderate/strong studies [18]. Results: Infant Development Global Indices of Infant Development (Table 1) General Overview of Studies (n = 2) Two longitudinal studies from Finland and New Zealand evaluated the influence of maternal distress on global indices of infant development at 12 months [19, 20]. One study was community-based with half of the mothers having term, small for gestational age infants [20],

Child Psychiatry Hum Dev (2012) 43:683 714 687 *Articles retrieved through search strategy; title and abstract review (n=17,792) Studies excluded (n=17,620) due to: (a) not a primary study or a case study (n=5,753); (b) wrong exposure and outcome (n=8,684); (c) wrong exposure (n=2,660); (d) wrong outcome (n=473); (e) correct exposure and outcome but in older population (n=50) Studies retrieved for more detailed evaluation (n=172) Studies excluded due to recruitment in developing countries; maternal distress measured after 12 months postpartum; maternal and paternal measures of distress not analyzed separately; children <8 years combined with older group (n=108) Studies included in the comprehensive review (n=64) Studies involving outcomes of children aged 13 months to 8 years (n=46) Studies summarized in this paper (n=18) (e.g., outcomes included global indices of infant development, behaviour, socioemotional development, cognitive development, and psychomotor development for infants aged birth to 12 months) Fig. 1 Flow diagram of included and excluded studies. *Note The findings presented in this paper were part of a larger systematic review of studies that examined maternal distress on children s development from birth to age 8. Forty-six studies involved children aged 13 months to 8 years, and this paper summarized the findings of 18 studies of children from birth to 12 months whereas the other recruited half of its participants from infertility clinics [19]. The follow-up sample sizes were 520 [19] and 744 [20] (total N = 1,264).These studies assessed the effect of prenatal trait anxiety in the second trimester (18 20 weeks) [19], prenatal/postnatal perceived stress in late gestation through to the immediate postpartum period [20], parenting stress at 12 months (maternal report) [20], and PPD at 2 and 12 months [19]. Quality The overall quality rating was weak for the Punamaki et al. [19] and moderate for Slykerman et al. [20]. Attrition rates were similar at 25% [20] and 30.2% [19]. The main

688 Child Psychiatry Hum Dev (2012) 43:683 714 Table 1 Key aspects of studies of global indices of infant development (n = 2) Citation/ Quality rating Study design and sample Measure and timing of exposure Infant outcome (assessor) Results (effect size) a Adjusted for key potential confounders Punamaki 2006 19 (n = 520) weak Slykerman 2007 20 (n = 744) moderate (Finland) Community (subsample of women with/without ART) Mean age 33.0 10.4 13.1% unskilled work Community (half SGA) (New Zealand) Mean age 31.3 (no delay) versus 31.6 years (delay) Low SES 64.6% (no delay) versus 36.4% (delay) Single 62.8% (no delay) versus 37.2% (delay) Prenatal anxiety (Spielberger Trait Anxiety Inventory; General Health Questionnaire) and depression (Beck Depression Inventory) (2nd trimester) PPD (Beck) (2 and 12 months) Prenatal perceived stress (Perceived Stress Scale) (assessed after delivery for past month, 3rd trimester); parenting stress at 12 months (single item) Developmental achievement at 12 months (maternal report of infants sleep, standing, walking, words) Developmental milestones at 12 months (Revised Denver Prescreening Developmental Questionnaire) (mother) NS (prenatal anxiety, depression; PPD) Neonatal health a significant mediator (prenatal anxiety? poor neonatal health? infant development) NS perceived stress (AOR 1.45, CI 0.92 2.29); NS parenting stress (AOR 1.47, CI 0.93 2.31) In high perceived stress group, 37.6% infants had development delay; in normal/low stress group, 30.5% had delay In extreme/moderate parenting stress group, 36.8% infants delayed; in low/no parenting stress group, 28.1% delayed Current distress; PPD; neonatal health (S); birth complications; child characteristics Education; marital status (S); occupation; gestational age; birthweight; smoking in pregnancy (S); smoking in first year; pregnancy hypertension; parity; social support at 1 year; breastfeeding; pregnancy marijuana; parenting satisfaction (S) Reported if available or calculable; NS statistically non-significant (p C.05), S statistically significant (p \.05), PPD postpartum depression a

Child Psychiatry Hum Dev (2012) 43:683 714 689 limitation in both studies was the lack of blinding of the outcome assessor (i.e., mother). In both studies, maternal distress was measured using psychometrically tested self-report instruments (Spielberger Trait Anxiety, Beck Depression Inventory [19]; Perceived Stress Scale [20]). Slykerman et al. [20] assessed maternal perceived stress shortly after delivery for the previous month, a period encompassing largely the prenatal, but also early postpartum, periods. Both studies controlled for a number of potential confounders, but only Punamaki et al. [19] adjusted for postpartum and current distress. While Slykerman et al. [20] used the widely validated Revised Denver Prescreening Developmental Questionnaire for infant outcomes, Punamaki et al. [19] utilized investigator-developed questions. Main Findings The proportions of developmental delay in the Slykerman study [20] were 33.8% in appropriate for gestational age (AGA) infants and 32.4% in small for gestational age (SGA) infants (NS difference), and were not reported by Punamaki et al. [19]. Neither study found a significant direct effect of maternal distress in the whole sample; however, in a subsample of SGA infants, Slykerman et al. [20] found that those of mothers with high parenting stress had over twice the odds of developmental delay. Punamaki et al. [19] found that prenatal anxiety played an indirect role in delayed development (see moderators and mediators). Neither study found a significant association between the other forms of distress, including perceived stress, prenatal trait anxiety, or current PPD and global indices of development. Moderators and Mediators Using structural equation modeling, Punamaki et al. [19] found that prenatal anxiety was mediated by neonatal health (i.e., birthweight, apgar scores, intensive care admission, maternal report of infant health). In other words, prenatal anxiety during the second trimester was related to poor neonatal health, which then was associated with developmental problems. In the subsample of women who used assistive reproductive technology (ART) [19], the relationships were somewhat different. Both prenatal depression and anxiety were associated with PPD at 2 months, which was related to developmental problems at 12 months. No direct or indirect effect of current PPD was found. Other Potential Confounders Although both studies assessed a large number of potential confounders, only prenatal smoking (n = 1) [20] and parenting satisfaction (n = 1) [20] were significantly related to child development. Other factors that were not related included socioeconomic status (education, marital status, occupation), birthweight, pregnancy-related hypertension, parity, social support at 1 year, breastfeeding duration, marijuana use in pregnancy, maternal medical problems, gestational age [20], birth complications [19, 20], child temperament, and PPD [19]. Summary of Effect of Maternal Distress on Global Indices of Infant Development One study reported rates of global developmental delay among SGA infants of 37.6% for those whose mothers had high perceived stress compared to 30.5% for those of mothers

690 Child Psychiatry Hum Dev (2012) 43:683 714 with low stress. The authors noted that the high rates of delay observed in this study may be related to their assessment instrument, which was a screening tool intended to identify infants requiring further assessment. Of the 2 studies assessing the effect of maternal distress on global indices of infant development at 12 months, only one showed a small, indirect effect of prenatal anxiety. Based on this single study with an overall quality rating of weak, there is insufficient evidence to support an association between maternal prenatal or postnatal distress and global indices of infant development. Infant Behavior (Table 2) General Overview of Studies (n = 2) One longitudinal [21] and one cross-sectional study [22] explored the relationship between prenatal maternal distress and infant behavior. These community-based studies involved participants from the Netherlands [21] and Spain [22] with respective sample sizes of 131 and 163 (Total N = 249). They explored the effect of a variety of forms of prenatal distress, including third-trimester state-trait anxiety [21, 22], perceived stress, and socioemotional stability [22]. Neither study assessed the effect of postpartum distress. Hernandez-Martinez et al. [22] measured distress during the first few days post-delivery for the period encompassing the previous month (the third trimester of pregnancy). Some distinction was made in the definition of state-trait anxiety. In particular, Brouwers et al. [21] defined high prenatal anxiety as either high state or trait anxiety, whereas Hernandez- Martinez et al. [22] distinguished between these in the analysis. Infant behavior was assessed throughout infancy, including the first few days of life [22] and 3 weeks and 12 months of age [21]. In all cases, the developmental assessment was performed by a researcher or examiner. The early examinations were conducted using the same instrument (Neonatal Behavioral Assessment Scale). Quality These studies achieved overall quality ratings of strong [21] and moderate [22]. The attrition rate for the longitudinal study was less than 20% [21]. In both studies, selfreported maternal distress (State-Trait Anxiety Inventory [21, 22]; Perceived Stress Scale [22]) and infant behavior (Neonatal Behavioral Assessment Scale [21, 22]; Bayley Scales of Infant Development [21]) were assessed using psychometrically evaluated instruments. The retrospective measures of prenatal distress used by Hernandez-Martinez et al. [22] represented a source of limitation in this study. Although confounders were adjusted for in other outcomes of the Brouwers study [21], statistical adjustment was not used in the analysis of infant behavior. Neither study adjusted for the potential influence of postpartum distress on the relationship between prenatal distress and infant behavior. Main Findings The proportion of infants with behavior problems was not reported in either study. Both studies reported significant effects of prenatal anxiety in the third trimester on behavioral outcomes. Specifically, infants of mothers with high state/trait [21] and trait anxiety [22] had lower scores on orientation (e.g., infant attention, alertness) in both studies at 2 3 days [22], 3 weeks, and 12 months [21]. In addition, moderate levels of prenatal trait anxiety

Child Psychiatry Hum Dev (2012) 43:683 714 691 Table 2 Key aspects of studies of infant behavior (n = 2) Citation/ Quality rating Study design and sample Measure and timing of exposure Infant outcome (assessor) Results (effect size) a Adjusted for key potential confounders Brouwers 2001 21 (n = 131) strong Hernandez- Martinez 2008 22 (n = 163) moderate (Netherlands) Community Mean age 30.3 30.9 years; years education 10.2 11.0 Cross-sectional (Spain) Community Mean age 31.8 years; mean weeks gestation 39.2 (SD 1.38) Prenatal anxiety (State-Trait Anxiety Inventory) (3rd trimester) Prenatal stress (Perceived Stress Scale); anxiety (State-Trait Anxiety Inventory, STAI); socioemotional stability (STAI) (2 3 days post-delivery for past month, 3rd trimester) Infant behavior at 3 weeks (Neonatal Behavioral Assessment Scale) and 12 months (Bayley Scales of Infant Development) (researcher) Neonatal behavior at 2 3 days post-birth (Neonatal Behavioral Assessment Scale, NBAS) (NBAS examiners) S (3 weeks and 12 months) 3 weeks: Infants of anxious mothers scored S lower on orientation (anxious mothers M = 6.0, SD 1.1 (anxious) versus M = 6.5, SD 0.9 (nonanxious) 12 months: Infants of anxious mothers scored S lower than non-anxious mothers on task orientation and motor coordination (data not shown) S (socio-emotional stability on infant regulation, B = 0.178, p =.04; quality of alertness, B = 0.196, p =.02; reduced irritability, B = 0.184; p =.03) (small) S (trait anxiety on social interaction, B = -0.212, p = 0.02; quality of alertness, B =-0.172, p = 0.04) (small) NS (perceived stress; state anxiety) Moderator: Gender NS None for this outcome Gestational age; delivery (normal vs. difficult); maternal age; prenatal nicotine/alcohol use; birthweight; gender; infant age (hours); feeding method (significance not reported) Reported if available or calculable; NS statistically non-significant (p C.05), S statistically significant (p \.05), PPD postpartum depression a

692 Child Psychiatry Hum Dev (2012) 43:683 714 were associated with reduced social interaction scores at 2 3 days post-delivery [22] and high state/trait anxiety was related to less motor coordination at 12 months [21]. Socioemotional stability (i.e., the absence of mood changes) was associated with improved infant self-regulation [22]. Of all the measures of prenatal distress assessed, perceived stress and state anxiety did not have an impact on infant development. The finding that trait anxiety (and combined trait/state) and socio-emotional stability were related to infant development may suggest that it is the more enduring forms of prenatal distress that are implicated in infant behavior. Hernandez-Martinez et al. [22] found these effects to be small, whereas Brouwers et al. [21] did not report the magnitude of effect or provide data to derive it. Moderators and Mediators Hernandez-Martinez et al. [22] found that infant sex did not moderate the relationship between maternal prenatal distress and behavioral outcomes. Other Potential Confounders Although Hernandez-Martinez et al. [22] controlled for a wide variety of potential confounders, their individual significance was not reported. Summary of Effect of Maternal Distress on Infant Behavior No studies reported the prevalence of infant behavior problems. Both studies evaluating the effect of prenatal distress on infant behavior in infants 48 h to 12 months of age reported significant findings. Based on these 2 community-based studies of moderate and strong quality, there is some evidence that third trimester anxiety and socio-emotional stability are related to behavior problems in infants. No studies assessed postpartum distress. Infant Cognitive Development (Table 3) General overview of studies (n = 7) Seven longitudinal studies evaluated the effects of prenatal (n = 3) and postpartum (n = 5) distress on infant cognitive development at 3 12 months. These studies recruited women from a variety of countries, including the Netherlands (n = 3), Australia (n = 2), the United Kingdom (UK) (n = 1), and the United States (US) (n = 1). Four represented community samples [21, 23 25], whereas three focused on unique populations including women attending a residential parentcraft program for infant difficulties [26]; those with high risk pregnancies complicated by pre-eclampsia, HELLP syndrome (i.e., hemolysis, elevated liver enzymes, and low platelets), or fetal growth restriction [27]; and chronic cocaine users [28]. Sample sizes ranged from 71 to 170 for all studies except Reilly et al. (n = 1,591) [25] (Total N = 2,327). The studies that assessed prenatal distress measured a variety of types spanning the second and third trimesters of pregnancy, including state-trait anxiety (32 weeks) [21]; stressful life events, perceived stress, maternal cortisol levels (15 17, 27 28, and 37 38 weeks) [23]; and depression (last trimester) [28]. Of the 5 studies evaluating postpartum distress, most assessed PPD [24, 26, 28], with 2 assessing non-specific

Child Psychiatry Hum Dev (2012) 43:683 714 693 Table 3 Key aspects of studies of infant cognitive development (n = 7) Citation/ Quality rating Study design and sample Measure and timing of exposure Infant outcome (assessor) Results (effect size) a Adjusted for key potential confounders Buitelaar 2003 23 (n = 170) moderate Murray 1992 24 (n = 111) strong Beckwith 1999 28 (n = 71) weak (Netherlands) Community primiparous women, low risk pregnancies (United Kingdom) Community Mean age 28.0 years; 40% manual labour/ unemployed (United States) Chronic cocaine users Mean age 28.9 years; 83.3% minority status; 84.6% single Prenatal stress in 2nd trimester (Everyday Problem List, 15 17 weeks) and 3rd trimester (Pregnancy Related Anxieties Questionnaire and Perceived Stress Scale, 27 28 weeks) PPD at 2 3 months (Edinburgh Postnatal Depression Scale; Standardized Psychiatric Interview if EPDS C 13) Prenatal depression 3rd trimester (Millon Clinical Multiaxial Inventory I); PPD at 6 months (Beck Depression Inventory) Cognitive development at 3 and 8 months (Bayley Scales of Infant Development, Mental Developmental Index, MDI) (researcher) Cognitive development at 9 months (Piaget s Object concept task) (psychologist) Cognitive development at 6 months (Bayley Scales of Infant Development Index, MDI) (researcher) S for 2nd trimester stress on 8 month MDI scores (AOR = 1.1, 95% CI 1.02 1.18) (small) NS (3rd trimester) S (First onset PPD related to poor performance on object concept task (even significantly poorer than mothers with history depression and PPD, p \.01) Severity: infants of mothers with major PPD performed more poorly than those with minor PPD S (prenatal? PPD Mean MDI = 96.0, SD 14.9; PPD alone) (small) NS (prenatal depression mean MDI = 114.4, SD 10.3 versus never depressed mean MDI = 110.9, SD 11.2) Moderator: women with severe prenatal depression but no depression at 6 months had infants with better MDI scores that women depressed pre- and postpartum; chronic depression negative impact on caregiving Birthweight; gestational age; prenatal smoking/alcohol; biomedical risk factors; breastfeeding? postpartum stress at 20 days, 3 and 8 months? PPD? postpartum psychological well-being Gender; education; unplanned pregnancy; obstetric complications; prenatal anxiety; marital friction; social support; housing; employment; social class; paternal psychiatric history Mother child interaction; postnatal depression

694 Child Psychiatry Hum Dev (2012) 43:683 714 Table 3 continued Citation/ Quality rating Study design and sample Brouwers 2001 21 (n = 131) strong Cornish 2005 26 (n = 112) moderate Kaspers 2009 27 (n = 141) strong (Netherlands) Community mean age 30.3 30.9; years education 10.2 11.0 (Australia) Clinical (recruited from centre for infant problems) mean age 31.4 years; 11% B high school; 93% Caucasian (Netherlands) Women with high risk pregnancies Mean age 30.4 30.8 years; 31 36% high education; 78 81% Caucasian Measure and timing of exposure Prenatal anxiety (State- Trait Anxiety Inventory) (3rd trimester) PPD (Center for Epidemiological Studies Depression Scale; The Composite International Diagnostic Interview) (4 and 12 months) Postpartum psychological symptoms (Symptom Check list-90) (sum of 0, 3, 12 months) Infant outcome (assessor) Results (effect size) a Adjusted for key potential confounders Cognitive development at 12 months (Bayley Scales of Infant Development, Mental Development Index, MDI) (researcher) NS (In high anxiety group, mean MDI = 97 (SD 14) (high anxiety) vs mean MDI = 103 (SD 14) (no anxiety) (p =.07) Apgar 1 and 5 min; home environment; education; maternal age; prenatal smoking/ alcohol; type of delivery; breastfeeding; gender; birthweight; gestational age; PPD at 12 months Language development at 12 months (Receptive- Expressive Emergent Language Test) (mother) NS (no depression M = 104.2, SD 11.2; brief depression M = 104.9, SD 13.4; chronic depression M = 102.4, SD 11.3) (NS) Moderator: gender (NS) Gender; education; maternal age; mother bilingual Bayley Scales of Infant Development, Mental Development Index (MDI) at 12 months (psychologist) NS (Mean MDI = 87 (range 59 102), high psychological symptoms versus MDI = 89 (range 66 124), low psychological symptoms) Maternal disease; SES; education; antenatal corticosteroids; gestational age; birthweight; apgar scores; gender; neonatal disease

Child Psychiatry Hum Dev (2012) 43:683 714 695 Table 3 continued Citation/ Quality rating Study design and sample Measure and timing of exposure Infant outcome (assessor) Results (effect size) a Adjusted for key potential confounders Reilly 2006 25 (n = 1591) moderate (Australia) Community 9.3% low education; 3.1% non- English speaking background Postpartum: psychological distress (Non-specific Psychological Distress Scale) (8 months) Infant communication at 8 and 12 months (Communication and Symbolic Behavior Scales, CSBS) (mother) NS (8 months, M = 99.2, SD 14.3; 12 months, M = 96.4, SD = 13.1) Prematurity, sex (S), multiple birth (S), SES (S), maternal vocabulary and education; non-english speaking background; family history of speech difficulties (S) Reported if available or calculable; NS statistically non-significant (p C.05), S statistically significant (p \.05), PPD postpartum depression a

696 Child Psychiatry Hum Dev (2012) 43:683 714 psychological distress [25, 27]. In terms of infant outcomes, 5 of these studies assessed global cognitive infant development, and 2 focused on language development [25, 26]. Quality Among these 7 studies, most were rated as strong [21, 24, 27] or moderate [23, 25, 26], with only one study receiving a weak rating [28]. Attrition rates varied from less than 10 54%. The main limitation among these studies was the lack of assessment of potential confounders. All studies used psychometrically evaluated measures of maternal distress and child cognitive development. Four studies used self-report measures of maternal distress (State-Trait Anxiety Inventory [21]; Everyday Problem List, Pregnancy Related Anxieties Questionnaire, Perceived Stress Scale [23]; Non-specific Psychological Distress Scale [25]; Symptom Check list-90 [27]), and the remaining 3 used a combination of selfreport (Edinburgh Postnatal Depression Scale [24]; Center for Epidemiological Studies Depression Scale [26]; Beck Depression Inventory [28]) and structured interview by a psychologist. In most of the studies, assessments of cognitive development were conducted by a researcher/psychologist [21, 23, 24, 27, 28] with only 2 using maternal report [25, 26]. Global cognitive development was conceptualized quite consistently in these studies, with 4 studies using the Mental Development Index (MDI) of the Bayley Scales of Infant Development to assess cognitive outcomes [21, 23, 27, 28]. Main Findings The prevalence of cognitive delay ranged widely from 7% in infants of non-distressed women to 25% in those of distressed women in a community sample [21]. Among growthrestricted infants whose mothers experienced high risk pregnancies, the rates of cognitive delay were 34% (distressed women) and 38% (non-distressed women) [27]. Three of the 7 studies reported small, significant influences of maternal distress experienced in the prenatal (n = 1) [23] or postnatal (n = 2) [24, 28] periods. None of the studies of cognitive outcomes at 12 months were significant, and neither study focusing on language development reported significant associations. The overall quality ratings of the 3 studies with significant findings varied from weak to strong. Each of these significant studies controlled for postpartum distress and a variety of additional potential confounders. By form of distress, first trimester prenatal stress and high cortisol in the third trimester were associated with small reductions in cognitive scores at 3 months [23]. First onset PPD (at 2 3 months) was also associated with poor functioning on a cognitive object task test in 9-month old infants, even compared to mothers who had experienced both a history of depression and a previous PPD [24]. Adjusting for very few confounders, a significant association was also found between clinically diagnosed postpartum dysthymia in cocaine-abusing women and low cognitive scores in 6-month old infants, as well as with a combination of prenatal and postpartum dysthymia [28]. In this study, no relationship was found between prenatal dysthymia alone and cognitive development. Finally, although Kaspers et al. [27] found no difference in MDI scores of infants of women with high (MDI M = 87; Range 59 102) and low levels of psychological symptoms (MDI M = 89; Range 66 124), all infants in this study were born to mothers with severe complications (e.g., pre-eclampsia, HELLP syndrome) and the mean MDI scores were below the population mean of 100. Neither study that assessed cognitive outcomes by maternal report was significant [25, 26].

Child Psychiatry Hum Dev (2012) 43:683 714 697 Moderators and Mediators Beckwith et al. s [28] study of substance-abusing mothers with high levels of dysthymia found that maternal-child interaction was not a significant mediator of the relationship between prenatal dysthymia and cognitive development. In this study, chronicity of dysthymia was a significant moderator where infants of women who were severely depressed during pregnancy and postpartum had poorer cognitive outcomes than those of women who were depressed during pregnancy but experienced recovery [28]. Murray et al. [24] found that severity of depression was also an important factor in that infants of mothers with major PPD scored more poorly on cognitive assessments than those with mothers who had a minor PPD. However, Cornish et al. [26] did not find that severity or chronicity of PPD had an effect on language development. Finally, one study found a non-significant gender interaction [26], indicating that the relationship between maternal distress and cognitive outcomes did not differ between boys and girls. Other Potential Confounders Numerous potential confounders were assessed in this group of studies, including child factors (gestational age, birthweight, Apgar scores), maternal demographics (age, income, education, social class), maternal behavior (prenatal smoking, prenatal alcohol, breastfeeding), obstetrical characteristics (parity, mode of delivery, complications, unplanned pregnancy), and social factors (social support, marital conflict, home environment). Overall, very few additional variables were significant predictors of cognitive development. In particular, global cognitive functioning was impacted by low maternal education in one study [24], whereas language development was predicted by low SES, multiple births (e.g., twins had poorer scores) gender (e.g., girls better cognitive scores than boys), and family history of language difficulties [25]. Three of the 7 studies did not report significance levels for the potential confounders. Summary of Effect of Maternal Distress on Infant Cognitive Development The prevalence of infant cognitive delay ranged from 7 to 38%. Given that both studies of moderate quality (including one community-based study, n = 1,591) did not find an association between postpartum distress and language outcomes, there is no evidence at present to support this relationship. No studies evaluated prenatal influences on language development. Three of 5 studies found small, significant effects of prenatal stress (n = 1) and PPD (n = 2) at 2 3 months on general cognitive development. Based on these studies of weak, moderate, and strong quality, some evidence exists for a small effect of prenatal and postpartum distress on infant cognitive development. Because the number of studies was limited and a variety of forms of maternal distress was studied, clear trends related to the timing and specificity of the form of prenatal maternal distress were not evident. However, some preliminary observations can be made. Specifically, of the 3 studies that assessed prenatal distress, one examined and found a significant second trimester effect on cognitive outcomes, whereas the 3 that explored a third trimester effect did not. Among the 3 studies that evaluated postpartum distress, those that explored PPD (n = 2) demonstrated a significant relationship with infant cognitive development, whereas the study of psychological symptoms did not. One study found that infants tended to have poorer cognitive development if their mothers experienced chronic depression across pregnancy and postpartum

698 Child Psychiatry Hum Dev (2012) 43:683 714 (versus brief), or severe PPD. Importantly, among the wide array of risk factors for cognitive delay assessed by this group of studies, maternal psychological distress was most consistently related to infant cognitive delay. Infant Socio-Emotional Development (Table 4) General Overview of Studies (n = 4) Four studies evaluated the influence of postpartum distress on infant socio-emotional development, including 3 longitudinal [29 31] and one cross-sectional study [32]. No studies explored the effect of prenatal exposure on infant outcomes. Four different countries were represented by these studies (Sweden, Israel, Australia, UK), and sample sizes ranged from 44 to 223 (total N = 481). Two of the 4 studies involved clinical samples [29, 30], one was community-based [31], and one blended community and clinical [32]. As such, these studies examined the effect of postpartum psychiatric illness [29 32], generalized anxiety disorder and social phobias [30, 31], major PPD [30], and postpartum mood [32] on infant socio-emotional development. Infant outcomes were assessed throughout the first year of life at 10 weeks [31], 9 and 10 months [29, 30] and between 13 and 52 weeks (mean 30.1 weeks) [32]. All studies focused on social behavior as the indicator of infant socio-emotional development. Three studies assessed sociability using a variety of laboratory-based tests [29 31] and one was based on a primary clinician s rating of the infant-clinician interaction during a routine child care visit [32]. As such, all studies used researchers or clinicians to assess infant socio-emotional development. Quality Among these studies, most received weak ratings [29, 30, 32], with the remaining study rated as strong [31]. The main limitations were having high potential for selection bias, lack of assessment of potential confounders, and lack of blinding of the outcome assessor. Attrition rates were less than 25% for the longitudinal studies. None of the studies controlled for prenatal mood in their analyses. One study measured maternal distress with a self-report measure (Edinburgh Postnatal Depression Scale [32]), two used structured psychiatric interviews [29, 31], and one used both approaches (Beck Depression Inventory, Structured Clinical Interview, State-Trait Anxiety Inventory, Parenting Stress Index [30]). Main Findings The inclusion criteria for each of the 4 studies ensured that rates of clinical maternal distress were high in these studies. The rates of poor sociability in 2 studies reporting these data ranged from 6 to 14.8% in non-distressed women and 27.3 to 55% in those experiencing distress [31, 32]. Three studies found a significant effect of postpartum distress involving generalized anxiety/social phobia [30], depression [30], postpartum mood [32] and psychiatric illness [29] on infant social development. Among the significant studies, all had an overall quality rating of weak (the single study rated as strong reported a nonsignificant association). Overall, these studies found moderate effects of PPD on reduced social engagement and increased fear in infants; however, the magnitude of this effect may be inflated by the lack of control for potential confounders.

Child Psychiatry Hum Dev (2012) 43:683 714 699 Table 4 Key aspects of studies of infant socio-emotional development (n = 4) Citation/ Quality rating Study design and sample Measure and timing of exposure Infant outcome (assessor) Results (effect size) a Adjusted for key potential confounders Albertsson- Karlgren 2000 29 (n = 114) weak Feldman 2009 30 (n = 100) weak Matthey 2005 32 (n = 44) weak (Sweden) Clinical women admitted to psychiatric unit; 47.2% post-secondary education; 83.3% married (Israel) Clinical subsample of community sample -mean age 30.7 years; completed mean 15.7 years education Cross-sectional (pilot) (Australia) Community? Clinical (parentcraft unit) mean age 28.5 years; all mothers partnered; 31.8% post-secondary education Postpartum psychiatric illness by 6 months (diagnostic interview) PPD (Beck Depression Inv); generalized anxiety disorder/social phobias (Structured Clinical Interview; State-Trait Anxiety Inventory) (2 days, 6 month, 9 month); parenting stress at 6 months (Parenting Stress Index) PPD since birth, assessed when infant 13 52 weeks old (Edinburgh Postnatal Depression Scale) Sociability and fear of stranger; approachwithdrawal at 10 months (stranger-wariness situation; Stevenson and Lamb sociability situation) (mother/ researcher) Social engagement at 9 months (lab tests involving play) (researchers; mothers) Social behavior at 13 52 weeks (Mean 30.1 weeks, SD 12.1) (Alarme Distress de Bebe Scale) (clinicians) S (Infants in psychiatric sample had more negative sociability (M = 3.9, SD = 2.3) vs non-depressed mothers (M = 5.4, SD = 2.0; p \.05); these infants also had lower approach-withdrawal scores (depressed M = 2.9, SD = 0.8 vs non-depressed M = 3.4, SD = 0.6), p \.05) (moderate) S (Social engagement in mothers with maternal depression M = 2.14, SD = 0.94 versus maternal anxiety M = 2.86, SD = 0.87 versus control M = 3.44, SD = 0.92). (moderate-anxiety; large-depression) Moderator: In depressed mothers, maternal sensitivity moderated effects on child social engagement; NS for anxiety S (55% women with PPD had infants with low scores (M = 1.6) versus 6% mothers no PPD (M = 5.6), p \.001) Controls matched on infant sex; parity; maternal education; paternal education Maternal sensitivity (S) None

700 Child Psychiatry Hum Dev (2012) 43:683 714 Table 4 continued Citation/ Quality rating Study design and sample Measure and timing of exposure Infant outcome (assessor) Results (effect size) a Adjusted for key potential confounders Murray 2007 31 (n = 223) moderate Community (subsample screened for GAD and social phobia) -7.4 16.7% low SES; 5.5 7.3% single; 85.7 100% Caucasian Postpartum social phobia; anxiety disorder at 10 weeks (Structured Clinical Interview for DSM-IV Axis 1 disorders) Social responsiveness at 10 weeks (laboratory face-to-face interactions and social stranger challenge) (researcher) NS (AOR 1.9, 95% CI 0.8 4.1; mothers with social phobia not less sensitive or more controlling than even though they were more anxious and withdrawn) Neonatal irritability; sex; degree to which mother encourage infant to interact with stranger Reported if available or calculable; NS statistically non-significant (p C.05), S statistically significant (p \.05), PPD postpartum depression a

Child Psychiatry Hum Dev (2012) 43:683 714 701 Moderators and Mediators In addition to having an independent, direct effect, Feldman et al. [30] found that maternal sensitivity moderated the relationship between maternal PPD and infant social engagement. In other words, in mothers with high levels of sensitivity to their infants, maternal depression had no effect on infant social development; however, infants of depressed mothers with low sensitivity were more likely to experience suboptimal social development. This moderating effect was not found for maternal anxiety. Other Potential Confounders Few potential confounders were assessed in this group of studies. Among these, infant sex, neonatal irritability, and the degree to which mother encouraged her infant to interact were not significant predictors of infant sociability. Only maternal sensitivity demonstrated a moderate effect on social engagement in infants [30]. Summary of Effect of Maternal Distress on Infant Socio-Emotional Development Among the studies of infant outcomes, socio-emotional development was defined in terms of infant social competence with prevalence rates reported as 6.0 14.8% in infants of nondistressed women, and 27.3 55.0% in those of distressed women. There is limited evidence for an effect of postpartum distress on social development from 2 to 10 months based on the findings of 3 small, weak studies. Although the effect sizes reported were moderate to large, this is likely a reflection of the lack of control of potential confounders in these studies. The study with non-significant findings was a larger, community-based sample of moderate quality. No studies assessed the effect of prenatal distress. Maternal sensitivity was a moderator; in other words, there was no effect of maternal depression on infant sociability in mothers with high levels of maternal sensitivity. Infant Psychomotor Development (Table 5) General Overview of Studies (n = 7) Seven longitudinal studies examined the effect of prenatal [21, 23, 33 36] and postpartum [27, 36] distress on infant psychomotor development. These 7 studies recruited women from the Netherlands (n = 4) and the US (n = 3). Sample sizes ranged from 131 to 200 with the exception of van Batenburg et al. (n = 2,724) [36] (total N = 3,690). Six of these studies were community-based, with the remaining study recruiting women for a larger study of women with pre-eclampsia, HELLP Syndrome, or severe fetal growth restriction [27]. Two of the community-based studies involved low-to-middle income women [34, 35], one-third of whom were African-American and single. A wide variety of prenatal influences were assessed, including state-trait anxiety [21], stressful life events [23], perceived stress [23], salivary cortisol [23], plasma cortisol and placental corticoptropin releasing hormone [33], anger [34], dysthymia and major depression [35], anxiety [36], and depression [36]. One study evaluated prenatal distress in the first trimester [23], 5 in the second [23, 33 36], and 2 in the third [21, 33]. Postpartum distress included exposure to psychological symptoms [27], anxiety, and depression [36], and was assessed within the first 3 months postpartum [27, 36] and at 12 months [27].