WILLIAMS COLLEGE LIBRARIES 
COPYRIGHT ASSIGNMENT AND INS1RUCTIONS FOR A STIJDENT THESIS 
Your unpublished thesis, submitted for a degree at Williams College and administered by the Williams College Libraries, will be made available for research use. You may, through this form, provide instructions regarding copyright, access, dissemination and reproduction ofyour thesis. The College has the right in all cases to maintain and preserve theses both in hardcopy and electronic format, and to make such copies as the Libraries require for their research and archival functions. 
X The faculty advisor/s to the student writing the thesis claims joint authorship in this work. 
_ I/we have included in this thesis copyrighted material for which Ilwe have not received permission from the copyright holder/s. Ifyou do not secure copyright permissions by the time your thesis is submitted, you will still be allowed to submit. However, ifthe necessary copyright permissions are not received, e-posting of your thesis may be affected. Copyrighted material may include images (tables, drawings, photographs, figures, maps, graphs, etc.), sound files, video material, data sets, and large portions oftext. 
I. COPYRIGHT An author by law owns the copyright to his/her work, whether or not a copyright symbol and date are placed on the piece. Please choose one ofthe options below with respect to the copyright in your thesis. 
_ Ilwe choose not to retain the copyright to the thesis, and hereby assign the copyright to Williams College. Selecting this option will assign copyright to the College. If the author/s wishes later to publish the work, he/she/they will need to obtain permission to do so from the Libraries, which will be granted except in unusual circumstances. The Libraries will be free in this case to also grant permission to another researcher to publish some or all ofthe thesis. Ifyou have chosen this option, you do not need to complete the next section and can proceed to the signature line. 
X I!we choose to retain the copyright to the thesis for a period of_iQ_ years, or until my/our death/s, whichever is the earlier, at which time the copyright shall be assigned to Williams College without need offurther action by me/us or by my/our heirs, successors, or representatives ofmy/our estate/s. 
Selecting this option allows the author/s the flexibility ofretaining his/her/their copyright for a period ofyears or for life. 


A Meta-Analytic and Qualitative Examination of Predictors of Mother-Infant 
Bonding 
CAROLINE C. KAUFMAN 
Laura E. Sockol, Advisor 
A thesis submitted in partial fulfillment 
of the requirement for the 
Degree of Bachelor of Arts with Honors 
in Psychology 

WILLIAMS COLLEGE Williamstown, Massachusetts 
12 May 2015 Abstract 
Mother-infant bonding is a mother-based perspective of the complex relationship between mother and infant. Research has examined predictors of mother-infant bonding utilizing a variety of experiment types across several populations. In this study, we systematically and narratively examine research assessing predictors of mother-infant bonding including psychiatric, demographic, interpersonal, and other predictors. We found that maternal anxiety, maternal depression, and female infant sex predicted impaired mother-infant bonding. We also found that interventions and fetal attachment predicted healthy mother-infant bonding. Our results have important implications for both health care providers and clinicians. 
A Meta-Analytic and Qualitative Examination of Predictors of Mother-Infant Bonding 
A great deal of research has investigated the relationship between mother and child that begins during pregnancy and blossoms throughout the first few years of life. Bowlby was one ofthe first to research and define this relationship. He introduced the concept of"attachment" in his 1958 essay "The Nature of a Child's Tie to His Mother." Bowlby proposed that infants instinctually engage in certain behaviors, which are meant to cement a bond between the infant and its caregiver (Bowlby, 1958). This theory proposed that an infant's feelings of attachment that develop during infancy and early life continue throughout the lifespan (Bowlby, 1958). According to Bowlby, mothers or caregivers also participant in certain behaviors that either encourage or discourage the strength of the mother-infant attachment (Bowlby 1958; Bowlby 1988). Mary Ainsworth sought to examine this aspect of the mother-infant relationship through the Strange Situation Paradigm (Ainsworth, 1978). The Strange Situation Paradigm rests on the assumption that an infant's or young child's behaviors and reactions are manifestations of the nature of the mother-infant attachment relationship and, by examining these behaviors, we can determine the nature of the relationship (Ainsworth, 1978; Kennel & Klaud, 1984). Over time, Bowlby, Ainsworth and other researchers have introduced other concepts to further the literature's understanding ofthe mother-infant relationship. Ainsworth and Bowlby introduced concepts such as "maternal sensitivity" and "maternal responsiveness" (Bowlby, 1988; Bell & Ainsworth, 1972). These concepts underscored the maternal role in developing the mother-infant attachment relationship (Bell & 
Ainsworth, 1972). Researchers also explored how an infant's characteristics can strengthen or weaken mother-infant attachment (George & Solomon, 1999). Certain infant behaviors and characteristics such as infant crying, eye contact, and facial expressions can serve to strengthen or challenge the relationship (Eibel-Eibesfedlt, 1971 ; Richards, 1971). 
While mother-infant attachment is sometimes used interchangeably with the similar concept of mother-infant bonding, mother-infant bonding refers to the relationship between mother and infant during the infant's first year oflife. Mother-infant bonding represents the mother's perspective ofthe relationship with her infant. Mother­infant bonding originally began as a perspective of mother-infant attachment, but later evolved into a concept distinct from infant attachment. Research examining mother­infant bonding began to proliferate during the 1970s (Muller, 1996). By the mid to late 1980s, researchers accepted the existence of mother-infant bonding as a distinct relationship that begins to occur prior to the development of infant attachment develops-before the infant is even born (Williams, Joy, Travis, et al., 1987). Although both mother-infant attachment and mother-infant bonding address the relationship between caregivers and early children, there are some important distinctions between the two concepts. Firstly, mother-infant bonding is a process that occurs before mother-infant attachment (Klaus, Kennel, & Klaus, 1995). The later development of mother-infant attachment is largely dependent upon the establishment of the mother-infant bond (Klaus, Kennel, & Klaus, 1995). Secondly, the mother-infant bond is a progressive process that is not completely solidified at birth (Figueiredo, Costa, Marques, et al. , 2007). Some research indicates that the mother-infant bond is constantly growing and changing 
(Fleming, Rubble, Krieger, et al., 1997). Mothers tend to respond more affectionately towards their newborn infants as they progress through the postpartum period (Klaus, Kennel, & Klaus, 1995). Thirdly, mother-infant bonding differs from mother-infant attachment in that mother-infant bonding focuses on the mother's perspective ofthe interactions between herself and her infant (Hornstein et al., 2006). Despite differences between the two concepts, research indicates positive associations between mother-infant bonding and mother-infant attachment (Bell & Ainsworth, 1972; Elster, McAnarney & Lamb, 1983; Feldman, Weller, Leckman, Kuint, & Eidelman, 1999). 
For the purposes of this study, mother-infant bonding is defined as a mother­based perspective of the early relationship between infant and mother (Kennel & Klaus, 1984). Mother-infant bonding refers to the relational tie from mother to child that is specific and continues over time (Klaus & Kennel, 1995; Hornstein et al., 2006). Although the concept of mother-infant bonding is particularly reliant upon the mother's perception of the relationship, the relationship between mother and infant is bidirectional (Kennel & Klaus, 1984). For instance, a mother's perception of the emotional relationship between herself and a very difficult baby is likely to be different from a mother's perception of her relationship with an easy baby. 
Different methods are utilized to examine the strength or nature of the mother­infant bond. Some researchers utilize questionnaires that measure the mother's feelings about the infant and the relationship (e.g., Brockington et al., 2001; Condon & Corkindale, 1998; Muller, 1996; Shin, Park, & Kim, 2006). These questionnaires assess where the bond occurs on the continuum between "healthy" and "unhealthy" bonding. Some examples of questions include, "I can understand what my baby needs or wants" rated on a scale from almost always to almost never (Condon & Corkindale, 1998), "I feel confident caring for my baby" rated on a scale from always to never (Brockington et al., 2001), and "I feel angry at my baby" rated on a scale from always to never (Brockington et al., 2001). There are some weaknesses to mother-infant bonding self­report questionnaires, however. Hornstein and colleagues (2006) found that despite insignificant differences between psychotic and depressed mothers when maternal interactive behavior was assessed, significant differences were found between the two groups' perceptions ofthe mother-infant bond. Psychotic mothers expressed unreasonably positive beliefs about their bond with their infants on a self-report measure. And, clinically depressed mothers, on the other hand, expressed very little healthy bonding with infants on the self-report measure (Horntein et al., 2006). Unfortunately, mental illness may prevent some mothers from accurately evaluating their relationship with their infant. These results indicate that when working with clinical populations, self-report measures may not capture the full scope of the bond between mother and infant. 
There are limits to self-report measures even among mentally healthy mothers. Participants may not answer truthfully or may describe idealized feelings or thoughts about the relationship. Stigma and sensitivity to stigma are other limitations to mother­infant bonding self-report questionnaires. Some participants may not respond to questionnaires truthfully because they are aware of stigma against mothers who express some dissatisfaction with motherhood (Keneko & Honjo, 2014). Participants may also be afraid that nurses, researchers, or experimenters will judge them for their responses to questionnaires and may answer dishonestly. Kaneko and Honjo (2014) found that Japanese mothers were likely to downplay bonding impairment because of their acknowledgement and fear of stigma. As a result of stigma and shame, bonding impairment may be under-reported in the general population and across specific populations. 
Other measures assess behavioral interactions between infant and mother as manifestations of mother-infant bonding (e.g., Figueiredo et al., 2009). Bonding may manifest itself via interactions such as responsiveness, emotional involvement, interest in interactions with the infant, infant crying, eye contact, and facial expressions (Figueiredo et al., 2009). Figueiredo and colleagues (2009) propose that behavioral interactions provide opportunities for both mother and infant to show their bond and interest in the relationship to the other party. One ofthe more widely utilized behavioral interaction measures is the Bethlem Mother-Interaction Scale (Hipwell & Kumar, 1996). This scale is utilized by nurses to measure six variables between mother and infant: eye contact, physical contact, mood, general routine, risk to baby, and baby's condition. The overall impressions ofmother-infant interactions are rated by nurses on a four-point scale (Hipwell & Kumar, 1996). Other measures assess mothers' reaction times to infants, eye contact, touching, breastfeeding, and other relevant variables (Cohn, Campbell, Matias, et al., 1990). Despite the fairly wide use of these observational measures, a study conducted by Bienfait and colleagues (20 11) found that the majority of nurses in the study found evaluating the bond between mother and infant using this approach difficult. Additionally, most of these nurses also felt that third-party measures of bonding should have clearer and more precise guidelines and criteria. The results ofthe study did not indicate significant associations between the mothers' perception of the bond and the 
nurses' perception ofthe bond (Bienfait et al., 2011). Although Bienfait (2011) is a 
single study, it underscores the important of using both self-report and observational 
measures to fully capture the mother-infant bond. 
The importance of developing a healthy mother-infant bond and subsequent healthy mother-infant attachment is documented throughout the research literature. Brockington (2004) specifically suggests that the mother-infant bonding relationship is the single most important psychological process ofthat infant's early life. As discussed above, mother-infant bonding is a precursor to the development of the mother-infant attachment relationship (Williams, Joy, Travis, et al., 1987) and both relationships are significantly correlated (Bell & Ainsworth, 1972; Elster et al., 1983; Feldman et al., 1999). Therefore, mother-infant bonding is firstly important simply because it is a precursor and related to healthy infant attachment. 
Secondly, research across studies has indicated that the mother-infant relationship can be predictive or indicative of later problems, including emotional, behavioral, and cognitive delays as well as attachment disorders (Goldberg, 1983; Egeland & Vaughn, 1981; Campbell, Brownell, Hungerford, et al., 2004). The importance oflater infant attachment is also highly important. Many studies have underscored the importance of healthy infant attachment. Hubbs-Tait and colleagues (1996) reported that attachment at 44 months predicted attachment and externalizing problems up to ten months later. A meta-analysis conducted by Fearon and colleagues (20 1 0) found that insecure attachment predicts externalizing problems among both girls and boys. Additionally, Lounds, Bokowski, Whitman, Maxwell, & Weed (2005) found that healthy bonding and "good" 
maternal practices were related to a child's attachment at five years. 
The relationship between mother-infant bonding, mother-infant attachment, and other outcomes has lead some researchers to suggest that mother-infant bonding is related to the absence or presence of child maltreatment (Bloom, 1995; Milner, 1994; Williams et al., 1987). Impaired or difficult bonding is often proposed as being a serious threat to the well being ofthe infant in the mother-infant relationship (Brockington, 2004). Several studies have found strong relationship between maladaptive attachment patterns and child maltreatment (Crittendon & Ainsworth, 1978; Spieker & Booth, 1988). Mothers who commit infanticide and child abuse are more likely to have experienced poor bonding with their infants (Bloom, 1995; Williams et al., 1987). Clearly, mother-infant bonding is an important aspect of early development and the mother-child relationship. 
In some cases, the bonding relationship between mother and infant fails to develop or does not fully develop (Brockington, 2004). For instance, Brockington, Aucamp and Fraser (2006), interviewed women at a Mother and Baby Service in a hospital in England and found that 29% of the mothers had an element of rejection in their bond with their infant, 21% felt more than mild pathological anger towards their infant, and 11% of mothers felt severe anxiety about their infant. These percentages suggest that a significant portion of mothers in psychiatric care experience some aspect of impaired bonding with their infant. Luft (1964) described these feelings in new mothers as a "perversion" of the maternal instinct (194). In cases where bonding fails, mothers may experience feelings of resentment, hatred or rejection toward their infant (Brockington, 2004). In extreme cases, the failure ofthe mother-infant bond to develop has been associated with infant abuse or neglect (Egeland & Vaughn, 1981; Cicchetti & 
Rizley, 1981). 
In the last thirty years, researchers have explored antecedents to mother-infant bonding failure and success and have arrived at the consensus that the bonding relationship is not dependent on a single predictor or antecedent, but instead could be the result of a myriad of factors (Klaus & Kennell, 1982). This research indicates that one important set of potential predictors of mother-infant bonding are maternal psychiatric characteristics. Maternal depression or postpartum depression is the most commonly studied factor associated with bonding difficulties (MUller, Teismann, Havemann, Michalak, & Seehagen, 2013). Approximately 10-15% of mothers experience postpartum depression (Gavin, Gaynes, Lohr et al., 2005). For example, studies indicate that depressed mothers interact differently with their infants than non-depressed mothers, showing slower reaction times to crying, displaying fewer facial expressions towards their infants, and failing to properly assess their infant's needs (Field et al., 1985). Mothers with postpartum depression have been shown to be less sensitive to their children, through slower reaction times and less responsiveness to their infant's facial expressions (Campbell et al. , 2004). 
Similar issues such as anxiety and psychosis are also predictors of mother-infant bonding. The lifetime prevalence of anxiety disorders is estimated to be as high as 25% (Kessler, McGonagle, Zhao et al., 1994). Studies examining maternal anxiety as a predictor of mother-infant bonding are less common. The existing literature indicates that mothers with anxiety interact with their infants differently than mothers without anxiety (Nicol-Harper et al., 2007). Mothers with anxiety are more withdrawn from their infants and have more difficultly interacting with their infants (Beebe et al., 2011; Murray et al., 2008; Stein et al., 2012). Anxious mothers interact with their infants differently than non­anxious mothers and may subsequently develop different mother-infant bonding relationships. Lifetime prevalence ofpsychosis is smaller than both depression and anxiety (2.2%) and is studied less as a predictor ofmother-infant bonding (Kendler, Gallagher, Abelson et al., 1996). Psychosis interferes with one's perception of reality and could consequently interfere with the development of a health mother-infant bond. 
Demographic factors such as maternal age, infant sex, and maternal education have also been studied with mother-infant bonding. Demographic factors such as socioeconomic status, marriage status, education level, maternal age, infant weight, and infant sex are all important factors related to mother-infant bonding because they can either encourage or hinder mother-infant bonding. Mothers who are married or living with a partner may receive more economic support and logistical support and, consequently, may bond with their infants more easily. Younger mothers may have more energy to devote to their infants or they may be constrained by few economic resources and less life experience. Depending on the cultural context, one infant sex may be more desirable than the other and, therefore, may help or hinder mother-infant bonding (Booth, Verma, & Beri, 1994). 
Although many studies have explored the strength of the relationship between certain predictors and mother-infant bonding, the strength and direction of these relationships across the literature is not clear. A systematic review of the literature on mother-infant bonding has yet to be conducted to determine the "true" relationship between predictors and mother-infant bonding. For instance, some studies have found a relationship between depressive symptoms and mother-infant bonding (e.g. Edhborg, Matthiesen, Lundh, & Wistrom, 2005; Figeuiredo, Costa, Pacheco & Pais, 2009); however, other studies have not found such a relationship (e.g. Pearson, Lightman, & Evans 2011; Bienfait, Maury, Haquet, et al., 2011). Similarly, some research indicates adult attachment and mother-infant bonding are significantly related (Gunning, Waugh, Robertson & Holmes, 2011) while other research does not indicate find any significant association (Bienfait, Maury, Haquet et al., 2011). Other contradictions are found among research on maternal age and mother-infant bonding. Some studies indicate a relationship (Bienfait, Maury, Haquet et al., 2011) and others do not (Kitamura, Yamashita & Yoshida, 2009; Loh & Vostanis, 2004; Reck, Klier, Pabst et al., 2006). 
Meta-analyses, unlike narrative reviews, allow researchers to statistically identify patterns across many different studies (Borenstein, 2009). Narrative reviews provide interesting and useful information to readers, but narrative reviews force readers to provide generalizations across the literature. Unlike narrative reviews, meta-analyses utilize effect sizes for each relevant study and combine the effect sizes across all relevant studies, then weight these effects by study quality to estimate an overall effect size for a group of variables (Borenstein, 2009). This technique of computation will allow us to determine whether effect sizes are consistent across studies and to generalize the magnitude of the effect across all relevant studies (Borenstein, 2009). Specifically, we aim to understand how effective an intervention is or how strong a relationship is, not simply whether an intervention is effective or whether a relationship exists. Meta­analyses, unlike narrative reviews, do not simply compile the data and force the reader to judge trends across studies. Instead, meta-analyses make trends and relationships systematically clear for the reader and provides an actual statistical test of consistency of 
findings. 
Our study sought to explore the literature on mother-infant bonding to identify predictors of mother-infant bonding and to systematically assess the relationship between predictors and mother-infant bonding. We utilized our results to make suggestions for future research and clinical practice. 
Method Search Procedure and Selection of Studies 
Relevant studies were identified through searches of databases through the fall of 2014, including PubMed and Psyclnfo, using the phrase mother-infant bonding as a keyword. The reference lists of retrieved articles and relevant reviews were inspected for further studies. We also reviewed all studies citing common measure of mother-infant bonding, including the Mother-to-Infant Bonding Scale (MIBS; Taylor et al., 2005), the Postpartum Bonding Questionnaire (PBQ; Brockington et al., 2001 ), and the Maternal Postnatal Attachment Scale (MPAS; Condon & Corkindale, 1998). The abstracts ofthe studies identified during the literature search were then reviewed, articles whose abstracts suggest potential eligibility were obtained and reviewed for inclusion. 
To be included in the meta-analysis, studies had to meet several inclusion criteria. Studies had to be an empirical study ofhuman subjects (e.g. survey, chart review or retrospective chart review, case control design, randomized or quasi-randomized controlled trial). Case studies were excluded from the meta-analysis because they do not provide adequate information to calculate effect sizes. To be included, participants in the study must have been within 12 months postpartum. The postpartum period was limited so that we could ensure we were measuring predictors of the mother-infant bond rather than a mother-child bond. Studies included also had to have assessed mother-infant bonding using a validated self-report measure and have assessed at least one predictor of mother-infant bonding within 12 months postpartum. To be included, studies also had to assess a predictor before or at the same time mother-infant bonding was assessed. The authors of studies in which insufficient data was reported were contacted and asked for further data. 
A flow chart depicting the search process and exclusion of studies is presented in Figure 1. After the removal of duplicates, the search procedure yielded 43 8 studies. The 340 studies whose abstracts indicated potential eligibility for the meta-analyses were obtained and reviewed for inclusion. Ofthese 340 studies, 293 were excluded for the following reasons: not an empirical study of human subjects (n = 51), ineligible design (case report or case series; n = 30), participants were not within 12 months postpartum (n = 72), did not assess bonding using a validated self-report measure (n = 122), predictor was assessed after mother-infant bonding (n = 3), or the study was written in language other than English (n = 15). 
Coding of Included Studies 
These 47 studies were coded for six variables, including country, number of participants, age of participants, study design, sample type, and measure of mother-infant bonding. Effect sizes and moderators of eligible studies were coded by two raters in order to establish reliability. Studies were coded for study design as cross-sectional observational, prospective observational, chart review/retrospective chart review, case control, open trial, quasi-randomized controlled trial, or randomized controlled trial. Studies were also coded for potential moderators including location, sample size, infant age (in weeks), and timing of assessment. Location was coded as either primarily English speaking or not primarily English speaking. Our rationale for this coding was due to the fact that the most common measures of mother-infant bonding were originally written in English and would have to be translated in countries that did not primarily speak English. The sample was coded as either community or clinical. Community samples are samples in which participants are selected from the population. Clinical samples are samples in which participants are seeking or receiving psychological treatment. When outcomes were reported using multiple measures, we used the primary outcome measure as reported by the authors. If the authors did not specify a primary assessment measure, we used the smallest effect size in order to conduct the most conservative analyses. When predictors and/or outcomes were assessed at multiple time points, we used the assessment(s) at the latest point during the 12-month postpartum window for calculating effect sizes. We also coded the measure of mother-infant bonding utilized in the study. We identified all predictors or mother-infant bonding that had been reviewed in each eligible study and coded data necessary for the calculation of effect size(s). 
Any disagreement between the two coders on of these variables was resolved by consensus and was reviewed by the thesis advisor. The two coders agreed 92% of the time in identifying the country in which the study took place, 90% of the time in identifying the number of participants enrolled in the study, 100% of the time in identifying the age of participants, 78% of the time in identifying the study design, 96% of the time in identifying the sample type, and 98% ofthe time in identifying the measure of mother-infant bonding used in the study. 
Data Analysis 
We required at least three studies of a predictor to conduct a meta-analysis. The variables that were analyzed and the number of studies that assess that variable are as follows. We assessed six demographic variables, including maternal employment (k = 3), maternal age (k = 4), maternal education (k = 5), infant sex (k = 4), infant age (k = 4), and parity (k = 3). We assessed two interpersonal variables including, fetal attachment (k = 4) and social support (k = 2). We assessed three psychological variables, including depressive symptoms (k = 17), maternal anxiety (k = 8), and post-traumatic stress disorder (k = 3). The effect of interventions on mother-infant bonding was also assessed (k = 8). When there were not sufficient studies assessing a variable the available literature for some relationships a narrative review was conducted. 
We assessed the relationship between continuous predictors of mother-infant bonding using correlation coefficients (Lipsey & Wilson, 2001 ). Categorical predictors of mother-infant bonding were assessed using the standardized mean difference (Lipsey & Wilson, 2001 ). Overall effect sizes were estimated using random effect size models, based on the statistical assumption that the included studies represent the distribution of true effect sizes (Lipsey & Wilson, 2001 ). The heterogeneity of effect sizes was assessed using the Qstatistic and the Pindex (Lipsey & Wilson, 2001 ). Qstatistics that were significant indicate a larger range of effect sizes than is expected by within-study variance (Lipsey & Wilson, 2001). The Pvalue indicates the variance in effect sizes that is accounted by between-study variance (Higgins et al., 2003). 
Outliers for each analysis were identified using the sample adjusted meta-analytic deviance statistic (SAMD; Huffcutt & Arthur, 1995). We used a conservative SAMD cut­off score of 2.58 to considerexcluding studies from the analyses because extreme study results can result from anomalies, error, or the true differences (Beal, Corey, & Dunlap, 2002). SAMD scores were rank-ordered and scree plots were visually examined to identify outliers and confirm the presence ofoutliers. In cases in which the SAMD score was above the cut-off score, but the scree plot suggested the SAMD score was within the normal distribution, the study was retained. 
Publication bias was assessed using several methods. We used the fail-safe N value to determine the number of studies with null findings that would be necessary to produce an insignificant overall effect size. We used Rosenthal's ( 1991) recommended value of 5K+10, where K is the number of observed studies, to determine the cutoff for an unlikely number of studies. We visually examined the funnel plots ofthe distribution of effect sizes. An asymmetric funnel plot distribution suggests missing studies due to publication bias (Lipsey & Wilson, 2001). Duval and Tweedie's trim-and-fill procedure was used to identify skewed distributions of effect sizes. In cases in which a study's funnel plot was asymmetric, Duval and Tweedie's trim-and-fill procedure was used to assess whether any studies were missing from the distribution and provided an overall estimate for the model that corrects for missing studies (Duval and Tweedie, 2000). 
Moderator analyses were conducted on significant meta-analyses with significant heterogeneity. Continuous moderators were assessed using the meta-regression. Categorical moderators were assessed using mixed model ANOV As. Calculations of mean effect sizes, heterogeneity, and moderators were assessed using Comprehensive Meta-Analysis version 2.2.046 (Borenstein, Hedges, Higgins, & Rothstein, 2005). 
Results 
Characteristics of the included studies are presented in Table 1. Of the included studies 17% (n = 7) were randomized controlled trial, 2% (n = 1) quasi-randomized controlled trial, 74% were prospective observational (n = 36), and 9% (n = 4) were cross­sectional studies. The studies were conducted in a variety of countries. Almost half of the studies were either conducted in the United Kingdom (n = 11; 23%) or the United States (n = 10; 21 %). The rest ofthe studies were conducted were conducted in a variety of countries including Australia, Bangladesh Germany, Israel, Japan, Jordan, New Zealand, Norway, Portugal, Scotland, Sweden, South Korea, Taiwan, and Turkey. Of the studies, 53% (25) were conducted in countries that primarily spoke English and 47% (22) were conducted in countries that did not primarily speak English. The mean weeks postpartum at the timing of assessment was 18.09 (SD = 15.94). Almost half of the included studies ( 4 7%) utilized the PBQ (Brockington et al., 2001) to assess mother-infant bonding. The next two most utilized measures in the included studies were the Mother Attachment Scale (MUller, 1994) and the MIBS (Taylor et al., 2005). 

Demographic Predictors 
Infant Age. Two studies examined infant age as a predictor of mother-infant bonding. A meta-analysis was not conducted on infant age as a predictor of mother-infant bonding because too few studies examined this predictor. 
Gharaibeh and Hamlan (2011) examined infant age as a predictor of mother­infant bonding in a cross-sectional study with 220 primiparous Jordanian mothers. Participants were recruited from various health centers in one city within Jordan. Mother-infant bonding was measured using the Maternal Attachment Inventory (MAl; Muller, 1994). Mother-infant bonding did not differ between infants older than four 
months and infants younger than four months. 
Oriin, Yalc;m, and Mutlu (2013) assessed the relationship between infant and mother-infant bonding with community participants in Turkey. For these analyses, mother-infant bonding was assessed using the MIBS (Taylor, Atkins, & Kumar, 2005) and the PBQ (Brockington, 2001). Results found no significant relationship between infant age and mother-infant bonding; older infants were no more likely to experience healthy mother-infant bonding than younger infants. 
Infant Sex. Four studies assessed infant sex as a predictor of mother-infant bonding. 
In a prospective observational study, Figuieredo, Costa, Pacheco, and Pais (2009) examined the relationship between infant sex and mother-infant bonding. Participants were 315 community mothers in Portugal and were within the first week postpartum. Over 90% of the participants were Caucasian and identified as Catholic. Mother-infant bonding was assessed using a revised and expanded Portuguese version of the Mother-to­Infant Bonding Scale (MIBS; Figueiredo et al., 2005). We utilized the "Positive Bonding" sub scale of this measure to conduct this analysis. In this study, mothers of male infants had significantly better bonding than mothers of female infants. 
Gharaibeh and Hamlan (20 11) also examined infant sex as a predictor of mother­infant bonding in the previously described study of Jordanian mothers. In this study, healthy mother-infant bonding did not differ significantly between mothers of male and female infants. 
A correlational study conducted by Moehler and colleagues (2006) examined mother-infant bonding among 101 community participants recruited from local obstetric units in Germany. Participants were assessed at two, six, and 24 weeks postpartum. Mother-infant bonding was assessed using the PBQ (Brockington, Oates, George, et al., 2001). Healthy mother-infant bonding did not significantly differ between mothers of male and female infants. 
Ortin, Yal<;:m, and Mutlu (2013) also assessed the relationship between infant sex and mother-infant bonding in the previously described study of community participants in Turkey. Mothers of female children experienced significantly worse bonding relationships with their infants than did mothers of male children. 
A meta-analysis was conducted to determine the strength of the relationship between infant sex and mother-infant bonding. Effect scores were calculated so that positive effect sizes represent greater levels of bonding when the infant is male. Table 2 presents the results of the random effects model for the relationship between infant sex and mother-infant bonding. The analysis indicated that mothers of male infants experience significantly better bonding than mothers of female infants (Cohen's d = 0.20, CI 0.06-0.35,p < 0.01). None ofthe SAMD values ofthe included studies were above the absolute value of2.58 and consequently were not excluded from the analyses. 
Results of publication bias did not indicate potential publication bias in the included studies. The fail-safe N value was 4, which did not exceed the tolerance value of 
30. The Q statistic indicated that there was not significant heterogeneity among the included studies (p = 0.58) and the l value ofO.OO indicated a low level of heterogeneity. 
The funnel plot was symmetric (see Figure 2) and the trim and fill procedure did not 
suggest any missing studies. 
Timing of assessment was assessed as a potential moderator (see Table 11 ). Timing of assessment was not a significant moderator of the relationship between infant sex and mother-infant bonding (slope= -0.01,p = 0.48). Bonding measure, location, sample type could not be assessed as potential moderators due to insufficient studies in subgroups. 
Maternal Age. Three studies examined maternal age as a predictor of mother­infant bonding. A meta-analysis assessing maternal age as a predictor of mother-infant bonding could not be conducted due to differences in how data was reported and analyzed. 
In a prospective observational study, Cooklin, Rowe, and Fischer (2012) examined the relationship between maternal age and mother-infant bonding in 129 community participants in Australia. Participants were recruited from one private and one public hospital and were assessed during pregnancy and at 12 and 30 weeks postpartum. Mother-infant bonding was assessed using the Postnatal Attachment Questionnaire (PAQ; Condon & Corkindale, 1998). Demographic questions assessed maternal age; mothers over the age of 30 years were compared to mothers under the age of 30. The study did not find a significant difference in mother-infant bonding between older and younger mothers. 
In the previously described study, Figuieredo, Costa, Pacheco, and Pais (2009) also examined the relationship between maternal age and mother-infant bonding. The study found no significant difference in mother-infant bonding between mothers under 
the age of 20 and mothers over the age of 20. 
Kokubu, Okano, and Sugiyama (2012) examined mother-infant bonding among community Japanese participants in a cross-sectional study. Ninety-nine participants were assessed during late pregnancy, five days postpartum, and four weeks postpartum. Mother-infant bonding was assessed using a Japanese version ofthe MIBS (Taylor, Atkins, & Kumar, 2005). The results did not indicate a significant relationship between maternal age and bonding at five days or four weeks postpartum. 
Maternal Education. Five studies examined maternal education as a potential predictor of mother-infant bonding. Of these five studies, four were meta-analyzed. 
In the previously described study, Cooklin, Rowe, and Fischer (2012) also examined the relationship between maternal education and mother-infant bonding. Demographic questions assessed the highest level of education the mother had completed; mothers who had completed an undergraduate or postgraduate degree (tertiary qualification) were compared to mothers who had completed secondary school or a partial undergraduate program (no tertiary qualification). There was not a significant difference in mother-infant bonding between mothers who had earned a tertiary qualification and mothers who had not earned a tertiary qualification. 
In another prospective observational study, Dubber, Reck, Muller, and Gawlik (2014) assessed the relationship between maternal education and mother-infant bonding with community participants in Germany. Participants were recruited from Heidelberg University Women's Hospital and were assessed at 32 weeks gestation and 
approximately 12 weeks postpartum. Researchers coded maternal education was coded as 
having a university degree, having completed a university entrance exam, having a high 
school education, or having completed some high school. Mother-infant bonding was 
assessed using the PBQ (Brockington et al., 2001 ). Level ofeducation was significantly 
correlated with bonding; mothers with higher levels of education reported significantly 
more impaired bonding. 
In the prospective observational study described previously, Figuieredo, Costa, Pacheco, and Pais (2009) also examined the relationship between maternal education and mother-infant bonding. Maternal education was assessed using a questionnaire assessing the highest level of maternal education received; the authors compared mothers who had a ninth grade education or higher to mothers who had not completed ninth grade. Mothers who had education past grade nine had significantly more positive mother-infant bonding than mothers who did not receive education past grade nine. 
The previously described study by Ortin, Yalym, and Mutlu (2013) also assessed the relationship between maternal education and mother-infant bonding. Maternal education was assessed using with a questionnaire concerning the highest level of maternal education received; mothers who had an educational level greater than eighth grade were compared to mothers who had an educational level of eighth grade or less. The study found no significant difference in mother-infant bonding between mothers who had an eighth-grade education or lower compared to mothers who had completed more education. 
Sockol, Battle, Howard and Davis (20 14) also explored the relationship between maternal education and mother-infant bonding among 180 participants in the United States who were enrolled in a partial hospitalization program. Participants were assessed at an average of nine weeks postpartum. Maternal education was assessed using a questionnaire concerning the highest level of maternal education received; mothers who had a college degree were compared to mothers who did not have a college degree. Mother-infant bonding was assessed using the PBQ (Brockington, et al., 2001). Mothers who had received a college degree reported significantly more impaired bonding than mothers who had not completed a college degree. 
A meta-analysis was conducted in order to assess the relationship between maternal education and mother-infant bonding. Effect scores were calculated so that positive effect sizes represent greater levels of healthy bonding when there are higher levels of education. Table 3 presents the results of the random effects model of three assessing studies the relationship between maternal education and postpartum bonding. There was not a significant difference between higher and lower education groups (d =­0.12, 95% CI -0.89-0.64,p = 0.75). Two of the studies had SAMD values greater than 
2.58 (Figueiredo et al. , 2009; Sockol et al., 2014). Visual examination of the scree plot indicated that the value for the study by Figuieiredo and colleagues (2009) was discrepant from the overall distribution. After exclusion of this outlier, results indicated that impaired bonding was significantly greater among mothers with higher levels of education compared to mothers with lower education (d= -0.48, 95% CI -0.76-(-0.20),p = 0.001). 
Results of tests for publication bias indicated potential publication bias in the included studies. The fail-safe N value was zero, which does not exceed the tolerance value of 25. The Q statistic indicated that there was significant heterogeneity among the effect sizes (p < 0.001), and the Pvalue of93.35 indicated a large level of heterogeneity. 
The funnel plot was asymmetric (see Figure 2), but the trim-and-fill correction did not 
suggest any missing studies. 
Moderator analyses could not be conducted due to insufficient studies in subgroups. A meta-regression to assess timing of assessment as a moderator could not be conducted due to the too small number of studies. 
Maternal Employment. Three studies examined maternal employment as a predictor of mother-infant bonding. Sufficient data from these studies was not available to conduct a meta-analysis of differences in mother-infant bonding between unemployed and employed mothers. 
In the previously described prospective observational study, Cooklin, Rowe, and Fischer (2012) also examined the relationship between maternal employment and mother­infant bonding. The study found no significant differences in mother-infant bonding between employed mothers and unemployed mothers. 
In the previously described prospective observational study, Figuieredo, Costa, Pacheco, and Pais (2009) also examined the relationship between maternal employment and mother-infant bonding. The study found that unemployed mothers reported lower levels of positive bonding with their infants, but found no significant differences on other elements of the mother-infant bond. 
The previously described study of Turkish mothers by Oriin, Y al«;m, and Mutlu (2013) also assessed the relationship between maternal employment and mother-infant bonding. The study found no significant difference in mother-infant bonding between employed mothers and housewives. 
Parity. In a cross sectional observational study, Aiello & Lancaster (2007) examined mother-infant bonding among 71 community adolescent mothers in a longitudinal study conducted in Australia. Participants were assessed during pregnancy and at approximately eight, 24, 52, and 60 weeks postpartum. Mother-infant bonding was measured using the MPAS (Condon & Corkindale, 1998). Mothers who had had a previous pregnancy reported significantly healthier mother-infant bonding at 24 weeks postpartum. 
In the previously described prospective observational study, Figuieredo, Costa, Pacheco, and Pais (2009) also examined the relationship between parity and mother­infant bonding. Healthy mother-infant bonding did not significantly differ between primiparous and multiparous mothers. 
Kokubu, Okano, and Sugiyama (2012) also examined mother-infant bonding among community Japanese participants in the previously described study. There was no significant difference in mother-infant bonding between primiparous mothers and multiparous mothers. 
Sockol, Battle, Howard and Davis (20 14) also explored the relationship between parity and mother-infant bonding in a previously described study. There was not a significant relationship between the number of previous births and mother-infant bonding. 
A meta-analysis was conducted to determine the strength of the relationship between parity and mother-infant bonding. Table 4 presents the results of the random effects model for three studies assessing the relationship between parity and mother­infant bonding. Effect scores were calculated so that positive effect sizes represent better mother-infant bonding when mothers have more children. Parity was not significantly 
associated with mother-infant bonding (r = 0.12, 95% CI -0.13-0.35,p = 0.35). The 
SAMD values for two studies (Aiello & Lancaster, 2007; Sockol et al., 2014) exceeded 
2.58, but visual examination ofthe scree plot did not indicate that these values were 
discrepant from the distribution so they were retained in subsequent analyses. 
Results of tests for publication bias did not indicate potential bias in the included studies. The fail-safe Nvalue was zero, which did not exceed the tolerance value of25. The Q statistic indicated significant heterogeneity among the effect sizes (p < 0.01) and the Pvalue of 79.55 indicated a large level of heterogeneity. The funnel plot was symmetrical (see Figure 2) and Duval and Tweedie's trim-and-fill correction did not indicate any missing studies. 
Bonding measure, sample type, and location were not assessed as potential moderators as there were insufficient studies for subgroup analyses. Timing of assessment was a significant moderator of the relationship between parity and mother­infant bonding (slope= 0.02,p < 0.01). The later mother-infant bonding was assessed, the stronger the observed relationship between parity and mother-infant bonding (see Table 11). Interpersonal Factors 
Fetal Attachment. The bond a mother feels towards her unborn child is often referred to as fetal attachment (Edhborg, Nasreen, & Kabir, 2011). Some researchers posit that fetal attachment is likely related to mother-infant bonding because it is likely similar to the relationship between mother and infant (e.g., Edhborg, Nasreen, & Kabir, 2007; Shin & Kim, 2007). Fetal attachment as a relationship may be a precursor to the mother-infant bonding relationship. Four studies examined fetal attachment as a potential predictor of mother-infant bonding. 
In a prospective observational study, Damato (2004) assessed the relationship between fetal attachment and mother-infant bonding in community mothers oftwins. Participants were recruited support groups for mothers of twins. Of the 168 women who agreed to be involved in the study, 139 participants fully completed the questionnaires and were included in the analyses. Questionnaires assessing fetal attachment were mailed to the participants during pregnancy and mother-infant bonding assessments were completed at one month postpartum. Prenatal attachment was assessed using the Maternal Fetal Attachment Scale (MF AS; Cranley, 1981) and postpartum bonding was assessed using the PBQ (Brockington et al., 2001 ). The study found better prenatal bonding was significantly associated with healthy postpartum bonding. 
In the previously described study, Dubber, Reck, Muller, and colleagues (2014) also assessed the relationship between fetal attachment and mother-infant bonding with community participants in Germany. Prenatal attachment was assessed using the revised version of the Maternal Fetal Attachment Scale (MFAS-R; Van den Bergh, 1989). The study found that mothers who experienced better fetal attachment were significantly more likely to experience healthier mother-infant bonding. The study found that positive fetal attachment was significantly associated with lower rates of impaired postpartum bonding. 
Edhborg, Nasreen, and Kabir (2011) assessed the relationship between fetal attachment and mother-infant bonding with community participants recruited from a longitudinal study conducted in rural Bangladesh. Fetal attachment was assessed using the Prenatal Attachment Inventory (Muller, 1993) and postpartum bonding was assessed using the PBQ (Brockington, et al., 2001 ). The study found that stronger fetal attachment during pregnancy was associated with lower levels of impaired bonding at two to three months postpartum. 
Finally, in a prospective observational study, Shin & Kim (2007) also examined the relationship between fetal attachment and mother-infant bonding among 196 community participants in South Korea. Participants were recruited from four community hospitals in Korea. The mothers were assessed during pregnancy and within six weeks postpartum. Fetal attachment was assessed using the Korean version of the Maternal Fetal Attachment Scale (Kim, 2000) and postpartum bonding was assessed using the Maternal Attachment Inventory (MAl; Muller, 1994). The study found that better fetal attachment during pregnancy was significantly associated with better postpartum bonding. 
A meta-analysis was conducted to determine the strength of the relationship between fetal attachment and mother-infant bonding. Effect scores were calculated so that the positive effect sizes represent greater levels of healthy bonding when there are greater levels of fetal attachment. Table 5 presents the results ofthe random effects model for the correlation between fetal attachment and postpartum bonding. There was a significant overall relationship between fetal attachment and mother-infant bonding, with an effect size in the small to moderate range (r = 0.35, 95% CI 0.16-0.51,p < 0.001). The SAMD scores indicated that the values for the studies by Edhborg and colleagues (2011) and Shin and Kim (2007) exceeded 2.58. Visual examination of the scree plot indicated that the study by Edhborg and colleagues (20 11) was discrepant from the distribution and was excluded from subsequent analyses. The relationship between fetal attachment and mother-infant bonding was significant after exclusion of this outlier; the average effect 
size excluding this outlier was in the moderate range (r = 0.43 95% CI 0.34-0.51, p < 
0.001). 
The Q statistic indicated that there was not significant heterogeneity among the effect sizes (p = 0.68), and the Pvalue ofO.OO indicated a low level of heterogeneity. Results of tests for publication bias indicated potential bias in the included studies. The fail-safe N value was 4 7, which greatly exceeded the tolerance of 25. The funnel plot was asymmetric (see Figure 2), and the trim-and-fill correction suggested two studies missing to the right ofthe mean. After correcting for publication bias, the overall effect size was 0.46 (95% CI 0.39-0.53). 
Timing of assessment was assessed as a potential moderator. Location and bonding measure could not be assessed as moderators because of insufficient studies in subgroups. Timing of assessment was significantly associated with effect size (slope=­0.03,p < 0.05). The relationship between fetal attachment and mother-infant bonding was weaker in studies which assessed mother-infant bonding later during the postpartum period (see Table 11). 
Social Support. Two studies assessed the relationship between social support and mother-infant bonding. In the prospective observational study, Herunger and colleagues (2014) examined the relationship between social support and mother-infant bonding with Turkish mothers who had either had a vaginal or caesarean delivery. Mother-infant bonding was assessed using the MAS (Muller, 1994) and social support was assessed using the Multidimensional Scale of Perceived Social Support (MSPSS; Zimet, Dahlem, Zimet, et al., 1988). The study found a significant association between mother-infant bonding and social support. Higher levels of social support were significantly associated 
with mother-infant bonding. 
The previously described study conducted by Sockol and colleagues (2014) also assessed the relationship between marital status and impaired mother-infant bonding. The results indicated that married and partnered mothers had significantly more impaired mother-infant bonding than single, divorced, and widowed mothers. 
Psychiatric Predictors Anxiety. Six studies examined the relationship between maternal anxiety and mother-infant bonding. 
In the previously described study, Dubber, Reck, Muller, and Gawlik (2014) also assessed the relationship between anxiety and mother-infant bonding. Anxiety was assessed using the State Trait Anxiety Inventory (ST AI; Speilberger et al., 1970). The study found that trait anxiety was significantly associated with impaired mother-infant bonding, and there was a trend for an association between state anxiety and impaired mother-infant bonding. 
The previously described study by Edhborg, Nasreen, and Kabir (2011) also assessed the relationship between anxiety and mother-infant bonding. Anxiety was assessed using the state form of the STAI (Spielberger, 1983). The study found that state anxiety was significantly associated with impaired mother-infant bonding. 
The previously described study by Oriin, Yalc;m, and Mutlu (2013) also assessed the relationship between anxiety and mother-infant bonding. For these analyses, mother­infant bonding was assessed using the MIBS (Taylor, Atkins, & Kumar, 2005) and the PBQ (Brockington, 2001) and anxiety was assessed using the anxiety sub scale ofthe Brief Symptom Inventory (BSI; Derogatis, 1992). The study found that anxiety was 
significantly associated with impaired mother-infant bonding as assessed by the MIBS. 
The previously described study by Sockol, Battle, Howard and Davis (2014) also explored the relationship between anxiety and mother-infant bonding. Mothers with a diagnosed anxiety disorder reported comparable mother-infant bonding to mothers with other psychiatric disorders. 
Tietz, Zietlow, and Reck (2014) conducted a prospective observational study comparing 48 normal mothers to 30 mothers with postpartum anxiety disorders. Participants with anxiety disorders were recruited from mother and infant units and normal participants were recruited from community populations; all participants were recruited from Germany. Mother-infant bonding was measured using a German version of the PBQ (Brockington, Oates, George, et al., 2001) and anxiety was assessed using the Agoraphobic Cognitions Questionnaire (ACQ; Chambless, Caputo, Bright, & Gallagher, 1984), Body Sensations Questionnaire (BSQ; Chambless, Caputo, Bright, & Gallagher, 1984), and the Mobility Inventory (MI; Ehlers et al., 2001). The study found that anxiety­related cognitions, physical symptoms, and avoidance behaviors were all significantly associated with impaired mother-infant bonding. 
Turner, Wittkowski and Hare (2008) examined mother-infant bonding and anxiety among 64 community mothers in the United Kingdom in a prospective observational study. Mother-infant bonding was measured using the PBQ (Brockington, Oates, George, et al., 2001) and anxiety was assessed using the Beck Anxiety Inventory (BAI; Beck, 1990). The study found that anxiety was significantly associated with impaired mother­infant bonding. 
Five of the studies reported sufficient data for inclusion in the meta-analysis. Effect sizes were coded so that positive effect sizes represent a positive relationship between symptoms of anxiety and mother-infant bonding. Table 6 represents the results of the random effects model for the relationship between symptoms of anxiety and mother-infant bonding. There was a significant relationship between anxiety and mother­infant bonding, with higher levels of anxiety associated with moderate decreases in mother-infant bonding (r =-0.32, 95% CI -0.42-(-0.10),p < 0.001). 
The SAMD scores for two studies (Edhborg, Nasreen & Kabir, 2011; Orlin, Yal<;:m, & Mutlu, 2013) exceeded 2.58 and visual examination of the scree plot indicated that both of these scores were discrepant from the distribution and were consequently excluded from further analyses. After exclusion of these outliers, the relationship between symptoms of anxiety and mother-infant bonding remained significant (r = -0.33, 95% CI -0.44-(-0.20),p < 0.001) 
The Q statistic indicated that there was riot significant heterogeneity among the effect sizes (p = 0.93) and the i value of 0.00 indicated a very low level of heterogeneity. Results of tests for publication bias did not indicate potential bias in the included studies. The fail-safe N value was 16 which did not exceed the tolerance value of 30. The funnel plot was symmetric (see Figure 2). Duval and Tweedie's trim and fill procedure suggested no missing studies. 
Timing of assessment was assessed as a moderator ofthe relationship between anxiety and mother-infant bonding using meta-regression (see Table 11). Timing of assessment was not significantly associated with effect size (r = -O.OO,p = 0.93). 
Location, bonding measure, and anxiety measure could not be assessed because of insufficient studies within the subgroups. Depressive Symptoms. Seventeen studies examined maternal depression as a potential predictor of mother-infant bonding. 
The previously described study by Edhborg, Matthiesen, Lundh and Widstrom (2005) also examined the relationship between depressive symptoms and mother-infant bonding among community participants in Sweden. Depressive symptoms were assessed using the Edinburgh Postnatal Depression Scale (EPDS; Cox, Holden, & Sagovsky, 1987), and researchers also assessed whether women had received a previous diagnosis of depression. The study found that both symptoms of depression and a previous diagnosis of depression were significantly associated with impaired mother-infant bonding at two months postpartum. 
Edhborg, Nasreen, and Kabir (2011) also assessed the relationship between depression and mother-infant bonding in the previously described study of community participants in rural Bangladesh. Depression was assessed using the EPDS (Cox et al., 1987). The study found that higher levels of depressive symptoms were significantly associated with higher levels ofimpaired mother-infant bonding at two to three months postpartum. 
The previously described study conducted by, Figuieredo, Costa, Pacheco, and Pais (2009) also examined the relationship between depression and mother-infant bonding. Depressive symptoms were assessed using the EPDS (Cox et al., 1987) and mothers were classified as depressed if they reported scores of 13 or greater. Depressed mothers reported significantly more impaired bonding than non-depressed mothers; specifically, depressed mothers reported higher levels of negative bonding than non­
depressed mothers but the two groups did not differ on positive bonding. 
In a prospective observational study, Hornstein, Trautmann-Villalba, Hohrn and colleagues (2006) examined mother-infant bonding in 35 mothers with either a diagnosis of depression or a diagnosis of a psychotic disorder. Participants were recruited from a mother and baby psychiatric unit in Germany and had infants between one week and 24 weeks postpartum. Participants were diagnosed as either depressed or psychotic by a psychiatrist using the Clinical Global Impression scale and the DSM-IV-TR. Bonding was assessed using the PBQ (Brockington, Oates, George, et al., 2001). The study found that depressed mothers experienced significantly more impaired mother-infant bonding than psychotic mothers. 
Kaneko and Honjo (2014) examined mother-infant bonding among community Japanese mothers in a prospective observational study. The 1786 participants were recruited from local public health centers at the infant's three-month check-up. Mother­infant bonding was measured using the PBQ (Brockington, Oates, George, et al., 2001) and depressive symptoms were measured using the EPDS (Cox et al., 1987). Mothers with higher depressive symptoms reported significantly higher levels of bonding impairment. 
The previously described study by Kokubu, Okano, and Sugiyama (2012) also examined the relationship between depression and mother-infant bonding among community Japanese participants. Depressive symptoms were assessed using the Hospital Anxiety and Depression Scale (HADS; Zigmong & Snaith, 1983). The study found a significant association between depressive symptoms during late pregnancy and at one 
month postpartum and impaired mother-infant bonding at one month postpartum. 
A correlational study conducted by Moehler and colleagues (2006) examined mother-infant bonding among 101 community participants recruited from obstetric units in Heidelberg, Germany. Participants were assessed at two weeks, six weeks, and 24 weeks postpartum. Mother-infant bonding was assessed using the PBQ (Brockington, Oates, George, et al., 2001) and depressive symptoms were assessed using the EPDS (Cox et al., 1987). Maternal depressive symptoms at four months postpartum were significantly associated with impaired bonding at four and fourteen months postpartum. 
A cross-sectional study conducted by Muzik and colleagues (2012) examined the relationship between mother-infant bonding and depression in 150 participants. The sample consisted of 97 women who had experienced child abuse or neglect and 53 mothers who had not experienced child abuse or neglect. Of these participants, 55% had a diagnosis ofPTSD, 20% major depression and 11% had a comorbid diagnosis ofPTSD and major depression. Participants were assessed at six weeks, 16 weeks, and 24 weeks postpartum. Depression symptoms and diagnosis were assessed using the Postpartum Depression Screening Scale (Beck and Gable, 2000) and mother-infant bonding was assessed using the PBQ (Brockington, Oates, George, et al., 2001). Women without a diagnosis of a major depressive disorder (MDD) reported significantly more impaired bonding at six weeks, four months and six months postpartum than women without an MDD diagnosis. 
Ohoka, Koide, Goto, Murase, and colleagues (2014) examined the relationship 
between mother-infant bonding and depressive symptoms among 389 Japanese 
community participants. Participants were recruited from two obstetric hospitals in Japan and were assessed during early and late pregnancy, five days postpartum, and four weeks postpartum. Mother-infant bonding was measured using the MIBS (Taylor, Atkins, & Kumar, 2005) and depressive symptoms were measured using the EPDS (Cox, et al., 1987). The study found a significant association between depressive symptoms and impaired mother-infant bonding at all four time points. 
In the previously described study, Ortin, Yal<;m, and Mutlu (2013) also examined the relationship between depressive symptoms and mother-infant bonding. Depressive symptoms were assessed using the EPDS (Cox et al., 1987). The study found a significant association between depressive symptoms and impaired mother-infant bonding at two months postpartum as assessed by both the PBQ and the MIBS. 
In a prospective observational study, Pearson, Lightman, and Evans (2011) examined the relationship between mother-infant bonding and depressive symptoms in 49 community participants in the United Kingdom. Participants were recruited from community midwives and were assessed during late pregnancy and between 12 and 24 weeks postpartum. Mother-infant bonding was assessed using the PBQ (Brockington, Oates, George, et al., 2001) and depressive symptoms were assessed using the EPDS (Cox et al., 1987). The study found that depressive symptoms were not significantly associated with mother-infant bonding. 
Seng and colleagues (2013) examined the relationship between mother-infant bonding and postpartum depression in a cross-sectional observational study with 566 community participants recruited from three prenatal clinics. Participants were recruited in three cohorts: women with a history of trauma exposure who met criteria for a lifetime history ofPTSD, women with a history oftrauma exposure who did not meet criteria for a lifetime PTSD diagnosis, and women without a history of trauma exposure. Participants were assessed over the phone during early and late pregnancy and at six weeks postpartum. Mother-infant bonding was assessed using the PBQ (Brockington, Oates, George, et al., 2001) and depressive symptoms were assessed using the Postpartum Depression Screening Scale (PDSS, Beck & Gable, 2000). The study found a significant association between depression diagnosis and impaired mother-infant bonding at six weeks postpartum, and having a diagnosis of MDD prior to birth was also significantly associated with impaired mother-infant bonding. 
The previously described study conducted by Sockol, Battle, Howard and Davis (2014) also examined the relationship between depressive symptoms and mother-infant bonding. Depression was assessed using the Edinburgh Postnatal Depression Scale (EPDS; Cox et al., 1987) and compared women who met criteria for MDD to women with other psychiatric diagnoses. The study found no significant differences in mother­infant bonding between mothers with depression and mothers with other psychiatric diagnoses. Among women with MDD, there was no difference in mother-infant bonding between women with moderate MDD and women with severe MDD. The study also found a significant association between depressive symptoms and impaired mother-infant bonding. 
In the previously described study conducted by Tietz, Zietlow, and Reck (2014) also assessed the relationship between depressive symptoms and mother-infant bonding. Depressive symptoms were assessed using the German version ofthe EPDS (Bergant et al., 1998). The study found a significant association between depressive symptoms and 
impaired mother-infant bonding. 
Turner, Wittkowski and Hare (2008) examined the relationship between mother­infant bonding and depressive symptoms among 64 mothers in the United Kingdom in an observational study of mothers with infants under one year old. Mother-infant bonding was measured using the PBQ (Brockington, Oates, George, et al., 2001) and depressive symptoms were measured using the Beck Depression Inventory-II (Beck, Steer, & Brown, 1996). The study found a significant association between depressive symptoms and impaired mother-infant bonding. 
A cross-sectional study conducted by Wittkowski, Williams and Wieck (2010) examined the relationship between mother-infant bonding and depressive symptoms among 132 clinical participants recruited from a psychiatric mother and baby unit in the United Kingdom. Approximately one-third ofthe participants were diagnosed with depression, and almost half of the participants experienced either moderate or severe depression. Mother-infant bonding was measured using the PBQ (Brockington, Oates, George, et al., 2001) and depressive symptoms were measured using the Beck Depression Inventory-II (Beck, Steer, & Brown, 1996). The study found a significant association between depressive symptoms and impaired mother-infant bonding. 
Yoshida, Yamashita, Conroy, Marks and Kumar (2012) examined the relationship between mother-infant bonding and depressive.symptoms in a cross-sectional observational study of 554 Japanese community participants. Participants were assessed at five days, four weeks and 16 weeks postpartum. Mother-infant bonding was measured 
using the MIBS (Taylor, Atkins, & Kumar, 2005); based on an exploratory factor 
analysis, the authors report outcomes for two subscales: lack of affection and anger and 
rejection towards the baby. Depressive symptoms were measured using the EPDS (Cox et 
al., 1987). The study found a significant association between depressive symptoms and 
both the lack of affection and anger and rejection subscales ofthe bonding measure. 
A meta-analysis of 13 studies was conducted to assess the relationship between maternal depressive symptoms and mother-infant bonding. Effect scores were calculated so that positive correlations represent higher levels of mother-infant bonding when mothers report higher levels of depressive symptoms. Table 7 presents the results of the random effects model for this meta-analysis. There was a significant overall relationship of medium magnitude between depression and mother-infant bonding (r = -0.41, 95% CI -0.48-(-0.34),p < 0.001). Three studies had SAMD values greater than 2.58 (Kaneko & Honjo, 2014; Wittkowski et al., 2010; Yoshida et al., 2012). Visual examination of the scree plot indicated that the values for two studies (Wittkowski et al. , 201 0; Yoshida et al. , 2012) were discrepant from the distribution and were subsequently excluded from the analyses. After these outliers were removed, the relationship between depressive symptoms and mother-infant bonding remained significant (-0.41, 95% CI -0.46-(-0.36), 
p < 0.001).. 
Results of tests for publication bias did not indicate potential bias in the included studies. The fail-safe N value was 1633, which greatly exceeds the tolerance value of 65. The Q statistic indicated there was significant heterogeneity among the effect sizes (p < 0.01) and Pvalue of 60.75 indicated a medium level of heterogeneity. The funnel plot was symmetrical (see Figure 2) and Duval and Tweedie's fill-and-trim correction suggested no missing studies. 
Three categorical characteristics ofthe included studies were assessed as potential moderators: measure of mother-infant bonding, sample type, and language (see Table 1 0). There was no significant difference in the strength of the relationship between mother-infant bonding and depression in studies that used the MIBS compared to studies that used the PBQ (p = 0.89). Study location was not a significant moderator of effect sizes. There was no difference in effect sizes in studies conducted in countries in which English was the primary language compared to studies conducted in languages in which English was not the primary language (p = 0.26). There was no significant differences in effect sizes between studies with clinical participants and studies with community participants (p = 0.75). Sample type could not be assessed as a potential moderator because of insufficient studies in subgroups. 
Timing of assessment was assessed as a potential moderator using meta­regression (see Table 11). Timing of assessment was significantly associated with effect size (slope= -0.01,p < 0.05). The association between depressive symptoms and mother­infant bonding was weaker in studies with later postpartum assessments. 
Post-Traumatic Stress Disorder. Three studies assessed post-traumatic stress as a predictor of mother-infant bonding. A meta-analysis assessing the relationship between PTSD and mother-infant bonding could not be conducted due to differences in how data was reported and analyzed. 
Davies, Slade, Wright, & Stewart (2008) conducted a prospective observational study examining the relationship between mother-infant bonding and PTSD among 211 community mothers. Participants were recruited from Sheffield maternity hospital and were assessed at six weeks postpartum. Mother-infant bonding was assessed using the Maternal Measures of Perceived Attachment (MP AS; Condon & Corkindale, 1998). Diagnosis ofPTSD was assessed using the Structured Clinical Interview for DSM-IV Axis I Disorders (SCID-PTSD Module) and PTSD symptoms were assessed using the Post-Traumatic Stress Disorder Questionnaire (PTSDQ; Watson, Juba, Manifold, Kucula, & Anderson, 1991 ). The study found that mothers who reported full or partial symptoms of PTSD reported significantly lower quality of attachment with their infants, including greater hostility towards their infants and less pleasure in interactions with their infants, than mothers without symptoms ofPTSD. Mothers who met full criteria for PTSD at six weeks postpartum also reported less desire for proximity to their infants. Symptoms of PTSD were also significantly associated with increased levels of impaired mother-infant bonding. 
In the previously described study by Muzik, Bockneck, Broderick and colleagues (2012) also examined the relationship between mother-infant bonding and PTSD in 150 participants diagnosed with either PTSD or depression. PTSD was diagnosed using the National Women's Study (NWS; Resnick et al., 1993). Mothers with a diagnosis of PTSD reported significantly greater impairments in mother-infant bonding at six weeks, four months, and six months postpartum. Symptoms of PTSD were also significantly associated with greater levels of impaired bonding. 
In the previously described study conducted Seng, Sperlich, Low, and colleagues (2013) also examined the relationship between mother-infant bonding and PTSD. PTSD symptoms and diagnosis were assessed using the National Women's Study (Resnick et al., 1993). The study found that symptoms of PTSD were significantly associated with impaired mother-infant bonding. While mothers with a previous diagnosis ofPTSD were not significantly more likely to experience impaired bonding with their infants; mothers who had a previous diagnosis of PTSD with comorbid major depressive disorder reported significantly higher levels of impaired mother-infant bonding. 

Interventions 
Eight studies examined the effects of interventions on mother-infant bonding. 
A randomized control study conducted by Ahn and Shin (20 1 0) examined the effects of a kangaroo care intervention on the mother-infant bond with 20 South Korean mothers of premature infants who were hospitalized in a Neonatal Intensive Care Unit (NICU). Mothers in the intervention group engaged in ten 60-minute sessions of close physical contact, or "kangaroo care," with their premature infants over a period of three weeks. The comparison group consisted of ten mother-infant pairs matched for gestational age and birth weight who did not participate in the intervention. Mother-infant bonding was assessed using a modified version ofthe MAl (Muller, 1994). Mothers in the intervention group experienced significantly better mother-infant bonding at post­treatment than mothers in the control group. 
A randomized controlled trial conducted by Garcia and colleagues (2010) examined the effect of repetitive transcranial magnetic stimulation on mother-infant bonding among nine mothers with postpartum depression. Repetitive transcranial magnetic stimulation is a non-invasive procedure that applies external electronic stimulation to the participant's head (Padberg, & Moller, 2003). Repetitive transcranial magnetic stimulation has been used to treat chronic major depression for decades (Greenberg & Kellner, 2005). Mothers met criteria for Postpartum Depression according to the DSM-IV-TR and scored greater than nine points on the EPDS (Cox, Holden, & Savogsky, 1987). Participants received repetitive transcranial magnetic stimulation five 
times a week for four weeks. Mother-infant bonding was assessed using the PBQ 
(Brockington et al., 2001) immediately after the intervention and 12 and 24 weeks after 
the intervention. Mothers experienced a significant improvement in mother-infant 
bonding from pre-treatment to post-treatment. 
Gtirol & Polat (20 12) examined the effects of a baby massage intervention on mother-infant bonding among Turkish community mothers. Mothers in the intervention group massaged their infants every day for 15 minutes for 38 days. Describe the control group, describe how large each group was. Mother-infant bonding was assessed using the MAl (MUller, 1994). Among mothers in the intervention group, mother-infant bonding increased significantly from pre-treatment to post-treatment. Mothers in the intervention group experienced significantly healthier mother-infant bonding at post-treatment than did mothers in the control group. 
Matthey and Speyer (2008) examined the effects of a treatment program for infant sleep problems on mother-infant bonding with 116 Australian mothers. Participants were recruited from residential units for mothers and infants with "baby-care difficulty." Mothers and their infants were admitted for five days in order to address and treat infant sleep problems. The intervention provided mothers with individualized information about their infants' sleep patterns and provided mothers with information as to how to cope with their infants' sleeping difficulties. Mother-infant bonding was assessed using the Being a Mother and Bonding Scale which was developed by the authors for the purpose ofthis study; participants were assessed at 5 weeks and 16 weeks post-discharge (BaMB; 
Matthey & Speyer, 2008). The study found that the number of mothers indicating 
mother-infant bonding impairment decreased significantly from 22.4% at pre-treatment to 
12.2% at post-treatment, and remained stable at four months post-discharge (12.5%). 
A randomized controlled trial conducted by Mulcahy and colleagues (2009) examined the effects of interpersonal therapy on mother-infant bonding. Participants were 50 Australian mothers diagnosed with postpartum depression. Participants were randomly assigned to either an interpersonal therapy (IPT) group in which participants received group therapy for eight weeks or a treatment as usual group. Participants in the interpersonal therapy intervention group met with a therapist for two individual sessions, participated in eight group therapy sessions, and participated in a meeting with the therapist and their partner. Mother-infant bonding was assessed using the MAl before the intervention, four weeks post-intervention, and eight weeks post-intervention (Muller, 1994). Mothers in the intervention group experienced significantly better mother-infant bonding with their infants at the 8-week assessment than did mother in the control group. 
O'Mahen and colleagues (2014) examined the effect of an internet-delivered behavioral activation intervention on mother-infant bonding among 249 women in the United Kingdom. Participants were recruited via "postpartum depression" chat rooms and Facebook pages. Participants were assessed within 12 months postpartum. In order to be eligible for inclusion in the study mothers had to score above a 13 on the EPDS (Cox, Holden, & Savogsky, 1987) and met the International Classification of Diseases, 1 01h Revision (lCD-I 0) criteria for Major Depressive Disorder. Participants in the intervention group completed a modified 12-session treatment course and a follow-up relapse prevention meeting online. Mothers in the intervention group completed guided 
behavioral interactions with their infants through an internet based intervention. Mother­
infant bonding was assessed using the PBQ immediately post-treatment and at a six 
month follow-up assessment (Brockington et al., 2001). Participants were randomly 
assigned to treatment as usual (control group) or the intervention group. The study found 
no significant difference in mother-infant bonding between the intervention and control 
groups at post-treatment. 
A randomized controlled trial conducted by Perry and colleagues (20 11) examined the effect of a perinatal depression prevention intervention on mother-infant bonding in 217 low-income Latina women. Participants were eligible for the study if they were considered at "high risk" for developing depression and scored 16 or above on the Center for Epidemiologic Studies . Scale (CES-D; Radloff, 1977). The intervention consisted of eight weeks of group therapy sessions and three individual meetings with a bilingual therapist. The therapy was designed to prevent the development of depression among the included participants. The control group received treatment as usual, which was characterized as usual care in the community. Participants were recruited from local clinics and mother-infant bonding was assessed using the MP AS twice during pregnancy and between six and eight weeks postpartum (Condon & Corkindale, 1998) between six and eight weeks postpartum. There was a trend for mothers in the intervention group to report better mother-infant bonding compared to mothers in the control group. 
Reay and colleagues (20 12) examined the effect of group interpersonal therapy on Australian mothers with postpartum depression in a randomized controlled trial. The intervention consisted of two individual therapy sessions and eight group therapy sessions. The control group received treatment as usual. Mother-infant bonding was assessed using the MAl (MUller, 1994) at baseline, week four of treatment, end of treatment, and 12 weeks and two years post-treatment. Mothers in the intervention group 
experienced significantly better bonding at post-treatment than did mothers in the control 
group. The study did not find any significant differences in mother-infant bonding 
between the intervention and control groups at three-month and two-year follow up 
assessments. 
Tsivos, Calam, Sanders and Wittkowski (2014) examined the effects of a parenting intervention on mother-infant bonding among British mothers with postpartum depression. Participants in the intervention group participated in a parenting intervention, "Baby Triple P," that focused on teaching mothers how to positively and effectively interact with their infants. The Baby Triple P intervention focused on providing mothers with positive parenting techniques and information about effective, positive parenting. The control group received treatment as usual. Mother-infant bonding was assessed using the PBQ (Brockington et al., 2001) at baseline, immediately post-treatment, and 12 weeks post-treatment. The study found that mother-infant bonding improved more from baseline to post-treatment and three months follow-up in the intervention group compared to the control group, but that these differences were not statistically significant. 
Two meta-analyses were conducted to assess the effect of interventions on mother-infant bonding. The first meta-analysis examined controlled effect sizes, the difference between intervention and control groups at post-treatment (Feske & Chambless, 1995). The second meta-analysis examined uncontrolled effect sizes, changes in the intervention group from pre-treatment to post-treatment (Feske & Chambless, 1995). 
The first meta-analysis was conducted to compare mother-infant bonding between control and experimental groups at post-treatment. Effect sizes were calculated so that positive effect sizes represent better mother-infant bonding in the intervention group at post-treatment. Table 7 presents the results of the random effects model for the relationship between intervention and mother-infant bonding. There was a trend for mothers in intervention groups to report better mother-infant bonding at post-treatment than mothers in control groups (d= 0.29, 95% CI -0.01-0.59,p = 0.06). The SAMD value for one study (O'Mahen et al., 2014) was above 2.58; visual examination ofthe scree plot indicated that this study was discrepant from the overall distribution so it was excluded from subsequent analyses .. After this outlier was removed, the difference in mother­infant bonding between mothers who received interventions and mothers in control conditions was significant (d= 0.39, 95% CI 0.20-0.57,p < 0.001). 
Results of tests for publication bias did indicated some potential for bias in the included studies. The fail-safe N value was 19 which did not exceed the tolerance value of 35. The Qstatistic did not indicate significant heterogeneity among the effect sizes (p = 0.69) and the Pvalue ofO.OO indicated a small level ofheterogeneity. The funnel plot was slightly asymmetric (see Figure 2), and Duval and Tweedie's trim-and-fill correction suggested two studies missing to the left of the mean. After the correction was applied, mothers in intervention groups still reported better mother-infant bonding than mothers in control groups (d = 0.34, 95% CI 0.16-0.51 ). 
Moderator analyses for bonding measure, sample type, and location could not be conducted due to insufficient studies in subgroups. A meta-regression was conducted to 
assess the relationship between timing of assessment and effect size. There were no 
differences in effect size associated with the timing of assessment (slope = -0.01, p = 
0.83). 
A second meta-analysis was conducted to determine the difference in mother­infant bonding between pre-treatment and post-treatment among mothers who received interventions (see Table 9). Effect sizes were calculated so that positive effect sizes indicate better mother-infant bonding at post-treatment compared to pre-treatment. Mother-infant bonding was significantly higher at post-treatment compared to pre­treatment (d= 0.72, 95% CI 0.34-1.10,p < 0.001). None ofthe SAMD scores were above 
2.58. 
Tests of publication bias indicated potential bias in the included studies. The fail safe N value was 55 which exceeded the tolerance of 35 studies. The Q statistic indicated significant heterogeneity among the effect sizes (p < 0.01) and the i value of72.84 indicated a medium level of heterogeneity among the effect sizes. The funnel plot (see Figure 2) was asymmetric, and Duval and Tweedie's trim-and-fill procedure suggested one study missing to the left of the mean. After the correction was applied, the analysis continued to indicate mothers reported higher mother-infant bonding at post-treatment compared to pre-treatment (d = 0.64, 95% CI 0.13-0.97). 
Moderator analyses could not be conducted for bonding measure, sample type, or location due to insufficient studies in subgroups. There was a trend for timing of assessment to be associated with effect size (slope= 0.04, p = 0.07). Studies with later postpartum assessments had greater changes in mother-infant bonding from pre-treatment to post-treatment. 
A second meta-analysis was conducted to determine the difference in mother­infant bonding between pre-treatment and post-treatment among mothers who received interventions (see Table 9). Effect sizes were calculated so that positive effect sizes indicate better mother-infant bonding at post-treatment compared to pre-treatment. Mother-infant bonding was significantly higher at post-treatment compared to pre­treatment (d= 0.72, 95% CI 0.34-1.10,p < 0.001). None ofthe SAMD scores were above 
2.58. 
Tests ofpublication bias indicated potential bias in the included studies. The fail safe N value was 55 which exceeded the tolerance of 35 studies. The Qstatistic indicated significant heterogeneity among the effect sizes (p < 0.01) and the Pvalue of72.84 indicated a medium level of heterogeneity among the effect sizes. The funnel plot (see Figure 2) was asymmetric, and Duval and Tweedie's trim-and-fill procedure suggested one study missing to the left of the mean. After the correction was applied, the analysis continued to indicate mothers reported higher mother-infant bonding at post-treatment compared to pre-treatment (d = 0.64, 0.13-0.97). 
Moderator analyses could not be conducted for bonding measure, sample type, or location due to insufficient studies in subgroups. There was a trend for timing of assessment to be associated with effect size (slope= 0.04, p = 0.07). Studies with later postpartum assessments had greater changes in mother-infant bonding from pre-treatment to post-treatment. 
Discussion 
Overall, our results revealed that several factors that significantly predicted mother-infant bonding. Psychiatric factors, demographic factors, an interpersonal factor, and interventions were significantly related to mother-infant bonding. 
Of the eight meta-analyses we conducted, our results revealed that six factors significantly predicted mother-infant bonding. We assessed four types ofpredictors: psychiatric, interpersonal, demographic, and interventions. Our analyses ofpsychiatric predictors revealed that maternal anxiety and maternal depression are significant predictors ofmother-infant bonding. Our analyses of interpersonal factors revealed that fetal attachment is a significant predictor of mother-infant bonding. Our analyses of demographic factors revealed that infant sex and maternal education were significant predictors of mother-infant bonding, but that parity was not a significant predictor of mother-infant bonding. Our analyses of interventions revealed that interventions significantly predicted mother-infant bonding in both controlled and uncontrolled meta­analyses. 
Our meta-analyses ofpsychiatric predictors revealed significant relationships of small to medium strength with mother-infant bonding. We found these results to be highly interesting because of the strength of these relationships and their clinical implications. 
We found that higher levels of depressive symptoms were associated with higher levels of impaired mother-infant bonding. These results are highly consistent with the research literature examining maternal depression and mother-infant relationships. Reviews of maternal depression have consistently found that depressed mothers 
demonstrate parenting difficulties (Downey & Coyne, 1990; Rutter, 1990). A meta­
analysis of maternal-depression and parenting behavior conducted by Lovejoy and colleagues (2000) found that maternal depression is significantly associated with irritability and hostility toward the child. The research also found that depressed mothers also reported significantly more disengagement from the child (Lovejoy et al., 2000). The negative parenting behaviors reported by depressed mothers likely contribute to the higher rates of bonding impairment among depressed mothers. The negative parenting behaviors associated with maternal depression likely prevent mothers from fully engaging with their infants and developing a healthy mother-infant bond. 
We also found that higher levels of maternal anxiety are associated with higher levels of impaired mother-infant bonding. Our results are highly consistent with the research literature assessing parenting between anxious mothers and their infants. Anxious mothers show less sensitive responsivity to their infants and use reduced emotional tone when interacting with their infants (Nicol-Harper, 2007). Research also indicates that anxious mothers are more withdrawn from their infants and have more difficult mother-infant interactions (Beebe et al., 20 II; Murray et al., 2008; Stein et al., 20I2). Because anxious mothers have more difficult interactions with their infants, they may also develop lower parenting self-efficacy (Bakersman-Kranenburg et al., 2013). This lowered self-efficacy could contribute to impaired mother-infant bonding by causing mothers to question their own control of interactions with infants. Strained interactions between anxious mothers and their infants may play a role in the development of impaired mother-infant bonding. Mothers who have difficult interactions with their infants may have more trouble developing positive feelings about the relationship with the infant and may subsequently have more difficultly developing a healthy mother-infant bond. 
Additionally, we also found that interventions effectively decreased impaired mother-infant bonding. Mothers in intervention groups experienced significantly better healthy mother-infant bonding at post-treatment than at pre-treatment. Similarly, mothers in intervention groups experienced significantly better bonding at post-treatment than did mothers in control groups. These results are particularly interesting because a variety of interventions were utilized in the meta-analysis including transcranial magnetic stimulation, kangaroo care, interpersonal therapy, Internet based behavioral interaction, agd educational parenting. The interventions also included a variety of populations including clinical, community, mothers with postpartum depression, mothers who did not speak English, and location. These results suggest that a variety of interventions would effectively reduce impaired mother-infant bonding for a variety of populations. 
Our results revealed that higher levels of fetal attachment were significantly associated with better mother-infant bonding. The definition of fetal attachment as proposed by Cranley (1981) suggests a relationship between fetal attachment and mother­infant bonding. Cranley (1981) defines fetal attachment as the extent to which women engage in behaviors that represent an affiliation and interaction with their unborn child" 
(p. 282). Using this definition, the relationship between fetal attachment and mother­infant bonding can be expected. Mothers who are more engaged with their unborn child would likely also engage more their infants, and encourage better mother-infant bonding. We also found that the relationship between fetal attachment and mother-infant bonding decreased the later the assessment occurred. Our results suggest that fetal attachment and mother-infant bonding are related, but are different relationships. The similarity between these relationships decreases as the infant grows older and the mother-infant bonding relationship definitively differs from the mother-fetal relationship. The results of our moderator analyses also have important implications for the development of a healthy mother-infant bond. Because the relationship between mother-fetal attachment and mother-infant bonding decreases over time, mothers who do not develop healthy fetal attachments may still be able to develop health mother-infant bonds. Mothers who do not develop positive attachments with their fetuses should be informed about the possibility of developing later healthy mother-infant bonds. 
Of the potential demographic factors, only infant sex, maternal education, and parity meta-analyses could be conducted. We found that mothers of male infants experienced significantly better bonding than mothers of female infants. The research on differences in mother-infant relationships between male and female infants is controversial. Some researchers suggest that female infants experience better relationships with their mothers (Jordan et al., 1991 ), while other researchers contend that male infants experience better relationships with their mothers (Eichenbaum & Orbach, 1991 ). The Strange Situation experiment, as designed by Ainsworth, revealed that more male infants tend to be categorized as having "disorganized" attachment than females (Carlson et al., 1989; Lyons-Ruth et al., 1997). However, a meta-analysis conducted in 1999 by van ljzendoorn and colleagues found no significant differences in mother-infant attachment between male and female infants. Consequently, our findings are a part of larger discussion about differences in the mother-infant relationship between male and female infants. Although van ljzendoorn and colleagues (1999) found no differences between male and female infants in their attachment patterns, the studies included in their meta-analysis utilized infants beyond 12 months postpartum. Because our meta-analysis only includes studies in which mothers are less than 12 months postpartum, we may have assessed a different relationship. There may not be significant differences in attachment between male and female infants outside of 12 months postpartum, but our findings suggest that mothers ofmale within the 12-month postpartum period experience better bonding than mothers of female infants. 
Infant sex preferences across culture and nation could also have influenced the results of our meta-analysis. Research has found that some cultures prefer one infant sex to the other (Arnold, 1997). In regions such as Latin America there is a slight preference for female infants to male infants (Arnold, 1997). In other regions such as the Middle East, Nepal, Turkey, and India male infants are preferred to female infants (Arnold, 1997). Other research has also replicated these results (Ben-Porath & Welch, 1976). There are some regions in the world in which there is no preference for one infant sex to the other (Arnold & Zhaoxiang, 1986). Of the studies included in our meta-analysis of infant sex, three of the four included studies were conducted in regions in which there is a documented preference for one infant sex over the other. Two of the included studies were conducted in regions in which male infants were preferred, one study was conducted in a region in which female infants were preferred, and one study was conducted in a region with no preference for infant sex. Mothers who have an infant of the less desirable sex may have a more difficult time forming a healthy bond with that infant. If mothers birth an infant with a less desirable sex, these mothers may feel slight 
resentment toward their infant and have difficulty bonding with their infant. 
Alternatively, mothers who do birth an infant of a more desirable sex may have an easier 
time bonding with their infant. 
We found that mothers with higher education experienced significantly more impaired mother-infant bonding. These results are highly surprising. Research indicates that children ofmothers with higher education are less likely to experience child mortality (Desai & Alva, 1998) and are less likely to experience physical health problems (Williams, Wake, Hesketh et al., 2005). Although more educated mothers may be more adept at handling and preventing child mortality and child health problems, education may not necessarily prepare them for or aid the development of the mother-infant bond. More educated mothers may have more experience in or more knowledge about certain aspects of motherhood, but knowledge and experience do not necessarily guarantee a healthy bond between mother and infant. Research also indicates that more educated mothers tend to wait longer to have children (Martin, Hamilton, Osternman et al., 2013). Because these mothers tend to wait longer to have children they may also have higher expectations of the mother-infant bond. These mothers may have higher hopes for the strength of the relationship. 
Our meta-analysis of parity did not reveal a significant relationship with mother­infant bonding. These results are surprising because research examining parity and the mother-infant relationship indicates that mothers of multiple infants are better at breastfeeding infants and are better at calming their fussy infants (Thoman, Turner, Leiderman et al., 1970; Crockenberg & Smith, 1982). Our lack of significant results could be due to the current dearth of studies examining parity as a predictor of mother­infant bonding. Alternatively, the better maternal behaviors associated with multiparous mothers may not necessarily guarantee a healthy mother-infant bond. Multiparous 
mothers may be at just a high a risk as primiparous mothers to develop an unhealthy 
mother-infant bond. 
We were unable to conduct meta-analyses ofPTSD, maternal employment, maternal age, and social support. Our narrative reviews did not definitively indicate whether maternal employment and maternal age are significant predictors of mother­infant bonding. Despite our inability to conduct meta-analyses for these predictors, our narrative reviews suggest that PTSD is likely associated with impaired mother-infant bonding. This conclusion is corroborated by the research literature on the relationship between PTSD symptoms and parenting (Lyons-Ruth & Block, 1996). Mothers who have been exposed to trauma and experience PTSD symptoms are significantly more likely to be hostile toward their infants and emotionally withdraw from their infants (Lyons-Ruth & Block, 1996; Brand, Engel, Canfield et al., 2006). Research also indicates that infants of mothers with PTSD symptoms are more likely to have difficult temperaments, be more wary of novel stimuli, and experience distress (Brand, Engel, Canfield et al., 2006). These maternal and infant qualities may make it more difficult for mothers and infants to form healthy mother-infant bonds. 
Our results have several important clinical implications. Firstly, mothers with mental illnesses such as depression, anxiety, and PTSD are more likely to develop impaired mother-infant bonds. Health care providers should screen for mental illnesses during pregnancy and the early postpartum period to assess whether mothers are coping with mental illnesses. Research on postpartum depression indicates that up to 50% of all cases ofpostpartum depression go undetected (Beck, & Gable, 2001). If maternal mental illnesses were properly diagnosed early in the postpartum period, perhaps mothers could 
be treated and encouraged to develop healthy mother-infant bonds. It is likely that other 
maternal mental illness in the postpartum period also go undiagnosed and, consequently, 
untreated. Research also indicates that using reliable and validated questionnaires to 
assess the mother-infant bond is particularly important (Klier, 2006). It is often very 
difficult for doctors and nurses to accurately assess the mother-infant bond and the 
severity of impaired mother-infant bonding (Klier, 2006). The difficulty in properly 
diagnosing and assessing maternal illness and the mother-infant bond suggests that 
mothers should be screened throughout the postpartum period using reliable and 
validated questionnaires. Screening procedures for mother-infant bonding and maternal mental illness should become a regular part of medical check-ups for the mother during pregnancy and during infant check-ups in the postpartum period. 
Our results indicate that depressed and anxious mothers are most at risk for developing impaired mother-infant bonds. If and when mothers are identified as depressed during pregnancy or the postpartum period, these mothers should be provided with resources to treat their mental illness and increase the chances of developing a healthy mother-infant bond. 
If and when screening procedures identify impaired mother-infant bonding relationships, our results indicate that a variety of interventions are effective at decreasing impaired mother-infant bonding among a variety of populations. Obstetric nurses, doctors, and clinicians should consider referring mothers with impaired mother-infant bonding relationships to interventions designed to increase healthy mother-infant bonding. 
Our results also revealed a significant association between fetal attachment and mother-infant bonding. These results suggest that if we can increase fetal attachment, we can increase the likelihood that mothers will develop healthy mother-infant bonds. Research indicates there may be specific ways to increase fetal attachment among mothers. Meta-analytic research conducted by Y archeski and colleagues (2009) found that social support, gestational age, and fetal ultrasound screening significantly predicted mother-fetal attachment. Health care providers should provide pregnant mothers with information about these factors in hopes that these factors will increase the likelihood of mother-fetal attachment and subsequently increase the likelihood of healthy mother­infant bonding. 
Despite the fairly large amount of research conducted on mother-infant bonding, we were only able to systematically assess six predictors of mother-infant bonding. Unfortunately, a fairly small amount of studies systematically assess predictors of mother-infant bonding provided sufficient data to conduct meta-analyses. We were unable to conduct meta-analyses on maternal age, social support, PTSD, and infant age due to insufficient effect size data. In addition, our meta-analyses suggested publication bias in almost half of our meta-analyses. Publication bias suggests that unpublished studies could alter our results and detract from our conclusions. These limitations suggest that more research should be conducted on potential predictors of mother-infant bonding. Researchers should note the lack ofresearch on certain predictors and should consider these predictors in future research. 
Some ofthe measures included in the studies in the meta-analysis could another 
potential limitation to our results. The PBQ (Brockington et al., 2001) was the most 
common measure of mother-infant bonding included in our studies. The questions in this measure may assess related, but different factors from mother-infant bonding. These factors could include maternal anxiety and maternal depression. The PBQ asks mothers to rate their answers from "always" to "never" on statements such as; "My baby winds me up" and "My baby makes me feel anxious." Questions such as these may not be assessing mother-infant bonding, but instead may be assessing maternal anxiety. This suggests that measures of mother-infant bonding may not actually be accurately assessing mother-infant bonding, but instead may be assessing problems such as mother-infant bonding. This suggests that some studies may not accurately assess severity of impaired mother-infant bonding because their measures are assessing maternal anxiety instead of impaired mother-infant bonding. 
Overall, our results indicate that psychiatric factors, demographic factors an interpersonal factor, and interventions are significant predictors of mother-infant bonding. These predictors can be assessed and treated to increase the chances of developing a healthy mother-infant bond. 
References References marked with an asterisk indicate studies included in the meta-analyses. 
*Ahn, H. Y., Lee, J., & Shin, H. J. (20 1 0). Kangaroo care on premature infant growth and maternal attachment and post-partum depression in South Korea. Journal of tropical pediatrics, 56(5), 342-344. 
*Aiello, R., & Lancaster, S. (2007). Influence of adolescent maternal characteristics on infant 
development. Infant mental health journal, 28(5), 496-516. Ainsworth, M.D. (1989). Attachment beyond infancy. American Psychologist, 44(4),496-516. Ainsworth, M.D., Blehar, M. C., Waters, E., & Walls, S. (1978). Patterns ofattachment: 
A 
psychological study ofthe strange situation. New Jersey, Lawrence Erlbaum Associates. Arnold, F. (1997). Gender preferences for children: Findings from demographic and health 
survey. Proceedings ofthe XXIII IUSSP General Population Conference, Beijing. Liege: IUSSP. Arnold, F., & Zhaoxiang, L. ( 1986). Sex preference, fertility, and family planning in 
China. Population and Development Review, 221-246. Bakermans-Kranenburg, M. J., Van Ijzendoorn, M. H., & Juffer, F. (2003). Less is more: meta-
analyses of sensitivity and attachment interventions in early childhood. 
Psychological bulletin, 129(2), 195. Beal, D. J., Corey, D.M., & Dunlap, W. P. (2002). On the bias ofHuffcutt and Arthur's (1995) 
procedure for identifying outliers in the meta-analysis of correlations. Journal of 
Applied Psychology, 87, 583-589. 
Beck, A. T., Steer, R. A., & Brown, G. K. (1996). Beck depression inventory-If. San Antonio, 
TX: The Psychological Corporation. Beck, C. T. & Gable, R. K. (2000). Postpartum depression screening scale: development and 
psychometric testing. Nursing Research and Practice, 29(5), 272-282. Beck, C. T., & Gable, R. K. (2001). Further validation of the postpartum depression screenmg 
scale. Nursing Research, 50(3), 155-164. Beebe, B., Steele, M., Jaffe, J., Buck, K. A., Chen, H., Cohen, P., ... & Feldstein, S. (2011). 
Maternal anxiety symptoms and mother-infant self-and interactive contingency. Infant Mental Health Journal, 32(2), 174-206. Bell, S. M., & Ainsworth, M. D. S. (1972). Infant crying and maternal responsiveness. 
Child 
Development, 43, 1171-1190. Ben-Porath, Y., & Welch, F. (1976). Do sex preferences really matter?. The Quarterly Journal of 
Economics, 285-307. Bergant, A. M., Nguyen, T., Heim, K., Ulmer, H., Dapunt, 0. (1998). German language versiOn 
and validation of the Edinburgh postnatal depression scale. translation: Deutsche Medizinische Wochenzeitschrift, 123(3), 35-40. 
Bienfait, M., Rideau, A., Picaud, J., Daude, H., Combes, C., & Franc, N. (2011). 
Pertinence of the self-report mother-to-infant bonding scale in the neonatal unit of a maternity ward. Early Human Development, 87(4), 281-287. 
Bloom, K. C. (1995). The development of attachment behaviors in pregnant adolescents. 
Nursing 
Research, 44 (5), 284-289. Booth, B. E., Verma, M., & Beri, R. S. (1994). Fetal sex determination in infants in Punjab, 
India: correlations and implications. Bmj, 309(6964), 1259-1261. Borenstein, M. (2009) Introduction to Meta-analysis. Chichester, U.K.: John Wiley & Sons. Borenstein, M., Hedges, L., Higgins, J., & Rothstein, H. (2005). Comprehensive meta­analysis 
version 2. Englewood NJ: Biostat. Bowlby, J. (1958). The nature ofthe child's tie to his mother. International Journal of Psycho-
Analysis, 39, 350-373. Bowlby, J. (1988). A secure base: Clinical applications ofattachment theory. London: 
Routledge. Brand, S. R., Engel, S. M., Canfield, R. L., & Yehuda, R. (2006). The Effect of Maternal PTSD 
Following in Utero Trauma Exposure on Behavior and Temperament in the 9 
Month Old Infant. Annals ofthe New York Academy ofSciences, 1071(1), 454­
458. 
Brockington, I. F., Oates, J., George, S., Turner, D., Vostanis, M., Murdoch, C. (2001). A screening questionnaire for mother-infant bonding disorders. Archives of Women 's Mental Health, 3(1), 133-140. 
Brockington, I. F. (2004). Postpartum psychiatric disorders. The Lancet, 363(9414), 
1078. 
Brockington, I. F., Macdonald, E., & Wainscott, G. (2006). Anxiety, obsessions and 
morbid 

preoccupations in pregnancy and the puerperium. Archives ofWomen's Mental 
Health, 9(5), 253-263. *Byers, J. F., Yovaish, W., Lowman, L. B., & Francis, J.D. (2003). Co-Bedding Versus Single 
Bedding Premature Multiple Gestation Infants in Incubators. Journal ofObstetric, 
Gynecologic, & Neonatal Nursing, 32(3), 340-347. Campbell, S., Brownell, C., Hungerford, A., Spieker, S., Mohan, R., & Blessing, J. (2004) The 
course of maternal depressive symptoms and maternal sensitivity as predictors of attachment security at 36 months. Developmental Psychology, 26, 15-23. Carlson, V., Cicchetti, D., Barnett, D., & Braunwald, K. (1989). Disorganized/disoriented attachment relationships in maltreated children. Developmental Psychology, 25, 
525­
531. Chambless, D. L., Caputo, G. C., Bright, P., & Gallagher, R. (1984). Assessment of fear of fear 
in agoraphobics: the body sensations questionnaire and the agoraphobic cognitions questionnaire. Journal ofconsulting and clinical psychology, 52(6), 1090. 
Cicchetti, D., & Rizley, R. (1981). Developmental perspective on the etiology, 
intergenerational transmission, and sequelae of child maltreatment. New Directions Child Development, II, 31-55. 
Cohn, J. F., Campbell, S. B., Matias, R., & Hopkins, J. (1990). Face-to-face interactions 
of postpartum depressed and nondepressed mother-infant pairs at 2 months. Developmental psychology, 26(1), 15. 
Cohn, J. F., Campbell, S. B., Matias, R., & Hopkins, J. (1990). Face-to-face interactions 
of postpartum depressed and nondepressed mother-infant pairs at 2 months. Developmental psychology, 26(1 ), 15. 
Cohn, J. F., & Tronick, E. Z. (1983). Three-month-old infants' reaction to simulated maternal 
depression. Child development, 185-193. Condon, J., & Corkindale, C. The assessment of parent to infant attachment: Development of a self-report questionnaire instrument. Journal ofReproductive and Infant 
Psychology, 16(1), 57-76. *Cooklin, A. R., Rowe, H. J., & Fisher, J. R. (2012). Paid parental leave supports breastfeeding 
and mother--infant relationship: a prospective investigation of maternal postpartum employment. Australian and New Zealand Journal ofPublic Health, 36(3), 249­
256. 
Cox, J. L., Holden, J. M., & Sagovsky, R. (1987). Detection ofpostnatal depression: Development of the 1 0-item Edinburgh Postnatal Depression Scale. British Journal ofPsychiatry, 150, 782-876. 
Cranley, M.S. (1981). Development of a tool for the measurement of maternal attachment 
during pregnancy. Nursing Research and Practice, 30,281-284. Crockenberg, S. B., & Smith, P. (1982). Antecedents of mother-infant interaction and infant 
irritability in the first three months of life. Infant Behavior and Development, 5(2), 105­
119. *Damato, E. G. (2004). Prenatal attachment and other correlates of postnatal maternal 
attachment to twins. Advances in Neonatal Care, 4(5), 274-291. *Davies, J., Slade, P., Wright, I., & Stewart, P. (2008). Posttraumatic stress symptoms 
following 
childbirth and mothers' perceptions of their infants. Infant Mental Health 
Journal, 29(6), 537-554. *Dearman, L. R., Musonda, P., Roberts, F. G., Bowles, K. M., & Morris, E. P. (2012). Bonding 
in women with postnatal anaemia: a pilot case control study looking at postnatal 
bonding in women who have been diagnosed with anaemia at a University Hospital in the East of England. Archives ofgynecology and obstetrics, 285(5), 1243-1248. 
Derogatis, L. R. (1992). The Brief Symptom Inventory-BSI administration, scoring and procedures manual-11. USA: Clinical Psychometric Research, Inc. Desai, S., & Alva, S. (1998). Maternal education and child health: Is there a strong causal relationship?. Demography, 35(1), 71-81. Downey, G., & Coyne, J. C. (1990). Children of depressed parents: an integrative review. Psychological bulletin, 1 08(1 ), 50. *Dubber, S., Reck, C., MUller, M., & Gawlik, S. (2014). Postpartum bonding: the role of perinatal depression, anxiety and maternal-fetal bonding during pregnancy. Archives of women's mental health, 1-9. Duval, S., & Tweedie, R. (2000). Trim and fill: A simple funnel-plot-based method of testing and adjusting for publication bias in meta-analysis. Biometrics, 56, 455­
463. 
*Edhborg, M., Matthiesen, A. S., Lundh, W., & Widstrom, A. M. (2005). Some early 

indicators 
for depressive symptoms and bonding 2 months postpartum-A study of new mothers and 
fathers. Archives ofWomen's Mental Health, 8(4), 221-231. *Edhborg, M., Nasreen, H. E., & Kabir, Z. N. (2011). Impact of postpartum depressive and 
anxiety symptoms on mothers' emotional tie to their infants 2-3 months 
postpartum: A population-based study from rural Bangladesh. Archives ofwomen's mental health, 14(4), 307-316. 
Egeland, B. & Vaughn, B. (1981 ). Failure of "bond formation" as a cause of abuse, neglect, and maltreatment. American Journal ofOrthopsychiatry, 51(1), 78-84. Egeland, B., & Farber, E. A. (1984). Infant-mother attachment: Factors related to its development and changes over time. Child development, 753-771. Ehlers, A., Margrad, J., Chambless, D. (2001). Fragebogen zu korperbezogenen Angsten, 
Kognitionen und Vermeidung (AKV). Beltz Test, Gottingen. Eibl-Eibesfeldt, I. (1971). Love and hate: The natural history ofbehavior patterns. New York, 
NY: Holt, Rinehart, & Winston, Inc. Eichenbaum, L., Orcbach, S. (1988). Between women. New York: Viking. Elster, A. B., McAnarney, E. R., & Lamb, M. E. (1983). Parental behavior ofadolescent 
mothers. Pediatrics, 71(4), 494-503. Eyer, D. E. (1994). Mother-infant bonding: A science fiction. Human Nature, 5(19),1 69­
94. 

Fearon, R. P., Bakermans-Kranenburg, M. J., Van IJzendoorn, M. H., Lapsley, A.M., 
& 
Reisman, G. I. (2010). The significance of insecure attachment and disorganization in the development of children's externalizing behavior: A meta­analytic study. Child Development, 81(2), 435-456. 
Feldman, R., Weller, A., Leckman, J., Kuint, & Eidelman, A. (1999). The nature of a 
mother's tie to her infant: Maternal bonding under conditions of proximity, separation and potential loss. Journal ofChild Psychology and Psychiatry, 40(6), 929-939. 
Feske, U. & Chambless, D. L. (1995). Cognitive-behavioral versus exposure only treatment for 
social phobia: A meta-analysis. Behavior Therapy, 26, 311-324. Field, T. (1992). Infants of depressed mothers. Development and psychopathology, 4(0 1 ), 49-66. Field, T. M., Sandberg, D., Garcia, R., Vega-Lahr, N., Goldstein, S., & Guy, L. (1985) 
Pregnancy problems, postpartum depression and early mother-infant interactions. 
Developmental Psychology, 21, 1152-1156. *Figueiredo, B., Costa, R., Pacheco, A., & Pais, A. (2007). Mother to infant bonding and father 
to infant initial emotional involvement. Early Child Development and Care, 177(5), 521-532. *Figueiredo, B., & Costa, R. (2009). Mother's stress, mood and emotional involvement 
with the infant: 3 months before and 3 months after childbirth. Archives ofWomen's 
Mental Health, 12(3),143-153. Fleming, A. S., O'Day, D. H., & Kraemer, H., & Wong, P. Y. (1997). Hormonal and experiental 
correlates of maternal responsiveness during pregnancy and the puerperium in human mothers. Hormones and Behavior, 31, 145-158. 
*Garcia-Toro, M., Pascual-Leone, A., Romera, M., Gonzalez, A., Mico, J., Ibarra, 0., & Tormos, J. M. (2001). Prefrontal repetitive transcranial magnetic stimulation as add on treatment in depression. Journal ofNeurology, Neurosurgery & Psychiatry, 71 (4), 546-548. 
Gerrard, J., Holden, J. M., Elliott, S. A., McKenzie, P., McKenzie, J., & Cox, J. L. 
(1993).A trainer's perspective of an innovative programme teaching health visitors about the detection, treatment and prevention of postnatal depression. Journal of Advanced Nursing, 18(11 ), 1825-1832. 
Gavin, N. I., Gaynes, B. N., Lohr, K. N., Meltzer-Brody, S., Gartlehner, G., & Swinson, 
T. (2005). Perinatal depression: a systematic review of prevalence and incidence. Obstetrics & Gynecology, 106(5, Part 1), 1071-1083. 
George, C., & Solomon, J. (1999). Attachment and caregiving: The caregiving behavioral system. In J. Cassidy & P. Shaver (Eds.), Handbook ofattachment: Theory, research and clinical applications (pp. 649-667). New York, NY: The Guilford Press. 
*Gharaibeh, M. K., & Hamlan, A.M. (2012). Factors influencing maternal attachment of first-
time Jordanian mothers. Journal ofResearch in Nursing, 17(3), 289-303 Goldberg, S. (1983). Parent-infant bonding. StLouis, MO: Mosby. Greenberg, R. M., Kellner, C. H. (2005). Electroconvulsive therapy: a selected review. 
The 
American Journal ofGeriatric Psychology, 13, 268-281. *Gunning, M. D., Waugh, H., Robertson, F., & Holmes, B. Emotional intelligence, attachment 
and bonding and communication. Community Practice, 84(3), 27-31. *Giirol, A., & Polat, S. (2012). The effects of baby massage on attachment between mother and 
their infants. Asian nursing research, 6(1 ), 35-41. 
Harmon, R. (1981). Perinatal influences on the family: Some preventive implications. J 
Prevent 
Psychiatry, 1(1), 132-139. *Hergiiner, S., <;ic;:ek, E., Annagiir, A., Hergiiner, A., & Ors, R. (2014). Association of delivery 
type with postpartum depression, perceived social support and maternal attachment. Dii~iinen Adam: The Journal ofPsychiatry and Neurological Sciences, 27(1), 15-20. 
Hettema, J. M., Neale, M. C., & Kendler, K. S. (2001). A review and meta-analysis of the 
genetic epidemiology of anxiety disorders. American Journal ofPsychiatry, 
158(10), 1568-1578. Hipwell, A. E., & Jumar, R. (1997). The impact of postpartum affective psychosis on the child. 
In L. Murray & P. J. Cooper (Eds.), Postpartum depression and child 
development (pp. 265-293). New York: Guilford Press. *H0ivik, M.S., Burkeland, N. A., Linaker, 0. M., & Berg Nielsen, T. S. (2013). The Mother and 
Baby Interaction Scale: a valid broadband instrument for efficient screening of postpartum interaction? A preliminary validation in a Norwegian community sample. Scandinavian journal ofcaring sciences, 27(3), 733-739. 
Hornstein, C., Trautmann-Villalba, P., Hohm, E., Rave, E., Wortmann-Fleischer, S., & Schwartz, 
M. (2006). Maternal bond and mother-child interaction in severe postpartum psychiatric disorders: Is there a link? Archives ofWomen's Mental Health 9(5), 279-284. 
Hubbs-Tait, L., Hughes, K. P., Culp, A.M., Osofsky, J.D., Hann, D. M., & Eberhart­Wright, A. ( 1996). Children of adolescent mothers: Attachment representation, maternal depression, and later behavior problems. American Journal ofOrthopsychiatry, 66(3), 416­
426. 
Huffcutt, A. I., & Arthur, W., Jr. (1995). Development of a new outlier statistic for meta­analytic 
data. Journal ofApplied Psychology, 80, 327-334. *Iida, M., Horiuchi, S., & Porter, S. E. (2012). The relationship between women-centred care 
and women's birth experiences: A comparison between birth centres, clinics, and 
hospitals in Japan. Midwifery, 28(4), 458-465. *Jordan, J. V., Kaplan, A. G., Miller, J. B., Stiver, I. P., & Surrey, J. L. (1999). Women's growth 
in connection. New York: Guilford. *Kaneko, H. & Honjo, S. (2014). The psychometric properties and factor structure of the 
postpartum bonding questionnaire in Japanese mothers. Psychology, 5, 1135­1142. Klaus, M. H., & Kennell, J. H. (1982). Parent-infant bonding, St. Louis, MO: Mosby. Klier, C. M. (2006). Mother-infant bonding disorders in patients with postnatal depression: the 
Postpartum Bonding Questionnaire in clinical practice. Archives ofwomen's 
mental health, 9(5), 289-291. Kendler, K. S., Gallagher, T. J., Abelson, J. M., & Kessler, R. C. (1996). Lifetime prevalence, 
demographic risk factors, and diagnostic validity of nonaffective psychosis as assessed in a US community sample: the National Comorbidity Survey. Archives 
ofgeneralpsychiatry, 53(11), 1022-1031. 
Kennell, J. H., & Klaus, M. H. (1984). Mother-infant bonding: Weighing the evidence. 
Developmental Review, 4(3), 275-282. *Kokubu, M., Okano, T., & Sugiyama, T. (2012). Postnatal depression, maternal bonding f 
failure, and negative attitudes towards pregnancy: a longitudinal study of pregnant 
women in Japan. Archives ofWomen's Mental Health, 15(3), 211-216. Kitamura, T., Takegata, M., Haruna, M., Yoshida, K., Murakami, M., & Goto, Y. (2013), The 
mother-infant bonding scale: Factor structure and psychosocial correlates of parental bonding disorders in Japan. Journal ofChild and Family Studies, 10, 1-9. Lamb, M. (1983). Early mother-neonate contact and the mother-child relationship. 
Journal of Child Psychology and Psychiatry, 171 , 175-181. Laux, L., Glanzman, P., Schaffner, P., & Spielberger, C. D. (1970). Das State-Trait­
Angstinventar. Beltz Test, Weinheim. Lipsey, M. W., & Wilson, D. (2001). Practical meta-analysis. Applied Social Research Methods 
Series, val. 49, Thousand Oaks, CA: SAGE Publications, Inc. *Loh, C. C., & Vostanis, P. (2004). Perceived mother-infant relationship difficulties in postnatal 
depression. Infant and Child Development, 13(2), 159-171 . Lounds, J. J., Borowski, J. G., Whitman, T. L., Maxwell, S. E., & Weed, K. (2005). Adolescent 
parenting and attachment during infancy and early childhood. Parenting: Science 
and Practice, 5(1), 91-117. Lovejoy, M. C., Graczyk, P. A., O'Hare, E., & Neuman, G. (2000). Maternal depression and 
parenting behavior: A meta-analytic review. Clinical psychology review, 20(5), 561-592. Lyons-Ruth, K., & Block, D. (1996). The disturbed caregiving system: Relations among 
childhood trauma, maternal caregiving, and infant affect and attachment. Infant 
Mental Health Journal, 17(3), 257-275. Lyons-Ruth, K., Zoll, D., Connell, D., & Grunebaum, H. U. (1986). The depressed mother and 
the one-year-old infant: Environment, interaction, attachment, and infant development. New Directions for Child and Adolescent Development, 34, 61-82. Luft H (1964) Die Wochenbettdepression. Klinik und pathogenetische Faktoren. 
Nervenarzt, 35 
185-194. Martin, J. A., Hamilton, B. E., Osterman, M. J., Curtin, S.C., & Matthews, T. J. (2015). Births: 
final data for 2013. National vital statistics reports: from the Centers for Disease Control and Prevention, National Center for Health Statistics, National Vital Statistics System, 64( 1 ), 1. 
*Mason, Z. S., Briggs, R. D., & Silver, E. J. (2011). Maternal attachment feelings 
mediate between maternal reports of depression, infant social-emotional development, and parenting stress. Journal ofReproductive and Infant Psychology, 29(4), 382-394. 
Milner, J. S. (1994). Assessing physical child abuse risk: The child abuse potential inventory. 
Clinical Psychology Review, 14(6), 547-583. *Moehler E, Brunner R, Wiebel A, Reck C, Resch F. (2006). Maternal depressive symptoms in 
the postnatal period are associated with long-term impairment of mother-child 
bonding. Archives ofWomen's Mental Health, 9(5), 273-8. *Muller, M. E. (1993). Development of the prenatal attachment inventory. Western Journal of 
Nursing Research, 15(3), 199-215. Muller, M. E. (1994) A questionnaire to measure mother-to-infant attachment. Journal of Nursing Research, 2( 1 ), 129-141. Muller, M. E. (1996). Prenatal and postnatal attachment: A modest correlation. Journal of 
Obstetric, Gynecologic, & Neonatal Nursing, 25(2), 161-166. Muller, D., Teismann, T., Havemann, B., Michalak, J., & Seehagen, S. (2013). Ruminative 
thinking as a predictor of perceived postpartum mother-infant bonding. Cognitive Therapy And Research, 37(1), 89-96. 
Murrary, L., Hipwell, A., & Hooper, R. (1996). The cognitive development of 5-year-old 
children ofpostnatally depressed mothers. Journal ofChild Psychology and Psychiatry, 37, 927-935. Murray, L., De Rosnay, M., Pearson, J., Bergeron, C., Schofield, E., Royal-Lawson, M., 
& 
Cooper, P. J. (2008). Intergenerational transmission of social anxiety: The role of 
social referencing processes in infancy. Child development, 79(4), 1049-1064. *Muzik, M., Bocknek, E. L., Broderick, A., Richardson, P., Rosenblum, K. L., Thelen, K.,& 
Seng, J. S. (2013). Mother-infant bonding impairment across the first 6 months postpartum: The primacy of psychopathology in women with childhood abuse and neglect histories. Archives ofwomen's mental health,16(l), 29-38. 
Nasreen, H. E., Kabir, Z. N., Forsell, Y., & Edhborg, M. (2011). Prevalence and 
associated factors of depressive and anxiety symptoms during pregnancy: A population based study in rural Bangladesh. BMC Women's Health, 11(1), 22. 
Nicol-Harper, R., Harvey, A. G., & Stein, A. (2007). Interactions between mothers and infants: 
Impact of maternal anxiety. Infant Behavior and Development, 30(1), 161-167. *Ohoka, H., Koide, T., Goto, S., Murase, S., Kanai, A., Masuda, T., ... & Ozaki, N. (2014). 
Effects of maternal depressive symptomatology during pregnancy and the postpartum period on infant-mother attachment. Psychiatry and clinical neurosciences, 68(8), 631-639. 
*O'Higgins, M., Roberts, I. S. J., Glover, V., & Taylor, A. (2013). Mother-child bonding 
at 1 year; associations with symptoms of postnatal depression and bonding in the first few weeks. Archives ofwomen's mental health, 16(5), 381-389. 
*O'Mahen, H. A., Richards, D. A., Woodford, J., Wilkinson, E., McGinley, J., Taylor, R. 
s.,& 
Warren, F. C. (2014). Netmums: a phase II randomized controlled trial of a guided Internet behavioural activation treatment for postpartum depression. Psychological medicine, 44(08), 1675-1689. 
*Ori.in, E., Yal9m, S. S., & Mutlu, B. (2013). Relations of maternal psychopathologies, 
social-obstetrical factors and mother-infant bonding at 2-month postpartum: a sample of Turkish mothers. World Journal ofPediatrics, 9( 4), 350-355. 
Padberg, F., & Moller, H. J. (2003). Repetitive Transcranial Magnetic Stimulation. CNS 
drugs, 17(6), 383-403. *Pearson, R. M., Lightman, S. L., & Evans, J. (2011), Attentional processing of infant emotion 
during late pregnancy and mother-infant relations after birth. Archive ofWomen's Mental 
Health, 14(1 ), 23-31. 
*Perry, D. F., Ettinger, A. K., Mendelson, T., & Le, H. N. (2011). Prenatal depression 
predicts postpartum maternal attachment in low-income Latina mothers with infants. Infant Behavior and Development, 34(2), 339-350. 
*Quinlivan, J. A., & Evans, S. F. (2005). Impact of domestic violence and drug abuse in pregnancy on maternal attachment and infant temperament in teenage mothers in the setting of best clinical practice. Archives ofWomen's Mental Health, 8(3), 191-199. 
*Reay, R. E., Owen, C., Shadbolt, B., Raphael, B., Mulcahy, R., & Wilkinson, R. B. 
(2012). Trajectories of long-term outcomes for postnatally depressed mothers treated with group interpersonal psychotherapy. Archives ofwomen's mental health, I 5(3), 217-228. 
Reck, C., Klier, C. M., Pabst, K., Stehle, E., Steffenelli, U., Struben, K., & Backenstrass, 
M. (2006). The German version of the Postpartum Bonding Instrument: Psychometric properties and association with postpartum depression. Archives ofWomen's Mental Health, 9(5), 265-271. 
Resnick, H. S., Kilpatrick, D. G., Dansky, B.S., Saunders, B. E., Best, C. L. (1993). 
Prevalence of civilian trauma posttraumatic stress disorder in a representative sample of women. Journal ofConsulting and Clinical Psychology> 61(6), 984-991. 
Richards, M.P. M. (1971). Social interaction in the first few weeks ofhuman life. 
Psychiatric 
Neurology Neurochirurgia, 74, 35-42. *Robakis, T. K., Williams, K. E., Crowe, S., Kenna, H., Gannon, J., & Rasgon, N. L. (2014). 
Optimistic outlook regarding maternity protects against depressive symptoms 
postpartum. Archives ofwomen's mental health, 1-12. Rosenthal, R. (1991). Meta-analytic procedures for social research. Newbury Park, CA: Sage. Rutter, M. (1990). Commentary: Some focus and process considerations regarding effects of 
parental depression on children. Developmental Psychology, 26(1), 60. Schroder, K. E., Carey, M.P., & Vanable, P. A. (2003). Methodological challenges in research 
on sexual risk behavior: II. Accuracy of self-reports. Annals ofbehavioral 
medicine, 26(2), 104-123. *~en, S., & Kavlak, 0. (2012). Transgenerational attachment in Manisa, Turkey.Contemporary 
nurse, 41(1), 126-132. *Seng, 1. S., Low, L. K., Sperlich, M., Ronis, D. L., & Liberzon, I. (2011). Post-traumatic stress disorder, child abuse history, birthweight and gestational age: a prospective cohort study. BJOG: An International Journal ofObstetrics & Gynaecology, 118(11), 1329-1339. 
*Shin, H., Park, Y. 1., & Kim, M. J. (2006). Predictors ofmaternal sensitivity during the early 
postpartum period. Journal ofAdvanced Nursing, 55(4), 425-434. *Sockol, L. E., Battle, C. L., Howard, M., & Davis, T. (2014). Correlates ofimpaired mother-
infant bonding in a partial hospital program for perinatal women. Archives of 
women's mental health, 17(5), 465-469. Speiker, S. J., & Booth, C. (1988). Family risk typologies and patterns of insecure attachment. In 
J. Belsky & T. Nezworski (Eds.) Clinical implications ofattachment (pp. 95-135). 
Hilsdale, NJ: Erlbaum. Spielberger, C. D., Gorsuch, R. L., & Lushene, R. E. (1970). State-Trait Anxiety Inventory, 
manual for the State-Trait Anxiety Inventory. Consulting Psychologist Press, Palo Alto. Stein, A., Craske, M.G., Lehtonen, A., Harvey, A., Savage-McGlynn, E., Davies, B., & 
Counsell, N. (2012). Maternal cognitions and mother-infant interaction in postnatal depression and generalized anxiety disorder. Journal ofabnormal psychology, 121(4), 795. 
Taylor, A., Atkins, R., & Kumar, R. (2005). Mother-to-infant bonding scale: Links with early 
maternal mood. Archives ofWomen 's Mental Health, 81, 45-51. Thoman, E. B., Turner, A.M., Leiderman, P. H., & Barnett, C. R. (1970). Neonate­mother 
interaction: Effects of parity on feeding behavior. Child Development, 1103-1111. *Tietz, A., Zietlow, A. L., & Reck, C. (2014). Maternal bonding in mothers with postpartum 
anxiety disorder: the crucial role of subclinical depressive symptoms and maternal 
avoidance behaviour. Archives ofwomen's mental health, 17(5), 433-442. *Tikotzky, L. (2014). Postpartum Maternal Sleep, Maternal Depressive Symptoms and Self-
Perceived Mother-Infant Emotional Relationship. Behavioral sleep medicine, 1­
19. *Turner JM, Wittkowski A, Hare DJ. (2008). The relationship of maternal mentalization and 
executive functioning to maternal recognition of infant cues and bonding. British Journal ofPsychology, 99(4), 499-512. Van den Bergh, B. R. (1989). De emotionele toestand van de (zwangere) vrous, obstetrische complicaties en het gedrag en de ontwikkeling van de foetus en van het kind tot 
de 
leefitjd can zeven maanden. Unpublished doctoral thesis, Catholic University of Leuven. Van Ijzendoorn, M. H., & Kroonenberg, P. M. (1988). Cross-cultural patterns of attachment: A 
meta-analysis ofthe strange situation. Child Development, 59, 147-156. Watson, C. G., Juba, M.P., Manifold, V., Kucula, T. & Anderson, P. E. D. (1991). The PTSD 
interview: Rationale, description, reliability, and concurrent validity on a DSM-III 
based technique. Journal ofClinical Psychology, 47, 205-214. Williams, T. M., Joy, L.A., Travis, L., Gotowiec, A., Blum-Steele, M., & Aiken, L. S. (1987). 
Transition to motherhood: A longitudinal study. Infant Mental Health Journal, 
8(3), 251-265. Williams, J., Wake, M., Hesketh, K., Maher, E., & Waters, E. (2005). Health-related quality of 
life of overweight and obese children. Jama, 293(1 ), 70-76. *Wittkowski, A., Williams, J., & Wieck, A. (2010). An examination of the psychometric 
properties and factor structure of the Post-partum Bonding Questionnaire in a clinical I npatient sample. British Journal ofClinical Psychology, 49(2), 163-172. Yarcheski, A., Mahon, N. E., Yarcheski, T. J., Hanks, M. M., & Cannella, B. L. (2009). A meta-
analytic study of predictors of maternal-fetal attachment. International Journal of Nursing Studies, 46(5), 708-715. 
*Yoshida, K., Yamashita, H., Conroy, S., Marks, M., & Kumar, C. (2012). A Japanese version of Mother-to-Infant Bonding Scale: Factor structure, longitudinal changes and links 
with maternal mood during the early postnatal period in Japanese mothers. Archives of Women's Mental Health, 15(5), 343-352. 
*Yuan, S. Y., & Freeman, R. (2010). Can social support in the guise of an oral health 
education intervention promote mother-infant bonding in Chinese immigrant mothers and their infants?. Health Education Journal. 
Zigmond, A. S., & Snaith, R. P. (1983). The hospital anxiety and depression scale. Acta 
psychiatrica scandinavica, 67(6), 361-370. Zimet, G. D., Dahlem, N. W., Zimet, S. G., & Farley, G. K. (1988). The multidimensional scale 
of perceived social support. Journal ofpersonality assessment, 52(1), 30-42. 
META-ANALYSIS OF MOTHER-INFANT BONDING Table 1 
Descriptions ofthe Included Studies 
Bonding  Timing of  
Stud~  Study Design  Location  Samrle Tyre  Measure  Assessment  Predictors Assessed  
Ahn et al. (20 1 0)  Randomized controlled  South  Mothers of  MIBS  Not specified  Intervention (kangaroo care)  
trial  Korea  preterm infants  
Byers et al. (2005)  Randomized controlled  USA  Mothers of  MAl  Not specified  Intervention (co-bedding pre-term infants)  
trial  preterm infants  
Chen et al. (2011)  Prospective observational  Taiwan  Mothers  MAl  4 weeks  Fetal attachment  
treated for  8 weeks  
infertility  
Cooklinet al. (2012)  Prospective observational  New  Community  PAQ  40 weeks  Maternal anxiety  
Zealand  
Damato (2004)  Prospective observational  USA  Mothers of  MAl  4 weeks  Prenatal attachment  
twins  
Davies et al. (2008)  Prospective observational  UK  Community  MPAS  4 weeks  PTSD total scales and subscales  
Dearman et al. (20 12)  Cross-sectional  UK  Anemic  PBQ  4 weeks  Maternal age  
observational  mothers  Parity  
Infant age  
Mode of delivery  
Dubber et al. (2014)  Prospective observational  Germany  Community  PBQ  12 weeks  Fetal attachment  
Depressive symptoms  
Maternal anxiety  
Pregnancy related anxiety questionnaire  
Edhborg et al. (2005)  Prospective observational  Sweden  Community  PBQ  1 week  Previous depression  
8 weeks  Bonding at one week  
Depressive symptoms  
Edhborg et al. (20 11)  Prospective observational  Bangladesh  Mothers with  PBQ  At childbirth  Maternal anxiety  
anxiety or  8-12 weeks  Depressive symptoms.  
depression, or  
mentally well  

META-ANALYSIS OF MOTHER-INFANT BONDING 
Bonding  Timing of  
Stud~  Stud~ Design  Location  Sam12le T~J2e  Measure  Assessment  Predictors Assessed  
Figueiredo et al. (2009)  Prospective observational  Portugal  Community  MIBS  l week  Maternal age  
Maternal education  
Employment status  
Adverse life events  
Parity  
Infant sex  
Gestational age  
Depressive symptoms  
Neonatal problems  
Figeuiredo et al. (2009)  Prospective observational  Portugal  Community  MIBS  Antenatal  Prenatal bonding  
12 weeks  
Garcia et al. (20 1 0)  Prospective observational  USA  Mothers with  PBQ  4-52 weeks  Intervention (rTMS)  
PPD  
Gharaibeth et al. (2012)  Prospective observational  Jordan  Community  MAl  4-24 weeks  Infant age  
Infant sex  
Planning of pregnancy  
Gunning et al. (20 11)  Cross-sectional  Scotland  Mothers with  PBQ  16 weeks  Attachment anxiety  
observational  low  Attachment avoidance  
socioeconomic  
status  
Gural et al. (2012)  Prospective observational  Turkey  Community  PBQ  24-48 hours  Intervention (baby massage)  
4 weeks  
Herunger et al. (2014)  Prospective Observational  Turkey  Community  MAS  Participants  Delivery type  
who delivered  Depression symptoms  
via vaginal  Social support  
delivery were  
M = 24.22, SD  
= 2.46 weeks  
Participants  
who delivered  
via cesarean  
section M =  
23.72, SD =  
2.74 weeks  
Hiyivik et al. (2013)  Prospective observational  Norway  Community  PBQ  8 weeks  Mother-infant interactions  
16 weeks.  
Iida et al. (2012)  Cross-sectional  Japan and  Community  MAl  1-5 days  Women centered care  
observational  Tokyo  

META-ANALYSIS OF MOTHER-INFANT BONDING 
Bonding  Timing of  
Stud~  Stud~ Design  Location  Sam2le T~2e  Measure  Assessment  Predictors Assessed  
Jordan et al. (2014)  Prospective observational  Australia  Mothers of  MPAS  16 weeks  Maternal depressive symptoms  
infants who  Maternal employment  
had cardiac  
surgery  
Kaneko et al. (2014)  Prospective observational  Japan  Community  PBQ  M = 17.66, SD  Depressive symptoms.  
= 1.37 weeks  
Kokubu et al. (2012)  Prospective observational  Japan  Community  MIBS  At birth  Mother' s response to pregnancy  
5 days  Unwanted pregnancy  
4 weeks  Anxiety late pregnancy  
Depression late pregnancy  
Bonding failure at 5 days postpartum  
Maternity blues  
Parity  
Loh et al. (2004)  Prospective observational  UK  Mothers with  PBQ  M = 13.7 and  Infant physical problems  
PPD  SD = 13 weeks  Birth weight  
Mason et al. (20 11)  Prospective observational  USA  Community  MPAS  8 weeks  Infant social-emotional development,  
24 weeks  Mother-infant interactions  
Depressive symptoms  
Mulcahy eta!. (20 1 0)  Randomized controlled  USA  Community  MAl  Not specified  Intervention (interpersonal group therapy)  
trial  
Moehler et al. (2006)  Prospective observational  Germany  Community  PBQ  2 weeks  Depressive symptoms  
6 weeks  Global functioning  
16 weeks  
60 weeks  
Muzik et al. (2013)  Prospective observational  USA  Mothers had  PBQ  6 weeks  Maternal child abuse  
experienced  16 weeks  
physical,  24 weeks  
sexual, or  
emotional  
abuse, or no  
abuse  
Ohoka et al. (2014)  Prospective observational.  Japan  Community  MIBS  Antenatally  Depressive symptoms  
5 days  
4 weeks  
O'Higgins et al. (2013)  Prospective observational  UK  Some  MIBS  1-4 weeks  Depressive symptoms  
depressed  9 weeks  Timing of assessment  
mothers  16 weeks  
52 weeks  

META-ANALYSIS OF MOTHER-INFANT BONDING 
Bonding  Timing of  
Study  Stud~ Design  Location  Sam12le T~12e  Measure  Assessment  Predictors Assessed  
O'Mahen eta. (2014)  Randomized controlled  UK  Community  PBQ  Not specified  Intervention (internet behavioral  
trial  activation)  
Orlin et at. (20 13)  Prospective observational  Turkey  Community  PBQ and  3 days  Depressive symptoms  
MIBS  8 weeks  Subscales ofthe BSI  
Pearson et at. (20 11)  Prospective observational  UK  Community  MAS  Antenatally  Depressive symptoms  
12-24 weeks  
Perry et at. (20 11)  Prospective observational  USA  Low-income  MPAS  24-32 weeks  Intervention (perinatal depression  
Latina mothers  prevention trial)  
Quinlivan et al. (2005)  Prospective observational  Australia  Some mothers  MAS  24 weeks  Exposure to domestic violence  
had  
experienced  
domestic  
violence  
Reay et al. (2012)  Prospective observational  Australia  Mothers with  MAl  Not specified  Intervention (interpersonal group therapy)  
with  PPD  
participants  
who have  
postpartum  
depression.  
Robakis et al. (2014)  Prospective observational  USA  Community  MIBS  4 weeks  Antenatal optimism  
Sen et al. (2012)  Prospective observational  Turkey  Community  MAl  4-16 weeks  Anxiety  
Avoidance  
Seng et al. (2013)  Prospective observational  USA  Women who  PBQ  6 weeks  Childhood maltreatment  
had  PTSD symptoms  
experienced  Pre-existing major depressive disorder  
trauma and  Postpartum PTSD  
developed  
PTSD, women  
who had  
experienced  
trauma and did  
not develop  
PTSD, and  
women with  
no exposure to  
trauma.  
Shin et al. (2007)  Prospective observational  South  Community  MAl  6 weeks  Fetal attachment  
Korea  Maternal sensitivity  

META-ANALYSIS OF MOTHER-INFANT BONDING 
Bonding  Timing of  
Study -­ Study Design  Location  Sample Type  Measure  Assessment  Predictors Assessed  
Sockol et al. (20 14)  Prospective observational  USA  Mothers were  PBQ  9 weeks  Depressive symptoms  
in a partial  Depressive disorder  
hospitalization  Major depressive disorder severity,  
program  Anxiety disorder  
Substance abuse disorder  
Race/ethnicity  
Marital status  
Education  
Mode of delivery  
Tietz et al. (20 14)  Prospective observational  Germany  Mothers either  PBQ  4-32 weeks  Anxiety  
had  Depressive symptoms  
postpartum  
anxiety or no  
mental health  
disorder  
Tikotzky et al. (20 12)  Prospective observational  Israel  Community  MPAQ  12 weeks  Sleep patterns  
24 weeks  
Tsivos et al. (20 14)  Randomized controlled  UK  Community  PBQ  M=24.8, SD  Parenting intervention  
trial  = 12.8 weeks  
Turner et al. (2008)  Prospective observational  UK  Community  PBQ  M =4.23, SD  Depressive symptoms  
= 2.85 weeks  Anxiety  
Wittkowski et al. (2010)  Cross-sectional  UK  Mothers were  PBQ  M = 12.28, SD  Depressive symptoms  
observational  in Mother and  = 12.52 weeks  
Baby Unit  
Yoshida et al. (2012)  Prospective observational  UK  Community  Japanese­ 5 days  Depressive symptoms  
MIBS  4 weeks  
16 weeks  
Yuan et al. (2010)  Quasi-controlled trial  UK and  Community  PBQ  24 weeks  Intervention (oral health)  
China  52 weeks  

Note. USA= United States of America, UK= United Kingdom, MIBS =Mother Infant Bonding Scale, MAl= Mother Attachment Inventory, PAQ =Postnatal Attachment Questionnaire, MP AS = Maternal Postnatal Attachment Scale, PBQ = Postpartum Bonding Questionnaire. 

META-ANALYSIS OF MOTHER-INFANT BONDING 
Table 2 
Random Weighted Controlled Effect Sizes from Studies Assessing Infant Sex as a Predictor of Mother-Infant Bonding 
Study  n  d  95%CI  SAMD  
Figuieredo (2009) Gharaibeh and Hamlan (20 11)  277 220  0.24* 0.20  -0.00-0.47 -0.07-0.47  0.40 -0.03  
Moehler et al. (2006) Orun et al. (20 13)  101 189  -0.04 0.29*  -0.44-0.35 0.00-0.58  -1.35 0.76  
k  d  95%CI  Q(dD  1  
Total (all studies included)  4  0.20**  0.06-0.35  1.94(3)  0.00  
* p < 0.05 ** p < 0.01  
*** p < 0.001 a outlier excluded  

META-ANALYSIS OF MOTHER-INFANT BONDING 
Table 3 
Random Weighted Controlled Effect Sizes from Studies Assessing Maternal Education as a Predictor ofMother-Infant Bonding 
Study  n  d  95%CI  SAMD  
Cocklin et al. (2012)  100  -0.34  -0.75-0.08  -1.51  
Figueiredo et al. (2009)  291  0.53***  0.30-0.77  8.02 a  
Sockol et al. (20 14)  163  -0.61 **  -0.22--0.23  -4.42  
k  d  95%CI  Q(df)  

Total (all studies)  3  -0.12  -0.89-0.64  30.05(2)* * *  93.35  
Total (outlier excluded)  2  -0.48**  -0.76--0.20  0.86(1)***  0.00  
*p < 0.05  
** p < 0.01  
*** p < 0.001  
a outlier excluded  

META-ANALYSIS OF MOTHER-INFANT BONDING 
Table 4 
Random Weighted Controlled Effect Sizes from Studies Assessing Parity as a Predictor of Mother-Infant Bonding 
Study  n  r  95%CI  SAMD  
Aiello & Lancaster (2007)  71  0.39**  0.17-0.57  2.73  
Kokubu et al. (20 11)  99  0.00  -0.20-0.20  -0.23  
Sockol et al. (2014)  172  -0.02  -0.170.13  -3.26  
k  r  95%CI  Q(d/)  p  
Total (all studies included)  3  0.12  -0.13-0.35  9.78(2)**  79.55  
* p < 0.05  
**p<0.01  
*** p < 0.001  

META-ANALYSIS OF MOTHER-INFANT BONDING 
Table 5 
Random Weighted Controlled Effect Sizes from Studies Assessing Fetal Attachment as a Predictor ofMother-Infant Bonding 
Study  n  r  95%CI  SAMD  
Damato (2004)  0.38***  0.23-0.51  0.57  
Dubber et al. (2014)  0.40*  0.05-0.67  0.40  
Edhborg et al. (2011)  0.17***  0.10-0.24  -6.62a  
Shin & Kim (2007)  0.46***  0.34-0.56  2.87b  

k  r  95%CI  Q(d[)  l  
Total (all studies included)  4  0.35***  0.16-0.51  19.25(3)***  84.42  
Total (outliers excluded)  3  0.43***  0.34-0.51  0.78(2)  0.00  
Total (trim-and-fill)  0.46  0.39-0.53  
* p < 0.05  
**p<O.Ol  
*** p < 0.001  
a outlier excluded  

META-ANALYSIS OF MOTHER-INFANT BONDING 
Table 6 
Random Weighted Controlled Effect Sizes from Studies Assessing Anxiety as a Predictor of Mother-Infant Bonding 
Study  n  r  95%CI  SAMD  
Dubber et al. (2014)  77  -0.32**  -0.50-( -0.1 0)  0.03  
Edhborg et al. (20 11)  663  -0.40***  -0.47-(-0.34)  -3.95 a  
Ortin et al. (2013)  189  -0.19*  -0.32-(-0.04)  2.90a  
Tietz et al. (2014)  78  -0.35**  -0.53-( -0.14)  -0.43  
Turner et al. (2008)  64  -0.30*  -0.51-( -0.06)  0.19  

 k  r  95% CI  Q(c{{)  p  
Total (all studies included)  5  -0.32***  -0.42-( -0.22)  8.83(4)  54.71  
Total (outliers excluded)  3  -0.33***  -0.44-( -0.20)  0.14(2)  0.00  
* p < 0.05  
** p < 0.01  
*** p < 0.001  
a outlier excluded  

META-ANALYSIS OF MOTHER-INFANT BONDING 
Table 7 
Random Weighted Controlled Effect Sizes from Studies Assessing Depressive Symptoms as a 
Predictor ofMother-Infant Bonding  
Study  n  r  95% CI  SAMD  
Edhborg et al. (20 11)  671  -0.34***  -0.40-( -0.27)  2.38  
Kaneko & Honjo (2014)  1786  -0.47***  -0.51-( -0.43)  -3.03  
Kokubu et al. (2012)  99  -0.46***  -0.60-( -0.29)  -0.65  
Moehler et al. (2006)  101  -0.35***  -0.51-(-0.17)  0.80  
Ohoka et al. (2014)  389  -0.40***  -0.48-( -0.31)  0.47  
Orlin et al. (2013)  189  -0.38***  -0.49-( -0.25)  0.64  
Pearson et al. (20 1 0)  49  -0.20  -0.46-0.09  1.69  
Seng et al. (20 13)  566  -0.47***  -0.53-( -0.41)  -1.94  
Sockol et al. (2014)  177  -0.29***  -0.42-( -0.14)  2.09  
Tietz et al. (2014)  78  -0.49***  -0.64-( -0.30)  -0.99  
Turner et al. (2008)  64  -0.53***  -0.69-(-0.33)  -1.36  
Wittkowski et al. (20 1 0)  132  -0.71 ***  -0.79-(-0.62)  -6.46 a  
Yoshida et al. (20 12)  544  -0.20***  -0.28-(-0.12)  5.71 a  
k  r  95%CI  Q(dj)  p  

Total (all studies included)  13  -0.41 ***  -0.48-( -0.34)  83 .63(12)* * *  85.65  
Total (outliers excluded)  11  -0.41 ***  -0.46-( -0.36)  25.48(10)**  60.75  
* p < 0.05  
** p < 0.01  
*** p < 0.001  
a outlier excluded  

META-ANALYSIS OF MOTHER-INFANT BONDING 
Table 8 
Random Weighted Controlled Effect Sizes from Studies Assessing Interventions as a Predictor of Mother-Infant Bonding 
Study  n  d  95%CI  SAMD  
Ahnet al. (2010)  20  0.89t  -0.03-1.81  1.36  
Gtirol & Polat (2012)  117  0.43*  0.06-0.80  0.84  
Mulcahy et al. (2009)  50  0.59*  0.02-1.16  1.18  
O'Mahen (2014)  73  -0.42t  -0.89-0.05  -3.25a  
Perry (2011)  217  0.29*  0.02-0.56  -0.14  
Reay et al. (2012)  45  0.33  -0.26-0.92  0.11  
k  d  95%CI  Q(dfl  1  
Total (all studies included)  6  0.29t  -0.01-0.59  12.01(5)*  58.38  
Total (outliers excluded)  5  0.39***  0.20-0.57  2.26(4)  0.00  
Total (trim-and-fill)  0.34  0.16-0.51  
p < 0.10  
* p < 0.05  
** < 0.01  
*** < 0.001  
a outlier excluded  

META-ANALYSIS OF MOTHER-INFANT BONDING 
Table 9 
Random Weighted Uncontrolled Effect Sizes from Studies Assessing Interventions as a Predictor ofMother-Infant Bonding 
Study  n  d  95%CI  SAMD  
Ahn et al. (20 1 0)  10  1.03**  0.26-1.80  0.48  
Gtirol & Polat (2012)  57  0.20  -0.06-0.47  -2.14  
Mulcahy et al. (2009)  23  1.04***  0.53-1.55  0.87  
O'Mahen (2014)  31  0.62**  0.24-1.01  -0.36  
Reay et al. (2012)  22  0.98***  0.47-1.49  0.68  
k  d  95%CI  Q(dj)  p  
Total (all studies included)  5  0.72***  0.34-1.10  14.73(4)**  72.84  
Total (trim-and-fill)  0.64  0.31-0.97  
* p < 0.05  
** < 0.01  
*** < 0.001  

META-ANALYSIS OF MOTHER-INFANT BONDING Table 10 
Categorical Moderator Analyses for Depression 
Moderator  k  r  95%CI  Q(d/)  p  
Bonding  0.02(1)  0.89  
Measure  
MIBS  3  -0.40***  -0.46-(-0.33)  
PBQ  8  -0.41 ***  -0.47-(-0.34)  
Location  
English Speaking  5  -0.38***  -0.47-(-0.28)  
Non-English Speaking  6  -0.44***  -0.48-( -0.40)  
Sample Type  0.10(1)  0.75  
Clinical  2  -0.38**  -0.56-(-0.16)  
Community  9  -0.41 ***  -0.46-( -0.36)  
*p < 0.05  
** p < 0.01  
*** p < 0.001  

META-ANALYSIS OF MOTHER-INFANT BONDING 
Table 11 
Continuous Moderator Analyses ofTiming ofAssessment 
Outcome  k  Slope  
Anxiety  6  -0.00  
Depression  13  -0.01 *  
Fetal Attachment  4  -0.03*  
Infant Sex  4  -0.01  
Interventions (controlled)  6  o.o4t  
Interventions (uncontrolled)  5  -0.01  
Maternal Employment  3  0.02**  
Parity  3  0.02**  
* < 0.05  
** < 0.01  
*** < 0.001  
tp<O.lO  

META-ANALYSIS OF MOTHER-INFANT BONDING 


-£ 
:.E 
:~ ~ 
Records identified through database 
searching & reference review: 
(n = 46R) 

Records after duplicates removed: 
(n = 438) 

Full-text articles excluded (n = 340) 
Full-text atiicles assessed for eligibility: 
(n = 293): 
Measure (n = 122) 
Participants (n = 72) 
Manuscript Type (11 = 51) 
Study Design (n = 30) 
Language (n = IS) 
Timing (n = 3) 

"C 
~ 
"C 
(j= 
=

-

Figure 1. PRJ SMA flow chart illustrating identification of included studies. Studies included in narrative review: (n = 47) 
Studies included in 
meta-analyses: 
(n = 33) 


META-ANALYSIS 

Ftrnel Plol of Standard Error by Stddiff in means 
//1\\ jb\
J  4 0  \  
,:,  \  
I  \  
/  i  
III  ' \ .  

Infant Sex 
Funnel PloC of Standard Error by Fishe(' Z 
•I 
i\ 
i \ 
! \ 

I1~ 
J \ 
\

..., ! 
F eta! Attachment 
Ftrnel Plot of Standard Error by Std diff in m4H•na 


/;I\\ 
I I

/'r \ 

'  o  0  \ I  
\  
I  \  

OF 
! 
l ~ 
\ 
\ 
Maternal Education 
Funnel Plot of Standard Error by Fishel's Z 
p1'1\~ 
I\ 
i t 
I ' 
I \ 
I \ 
I :) \ 
I \ 
I \ 
I \ 
I I 
Anxiety 
Funnel Plot of Standard Error by std diff in meam 
!;'I\I 
I\ 
i i 
! I \ 

/I-\ 

I I 
MOTHER-INFANT 
Funnel Plot of Standard Error by Sid diff in means 
/\
j \ 
\ 
BONDING 

Funnel Plot of Standard Error by Fisher's Z 
h 
1\ II i'\ I fc \ 
!
! 0 I \ \0 
I \ 

I \ 
Pari 
Funnel Plot ot Standard Error by Fisher's Z 
J',,~l 
pf
' P• 
il l 
f!cb Iclo\ 
/ oj \ 
1c I \ 
I i d
I i \ 
I I \ 

' ' 
Depression 
Controlled Interventions Uncontrolled Interventions 

Signature Redacted 
Signature Redacted 

Signature Redacted