Content uploaded by Larissa Rossen
Author content
All content in this area was uploaded by Larissa Rossen on Aug 10, 2020
Content may be subject to copyright.
Social Development. 2020;29:3–20. wileyonlinelibrary.com/journal/sode
|
3
© 2019 John Wiley & Sons Ltd
Received: 13 November 2018
|
Revised: 4 June 2019
|
Accepted: 21 Ju ne 2019
DOI : 10.1111 /sod e.12392
REVIEW
The role of antenatal and postnatal maternal
bonding in infant development: A systematic
review and meta‐analysis
Genevieve A. Le Bas1 | George J. Youssef1,2 | Jacqui A. Macdonald1,2,3 |
Larissa Rossen4,5 | Samantha J. Teague1 | Emily J. Kothe1 |
Jennifer E. McIntosh1,2,3 | Craig A. Olsson1,2,3 |
Delyse M. Hutchinson1,2,3,5
1School of Psychol ogy, Facult y of Health,
Centre for Social a nd Earl y Emotion al
Develop ment, D eakin Un iversity Geelong,
Victoria, Australia
2Murdoch Children's Research
Instit ute, Royal C hildren's Hospit al,
Melbou rne, Victoria , Austr alia
3Depar tment of P aediatrics, Universi ty
of Melbourne, Royal Children's Hospital,
Melbou rne, Victoria , Austr alia
4School of Psychiat ry, Facult y of
Medicine, University of New South Wales,
Sydney, Victoria, Australia
5Nationa l Drug an d Alcoho l Research
Centre, University of New S outh Wales,
Sydney, New South Wales, Australia
Correspondence
Delyse M . Hutchinson, School of
Psychology, Facult y of Health, Cent re for
Social and Early Emotional Development,
Deakin University, Burwood, Victoria 3125,
Australia.
Email: delyse.hutchinson@deakin.edu.au
Abstract
The purpose of this study was to systematically review the
empirical literature on maternal bonding and associations
with infant physical, psychological, and social development.
Nineteen articles met inclusion criteria and were included
in a qualitative synthesis (79 effect sizes); 15 articles were
suitable for aggregation in a series of 14 meta‐analyses (51
effect sizes). All mean effects were in the expected direc‐
tion, with higher maternal bonding contributing to infant de‐
velopmental outcomes, including higher attachment quality
(r = .35) and parent‐reported lower colic rating (r = .22), eas‐
ier temperament (r = .19), and positive infant mood (r = .27).
Consistent with theoretical explanatory models, the review
provides suppor t for the hypothesis that maternal bonding
plays a role in fostering more optimal infant development.
The review also identifies a paucity of empirical work on this
topic and provides directions for future research.
KEYWORDS
affect, early experience, longitudinal studies, mother–child
relations, mothers
4
|
LE BAS Et AL.
1 | INTRODUCTION
The powerful affectional bond bet ween a mother and her child first develops during pregnancy, permeating ma‐
ternal behaviour, and care through the antenatal and postnatal periods (Condon, 1993; Korja et al., 2010). A
positive affectional connection between mother and infant facilitates the provision of care by the mother, and the
maintenance of an affective environment in which the infant feels safe to explore, learn, and develop age‐appro‐
priate autonomy. Theoretically, the connection is reciprocal and vital for optimal infant development (Ainswor th,
Blehar, & Waters, 2014; Winnicott, 1965). To date, no systematic meta‐analytic reviews of which we are aware
provide empirical data on these associations; yet such research has potential to guide prevention and targeted
intervention strategies to improve maternal bonding and in turn, infant development.
Maternal bonding refers to a mother's self‐reported emotional connection to her child. During pregnancy,
mothers typically develop internalized representations of their unborn child accompanied by an affectional tie to
the foetus, often referred to as antenatal maternal bonding (Condon, 1993). A positive bond during pregnancy is
associated with engaging in behaviours of af filiation and affection, such as talking to the unborn child, and posi‐
tive health practices, such as abstaining from drugs (Condon, 1993; Cranley, 1981; Van der Zalm, 1995). Postnatal
maternal bonding refers to a mother's emotional tie to her infant. Maternal perceptions of a positive postnatal
bond are indicated by a mother's felt pleasure in interacting with her infant, developing competence in her own
capacity to understand and meet her infant's needs, and acceptance and tolerance of the demands of the mater‐
nal role (Condon & Corkindale, 1998). Maternal perceptions of the postnatal affectional bond reflect pleasure in
maternal–infant interactions, which influence an infant's capacity for affective signalling (Bornstein, 1989; Grant,
McMahon, Reilly, & Austin, 2010). Higher bonding is theoretically expressed through behaviours such as maternal
sensitivity and emotional availability (Feldman, Weller, Leckman, Kuint, & Eidelman, 1999), which invite interac‐
tion and encourage healthy social‐emotional behaviours (e.g., interest, touch, smiles, play, adaptivity, approach,
and emotional regulation) and development.
Condon (1993) suggests that a mother's internalized representation of her child is increasingly elaborated
upon during pregnancy, with antenatal bonding providing structure for the subsequent development of postnatal
bonding (Condon & Corkindale, 1998). In a sample of 372 mothers, Rossen et al. (2016) found that higher antena‐
tal bonding prospectively predicted higher postnatal bonding at all pregnancy trimesters, after controlling for a
range of demographic and postnatal covariates. Although postnatal bonding appears to have its roots in antenatal
bonding, birth marks the introduction of additional complexities; a mother is required to integrate her perceptions
of motherhood and her child with new realities, including the characteristics of her infant (Condon, 1993; Solomon
& George, 1996). Given the complex ities unique to each period, it is important to examine how the mater nal bond
in both phases might shape development in infancy.
Maternal bonding is underpinned by the integration of oxy tocin and dopamine in striatum. Oxytocin and
dopamine systems support multiple motivational behaviours, including social orienting and seeking, and contact
maintaining. Connectivity among these systems during bond formation enables plasticity of the brain reward
system and reorganization of neural networks (Feldman, 2017). This process of the mother forming a selec tive
and enduring bond with her child is accompanied by mental, emotional, and behavioural changes that facilitate
physical and psychological proximity (Feldman et al., 1999; Insel & Young, 2001; Kendrick, 2000). During interac‐
tions, mothers and infants have been shown to demonstrate synchronized nonverbal behaviour (i.e., gaze, affect,
vocal, touch), heart rate, oxytocin response, cortisol response to stress, and brain oscillations in alpha and gamma
rhythms (Feldma n, 2017). “ The capacity to give and receive love and maintain long‐term bonds is increasin gly rec‐
ognized as key to human thriving, impacting well‐being, positive outlook in the face of adversity, physical health,
and bet ter ageing” (Feldman, 2017, p. 94). Individual dif ferences related to the timing and intensity of bonding re‐
main unclear, though an amalgam of psychological, interpersonal, and biological mechanisms is likely. For example,
perinatal depression and anxiet y are associated with bonding, which may be related to difficulties with emotion
|
5
LE BAS Et A L.
regulation being a common feature amongst these symptom profiles (Brockington, 200 4; Rossen et al., 2016).
Stable personalit y factors may also contribute to self‐reported bonding, with research indicating links between
emotional stability and extraversion and stronger maternal–infant bonds (de Cock et al., 2016).
The constructs of bonding and attachment are often confused or conflated. This is compounded by the fact
that many bonding assessment instruments are referred to as “attachment” measures (Condon, 1993; Condon &
Corkindale, 1998; Cranley, 1981). Indeed, researchers often refer to a mother's felt bond towards her infant as
attachment, as this relationship relates to the development of feelings of love and protection towards the infant
(Walsh, 2010). Bonding and attachment are interrelated constructs, however they are motivated by different goals
and governed by different behavioural systems. Bowlby (1969, p. 377) argues “There is a strong case… for restrict‐
ing the term attachment to the behaviour typical of the child to parent and the behavioural system responsible
for it, and to avoid using it to describe the complementary behaviour and behavioural system of the parent”. More
specifically, bonding is a function of the caregiving behavioural system, serving to protect and comfort the infant
in times of distress or threat. The attachment behavioural system is activated when an individual feels threat or
distress and seeks to elicit care and protection. At tachment represents the patterning of the infant's behaviour
and expectations of the mother's caregiving responses assessed observationally (Ainswor th, 1989; Cassidy &
Shaver, 2016). In contrast, bonding specifically refers to the mother's self‐repor ted emotional connection to her
child. This confusion in the literature regarding the definition and measurement of bonding may explain why
numerous studies have established associations between infant attachment and development (Ainsworth, 1989;
Bowlby, 1969; Cassidy & Shaver, 2016; Groh, Fearon, van IJzendoorn, Bakermans‐Kranenburg, & Roisman, 2017),
yet few studies have explored the influence that bonding may have on infant social‐emotional development.
Branjerdporn, Meredith, Strong, and Garcia (2017) recently synthesized the available research on a mother's
felt emotional connection to her infant during pregnancy and infant developmental outcomes. Whereas the au‐
thors termed this construct maternal–infant attachment, the conceptual content within this review closely ap‐
proximates our definition of antenatal bonding. Data were qualitatively synthesized according to developmental
domain: infant temperament (n = 5), ad ap tive be ha vi our (n = 2), and mil est on e at tainm en t (n = 1) . Their study fo un d
evidence that higher antenatal bonding was associated with more optimal infant development in infants aged
0–26 months, with outcomes including: normal levels of activity (i.e., not overactive; Davoudi, 2012); regular phys‐
iologic functions, such as sleep and hunger (Priel & Besser, 2000); approach to new stimuli (Davoudi, 2012; Priel
& Besser, 2000); adaptable temperament (Davoudi, 2012; Priel & Besser, 2000); less intense temperament (i.e.,
less high energy responses; Davoudi, 2012; Priel & Besser, 2000); positive overall mood (Davoudi, 2012; White,
Wilson, Elander, & Persson, 1999); attentive temperament (Davoudi, 2012); higher sensory threshold (i.e., more
stimulation required to evoke a response; Davoudi, 2012); lower colic rating (Escallier, 1995); and increased devel‐
opmental milestone attainment (Alhusen, Hayat, & Gross, 2013). Only one association was identified that linked
higher maternal bonding with a less optimal outcome—lower (total) sleep time in infancy (Speltzer, O’Beirne, &
Bishop, 2008). Taken together, the authors found evidence to suggest that higher maternal bonding was generally
associated with more optimal infant development, but concluded that results should be interpreted with caution
due to the limited number and low quality of many of the available studies.
The Branjerdporn et al. (2017) review provided a much needed synthesis of the extant research linking ante‐
natal bonding and infant development; however, three specific limitations are noted here. Firstly, the review was
confined to associations between antenatal bonding and infant development. Examination of both the antenatal
and postnatal periods remains important, given: (a) the complexities unique to each period (Condon, 1993; Rossen
et al., 2017; Solomon & George, 1996), and (b) the paucity of available research on bonding and infant develop‐
ment. Secondly, the review reported only on statistical significance and direction of effects rather than reporting
specific correlation values (i.e., no information provided on the magnitude of associations) nor any meta‐analytic
es tim ate of th e pop u la tio n eff e ct. Thir dly, th e re vie w res tri c te d the sea rch to a nar row arr ay of de vel opm ent al ou t‐
come domains, and within these, did not have a sufficient number of studies for meta‐analysis. Addressing these
limitations would allow more studies to be retrieved and for the results to be examined through meta‐analysis.
6
|
LE BAS Et AL.
1.1 | The current review
The aims of this systematic review and meta‐analysis were to: (a) identify studies that prospectively examined the
impact of antenatal or postnatal maternal bonding on infant developmental outcomes; (b) extract effect sizes for
these associations; (c) group similar infant developmental outcomes together and assess the design and qualit y of
the studies included; and, (d) conduct a meta‐analysis for each infant outcome domain (depending on study avail‐
ability), including information on the strength and direction of observed ef fects. Outcomes of maternal bonding
were restricted to the period of infancy (the first 24 months of postnatal life) given: (a) it is proximal to the meas‐
ure ment of ante natal and pos tnatal bo nding ; (b) it is a crit ic al time of development chara ct erized by ma rked physi‐
ological, physical, and psychological change; and, (c) early life events influence the child's capacity to cope with
its environment in later life (Bornstein, 2014; Fraser et al., 2012; Gluckman, Hanson, & Mitchell, 2010). Finally,
we note that although paternal bonding is also likely to play a role in infant social‐emotional development both
directly, and indirectly via the partner relationship (e.g., fathers play an import ant role in supporting the mother's
well‐being; Barker, Iles, & Ramchandani, 2017), we chose to focus on maternal bonding given the limited research
available on paternal bonding and child development.
2 | METHOD
2.1 | Data sources
The method adhered to Preferred Reporting Items for Systematic Reviews and Meta‐Analyses (PRISMA) guide‐
lines (Moher, Liberati, Tetzlaff, & Altman, 2009). The systematic search procedure included an electronic and
grey literature search, as well as forward and backward citation analysis. The following electronic databases were
searched for peer‐reviewed articles (including theses): PsycINFO (EBSCOhost platform); MEDLINE Complete
(EBSCOhost plat form); Informit; CINAHL Complete (EBSCOhost platform); and, Embase. Search terms were de‐
veloped for four concepts: maternal, bonding, infant, and longitudinal. Search terms for the four concepts in‐
cluded both free text terms and subject headings. Including subject headings allowed for increased specification
of search terms; including key free text terms allowed for records without subject headings to be captured (e.g.,
records recently added to the database). Due to discrepancies in terminology, we included a wide range of search
terms to describe bonding, including terms such as “attachment”. See Supporting Information Table S1 for search
terms and Table S2 for search syntax (for all databases). Search terms on infant developmental outcomes were
not included so as to retrieve records on all possible outcomes. No limits were applied to searches so that records
without categorization were captured. The grey literature search involved the examination of the first 100 cita‐
tions (10 pages) in a Google search. It also involved contacting authors of studies where the relevant variables
were included in an article but the associations were not reported. Finally, Web of Science was used to review
citation and reference lists of: (a) included studies; and, (b) relevant reviews on maternal bonding.
2.1.1 | Eligibility criteria
Eligible studies were identified by first screening titles and abstract s, and then reviewing full text. For both stages
of screening, G.L. screened 100% of articles and L.R. screened 10% of articles for reliabilit y analysis; some articles
also required discussion [G.L., G.Y., L.R ., J.A.M., and D.H.] to determine whether they met eligibility criteria. Studies
were included in the review if they met the following criteria: (a) available in English; (b) human sample; (c) empirical
and peer‐reviewed (i.e., conference abstracts, commentaries, editorials, literature reviews, and case studies were
excluded; theses were included); (d) included a measure of maternal bonding (self‐reported or inferred through
interview) during the antenatal or postnatal period (gestation‐24 months postpar tum); (e) included a measure of
developmental outcome during infancy (0–24 months); (f) longitudinal design, where maternal bonding was tested
|
7
LE BAS Et A L.
as the predic tor and infant development as the outcome, with a minimum of 1 month between assessment time
points; (g) not a parenting intervention study, unless findings of a control group were reported (because parent‐
ing intervention may impact bonding; Feldman, Eidelman, Sirota, & Weller, 2002; Willinger, Diendorfer‐Radner,
Willnauer, Jörgl, & Hager, 20 05); and, (h) reported (or provided on request) an unadjusted correlation bet ween
maternal bonding and later infant outcome.
2.1.2 | Measurement of maternal bonding
We [G.L., J.A.M, L.R ., D.H.] assessed (with consensus agreement) whether purported bonding measures were a
measure of the mother's felt bond to her child, or a measure of another related yet different construct (e.g., infant
or adult attachment). Consistent with Condon and Corkindale (1998) and Lamb (1974), we argue that measures of
antenata l and pos tnata l ma te rnal bo nd in g requi re a foc us on the mot he r's sel f‐r ep ort ed af fe cti ve bon d to her chil d
(self‐report measure or inferred through interview) rather than factors likely to mediate or indicate the presence
of bonding, such as maternal attitudes, beliefs, or behaviours. We reviewed 77 potential measures of maternal
bonding. Of these 77 measures, 47 were excluded because they were assessed as measuring a different construct;
the remaining 30 measures progressed to item‐level analysis. These 30 measures were compared at the item‐level
with prototypical measures of bonding to determine the suitability of the article being included (17 further meas‐
ures were excluded). At this stage, we identified 13 measures of maternal bonding. Only eight measures were
included in the current review, as some studies were excluded for reasons unrelated to the bonding measure. The
bonding measures are described in Supporting Information Table S3, including information on the psychometrics
from the original measure development studies.
2.2 | Data extraction
Using a standardized, pilot‐tested extraction sheet, data were extracted and collated from studies that met inclu‐
sion criteria (N = 19). Data were ext racte d by G.L . and indepe ndently cr oss‐referenced by G.Y. for all stu dies. Data
extracted includes: study characteristics; maternal bonding measure and time of assessment; infant outcome
measure and time of assessment; and ef fect size, p value, and sample size for the association between maternal
bonding and infant outcome. Standard effect sizes (Pearson's r) were extracted from 14 studies. For the remain‐
ing five studies, effect sizes were converted from Cohen's d or chi‐square to Pearson's r using standard formulas
(Borenstein, Hedges, Higgins, & Rothstein, 2011). Where multiple publications reported on the same sample and
finding, preference was given to studies that reported on total bonding (as opposed to only including bonding
subscales), and then studies with the larger sample size.
2.3 | Screening for meta‐analysis eligibility
With consensu s ag re em en t, we [G .L ., G.Y., J.A.M., D.H.] groupe d st ud ie s fo r me ta ‐analyses accor di ng to infant de‐
velopmental outcomes. For each infant outcome able to be grouped (association with bonding reported on in two
or more studies), a meta‐analysis was conduc ted. Some outcomes clearly measured the same construct, and were
easily grouped; these include: attachment secure/insecure classifications only, colic, and the domains of tem‐
perament (activity, adaptability, approach, distractibility, intensity, mood, persistence, rhythmicity, and sensory
threshold). We also included three additional aggregate “superordinate” meta‐analyses examining bonding and:
(a) attachment quality (secure/insecure and disorganized/not‐disorganized classifications); (b) social‐emotional
outcomes (explorator y activity, negative affect, positive affect, solicitation of attention, social‐emotional compe‐
tence, externalizing behaviours, and internalizing behaviours); and, (3) temperamental difficulty (where available,
all effects for each temperamental domain were included; otherwise, we included the effect for total tempera‐
mental difficulty). These superordinate meta‐analyses looked at the broadest categories possible and likely had
8
|
LE BAS Et AL.
TABLE 1 Summary of meta‐analytic findings for effect sizes and heterogeneity (N = 15 studies)
Predictor Outcome N effects Included studies
Combined
N of
participants r (95% CI)
p
value
for r I2 (95% CI)
Higher bonding as‐
sociated with more
optimal outcome
Antenat al and
postnatal
bondinga
Attachmentb4 Benoit , Parker, and Zeanah (1997);
Crawford and Benoit (2009); Hall
et al. (2015); Huth‐Bocks, Theran,
Levendosky, and Bogat (2011)
477 .35 (0.16, 0.52) .001** 78 (21, 99) More secure
attachment
Antenatal
bonding only
Attachment: Secure/
insecure classifica‐
tions only
2Benoit et al. (1997); Huth‐B ocks et
al. (2011)
232 .30 (−0.0 4,
0.58)
.080 85 (24,
100)
More secure
attachment
Antenatal
bonding only
Colic/non‐colic 2Escallier (1995); Pinkus (1979) 141 .22 (0.03, 0.39) .025* 15 (0,
100)
Lower colic rating
Antenat al and
postnatal
bonding
Social‐emotional out‐
comesb: Exploratory
activity, negative
affect, positive
affect, solicita‐
tion of attention,
social‐emotional
competence, exter‐
nalizing behaviours,
and internalizing
behaviours
9Huth‐Bocks et al. (2011); Mason et
al. (2011); Sierau et al. (2016)
706 .13 (−0.13,
0.37)
.163 71cMore optimal
social‐emotional
behaviours
Antenat al and
postnatal
bondinga
Temperamental
difficultyb,d: Activity,
adaptability, ap‐
proach, distract‐
ibility, intensity,
mood, persistence,
rhythmicity, sensory
threshold
36 Davoudi (2012); Della Vedova
(2014); Hammarberg (20 06); Jones
(1996); Parfitt, Ayers, Pike, Jessop,
and Ford (2014); Priel and Besser
(2000); White et al. (1999)
749 .19 (0.03, 0.34) .029* 74 cEasier overall
temperament
(Continues)
|
9
LE BAS Et A L.
Predictor Outcome N effects Included studies
Combined
N of
participants r (95% CI)
p
value
for r I2 (95% CI)
Higher bonding as‐
sociated with more
optimal outcome
Antenatal
bonding only
Temperament:
Activity only
3Davoudi (2012); Della Vedova (2014);
Hammarberg (2006)
298 .18 (−0.43,
0.67 )
.583 96 (86,
100)
Normal levels
of activity (not
overactive)
Antenatal
bonding only
Temperament:
Adaptability only
4 Davoudi (2012); Della Vedova (2014);
Hammarberg (20 06); Priel and
Besser (2000)
416 .23 (−0.0 4,
0.47 )
.092 86 (50, 99) More adaptable
(easier to modify
reactions to stimuli
in a desired way)
Predictor Outcome N effects Included studies
Combined
N of
participants r (95% CI)
p value
for r
I2 (95%
CI)
Higher bonding
associated with
more optimal
outcome
Antenatal
bonding
only
Temperament:
Approach only
4 Davoudi (2012); Della Vedova
(2014); Hammarberg (20 06); Priel
and Besser (200 0)
418 .22 (−0.04,
0.45)
.102 85 (48,
99)
More approach to
new stimuli
Antenatal
bonding
only
Temperament:
Distractibility only
3Davoudi (2012); Della Vedova
(2014); Hammarberg (20 06)
303 .17 (−0.05,
0.37)
.135 69 (0, 99) More able to
distract from
fussing
Antenatal
bonding
only
Temperament:
Intensit y only
4 Davoudi (2012); Della Vedova
(2014); Hammarberg (20 06); Priel
and Besser (200 0)
418 .09 (−0.19,
0.35)
.524 87 (57,
99)
Less intense (less
high ener gy
responses)
Antenatal
bonding
only
Temperament: Mood
only
5Davoudi (2012); Della Vedova
(2014); Hammarberg (20 06); Priel
and Besser (200 0); White et al.
(1999)
480 .27 (0.07,
0.45)
.009** 79 (36,
98)
More positive
overall mood
Antenatal
bonding
only
Temperament:
Persistence only
3Davoudi (2012); Della Vedova
(2014); Hammarberg (20 06)
303 .19 (−0.02,
0.38)
.074 66 (0, 99) More per sis‐
tent/attentive
temperament
TABLE 1 (Continued)
(Continues)
10
|
LE BAS Et AL.
Predictor Outcome N effects Included studies
Combined
N of
participants r (95% CI)
p value
for r
I2 (95%
CI)
Higher bonding
associated with
more optimal
outcome
Antenatal
bonding
only
Temperament:
Rhythmicity only
4 Davoudi (2012); Della Vedova
(2014); Hammarberg (20 06); Priel
and Besser (200 0)
418 .14 (−0.02,
0.3)
.095 63 (0, 97) More regular
physiologic
functions, such
as sleep and
hunger
Antenatal
bonding
only
Temperament:
Sensory threshold
only
3Davoudi (2012); Della Vedova
(2014); Hammarberg (20 06)
303 .14 (−0.49,
0.68)
.683 97 (88,
100)
High threshold
(more stimula‐
tion required
to evoke a
response)
Note: N = total number; r = mean effect size; CI = confidence interval; I2 = indicator of heterogeneity in percent ages.
aOnly one postnatal effect was available and included in this meta‐analysis.
bThis meta‐analysis was a superordinate meta‐analysis. See Suppor ting Information Table S4 for further information on included infant outcomes.
cCIs not available for I2 when using the Robumeta package in R.
dThis superordinate meta‐analysis included all effects relating to temperament. Where available, all effects for each temperamental domain were included; otherwise, we included
the effect for tot al temperamental difficult y.
*p < .05; **p < .01.
TABLE 1 (Continued)
|
11
LE BAS Et A L.
more heterogeneity; however, they allowed us to obtain meta‐analytic associations using a larger number of ef‐
fects. We considered including superordinate analyses for cognitive/language and physical/motor outcomes,
however, decided through consensus agreement, that the available infant outcomes were too heterogeneous.
A total of 15 studies reporting 51 suitable effect sizes were eligible for aggregation in the series of 14 meta‐
analyses, described in Table 1. Findings of outcomes unable to be grouped (association with bonding reported in
only one study for a specific domain) were included in qualitative synthesis but excluded from the meta‐analysis.
Suppor ting Information Table S4 provides detail on all studies included in quantitative and qualitative synthesis.
2.4 | Data analysis
2.4.1 | Meta‐analysis
The raw data file and code used to conduct the meta‐analysis can be found here: https ://osf.io/s5rqm/ . For meta‐
analyses that invo lved only one effe ct size per study, we used th e Met afor package v1.9.8 (Vie chtbauer, 2010) in R
software v3.3.1 (R Core Team, 2015) and included random effec ts to account for study heterogeneit y. For meta‐
analyses that involved more than one effect per study (i.e., the social‐emotional and temperamental difficulty
superordinate meta‐analyses), we used a robust variance meta‐analysis approach using the Robumeta package in
R (v3.3.2; Fisher & Tipton , 2015). This techni que accounts for mult iple de pendent effects via ro bust esti matio ns of
effect size weight s and standard er rors (Hedges, Tipton, & Johnson, 2010; Tanner‐Smith, Tipton, & Polanin, 2016).
Between‐study heterogeneity was tested using the I2 statistic, providing a percentage estimate of the amount of
to t al va ria b ili ty in ef fec t siz e esti m ate s tha t can be at tri but e d to he t ero g ene ity am o ng the tr u e effe c t s (V i ech tba u e r,
2010). We ca ution over‐interpretation, howe ver, due to the sm all number of effec t sizes identif ied per met a‐anal y‐
sis (Sterne et al. , 2011). Further, although we intended to test for bias through visu al inspection of funne l plots and
using Egge r's te st of as ymm etr y (Eg ger, Smith, Schneider, & Minde r, 1997), the small numb er of ef fec ts meant that
test power was too low to distinguish chance from real asymmetry (Sterne et al., 2011). Published guidelines were
used to interpret meta‐analytic effect sizes, where r = .1 is small, .3 moderate, and .5 large (Cohen, 1992).
2.4.2 | Risk of bias assessment
A components approach was adopted to assess risk of bias in this review, whereby the research in each article
was rated on individual criteria that reflect the methodological risk of bias (Higgins, Thompson, Deeks, & Altman,
2003). These criteria included: maternal bonding measure, type (self‐report or interview), and time of assessment;
and infant outcome measure, type (maternal‐report or administered), and time of assessment. These criteria were
recorded during the dat a extraction process and sensitivity and moderation analyses were conducted to examine
whether findings were robust to the quality of the methodological approaches of the articles included (Higgins et
al., 200 3). Moderation an alysis was onl y conduc ted when ther e were at least t wo ef fec ts per level of the mo der a‐
tor variable (i.e., two effec ts for one maternal bonding measure, and two effects for another). When a moderation
was not possible, we used sensitivit y analyses to examine whether removal of the effects from studies that em‐
ployed different methodological approaches influenced the magnitude of the meta‐analytic effect (when at least
two effects were remaining to be analysed).
3 | RESULTS
3.1 | Study Selection
The PRISMA diagram is shown in Figure 1. Of the 19,813 studies identified in the electronic and grey literature
search, 19 were included in qualitative synthesis, and 15 were included in the meta‐analyses. Notably, using the
12
|
LE BAS Et AL.
Prevalence‐Adjusted and Bias‐Adjusted Kappa statistic (PABAK; Byr t, Bishop, & Carlin, 1993), inter‐rater reliabil‐
ity was high for both title and abstract screening (PABAK = 0.94) and full text screening (PABAK = 0.91).
3.2 | Qualitative synthesis
In cl u de d stud ie s wer e pub li she d bet wee n 1979 and 2017, with the ma jor it y pub li she d fro m 200 8 onwar ds (n = 12).
Most samples were recruited from either the United States (n = 7) or Canada (n = 4). Remaining samples were
recruited from The Netherlands (n = 2), and the U.K., Israel, Sweden, Italy, Germany, and Australia (each n = 1).
Avera ge maternal age at Time 1 ranged from 21 to 34 years (M = 28.03, SD = 3.9). Sample sizes that were use d for
correlations within ea ch stu dy ranged fro m 34 to 499 (M = 145.16, SD = 115.89; sum of samples across all included
studies = 2,758). Study characteristics are further described in Supporting Information Table S4, along with study
predictors and outcomes, and findings specific to their association (for all 19 included studies).
Fourteen studies investigated the impact of antenatal bonding only on infant outcomes, three studies inves‐
tigated postnatal bonding only; and two included both antenatal and postnatal bonding. Antenatal bonding was
assessed using a range of measures, including the Maternal Antenatal Attachment Scale (MAAS; n = 5), Maternal‐
Fetal Attachment Scale (MFAS; n = 4), Prenatal Attachment Inventory (PAI; n = 3), Working Model of the Child
Interview (WMCI; n = 2), Working Model of the Child Interview—Disrupted (WMCI‐D; n = 1), and Social Distance
Scale (n = 1). These measures were all administered in the second or third trimester of pregnancy, which could be
because bonding quality is likely stronger towards the end of pregnancy (Rossen et al., 2016). Measures of postnatal
FIGURE 1 PRISMA diagram. Note . aThe 121 records here refer to those identified through later citation
analysis or Google search. bThe exclusion criteria are listed in hierarchical order—the first exclusion criterion that
a record met was considered its “reason for exclusion”
Records identified through
database searching
N= 29,125
Records identified through
citation analysis or Google search
N= 121
Records screened
N= 19,813
Full text articles assessed for
eligibility
N= 1,355
Records excluded
N= 18,458
Full text articles excludedb
N= 1,336
Duplicate (N= 7)
Non-English (N= 88)
Full text not available (N= 82)
Not peer-reviewed or empirical (N= 86)
No maternal bonding <24 months (N= 988)
No infant outcome <=24 months (N= 27)
Maternal bonding not the predictor (N= 38)
Intervention – no control finding (N= 3)
Did not report or unable to provide unadjusted
correlation between relevant variables (N= 12)
Same sample and same finding reported on in
>1 paper (N= 5) (selected paper that reported
on total bonding score, rather than bonding
subscales)
Studies included in qualitative
synthesis
N=19
Studies included in quantitative
synthesis
N=15
Records after duplicates
removed
N= 19,692 + 121a
|
13
LE BAS Et A L.
bonding included the Maternal Postnatal Attachment Scale (MPAS; n = 3), Postpartum Bonding Questionnaire
(PBQ; n = 1), and WMCI‐D (n = 1). These measures were administered from 2–12 months postpartum.
A range of infant outcomes were identified, and these were assessed from birth to 24 months postpartum.
Multiple studies investigated attachment, colic, social‐emotional outcomes, overall temperamental difficulty, and
the temperamental domains of activity, adaptability, approach, distractibility, intensity, mood, persistence, rhyth‐
micity, and sensory threshold. These findings comprised a total of 51 independent ef fect sizes that were able to
be included in meta‐analyses.
Infant outcomes with only one reported effect size (28 of 79 available effects) were not included in meta‐anal‐
ysis. These were: adverse neonatal outcomes; overall developmental milestone attainment; executive function;
behaviour during assessment; cognitive development; motor development; language development; and sleep.
Further, one total temperamental difficulty effect size was excluded due to the study Della Vedova (2014) also
reporting individual effect sizes for each temperamental domain. Of these studies, there was evidence to support
associations between higher maternal bonding and more optimal infant development across outcomes including:
neo natal develo pment (Alhusen , Gross , Hayat , Ros e, & Shar ps , 2012); chil d development (Alhu sen et al ., 2013); ex‐
ecutive func tion (de Cock et al., 2017); motor developme nt (Si era u et al., 2016); and langu age development (Sierau
et al., 2016). These effects sizes ranged from small to large, with the larger effects sizes belonging to neonatal
development (r = .52) and child development (r = .59). However, we note some studies found evidence for links
bet ween higher mate rnal bonding an d less optimal inf ant deve lopment, with outcome s including: less engaged be‐
haviour during assessment (Sierau et al., 2016), lower cognitive and language development (Sierau et al., 2016), and
le ss tota l sle ep time at on e wee k pos tpa r tu m (Sp elt ze r et al ., 20 0 8) . The se ef fe c t si ze s wer e all sma ll (r = −.1 to −.28).
There was no evidence in support of relations between maternal bonding and other variables in these independent
studies. Nevertheless, it is dif ficult to make any firm inferences on these qualitative findings, given that the major‐
ity of statistic ally significant effects belong to the one stud y (Sierau et al. , 2016). As such, meta‐analytic estimates,
which account for all available data provide a more appropriate summary measure of effects and we await future
research to provide insights into the relations between bonding and these additional independent variables.
3.3 | Meta‐analysis
Results from the series of 14 meta‐analyses are provided in Table 1. All mean effects were in the expected di‐
rection, where higher maternal bonding predicted more optimal infant developmental outcomes. The evidence
did not support the null hypothesis for four meta‐analyses: bonding (antenatal and postnatal) and attachment
(superordinate); antenatal bonding and colic; bonding (antenatal and postnatal) and temperamental difficulty (su‐
perordinate); and antenatal bonding and infant mood. Namely, higher bonding was associated with higher quality
infant attachment (superordinate), with a moderate effect size (N ef fects = 4 from four studies; combined N of
participants = 477; r = .35, 95%, CI: 0.16–0.52). Secondly, higher antenatal bonding predicted lower colic rat‐
ing, with a small effect size (N effec ts = 2 from two studies; combined N of participants = 141; r = .22, 95% CI:
0.03–0.39). Third, higher bonding was associated with easier infant temperament (superordinate), with a small
effect size (N effects = 36 from seven studies; combined N of participants = 749; r = .19, 95% CI: 0.03–0. 34). L ast,
within the specific temperament domains, higher antenatal bonding predicted more positive overall infant mood,
with a small‐moderate effect size (N effect s = 5 from 5 studies; combined N of par ticipants = 480; r = .27, 95% CI:
0.07–0.45). There was little evidence for non‐zero effect s for the remaining 10 meta‐analyses, with ef fect sizes
ranging from small‐moderate (r = .09–.30). These investigated attachment security (secure/insecure classifica‐
tions only); social‐emotional outcomes (superordinate); and the temperamental domains of activity, adaptability,
approach, distractibility, intensity, persistence, rhythmicity, and sensory threshold. Of note, confidence intervals
for heterogeneity statistics are not available for robust variance meta‐analyses, and for the standard meta‐analy‐
ses these CIs were large. As such interpret ation of heterogeneity is cautioned given the small number of effect
sizes in each meta‐analysis.
14
|
LE BAS Et AL.
3.4 | Sensitivity and moderation analyses
Few moderation or sensitivit y analyses were possible due to the low number of effects available per meta‐analy‐
sis. With respect to moderation analyses, where possible we examined the bonding measure employed, time of
exposure assessment (antenatal vs. postnatal), and time of outcome assessment (infant age). Of the analyses that
could be conducted, the meta‐analytic associations between bonding and attachment (superordinate), tempera‐
mental difficulty (superordinate), and the temperamental domains of approach, intensity, and rhythmicity, were
not moderated by ma terna l bonding measur e. However, ther e was a dif fe rence in the ass ociat ions be tw een bond‐
ing and inf an t mo od (p = .019); the magnitude was we aker in studies that incl ud ed the MA AS (k = 2, r = .07, 95% CI:
−0.05, 0.19), relative to the PAI (k = 2, r = .44, 95% CI: 0.11, 0.69). We conducted a moderation analysis examining
the timing of the maternal bonding assessment and infant social‐emotional outcomes (superordinate). There was a
difference (p = .027) in the mag nitud e of rela tion betwee n bonding and social‐emotional outcom es, with the ante‐
natal period (k = 2, r = .03, 95% CI: 0.03, 0.03) found to be lower than the postnatal period (k = 2, r = .21, 95% CI:
−0.11, 0.49). However, when examining the meta‐analytic correlation within these two periods separately, there
was only weak evidence for a non‐zero correlation between postnatal bonding and social‐emotional outcomes
(p = .076). We examined infant outcome time of assessment for all three superordinate analyses, but found no
significant relationships. No other moderation analyses were able to be performed.
With respect to sensitivit y analyses, when including only studies that used the MPAS (k = 2, r = .21, 95% CI:
−0 .11, 0. 49 ), th e met a‐a nal ytic co rre lat ion be t wee n bon d ing an d soc i al‐ e mot i ona l o utc o mes (sup ero r din a te) was no t
dif fe re nt to the magnit ud e of the or iginal me ta‐an al yt ic correlat ion (se e Tabl e 1). When inc lu ding on ly those studie s
that used the PAI, there was an increase in the meta‐analytic correlation between bonding and the temperamental
dom ains of activity (k = 2, r = .3825, 95% CI: −0.4, 0. 84), distrac ti bilit y (k = 2, r = .26, 95% CI: 0.04, 0.46), per siste nce
(k = 2, r = .31, 95% CI 0.15, 0.45), and sensory threshold (k = 2, r = .27, 95% CI −0.69, 0.88). However, we note that
the confidence intervals for these were large and overlapping with the original estimated ef fect size. Moreover, we
believe any differences were largely attributable to the sensitivity analyses comprising the study by Davoudi (2012),
which was found to have quite strong effect sizes. No other sensitivity analyses could be per formed.
4 | DISCUSSION
4.1 | Summary of evidence
This systematic review and meta‐analysis informs a significant gap in knowledge on the association of maternal
bonding with developmental outcomes in infancy. Our review identified a paucity of empirical work on this topic,
with the available studies being not only small (by sample size), but also highly heterogeneous in measurement,
making it difficult to draw firm conclusions about aetiological relations. Within this context, all mean meta‐ana‐
lysed effects linking bonding to infant development were in a positive direction, albeit effect sizes varied, ranging
from negligible to moderate in magnitude. Four meta‐analysed ef fect s were statistically significant. Specifically,
higher maternal bonding was correlated with: (a) higher quality attachment (superordinate meta‐analysis; N ef‐
fects = 4; r = .35) an d pa rent‐repor te d lower colic rating (N effect s = 2; r = .22), easier temperament (superordinate
meta‐analysis; N effects = 36; r = .19), and positive infant mood (N effects = 5; r = .27).
4.2 | Interpretation of findings
The cur rent re view re ported on unadj usted correlations, p rov iding suppor t for an asso ciation betwee n higher
maternal bonding and more optimal infant developmental outcomes. Results suggest that while infant devel‐
opment is multiply informed, a mother's perception of her felt emotional connection to her child is an impor‐
tant factor related to higher infant attachment quality and mood, and lower infant temperament dif ficult y
|
15
LE BAS Et A L.
and colic ra ti ng . Findings align wi th theoretica l perspec ti ve s, which sugges t that devel opmental ou tcomes are
influenced by shared affective states with the caregiver (Greenspan & Shanker, 2009). Infants have limited
capacity to regulate negative emotions, and as such, learn strategies for controlling and managing emotions
through interactions with caregivers (Bornstein, 1989; Grant et al., 2010). Maternal bonding may be reflected
in p le as ur e in maternal–infant int er ac tions , wh ic h may lea d to higher inf an t at tach ment qualit y and mood, an d
low er inf ant te mp er am ent diff ic ul ty and co lic ra ting. It is im po rta nt to ac kn ow le dge that t he se developmental
outcomes likely overlap; for example, maternal reports of colic may be accompanied by reports of difficult
temperament. Further, given all maternal and infant variables included in the meta‐analyses were assessed
via maternal self‐repor t (excluding attachment), result s may be influenced by maternal self‐reporting bias or
social desirability.
It is notable that effect sizes were generally small. A mother's subjective experience may not impact de‐
velopmental outcomes as strongly as the infant's real and observable experience of her caregiving behaviours,
par ticul ar ly under stress . Im po rtantl y, most includ ed stu dies exam in ed ante natal bonding ; it is possi ble that stro n‐
ger associations exist between postnatal bonding and infant development, given its direct ties to maternal–in‐
fant interactions. Findings may reflect developmental consolidation of experience dependent neural networks
(Bornstein, 2014; Schore & McIntosh, 2011), with early bonding experiences having a cascading influence on later
bonding, and later stages of development.
Despite all meta‐analytic effects being in the expected direction, there was little evidence to suggest
that antenatal bonding predicted the remaining domains of temperament (i.e., activity, adaptability, approach,
distractibility, intensity, persistence, rhythmicity, and sensory threshold). It is likely that the small number of
effects available for each meta‐analysis resulted in low statistical power (N ef fe ct s = 3–4). Nonet he less, bond‐
ing may have more of an influence on infant outcomes that are dependent on maternal–infant interactions and
affective signalling (e.g., mood and colic), compared with non‐social‐affective domains (e.g., temperamental
activity and intensity). It is possible that factors other than the mother's perceived bond may be more impor t‐
ant in influencing these outcomes.
There was limited evidence to suggest that bonding predicted the superordinate social‐emotional outcome
variable (r = .13) . Thi s res u lt may be ex pla i ne d by th e sma ll nu m ber of ava ila ble ef fec ts an d hete rog ene ity in infa nt
outcomes. Notably, one study focused on infant behaviour in free‐play with the mother (exploratory activity,
negative affec t, positive affect, and solicitation of attention; Huth‐Bocks, Theran, Levendosky, & Bogat, 2011),
another focused on externalizing and internalizing behaviours (Sierau et al., 2016), and the final study investi‐
gated social‐emotional competence (Mason, Briggs, & Silver, 2011). Further research including key measures
of social‐emotional development (such as the social‐emotional scales of the Ages and Stages Questionnaire or
Bayley Scales of Infant and Toddler Development) is recommended to enable firmer conclusions to be made
regarding it s as so ci at io n wi th bonding. Given the evi de nc e pr ov id ed for ass ocia ti on s be tween bon di ng and other
social‐affective outcomes (e.g., mood and colic), it is possible that bonding exerts a similar impact on overall
infant social‐emotional development. In particular, postnatal bonding may implicate infant social‐emotional de‐
velopment th rough its expre ssion in mater na l–in fant interaction s. On th e contr ar y, positive so cial an d em otion al
attributes may lead the mother to feel more emotionally connected to her child. In sum, further research is re‐
quired to disentangle the complex and interrelated pathways that are likely to exist between maternal bonding,
infant behaviour, and subsequent development in infancy and beyond.
4.3 | Limitations in the literature
Although the majority of findings demonstrated positive associations between maternal bonding and infant social
and emotional development, our results confirm the need for further evidence to strengthen the inferences that can
be made about the relationship between these t wo constructs. Consistent with the findings of Branjerdporn et al.
(2017), results highlighted the paucity of research on antenatal bonding and infant development (n studies = 16). It
16
|
LE BAS Et AL.
was pe rh aps more surp ri sing, however, that only fiv e st udies wer e id en tifie d in rel at io n to the post na tal bon d. To min‐
imize the risk tha t ar ticl es were not found in th e se arch, a ran ge of se ar ch ter ms were use d to des cr ibe mater na l bon d‐
ing. While this resulted in a large number of re cords (N records = 19,813), screening reduced this number markedly. It
is notable that the large majorit y of articles investigated attachment, or maternal factors likely to mediate or indicate
the presence of bonding (such as maternal attitudes or behaviours) rather than the mother–infant affectional bond.
Of those studies that did investigate maternal bonding, the majority focused on either the trajector y of bonding
acr oss the antenata l and pos tnatal pe ri ods, or links to other mater na l cha ra cter istics (r ather than to infant outco mes).
Of the studies that did investigate maternal bonding and infant outcomes (N studies = 77): 38 studies did not
include maternal bonding as a prospectively assessed predictor variable (i.e., the studies were cross‐sectional, or
treated maternal bonding as an outcome variable); three studies were parenting intervention studies that did not
report control group findings; 12 studies did not report on (or were unable to provide on request) the association be‐
tween mat er na l bonding and infant out comes, as it was not the focus of the st udy; and , fi ve studies re po rted the sam e
findings as other studies already included on our review (leaving 19 studies that met inclusion criteria). Taken together,
this highlight s a sig nific an t ga p in the ext ant li te rature. Indee d, it woul d appe ar tha t fu rth er res ea rch is nee ded on all of
the infant outcomes examined in the current review, as each outcome was investigated in no more than five studies.
The paucity of research on maternal bonding and infant development represents a significant gap in knowl‐
edge and affects the strength of inferences that can be drawn from the current meta‐analyses. Future research
should include prospective assessments with multiple waves across the antenatal and postnatal periods to en‐
hance knowledge of the relationship bet ween antenatal and postnatal bonding, and to explore whether these
constructs differentially predict infant developmental outcomes. It would also be valuable to explore the indepen‐
dent contribution of bonding to infant social‐emotional development, by including other predic tors, such as infant
attachment and maternal sensitivity. Replicating and extending the available research would allow for findings to
be pooled, and more meaningful conclusions to be drawn on the relations bet ween antenatal bonding, postnatal
bonding, and infant development.
This review further highlights confusion in the available literature on the measurement of maternal bond‐
ing. Indeed, we [G.L., J.A.M., L.R., D.H.] determined through a process of extensive review whether purpor ted
bonding measures were measures of the mother's felt bond to her developing foetus or infant, or measures of
another related yet different construct (e.g., maternal competence, attitudes, and adult attachment experiences).
In numerous cases (30 measures), we made comparisons at the item‐level with prototypical measures of bonding
to determine the suitability of an article being included. In other cases, measures were excluded because whilst
bonding (or an aspect of bonding) was measured, so too were other maternal characteristics, yet only total cor‐
related scores were provided by the authors. Several articles were also excluded because they assessed the moth‐
er's feelings about pregnancy or being a mother rather than the mother's affective bond to her child.
4.4 | Strengths and limitations of the current review
We extended the findings of Branjerdporn et al. (2017) by including studies on both the antenat al and the post‐
natal mother‐to‐infant bond, which is important given the complexities unique to each period. Further, including
post natal stu dies meant that we identifie d a notable gap in research, with only five studies identified on the post‐
natal bond. We also employed a systematic and extensive search strategy (outlined in the Method), which meant
that we included 19 relevant studies, compared to the eight studies included in the Branjerdporn et al. (2017)
review. For example, we contacted authors where relevant variables were included but not repor ted on. We also
extended the findings of Branjerdporn et al. (2017) by reporting on the magnitude of associations, and by includ‐
ing a series of meta‐analyses to quantitatively evaluate and statistic ally combine results of comparable studies.
We outline three limitations of the current review. The main limitation is that majority of the meta‐analyses
included only 2–5 effects, with confidence intervals for heterogeneity estimates being large. Related to this, it
was also difficult to make inferences about how methodological differences between studies may have influenced
|
17
LE BAS Et A L.
results bec aus e few meaningful mode rat ion and sensit ivity analyses could be conducte d due to the li mited number
of available effects. As such, we recommend a cautious approach to interpretation of findings given that the reliabil‐
ity of th e es ti mated met a‐ana ly tic ef fe ct sizes ma y be affec ted by un known study dif fe re nces (S ch roll, Mou st ga ar d,
& Gøtzsche, 2011). Finally, for both stages of screening (titles/abstracts and full text), only one author screened
100 % of ar ticle s, with a seco nd auth or screeni ng 10% of ar ticle s for reli ability analysis. This decision was a practica l
one given the large number of ar ticles retrieved due to our extensive search strategy. Although this approach is
considered standard and valid, full double screening may have increased reliability (Haddaway & Westgate, 2018).
4.5 | Conclusions and recommendations
The current review offers two key contributions to the literature on maternal bondin g and infant development. Firstly,
results highlight limitations in the extant literature and directions for future research. These include inconsistency in
the definition and measurement of bonding which has created some confusion in the literature. Further, there is a
marked paucity of research on bonding and infant social‐emotional development (particularly postnatal bonding), and
a lack of high‐quality longitudinal research. Within this context, and consistent with theoretical explanatory models,
the review provides some support for the hypothesis that antenatal and postnatal bonding are important in determin‐
ing more optimal development in a number of infant outcomes. Nota bly, this include s higher infant at tachme nt qualit y
and mood, and lower infant temperament difficulty and colic rating. There is a clear need for longitudinal studies that
include multiple antenatal and postnatal waves assessing bonding and a range of infant developmental outcomes.
Including multiple waves would enable isolation of effects (i.e., timing and dose) informing preventative intervention
and treatment approaches to bonding related difficulties and associated infant developmental sequelae, across the
antenatal and the postnatal periods.
CONFLICTS OF INTEREST
Gen evi eve A. Le Bas, Geo rge J. Youssef, Jacqui A. Macdonald, Larissa Rossen, Samant ha J. Teague, Emily J. Koth e,
Jennifer E. McIntosh, Craig A. Olsson, and Delyse M. Hutchinson declare that they have no conflict of interest.
AUTHORS CONTRIBUTION
All authors approved the final manuscript as submitted and agree to be accountable for all aspects of the work.
G. L . co nc ep tu alized and de si gn ed the st ud y, compl et ed dat a acquisiti on , co nt rib ut ed to da ta anal ys is and int er pre‐
tation, and wrote and submitted the manuscript.
G.Y. conceptualized and designed the study, supervised data acquisition, completed data analysis, contributed to
the interpretation of data, supervised the write‐up of results, and revised the manuscript.
J.A.M. supervised data acquisition, contributed to data analysis and interpretation, and revised the manuscript.
L.R. completed data acquisition, contributed to data interpretation, and revised the manuscript.
S.T. completed data acquisition, contributed to data interpretation, and revised the manuscript.
E.K. supervised data acquisition, contributed to data interpretation, and revised the methods and results sections
of the manuscript.
J.E.M. critically reviewed and revised the manuscript for important intellectual content.
C.O. critically reviewed and revised the manuscript for impor tant intellectual content.
D.H. conceptualized and designed the study, coordinated and supervised data acquisition, data analysis, and data
interpretation, and revised the manuscript.
DATA AVAIL ABI LIT Y S TATEMEN T
The raw data file and code used to conduct the meta‐analysis can be found here: https ://osf.io/s5rqm/ .
18
|
LE BAS Et AL.
ORCID
Delyse M. Hutchinson https://orcid.org/0000‐0003‐3221‐7143
REFERENCES
Ainswor th, M. (1989). Attachments beyond infanc y. American Psychologist, 44, 709. https ://doi.org/10.1037/0003‐066X.
44.4.709
Ainsworth, M., Blehar, M., & Waters, E. (2014). Patterns of attachment: A psychological study of the strange situation. New
York, NY: Psychology Pre ss.
Alhusen, J. L ., Gross, D., Hayat, M. J., Rose, L., & Sharps, P. (2012). The role of mental health on maternal‐fetal at‐
tachment in low‐income women. Journal of Obstetric, Gynecologic & Neonatal Nursing, 41, E71–E81. https ://doi.
org /10.1111/j.1552‐6909.2012. 013 85. x
Alhusen, J. L., Hayat, M. J., & Gross, D. (2013). A longitudinal study of maternal at tachment and infant developmental
outcomes. Archives of Women's Mental Hea lth, 16, 521–529. https ://doi. org /10.10 07/s00737‐013‐ 0357‐8
Barker, B., Iles, J. E., & Ramchandani, P. G. (2017). Fathers, fathering and child psychopathology. Currrent Opinion in
Psychology, 15, 87–92. https ://doi.org/10.1016/j.copsyc.2017.02.015
Benoit , D., Parker, K. C., & Zeanah, C. H. (1997). Mothers' representations of their infants assessed prenatally: Stability and
association with infants' attachment classifications. Journal of Child Psychology and Psychiatry, 38, 307–313. https ://
doi .or g/10.1111/ j.14 69‐7610.1997.tb01 5 15.x
Borenstein, M., Hedges, L. V., Higgins, J. P., & Rothstein, H. R. (2011). Introductio n to meta‐analysis. West Sussex, UK: John
Wiley & Sons.
Bornstein, M. H. (1989). Sensitive periods in development: Structural characteristics and causal interpretations.
Psychological Bulletin, 105, 179–197. https ://doi.org/10.1037//0033‐2909.105.2.179
Bornstein, M. H . (2014). Human infancy... and the rest of the lifespan. Annual Review of Psychology, 65, 121–158. https ://
doi.org /10.1146/annur ev‐psych‐120710‐100359
Bowlby, J. (1969). Attachment and loss, Vol. 1: Attachment. New York, NY: Basic Books.
Branjerdporn, G., Meredith, P., Strong, J., & Garcia, J. (2017). Associations between maternal‐foet al attachment and
infant developmental outcomes: A systematic review. Maternal and Child Health Journal, 21, 540–553. https ://doi.
org /10.1007/s10995‐016‐2138 ‐2
Brockington, I. (2004). Postpartum psychiatric disorders. The Lancet, 363, 303–310. https ://doi.o rg/10.1016/
S0140‐6736(03)15390‐1
Byrt , T., Bishop, J., & Carlin, J. B. (1993). Bias, prevalence and kappa. Journal of Clinical Epidemiology, 46, 423–429. https ://
doi.org/10.1016/0895‐4356(93)90018‐V
Cassidy, J., & Shaver, P. R. (2016). Handbook of attachment, third edition: Theory, research, and clinical applications. New
York, NY: Guliford Press.
Cohen, J. (1992). A power primer. Psychological Bulletin, 112 , 155. https ://doi.org/10.1037/h0045186
Condon, J. T. (1993). The assessment of antenatal emotional attachment: Development of a questionnaire instrument.
British Journal of Medical Psychology, 66, 167–183. https ://doi.org/10.1111/j.2044‐8341.1993.tb017 39.x
Condon, J. T., & Corkindale, C. J. (1998). The assessment of parent‐to‐infant attachment: Development of a self‐repor t
questionnaire instrument. Journal of Reprodu ctive and Infant Psychology, 16, 57–76. https ://doi.org/10.1080/02646
83980 8404558
Cranley, M. S. (1981). Development of a tool for the measurement of maternal attachment during pregnancy. Nursing
Research, 30, 281–284. https ://doi.org/10.1097/00006 199‐19810 9000‐00008
Crawford, A., & Benoit, D. (2009). Caregivers' disrupted representations of the unborn child predict later infant–caregiver
disorganized attachment and disrupted interactions. Infant Mental Health Journal: Of ficial Publication of the World
Association for Infant Mental Health, 30, 124–144. https ://doi.org/10.1002/imhj.20207
Davoudi, R . (2012). The relationship bet ween prenatal attachment and p ostnatal i nfant temperament (Unpublished doctoral
disser tation). Pacifica Graduate Institute, C arpinteria, CA.
de Cock, E. S., Henrichs, J., Klimstr a, T. A., Maas, A. J. B., Vreeswijk, C. M., Meeus, W. H., & van Bakel, H. J. (2017).
Longitudinal associations between parental bonding, parenting stress, and executive functioning in toddlerhood.
Journal of Child and Family Studies, 26, 1723–1733. https ://doi.or g/10.1007/s10826‐ 017‐0679‐7
de Cock, E. S., Henrichs, J., Vreeswijk , C. M., Maas, A. J., Rijk, C. H., & van Bakel, H. J. (2016). Continuous feelings of love?
The parent al bond from pregnancy to toddlerhood. Journal of Family Psychology, 30, 125. https ://doi.org/10.1037/
fam00 00138
Della Vedova , A. (2014). Maternal psychological state and infant's temperament at three months. Journal of Rep roductive
and Infant Psychology, 32, 520–534. https ://doi.org/10.1080/02646 838.2014.947472
|
19
LE BAS Et A L.
Egger, M., Smith, G. D., Schneider, M., & Minder, C. (1997). Bias in meta‐analysis detected by a simple, graphical test.
British Medical Journal, 315, 629–63 4. https ://doi.org/10.1136/bmj.315.7109.629
Escallier, L. A. (1995). Prenatal predictors of infant colic: Maternal‐fetal attachment, maternal state anxiety and maternal hope
(Unpublished doctoral dissertation). Adelphi University, New York, NY.
Feldman, R. (2017). The neurobiolog y of human attachments. Trends in Cognitive Sciences, 21, 80–99. https ://doi.
org/10.1016/j.tics.2016.11.007
Feldman, R ., Eidelman, A . I., Sirota, L ., & Weller, A. (2002). Comparison of skin‐to‐skin (kangaroo) and traditional care:
Parenting outcomes and preterm infant development. Pediatrics Springfield, 110, 16–26. https ://doi.org/10.1542/
peds.110.1.16
Feldman, R ., Weller, A., Leckman, J. F., Kuint, J., & Eidelman, A. I. (1999). The nature of the mother 's tie to her infant:
Maternal bonding under conditions of proximity, separation, and potential loss. The Journal of Child Psychology and
Psychiatry and Allied Disciplines, 40, 929–939. https ://doi.org/10.1111/1469‐7610.00510
Fisher, Z., & Tipton, E. (2015). Robumeta: An r‐package for robust variance estimation in meta‐analysis. Retrieved from
https ://arxiv.org/abs/1503.02220
Fraser, A., Macdonald‐Wallis, C ., Tilling, K., Boyd, A ., Golding, J., Davey Smith, G., … Lawlor, D. A. (2012). Cohort profile:
The avon longitudinal study of parents and children: Alspac mothers cohort. International Journal of Epidemiology, 42,
97–110. https ://doi.org/10.1093/ije/dys066
Gluc kman, P. D., Hanson, M. A., & Mitchell, M. D. (2010 ). Development al origins of health and dise ase: Reducing th e bur‐
den of chronic disease in the nex t generation. Genome Medicine, 2, 14. https ://doi.org/10.1186/gm135
Grant, K.‐A., McMahon, C., Reilly, N., & Austin, M.‐P. (2010). Maternal sensitivity moderates the impact of prenatal anx‐
iety disorder on infant mental development. Early Human Development, 86, 551–556. https ://doi.org/10.1016/j.earlh
umdev.2010.07.004
Greenspan, S. I., & Shanker, S. (2009). The first idea: How symbols, language, and intelligence evolved from our primate ances‐
tors to modern humans. Cambridge, MA: Da Capo Press.
Groh, A. M., Fearon, R. M. P., van IJzendoorn, M. H., Bakermans‐Kranenburg, M. J., & Roisman, G. I. (2017). Attachment
in the early life course: Meta‐analytic evidence for its role in socioemotional development. Child Development
Perspectives, 11, 70–76. https ://doi.org/10.1111/cdep.12213
Ha d d a w a y, N . R. , & Wes t g at e , M. (2 0 1 8 ) . Pr e d i ct in g th e ti me ne ed e d to co n d u c t an en v ir o n m en t al sy s te m a t i c re v i e w or sy s‐
tematic map: Analysis and decision support tool. Conservation Biology, 33, 434–443. https ://doi.org/10.1101/303073
Hall, R., Hoffenkamp, H., Tooten, A., Braeken, J., Vingerhoets, A., & Van Bakel, H. (2015). Longitudinal associations be‐
tween maternal disrupted representations, maternal interactive behavior and infant attachment: A comparison be‐
tween full‐term and preterm dyads. Child Psychiatry & Human Development, 46, 320–331. https ://doi.o rg/10.1007/
s10578‐014‐0 473‐3
Hammarberg, K . (2006). The experience of bir th and early mothering after assisted conception (Unpublished doctoral disser‐
tation). The University of Melbourne, Melbourne, Australia. Retrieved from https ://miner va‐access.unime lb.edu.au/
b i t s t r e a m / h a n d l e / 1 1 3 4 3 / 3 9 1 8 4 / 6 7 1 1 5 _ 0 0 0 0 2 8 7 0 _ 0 1 _ T h e s i s . p d f ? s e q u e n c e = 1
Hedges, L. V., Tipton, E., & Johnson, M. C. (2010). Robus t variance estimation in meta‐regression with dependent effect
size estimates. Research Synthesis Methods, 1, 39–65. https ://doi.org/10.1002/jrsm.5
Higgins , J. P., Thompson, S. G., Deeks, J. J., & Altman, D. G. (2003). Measuring inconsistency in meta‐analyses. B ritish
Medical Journal, 327, 557–560. https ://doi.org/10.1136/bmj.327.7414.557
Huth‐Bocks, A. C., Theran, S. A., Levendosky, A. A., & Bogat, G. A. (2011). A social‐contextual under standing of con‐
cordance and discordance between maternal prenatal representations of the inf ant and infant–mother attachment.
Infant Mental Health Journal, 32, 405–426. ht tps ://doi.org/10.1002/imhj.203 04
Insel, T. R ., & Young, L. J. (2001). The neurobiology of attachment. Nature Reviews Neuroscience, 2, 129. https ://doi.
org /10.10 06/h beh.2 001.1691
Jones, H. E. (1996). Predictors of a mother's feelings and perceptions about her infant: From pregnancy to postpartum
(Unpublished doctoral dissertation). C alifor nia School of Professional Psycholog y, Alhambr a, CA .
Kendrick, K. M. (2000). Oxytocin, motherhood and bonding. Experimental Physiology, 85, 111s–124s. https ://doi.
org/10.1111/j.1469‐445X.2000.tb000 14.x
Korja, R., Ahlqvist‐Björ kroth, S., Savonlahti, E., Stolt, S., Haataja, L., Lapinleimu, H., … PIPARI StudyGroup. (2010).
Relations between maternal attachment representations and the quality of mother–infant interaction in preterm
and full‐term infant s. Infant Behavior and Development, 33, 330–336. https ://doi.org/10.1016/j.infbeh.2010.03.010
Lamb, M. E. (1974). A defense of the concept of attachment. Human Development, 17, 376–385. ht tps ://doi.
org /10.1159/00 027 1359
Mason, Z., Briggs, R ., & Silver, E. J. (2011). Maternal attachment feelings mediate between maternal report s of depres‐
sion, infant social–emotional development, and parenting stress. Journal of Reprodu ctive and Infant Psychology, 29,
382–394. https ://doi.org/10.1080/02646 838.2011.629994
20
|
LE BAS Et AL.
Mo her, D., Li ber ati , A. , Tetz laf f, J., & Al tma n , D. G. (2009 ). Pr efe rre d rep orti ng it ems fo r sys tem at ic re v ie ws an d met a‐a nal ‐
yses: The prisma statement. Annals of Internal Medicine, 151, 264–269. https ://doi.org/10.1016/j.jclin epi.2009.06.005
Parfitt, Y., Ayers, S., Pike, A., Jessop, D., & Ford, E. (2014). A prospective study of the parent–baby bond in men and women
15 months after bir th. Journal of Reprod uctive and Infant Psychology, 32, 441–456. https ://doi.org/10.1080/02646
838.2014.956301
Pinkus, J. (1979). A prospective stud y of relationships between prenatal maternal parameters and early infant development
(Unpublished doctoral dissertation). University of Toronto, Toronto, C anada.
Priel, B ., & Besser, A. (20 00). Adult at tachment st yles, early relationships, antenatal at tachment, and percep‐
tions of infant temperament: A study of first‐time mothers. Personal Relationships, 7, 291–310. https ://doi.
org /10.1111/j.1475‐6811.200 0.tb 00 0 18.x
Rossen, L., Hutchinson, D., Wilson, J., Burns, L., A Olsson, C., Allsop, S., … Mattick, R. P. (2016). Predictors of postnatal
mother‐infant bonding: The role of antenatal bonding, maternal substance use and mental health . Archives of Wome n's
Mental Health, 19, 609–622. https ://doi.org/10.1007/s00737‐016‐0602‐z
Rossen , L., Hutchinson, D., Wilso n, J., Bur ns , L., Allsop, S ., Elliott , E., … Mat ti ck , R. (20 17). Materna l bo nd in g th roug h pr eg‐
nancy and postnatal: Findings from an australian longitudinal study. American Journal of Perinatology, 34, 808–817.
htt ps ://doi.org/10.1055/s‐0 037‐1599052
Schore, A., & McIntosh, J. (2011). Family law and the neuroscience of attachment , part i. Family Court Review, 49, 501–512.
https ://doi .org/10.1111/ j.174 4‐1617.2011.0 1387.x
Sc hr oll , J. B., Mo us tga ar d, R. , & Gøt zsche , P. C. (2011). Dea ling with sub sta nt ia l hete roge nei ty in coc hra ne revi ew s: Cros s‐
sectional study. BMC Medical Research Methodology, 11, 22–22. https ://doi.org/10.1186/1471‐2288‐11‐22
Sierau, S., Dähne, V., Brand, T., Kurtz, V., von Klitzing, K., & Jungmann, T. (2016). Effects of home visitation on mater‐
nal competencies, family environment, and child development: A randomized controlled trial. Prevention Science, 17,
40–51. https ://doi.org/10.1007/s11121‐015‐0573‐8
Solomon, J., & George, C . (1996). Defining the caregiving system: Toward a theory of caregiving. Infant Mental Health
Journal, 17, 183–197. https ://doi.org/10.1002/(SICI)1097‐0355(19962 3)17:3<183:AID‐IMHJ1 >3.0.CO;2‐Q
Speltzer, P., O'Beirne, M., & Bishop, A. (2008). Prenatal attachment and postnatal infant sleep. Journal of Prenatal &
Perinatal Psychology & Health, 23, 33–48. Retrieved from ht tps ://ww w.quest ia.com/libra ry/journ al/1P3‐17524
88201/ prena tal‐attac hment‐and‐postn atal‐infant‐sleep
Sterne, J. A. C., Sutton, A. J., Ioannidis, J. P. A., Terrin, N., Jones, D. R., Lau, J., … Higgins, J. P. T. (2011). Recommendations
for examining and interpreting funnel plot asymmetr y in meta‐analyses of randomised controlled trials. British Medica l
Journal, 343, d4002. https ://doi.org/10.1136/bmj.d40 02
Tanner‐Smith, E. E., Tipton, E., & Polanin, J. R. (2016). Handling complex meta‐analy tic data structures using robust
variance estimates: A tutorial in R. Journal of Developmental and Life‐Course Criminology, 2, 85–112. https ://doi.
org /10.1007/s408 65‐ 016‐ 00 26‐5
Van der Zalm, J. E. (1995). Accommodating a twin pregnancy: Maternal processes. Acta Geneticae Medicae et Gemellologiae:
Twin Research, 44, 117–133. https ://doi.org/10.1017/S0001 56600 0001781
Viech tb auer, W. (2010). Co nduc ting me ta ‐an al yses in r wit h th e metaf or packag e. Jou rnal of Statistical Software , 36, 1–48.
https ://doi.org/10.18637/ jss.v036.i03
Walsh, J. (2010). Definitions matter: If maternal–fetal relationships are not att achment, what are they? Archives of
Women's Mental Health, 13, 4 49–451. https ://doi.org/10.10 07/s00737‐010 ‐0152‐8
White, M. A ., Wilson , M. E., Elander, G., & Persson, B. (1999). The swedish family: Transition to parenthood. Scandinavian
Journal of Caring Sciences, 13 , 171–176. htt ps ://doi.o rg /10.1111/ j.1471‐6712.1999.tb 00 5 3 6.x
Willinger, U., Diendor fer‐Radner, G., Willnauer, R., Jör gl, G., & Hager, V. (2005). Parenting stress and parental bonding.
Behavioral Medicine, 31, 63–72. https ://doi.org/10.320 0/BMED.31. 2.63‐72
Winnicott, D. (1965). The maturational processes and the facilitating environment: Studies in the theory of emotional develop‐
ment. London, UK: Routledge.
SUPPORTING INFORMATION
Additional supporting information may be found online in the Supporting Information sec tion at the end of the
article.
How to cite this article: Le Bas GA , Youssef GJ, Macdonald JA, et al. The role of antenatal and postnatal
maternal bonding in infant development: A systematic review and meta‐analysis. Social Development.
2020;29:3–20. ht tp s ://doi.org /10.1111/so de.123 92
