British Journal of Medical Psychology (1997), 70, 359-372 Printed in Great Britain 359

0 1997 The British Psychological Society

The correlates of antenatal attachment in

pregnant women
John T. Condon” and Carolyn Corkindale
Department of Psychiatry, Flinders Medical Centre, Bedfmd Park, 5042 South Australia,
Australia

Maternal-foetal attachment represents the earliest and most basic form of human
intimacy, and has both theoretical and clinical significance. Utilizing a previously
published self-report questionnaire to assess antenatal attachment, the present paper
explores its correlates using 238 women in the third trimester of pregnancy. Extra-
polating from studies of maternal-infant attachment, it was hypothesized that depres-
sion and a lack of social support would be detrimental to the development of maternal
antenatal attachment. The findings confirmed these hypothesized effects. In particular,
the subgroup of women having low attachment was characterized by high levels of
depression and anxiety, low levels of social support (outside the partner relationship) and
high levels of control, domination and criticism within the partner relationship.
Antenatal attachment may be predictive for future maternal-infant attachment. The
findings suggest that negative mood states and lack of social support during pregnancy
may warrant greater attention than has previously been accorded them.

In this paper, the term ‘attachment’ is used to refer to the emotional tie or bond which
normally develops between the pregnant woman and her unborn infant. Condon (1985)
has described the time course of the development of this maternal-foetal relationship,
with particular reference to the steep increase in attachment following the first experience
of foetal movement.
More recently a theoretical model for understanding the components of maternal-
foetal attachment has been presented (Condon, 1993). This model places particular
emphasis upon several indicators of the presence and strength of maternal-foetal
attachment. These include a desire for knowledge about the foetus, pleasure in interaction
with the foetus (both in fantasy and reality) and a desire to protect the unborn baby and
meet hislher needs, even at the expense of the mother’s own.
A preliminary self-report questionnaire was developed on the basis of this model and
subjected to item analysis and reliability studies (Condon, 1993). Factor analysis of the
resultant 19-item Maternal Antenatal Attachment Scale (MAAS) suggested two under-
lying dimensions of the antenatal attachment construct. The first of these represents
quality of attachment and includes experiences of closeness, tenderness, pleasure in
interaction, distress at fantasized loss and the conceptualization of the foetus as a ‘little
person’. The second factor equates with strength or intensity ofpreoccupation with the foetus,
i.e. the extent to which the foetus occupies a central place in the woman’s emotional life.

*Requests for reprints.

360 John 7: Condon and Carolyn Cortindale
It includes the amount of time spent thinking about, talking to, dreaming about or
palpating the foetus. It includes the strength of the accompanying feelings, but not their
actual quality. In brief, these two factors could be considered to represent ‘quality of
involvement’ and ‘intensity of preoccupation’.
Maternal-foetal attachment is of potential significance from both theoretical and
clinical perspectives. It represents the development of the earliest, most basic form of
human intimacy involving an object characterized by a curious admixture of fantasy and
reality. Study of factors which facilitate or inhibit its development (and influence its
intensity) may provide important insights into the determinants of more complex
subsequent relationships such as the maternal-infant one.
From a clinical perspective, an understanding of this relationship provides a framework
for conceptualizing reactions to foetal loss in both early and late pregnancy (Condon,
1986a; Frost & Condon, 1996). In addition, the need to protect and safeguard an object
may be strongly influenced by the nature and intensity of the attachment to it. A ‘much-
loved’ foetus should be at far less risk of neglect or abuse, the mother being less likely to
indulge in behaviours, including alcohol and nicotine use, which potentially endanger
it (Condon, 19686, 1987; Condon & Hilton, 1988). Finally, it seems not improbable that
maternal-foetal attachment may be a predictor of future maternal-infant attachment
(Condon & Dunn, 1988), an issue which we will be able to address using our postnatal
data. Thus, interventions which improve the maternal-foetal relationship may also have
long-term beneficial consequences.
Previous research has suggested that approximately 10-1 5 per cent of women develop
minimal attachment to their foetus during pregnancy (Condon, 1985). The findings
presented in the present paper represent the first stage of a longitudinal study of the
determinants of maternal-foetal and maternal-infant attachment during pregnancy
and the first postnatal year. The testing of two hypotheses, derived from clinical
experience, was a central objective of this investigation. First, attachment will be hilitated
by social support both within and outside the partner relationship. Second, one of the
accompaniments of depression (even in its milder forms) will be emotional detachment, i.e.
depression has the capacity to specifically impair the development of attachment.
It was recognized from the outset that disentangling causal chains involving three
variable types (viz. attachment, depression and social support) would be a formidable
task. Throughout the present paper, the term ‘correlate’ is used rather than ‘determinant’
or ‘predictor’. The inference implied by both the latter terms is inappropriate if based
only upon cross-sectional data.
The two main independent variables in this study are depression and social support. A
large body of empirical evidence attests to the importance of these in influencing
maternal-infant attachment. In the case of maternal-foetal attachment, one could
postulate that depression may impair the woman’s ability to experience ‘pleasure in
proximity’ to the foetus, as well as other positive affective experiences which help
counteract the stresses of pregnancy. The depressed pregnant woman may experience the
foetus as a source of irritation or guilt, and feel overwhelmed, or even invaded, by the
foetal presence (Raphael-Leff, 1985, 1986).
Social support and depression interact in potentially complex bidirectional ways. In
addition, a plethora of other variables, including early developmental experiences,
impinge on both (Raphael-Leff, 1985, 1986). The pregnant woman may experience
 Antenatal attachment in pregnant women 361
increased need for social support from her own mother through identification, from her
partner to share and validate her experiences and from more experienced women for
reassurance. In contrast, as described by Raphael-Leff (1985, 1986), some pregnant
women’s adaptation may involve a need to experience their pregnancy ‘exclusively’ and
not share with others, such as their partner.
Both social support and depression are complex constructs and both present sub-
stantial difficulties in measurement in terms of reliability and validity. In the case of
depression, a number of authors (Boyle, 1985; Snaith, 1991, 1993) have emphasized that
the available evidence supports a heterogeneous notion of depression in contrast to the
concept of depression as a homogeneous construct. More specifically, instruments
assessing depression differ in the emphasis they accord to physiological (as opposed to
psychological) changes and alterations in day-to-day social/occupational coping. The
time period of inquiry differs between instruments. Some instruments include items
assessing anxiety and irritability and some use bipolar response formats (e.g. positive and
negative moods) as opposed to unipolar (neutral and negative moods). Because of this
diversity, and the ways in which pregnancy may inflate depression scores (e.g. as a result
of the normal physiological changes and alteration in usual day-to-day activities), the
decision was taken to use a number of different self-report measures of depression as
described below. A comparison of the results from these different measures (during
pregnancy and the postnatal period) will be the subject of a separate paper.
Social support is also a complex construct with different measuring instruments
emphasizing different facets. It was considered important to assess social support both
within and outside the partner relationship. The actgal level of social support may
increase (or decrease) as a result of pregnancy. The perceived level may decrease if the
woman’s need for support increases. As in the case of depression, the decision was taken to
utilize multiple measures of this construct as described below.

Method
Pregnant women in the third trimester of pregnancy were approached randomly in a general hospital
obstetric clinic and invited to participate in a longitudinal study of postnatal mental health. Of 280 women
approached, 260 agreed to do so and 238 completed batteries of questionnaires were received from women in
the third trimester. Women who initially failed to return questionnaires were given a single reminder phone
call.

Assessments
The women completed a battery of self-report questionnaires at home and returned these by prepaid mail.
The questionnaires included:
A series of questions concerning socio-dwnographic characteristic and obstetric history.
The Zung Self-rating Depression Scale (ZUNG; Zung, 1965). This is a widely used and extensively validated
20-item instrument. It contains some somatic items but has been frequently used in studies of postnatal
depression in North America.
The Hospital Anxiety Depression Scale (HADS; Zzgmond & Snaith, 1983). This 14-item scale contains seven
items assessing depressed mood (HADS-D), and a further seven assessing anxiety (HADS-A). It excludes
somatic changes since it was designed for use with physically ill populations. Thus, it is also quite
appropriate for the postnatal period.
The Edinburgh Postnatal Depression Scale (EPDS; Cox, Holden & Sagousky, 1987). This 10-item scale was
specifically designed as a screening instrument for postnatal depression. In particular, it avoids misinterpreting the
362 John 1: Condon and Carolyn Cortinhle
normal postnatal physiological changes (and sleep disturbance) as indicators of depression. There are no items
which would preclude its use during pregnancy.
The Profile of Mood States (POMS; McNair & Droppleman, 1981). This 65-item self-report instrument
comprises six subscales each assessing the extent to which a subject has experienced a particular feeling over
the past week. The subscales are depression/dejection; tensiodanxiety; anger/hostility; confusion/bewilder-
ment; fatigue and vigour. The vigour subscale has positively worded items (e.g. energetic, lively, vigorous)
and the fatigue subscale has negatively worded ones (e.g. worn-out, weary, exhausted). The authors of the
instrument found these to be essentially independent factors albeit with some negative correlation.
The Maternal Antenatal Attachment Scale (MAAS). As described above.
The Dyadic Adjustment Scale (DAS; Spanier, 1976). This 32-item self-report scale comprises four subscales
for assessing the quality of the partner relationship, viz. satisfaction subscale; consensus subscale (assessing
degree of agreement over day-to-day issues, decision, future goals, etc.); affectional expression subscale
(including sexual intimacy) and cohesion subscale (assessing capacity to share interests and activities).
The Intimate Bond Measure (IBM; Wilhelm G Parkeu, 1988). This 24-item self-report scale has two subscales
assessing the nature of the partner relationship. The care subscale deals with aspects such as affection,
consideration, understanding. The control subscale assessed the extent to which the subject feels dominated,
intruded upon, criticized and controlled by the partner.
The Social Support Questionnaire (SSQ; Sarason Levine, Basham C Sarason, 1983). Each item in their
questionnaire requires the participant to identify (by initials and relationship) up to nine people who could
provide support in a specific situation. The participants then rate their overall satisfaction with that level of
support. The results are computed as three subscale scores, viz. number (the number of supportive persons
available), satisfaction (with that level of support) and family support (the number of supportive first degree
relatives). Completion of the original 27-item version is very time consuming and the shortened 12-item
form was used.
Life Events Scale (LES; Barnett, Hanna G Parker, 1983). Barnett et al. have identified a set of life-events of
particular relevance to pregnant women. A subset of these was used in the LES questionnaire with provision
for women to list any other distressing events which may have occurred during pregnancy.
Parental Bonding Instrumental (PBI; Parker, Tupling G Brown, 1979).This scale contains 25 items relating to
each of the participant’s parents. Each parent is rated on two dimensions, viz. care and overprotection.

Data analysis
The approach to analysing the data has been influenced by an awareness that the main group of independent
variables (depression and social support scores) are highly intercorrelated. As emphasized by Norusis (1985),
multivariate statistical techniques can be unreliable under these circumstances. We have used two approaches
to analysis (viz: univariate and multivariate) and applied each of these to address two separate (but related)
questions, viz:

1 . What are the correlates of antenatal attachment for the sample considered as a whole? To address
this question we have used univariate correlation and multiple linear regression analysis.
2. What differentiates the woman falling in the lower quartile (25 per cent) on antenatal attachment
from the remainder, the lower quartile being the subgroup of potential clinical relevance. To
address this question we have used univariate (independent) t tests and discriminant function
analysis.

It was our hope that there would be sufficient agreement between these four approaches to enable us to have a
reasonable degree of confidence in any conclusions regarding the correlates of antenatal attachment.

Results

Sample characteristics
The sociodemogtaphic characteristic of the sample (N = 238) are summarized in Table 1.
At the time of completing the questionnaire the mean gestation was 33.7 weeks (SD = 7.0).
 Antenatal attachment in pregnant women 363
Table 1. Sample characteristics
Mean age (SD) 27.4 (4.7)
Socio-economic status (%)
Professional 24
Skilled 30
Semi-skilled 23
Unskilled 23
Mean years in partner relationship (SD) 4.2 (3.5)
% primigravidas 24
Number of previous children (%)
0 33
1 41
2 21
3 or more 5

Frequency distribution of attachment scores

Figure 1 shows the frequency distribution of the 19-item global attachment score. One
woman scored the lowest possible score of 19, however the remainder scored in the range
45 to 91 with a distribution approximating a normal distribution with some skewing
towards the high attachment end (M = 75.5; SD = 9.0). The 10-item quality of

19:OO ' 50:OO ' 5 8 1 0 0 . 6 4 1 0 0 . 68100 72:OO 76100 . 8O:OO . 8 4 : O O . 88:OO '
47.00 56.00 62.00 66.00 70.00 74.00 78.00 82.00 86.00 90.00

GLOBAL ATTACHMENT SCORE

Figure 1. Frequency distribution of global scores.

3 64 John 1: Condon and Carolyn Cortindale
30

C
a,
E
b
20

W
0

-z”
$I

w
3 10
01
W
6

EDINBURGH DEPRESSION SCORE

Figure 2. Frequency distribution for EPDS scores.

involvement subscale is also positively skewed (M = 45.2; SD = 3.8). The eight-item1

intensity of preoccupation subscale is normally distributed with no skewness (M = 26.4;
SD = 4.9).
It is important to note that, although the items in the two subscales were originally
derived from factor analysis, when subscale scores are computed by simple summation of
items the two subscales are correlated with a Pearson correlation of .54 ( p < .001). Hence,
the global attachment score is a meaningful measure of overall attachment.

Frequency distributions of ina!ependent variables

Space precludes detailed presentation of these distributions. The means and SDs of all
variables are shown in Table 4. Because of the special focus upon depression in this study
the EPDS and Zung distributions are shown in Figs 2 and 3 respectively. The spread of
scores is noteworthy with 10 per cent of the women scoring 13 or higher on the EPDS, a
level usually considered suggestive of clinically significant depression.

Correlation coefFcients
The correlation coefficients between maternal attachment scores and the main independent
variables are presented in Table 2. Pearson correlation coefficients are used where both
‘One item (item 7) of the questionnaire did not load sufficiently strongly on either subscale to warrant inclusion in the
subscale scores.
 Antenatal attachment in pregnant women 365
16-

14.

-5 12’

E
=
w
10
0

z ” 8
Y

*
8 6
3
01
w
!z
E 4

Figure 3. Frequency distribution of Zung scores.

variables are continuous and Spearman rank coefficients for discrete variables. In tables
such as this, containing many univariate comparisons, a significance level of .05 is
inappropriate (since 1 in 20 coefficients will be significant by chance). Consequently only
coefficients exceeding .01 significance have been designated as significant.
These data suggest that different types of variables are associated with the two
attachment subscale scores. For example, anxiety and depression appear to have a far
stronger correlation with quality of involvement than on intensity of preoccupation. The
latter is, as would be expected, more highly correlated with number of children and
number of recent life-events. The level of social support, both within and outside the
partner relationship, is correlated with both attachment subscales as well as global scores.
It is important to note those variables which are not significantly correlated with
attachment scores. These include age, socio-economic status, obstetric variables
(including miscarriages or termination of pregnancy) and duration of the partner
relationship.
In summary, this preliminary analysis suggests that a women’s psychological state (as
measured by the depression, anxiety and other POMS scores) powerfully impacts upon
the quality of her attachment experience. Her intensity of preoccupation with the baby
appears to be much more influenced by ‘external’forces such as other children, life-events,
vigour, etc., i.e. by how ‘busy’the woman is. Social support is important in terms of both
subscales, as is the degree to which the pregnancy was planned.
The scores on many of these independent variables are, as expected, highly intercorrelated.
366 John 1: Condon and Carolyn Cortindale
Table 2. Correlates of antenatal attachment using univariate analyses
Correlation coefficients (N= 236)
Attachment quality Attachment intensity Global score
Number of children ns. -.27** -.18*
Degree of planning of pregnancy -.23** -.18* -.24**
Depression scores
EPDS -.38** ns. -.22*
ZUNG -.27** n.s. n.s.
HAD-D --.42** -.29** -.33**
POMS-D -.36** n.s. -.27**
Social support (SSQ)
Number .28** n.s. .22*
Satisfaction .29** .25** .27**
Family .26** ns. .22*
Partner relationship
IBM (care) .lS* .20* .22*
IBM (control) -.22* n.s. -.25**
DAS (satisfaction) n.s. n.s. .17*
DAS (consensus) .17* n.s. n.s.
DAS (affection) n.s. n.s. ns.
DAS (cohesion) .18* .22* .22*
DAS (total) .20* .18* .22*
Total life-events score n.s. --.18* n.s.
Anxiety scores
HAD-A -.41** n.s. -.25**
POMS-A -.22* n.s. n.s.
Other POMS scores
Anger -.29** n.s. -.24**
Vigour .34** .21* .25**
Fatigue n.s. ns. ns.
Confusion -.26** n.s. n.s.
Total -.32** n.s. -.23*
Parental bonding instrument
Maternal care n.s. ns. ns.
Paternal care n.s. .21* .23*
Maternal protection n.s. n.s. ns.
Paternal protection n.s. n.s. ns.
*p < .01; **p < ,001, two-tailed

Thus, the consistent pattern of influence between the different instruments is not
unexpected, but nevertheless is reassuring.

Regression analysis
The results of stepwise linear regression analysis, using global attachment score as the
dependent variable and the scores on the other instruments as independent variables, are
summarized in Table 3. Seven variables emerged in the final equation which accounted
for 38 per cent of the variance in global attachment scores.
 Antenatal attachment in pregnant women 367
Table 3. Regression analysis for global attachment scores
Variable Beta weight t P
Number of children -.16 -2.1 .04
IBM (Control score) -.21 -2.6 .01
Social support (satisfaction scale) .22 2.6 .01
Depression
Poms-D -.52 -3.3 .001
Had-D -.45 -3.9 .001
POMS (other scales)
Vigour .2 1 1.6 .09
Total -35 -4.2 c.001
R2 = 0.38
F = 10.3
Signif. of F = <.001

The results from regression analysis can be unreliable when the independent variables are
strongly intercorrelated (Norusis, 1985). Hence attachment subscale scores were not examined.
The results are consistent with the univariate correlation analysis. They underline the
importance of depression and other negative mood states (i.e. POMS total score) upon
attachment, as well as social support both within and outside the partner relationship.
They also suggest that antenatal attachment tends to be higher in first pregnancy.
Nourisis (1985) has cautioned regarding overinterpreting the magnitude of the beta
weights as always reliably indicating the relative strength of the effect of a particular
variable. Nevertheless, these results, taken in conjunction with the correlational analysis,
would suggest that negative mood states have the most powerful detrimental effect,
followed by lack of social support (partner and external being approximately equal in
importance) followed by other variables.

Characteristics of the low attachment quartile

The analyses presented above reflect relationships between variables for the entire sample of
pregnant women. Of potential clinical interest is the subgroup of women having low
attachment scores. Consequently we have compared the 25 per cent (quartile) of women
having the lowest attachment scores with the remainder using independent t tests. The
results are presented in Table 4 .
Once again, because of the multiple comparisons, two-tailed significance level has been
reduced to .01 instead of .05.
As expected from the previous analyses, negative mood states emerge as powerful
predictors of membership of the low attachment subgroup. It is noteworthy that neither
the EPDS nor ZUNG scales differentiated this group. This is consistent with the
regression analysis which selected the HAD-D and POMS-D depression scores as most
strongly associated with low attachment.
A lack of external social support is associated with membership of this group, as are
low IBM care and high IBM control. None of the DAS scores were contributory to
defining the low attachment subgroup.
368 John T. Condon and Carolyn Cortindule
Table 4. Variables with significant differences between lowest quartile on global attachment and
the remainder
Mean (SD) lowest Mean (SD) Two-tail
Variable quartile remainder t significance
Number of children 1.3 (1.1) 0.9 (0.9) 3.0 .003
Degree of planning of pregnancy 2.8 (1.5) 2.2 (1.3) 2.8 .005
Depression scores
EPDS 8.5 (5.6) 7.0 (4.4) 2.1 ns.
ZUNG 49.5 (10.8) 47.2 (9.5) 0.1 n.s.
HAD-D 6.2 (3.1) 4.1 (2.8) 4.4 <.001
POMS-D 10.7 (9.4) 7.2 (7.3) 2.9 .004
Social support
Number 83.4 (50.5) 114.4 (57.0) -3.5 <.001
Satisfaction 132.8 (28.3) 148.7 (18.5) -4.7 <.001
Family 61.4 (43.3) 87.3 (45.3) -3.6 <.001
Partner relationship
IBM (care) 26.9 (6.6) 29.9 (6.4)-3.1 ,002
IBM (control) 7.4 (6.9) 5.0 (4.4) 3.1 .003
DAS (satisfaction) 40.5 (4.8) 41.8 (4.7) -1.7 ns.
DAS (consensus) 49.9 (5.3) 51.7 (6.7) -1.8 n.s.
DAS (affection) 9.1 (1.9) 9.4 (2.1) -0.7 n.s.
DAS (cohesion) 14.8 (4.0) 162 (4.0) -2.2 n.s.
DAS (total) 115.3 (12.2) 119.7 (13.7) -2.0 ns.
Anxiety scores
HAD-A 6.5 (3.4) 5.0 (3.3) 2.8 ,006
POMS-A 10.1 (6.2) 9.0 (5.4) 1.3 n.s.
Total life-events score 1.9 (1.7) 1.5 (1.4) 1.7 ns.
Other POMS scores
Anger 9.7 (6.9) 7.3 (6.5) 2.5 n.s.
Vigour 10.4 (5.6) 13.3 (5.6) -3.5 ,001
Fatigue 14.4 (7.0) 12.8 (6.0) 1.7 ns.
Confusion 8.1 (4.7) 7.2 (3.9) 1.5 ns.
Total 41.8 (32.1) 29.5 (26.9) 2.8 .006
Parental bonding instrument
Maternal care 25.0 (7.4) 27.1 (8.4) 1.6 n.s.
Paternal care 20.8 (9.5) 25.2 (8.2) -3.0 ,003
Maternal protection 12.3 (7.1) 12.2 (7.0) .1 n.s.
Paternal protection 13.3 (8.4) 11.6 (6.9) 1.3 ns.

Number of children and degree of pregnancy planning continue to emerge as

influential. Unexpectedly, low PBI paternal care appears t o characterize this group.
Life-events score and the other variables listed as having insignificant effects in the
previous analysis continue to exert no influence.
In summary, membership of the low attachment subgroup is associated with negative
mood states (depression, anxiety and lack of vigour), weak external supports and low care/
high control in the partner relationship. Larger number of children and unplanned
pregnancy seem also to be risk factors.
 Antenatal attachment in pregnant women 369
Table 5. Discrimination function analysis (lowest quartile versus the remainder)
Predictor variable Quality subscale Intensity subscale Global score
Number of children -0.24
Degree of planning of pregnancy 0.30
Depression scores
POMS-D -0.33
HAD-D -0.29 -0.73 -0.76
EPDS -0.46
Anxiety scores
HAD-A -0.84
POMS-A -0.57 -0.67
Social support
Satisfaction 0.38 0.60 0.61
Partner relationship
IBM (control) -0.29 -0.22 -0.40
% of women correctly classified 71.6 73.9 78.0
Significance of discriminant function p < .001 p < .001 p < .001

Discriminantfunction analysis
Discriminant function analysis was used to hrther explore the characteristic of the low
attachment quartile vis-2-vis the remainder. The caveats mentioned in the regression
analysis regarding intercorrelation of independent variables also apply here.
Table 5 shows the standardized discriminant function coefficients for the two subscales
and also for the global scores. Stepwise entry of variables was used. The magnitude of
these coefficients provide an indication of the discriminative power of the variable in
demarcating the lowest quartile from the remainder. The importance of the previously
identified constellation of depression/anxiety scores, perceived external social support and
control in the partner relationship is once again apparent. In addition, number of children
enters the equation for the intensity subscale, while degree of planning of pregnancy
enters that for quality of attachment. All three discriminant functions were capable of
correctly classifying over 70 per cent of the women.
In essence, the above-mentioned constellation represents the most parsimonious subset
of those identified in the t test analysis for characterizing the low attachment subgroup.
The parsimony of this solution presumably reflects the high correlation of the variables in
this constellation with the remaining variables in the mood and social support domains.

Discussion
The sample size in this study is sufficiently large and representative to permit cautious
generalization of the findings. The internal consistency (and split-half reliability) of the
antenatal attachment instrument (MAAS) have been previously published and are
acceptably high (Condon, 1993). The spread of MAAS scores suggests that social
desirability influences do not prevent a minority of women acknowledging negative or
indifferent feelings towards their unborn infant. We would speculate that, since parent-to-
foetal attachment has received less public attention then parent-to-infant attachment or
370 John T. Condon and Carolyn Cortindule
‘bonding’, expectant parents do not have the same expectations regarding it, and its
absence may not be stigmatized in the same way.
The suitability of the data for analysis by the multivariate techniques was checked by
the methods suggested by Norusis (1985), including plots of residual values. The
assumptions underlying the methods were verified.
The results provide evidence in support of both the previously stated hypotheses. First,
the detrimental impact of depression upon quality of antenatal attachment and global
attachment scores emerges consistently. Clinically, depressed patients frequently describe
a sense of detachment from those towards whom they normally feel affection. If
depression remits, feelings of attachment usually return. Thus, the impact of depression
upon antenatal attachment is not unexpected.
Second, a number of facets of social support appear to be important to antenatal
attachment. Satisfaction with external social supports appears more influential than the
number of such supports. In the partner relationship, high levels of care consistently
correlate positively with attachment and high levels of control correlate negatively.
In interpreting these findings it is important to emphasize the cross-sectional nature of
the data, and the complex, possibly bidirectional, relationships which may exist between
some of the three main variable types, viz: depression, social support and attachment. For
example, social support and depression may exert direct influences on attachment.
Depression may influence both perceived and actual support, and actual social support may
influence depression. Moreover, there may be many other factors such as personality and
social circumstances which impact upon all three domains.
The explanation of why some measures of the independent variable constructs emerged
as significant, while others did not, warrants consideration. Multivariate analyses will
always choose the most parsimonious subset of variables, rejecting variables correlated
with those already in the equations. However it is clear from the univariate analyses
(Tables 2 and 4), that in the case of depression, the four measures differed, the EPDS score
seemingly having less impact than the HAD-D score.
As previously mentioned, this probably reflects differences in item content. For
example, the EPDS includes items targeting anxiety, self-blame, impaired coping and
self-harm, none of which are covered in the HAD-D. The latter is almost exclusively
focused on quality of mood, especially the capacity for enjoyment. Given the centrality of
the ‘pleasure in proximity’ notion in most theories of attachment, the HAD-D would be
expected to have a stronger correlation.
Literature on the correlates of maternal-foetal attachment is sparse. This literature has
been reviewed elsewhere (Condon, 1993; Muller, 1992), and the findings have often been
quite inconsistent. As suggested by several researchers (Mercer, Ferketich, May, De
Joseph & Sollid, 1988; Muller & Ferketich, 1993; Muller, 1992) this most likely reflects
methodological difficulties in previous approaches to measurement of the attachment
construct. Some recent studies have sought to explore antenatal attachment in particular
subgroups such as pregnant adolescents (Koniak-Griffin, Lominska & Brecht, 1993), or
women who became pregnant following in vitro fertilization or undergoing amniocentesis
(e.g. McMahon, Ungerer, Beaurepaire, Tennant & Saunders (1995) who utilized the
MAAS). Zachariah (1994, using an older antenatal attachment scale (Cranley, 19Sl),
found a relationship between antenatal attachment and gestation, but not the quality of
the partner relationship. She also questioned the reliability and validity of the attachment
 Antenatal attachment in pregnant women 37 1
measure. We are not aware of any previous studies which have investigated the
relationship between antenatal attachment and depression.
As pointed out in Muller’s (1992) review, in most studies, demographic variables
appear to have little influence in antenatal attachment, and most studies have found a
positive correlation with satisfaction in the partner relationship.
We are aware of a number of studies which are currently in progress using the MAAS.
However none of these are complete at the time of writing.
The findings of the present study provide some evidence in support of the construct
validity of the MAAS and its subscales. Thus, the variables which correlate with
attachment scores are not counter-intuitive, and are those which are known to influence
maternal-infant attachment in similar ways. As highlighted by Muller’s (1992) review,
previous antenatal attachment measures have often produced counter-intuitive findings.
There is increasing recent evidence that depression is relatively common during
pregnancy, and possibly more common than in the postnatal period (e.g. Demyttenaere,
Lenaerts, Nijs & Vas Assch, 1995; Gotlib, Whiffen, Mount, Milne & Cordy, 1989;
O’Hara, Zekoski, Philipps & Wright, 1990). The scores presented in Figs 2 and 3 are
consistent with this notion. The potential negative impact of such depression upon
maternal-foetal attachment suggests that it may warrant greater clinical attention than it
has previously received. Depression and lack of social support, within and outside the
partner relationship, are intimately related. Longitudinal studies (during pregnancy) are
required if cause and effect relationships are to be clarified between antenatal attachment,
depression and social support.
The question of what range of scores constitutes a ‘healthy’ attachment to the foetus
cannot be answered from the present data. There is, as yet, no ‘gold standard’ against
which these self-report questionnaire scores can be compared to define a cut-off score for
‘unhealthy’ attachment. At present, abnormally low attachment can only be defined in a
statistical sense in terms of the extent to which a particular woman deviates below the
mean for a population of normal women.
Unlike maternal-infant interactions, observations of maternal-foetal behaviours have
very limited potential for providing a ‘gold standard’. A maternal-antenatal attachment
interview is under development. This may be useful for validating the self-report
questionnaire and also providing an indication of what level of scores may be indicative
of maternal-foetal psychopathology.

References
Barnett, B. E. W., Hanna, B. & Parker, G. (1983). Life events scales for obstetric groups. Journal of
Psychosomatic Research, 27, 313-320.
Boyle, G. J. (1985). Self-report measures of depression: Some psychometric considerations. BritishJournal of
Clinical Psychology, 24, 45-59.
Condon, J. T. (1985). The parental-foetal relationship-A comparison of male and female expectant parents.
Journal of Psychosomatic Obstetrics and Gynaecology, 4, 27 1-284.
Condon, J. T. (1986a). Management of established pathological grief reaction after stillbirth. American
Journal of Psychiatty, 143, 987-992.
Condon, J. T. (19866). The spectrum of fetal abuse in pregnant women.Journal of Nervoas and Mental Disease,
174,509-516.
Condon, J. T. (1987). ‘The battered fetus syndrome’. Preliminary data on the incidence of the urge to
physically abuse the unborn child. Jozlvnal of Nervous and Mental Disease, 175, 722-725.
372 John T Condon and Carolyn Cortindale
Condon, J. T. (1993). The assessment of antenatal emotional attachment: Development of a questionnaire
instrument. British Journal of Medical Psychology, 66, 167-183.
Condon, J. T. & Dunn, D. J. (1988). Nature and determinants of parent-to-infant attachment in the early
postnatal period. Journal of the American Academy of Child Adolescent Psychiatry, 27, 293-299.
Condon, J. T. & Hilton, C. A. (1988). A comparison of smoking and drinking behaviours in pregnant
women: Who abstains and why. Medical Journal of Australia, 148, 381-385.
Cox, J. L., Holden, J. M. & Sagovsky, R. (1987). Detection of postnatal depression; development of the 10-
item Edinburgh Postnatal Depression Scale. British Journal of Psychiatry, 150, 782-786.
Cranley, M. S. (1981). Development of a tool for the measurement of maternal attachment during pregnancy.
Nursing Research, 30, 281-284.
Demyttenaere, K., Lenaerts, H., Nijs, P. & Vas Assche, F. A. (1995). Individual coping style and
psychological attitudes during pregnancy predict depression levels during pregnancy and during
postpartum. Acta Psychiatrica Scandinavica, 91, 95- 102.
Frost, M. & Condon, J. T. (1996). The psychological sequelae of miscarriage: A critical review of the
literature. Australian and New Zealand Journal of Psychiatry, 30, 54-62.
Gotlib, I. H., Whiffen, V. E., Mount, J. H., Milne, K. & Cordy, N . I. (1989). Prevalence rates and
demographic characteristics associated with depression in pregnancy and the postpartum. Journal of
Consulting and Clinical Psychology, 57, 269-274.
Koniak-Griffin, D., Lominska, S. & Brecht, M. L. (1993). Social support during adolescent pregnancy:
A comparison of three ethnic groups. Journal of Adolescence, 16, 43-56.
McMahon, C. A,, Ungerer, J. A,, Beaurepaire, J., Tennant, C. & Saunders, D. (1995). Maternal and paternal
anxiety and fetal attachment after IVF conception. Paper presented at the World Association for Infant
Mental Health Pacific Rim Meeting, Sydney, 21-24 April.
McNair, Lorr, M. & Droppleman, L. F. (1981). Manual: Profile of Mood States. San Diego, CA: EDITS.
Mercer, R. T., Ferketich, S., May, K., De Joseph, J. & Sollid, D. (1988). Further exploration of maternal and
paternal fetal attachment. Research in Nursing and Health, 11, 83-95.
Muller, M. E. (1992). A critical review of prenatal attachment research. Scholarly Inquiry fm Nursing Practice:
A n International Journal, 6, 5-26.
Muller, M. E. & Ferketich, S. (1993). Factor analysis of the Maternal Fetal Attachment Scale. Nursing
Research, 42, 144-147.
Norusis, M. J. (1985). SPSS-X Advanced Statistics Guide. Chicago: McGraw-Hill.
O’Hara, M. W., Zekoski, E. M., Philipps, L. H . & Wright, E. J. (1990). Controlled prospective study of
postpartum mood disorders: Comparison of childbearing and nonchildbearing women. Journal ofAbnormal
P~ychology,99, 3 - 15 .
Parker, G., Tupling, H., Brown, L. B. (1979). A parental bonding instrument. British Journal of Medical
Psychology, 52, 1-10,
Raphael-Leff, J. (1985). Facilitators and regulators; participators and renouncers: Mothers and fathers
orientations towards pregnancy and parenthood.Journal of Psychosmutic Obstetrics and Gymzaolog~,4, 169-184.
Raphael-Leff, J. (1986). Facilitators and regulators: Conscious and unconscious processes in pregnancy and
early motherhood. British Journal of Medical Psychology, 59, 43-55.
Sarason, I. G., Levine, H. M., Basham, R. B. & Sarason, B. R. (1983). Assessing social support: The Social
Support Questionnaire. Journal of Personality and Social Psychology, 44, 127- 139.
Snaith, R. P. (1991). Measurement in psychiatry. British Journal of Psychiatry, 159, 78-82.
Snaith, R. P. (1993). What do depression rating scales measure. British Jourml ofPsychiafy, 163, 293-298.
Spanier, G. B. (1976). Measuring dyadic adjustment: New scales for assessing the quality of marriage and
similar dyads. Journal of Marriage and the Family, 38, 15-30.
Wilhelm, K. & Parker, G. (1988). The development of a measure of intimate bonds. Psychological Medicine,
18, 225-234.
Zachariah, R. (1994). Maternal-fetal attachment: Influence of mother-daughter and husband-wife
relationships. Research in Nursing and Health, 17, 37-44.
Zigmond, A. S. & Snaith, R. P. (1983). The Hospital Anxiety and Depression Scale. Acta Psychzatrzca
Scandinavica , 67, 36 1-370.
Zung, W. K. (1965). A self-rating depression scale. Archives of General Psychiatry, 12, 63-70.

Received 9 July 1996; revised version received 7 January 1997