Soc Psychiatry Psychiatr Epidemiol (2002) 37: 6067

Steinkopff Verlag 2002

ORIGINAL PAPER

A. Bifulco P. M.Moran C. Ball A. Lillie

Adult attachment style. II: Its relationship

to psychosocial depressive-vulnerability

Accepted: 21 September 2001

SPPE 511

Abstract Background A range of studies show adult

attachment style is associated with depressive-vulnerability factors such as low self-esteem, poor support and
childhood adversity. However, there is wide inconsistency shown in the type of insecure style most highly associated. Few studies have examined attachment style in
relation to clinical depression together with a range of
such factors in epidemiological series. The present study
uses an interview measure of adult attachment which
differentiates type of attachment style and degree of insecurity of attachment, to see: (a) if it adds to other vulnerability in predicting depression and (b) if there is
specificity of style to type of vulnerability. Method Two
hundred and twenty-two high-risk and 80 comparison
women were selected from questionnaire screenings of
London GP patient lists and intensively interviewed. The
Attachment Style Interview (ASI) differentiated five
styles (Enmeshed, Fearful, Angry-dismissive, Withdrawn and Standard) as well as the degree to which attitudes and behaviour within such styles were dysfunctional (non-standard). Attachment style was examined
in relation to low self-esteem, support and childhood experience of neglect or abuse, and all of these examined
in relation to clinical depression in a 12-month period.
Results The presence of any non-standard style was significantly related to poor support, low self-esteem and
childhood adversity. Some specificity of type of style
and type of vulnerability was observed. Logistic regression showed that non-standard Enmeshed, Fearful and
Angry-dismissive styles, poor support and childhood
neglect/abuse provided the best model for clinical depression. Conclusion Non-standard attachment in the
A. Bifulco P. M.Moran C. Ball A. Lillie
Lifespan Research Group
Dept of Social and Political Science
Royal Holloway, University of London
11, Bedford Square,
London WC1B 3RA, UK
Tel:. +44-20/73 07-86 00
Fax: +44-20/76 36-22 68
E-Mail: A.Bifulco@rhul.ac.uk

form of markedly Enmeshed, Fearful or Angry-dismissive styles was shown to be associated with other depressive-vulnerability factors involving close relationships, self-esteem and childhood adversity and added to
these in modelling depression.
Key words attachment style self-esteem support
childhood adversity

Introduction
Adult attachment style is inextricably linked with both
the quality of ongoing close relationships and generalised attitudes to relating. On the basis of childhood
classifications, insecure attitudes of Avoidance (involving fear or mistrust of others) or Anxious/ambivalence
(involving fear of abandonment) have been differentiated from Secure attitudes (involving belief in others
dependability and availability) (Ainsworth et al. 1978).
In attachment terms, the seeking of support is argued to
regulate feelings of insecurity. Thus, distress during
times of threat, stress, pain or illness relieved in childhood by behaviours designed to seek or maintain proximity to the caregiver, is generalised to other close relationships in adulthood (Bowlby 1973). This provides the
safe haven function of attachment relationships.Absent
or unpredictable support reduces feelings of security
and is, thus, argued to predispose to affective disorder
(Bowlby 1977; Holmes 1993).
Bowlby identified two key facets in the development
of secure attachment that the self is perceived as worthy of love and attention, and that others are viewed as
warm and responsive (Bowlby 1973). Children develop
internal working models or representations of attachment that allow them to predict and interpret the behaviour of attachment figures and view themselves in
relation to others. Attachment theory posits that these
working models serve as templates for the interpretation of later relationships throughout the lifespan. Thus,
early relationships are argued to exert long-term impact

61

on subsequent relationships by affecting the nature and

development of the mental models. Both self-esteem
and attachment style are deemed to be distorted by adverse childhood experience involving absent or unpredictable care from parent figures. Given that the elements involved in attachment theory (poor support, low
self-esteem and childhood adversity) have all additionally been identified as vulnerability factors for clinical
depression (Brown et al. 1990), the present paper examines adult attachment style in relation to these factors to
elaborate on an existing psychosocial model of depression (op cit).
Evidence is now emerging for the association between insecure attachment styles and depressive-related
vulnerability. Thus, individuals with insecure attachment styles report poor marital functioning, communication and problem solving (Kobak and Hazan 1991),
low flexibility and reciprocity in confiding (Mikulincer
and Nachshon 1991) and low support-seeking in situations where anxiety increases (Simpson et al. 1992). Insecure styles are associated with low self-esteem (Murphy and Bates 1997; McCarthy 1999) and a model
developed which identifies negative attitudes to self as a
central feature of the attachment classification (Griffin
and Bartholomew 1994). Childhood experiences such as
abuse (Alexander 1993) and parenting style (McCarthy
1999) are shown to be associated with insecure attachment style. Among women with adverse childhood experience, insecure attachment styles relate not only to
poorer functioning in partner relationships but also to
the presence of a deviant partner (McCarthy 1999).
However, while there is consistency in linking any insecure attachment style with such risks, there is virtually
no consistency in differentiating the type of attachment
style to any of them. For example, few studies are able to
differentiate insecure styles in terms of social support
(Bartholomew et al.1997).Findings in relation to self-esteem are contradictory: low self-esteem correlates most
highly to Fearfulness in some studies (Murphy and
Bates 1997),Anxious/ambivalence in others (McCarthy
1999) and Avoidance in others (Mickelson et al. 1997).
Similarly, childhood sexual abuse is variously related to
Fearful style (Alexander 1993) and Anxious/ambivalence (Whiffen et al. 1999). Poor parenting in contrast
relates to Avoidant attachment (McCarthy 1999).
Few studies have simultaneously examined vulnerability characteristics involving support, self-esteem and
childhood experience together with attachment style
and depression. However, one study showed insecure or
vulnerable styles including Ambivalent and Independent-hostile to be significantly related to poor support
from partner or close other (Harris and Bifulco 1991).
Vulnerable attachment styles occurred twice as often
among those with lack of parental care in childhood. Logistic regression showed that childhood lack of care,
poor support, helpless coping and vulnerable attachment style were required to model depression. These
findings have not yet been repeated with the benefit of
updated measures of attachment, the inclusion of self-

esteem and the examination of childhood abuse as well

as neglect.
A companion paper describes how a new attachment
interview differentiates five attachment styles (Enmeshed, Fearful, Angry-dismissive, Withdrawn and
Standard) with the first three significantly related to depression, but only at marked or moderate levels of
insecurity or non-standard levels (Bifulco et al. submitted). This relationship held when examined retrospectively in relation to depression in prior 12 months, but
also prospectively in relation to new onset of disorder.
Such non-standard styles were twice as common in series selected for vulnerability in terms of conflictful relationships or poor support, and childhood neglect or
abuse. The aim of the present report is to examine such
styles in relation to varieties of poor support (absence of
support figures vs conflictful or indifferent relationships), self-esteem and childhood adversity. A final
model will examine attachment style and other vulnerability in relation to clinical depression in a 12-month
period. The experience of early life depression in
teenage years has rarely been examined in relation to
adult attachment style and yet could prove important either as an alternative etiological factor (e. g. due to the
scarring effects of earlier disorder) or as a mediator for
adverse childhood experience in relating to adult vulnerability (Kessler and Magee 1993). Early life depression will, thus, be examined in a model including childhood experience, other adult vulnerability and insecure
attachment.
The following hypotheses were tested:
All non-standard (i. e. marked or moderate levels
of insecure) attachment styles are expected to be associated with poor support. However, the styles are
expected to be differentiated by marital status and
the quality of close relationships. Thus, those with
Fearful and Withdrawn styles are expected to have
fewer support figures and to be single. In contrast
those with Enmeshed and Angry-dismissive styles
are expected to have more conflictful close relationships, including those with partners.
Non-standard attachment styles are expected to relate to low self-esteem with Fearful style expected to
have the highest rates.
Non-standard attachment is expected to relate to
childhood neglect/abuse. Little discrimination between style is expected.
Non-standard Enmeshed, Fearful and Angry-dismissive styles are expected to model depression in
addition to poor support, low self-esteem and childhood adversity. This is expected to hold when controls are made for symptomatology at interview, and
prior depression in teenage years.

62

Subjects and methods

Sample
The sample was taken from a programme of research investigating
vulnerability to depression. It comprised a selected series of 302 community-based women, questionnaire-responders from registers of
patients of general practitioners in Islington, North London. Of these
three-quarters were selected for vulnerability involving either poor
ongoing relationships (conflict with partner or child or lack of close
confidant), low self-esteem, or neglect/abuse in childhood. The remainder formed a comparison series. The women were interviewed
in depth about their childhood experience, ongoing supportive relationships, self-esteem, attachment style and depression. The women
were selected into two series utilising different screening: (i) 104
mothers were selected for ongoing vulnerability in terms of problematic close relationships or low self-esteem and seen prospectively
over 12 months; and (ii) 118 women were selected for the presence of
neglect or abuse in childhood and 80 consecutive questionnaire responders for comparison. The latter groups comprised pairs of sisters, raised together and interviewed independently. The sample is described at length elsewhere (Bifulco et al. 1997; Bifulco et al. 1998a)
and in a companion paper (Bifulco et al. 2002). Interview compliance
in this high-risk series was 60 %, with 23 % refusals and a further 19 %
proving unobtainable, largely due to the sisters unavailability.
Overall, 32 % of the series were working-class, 61 % married/cohabiting and 69 % had children. A fifth of the women were single
mothers and half the women had been separated or divorced at some
point in the past. The average age was 34.6, with a range of 1950
years.
Measures
All women were contacted at home for intensive interview. Interviews
were tape-recorded and ratings conducted after the visit. Ratings
were made according to predetermined criteria, with consensus
meetings to check consistency of ratings and resolve anomalies. All
vulnerability ratings were checked by raters blind to psychiatric or attachment status.
Self Evaluation and Social Support Interview (SESS)
(OConnor and Brown 1984)
This semi-structured interview measure assessed self-esteem and the
quality of close relationships. Inter-rater reliability on all scales is satisfactory (OConnor and Brown 1984).
Negative Evaluation of Self (NES). NES is an index based on the rating of negative statements made in response to a range of questions
about the self concerning personal attributes (such as attractiveness,
intelligence, personality), role competence (for example, as mother or
worker) or lack of self-acceptance (more global statements denoting
self-dislike) (Brown et al. 1990). The presence of a high rating
(marked or moderate on a 4-point scale) on at least one of the three
negative scales indicated the presence of NES.
Support. Detailed assessments of relationship with partner and
with up to two very close others (VCO) was made in terms of confiding and support. For VCO, the degree of confiding and frequency
of contact was utilised in the support assessments. Partner support
was assessed on a global scale reflecting confiding, emotional support
and conflict. The overall scale utilised 7 points with 13 denoting very
good to good-average support and 47 denoting poor-average to
poor support. A further distinction was made in terms of whether
conflict (points 2, 4 and 6) or indifference (points 3, 5 and 7) was present.
Poor support. This was assessed in terms of: (a) the lack of a confidant described as very close and seen at least monthly (lack of true
very close other) and (b) the presence of an unsupportive partner

(points 47 on overall scale). The latter was further divided into conflictful, unsupportive relationships (ratings of 4 or 6 on the overall
support scale) or indifferent unsupportive relationships (rating of 5
or 7). A dichotomised index of poor support reflected the presence
of unsupportive partner for those married or cohabiting and the lack
of a close confidant for those single.
Attachment Style Interview (ASI) (Bifulco et al. 1998b)
The ASI measure is described in a companion paper (Bifulco et al.
2002). Overall attachment style was a global judgement of style based
on the ability to make intimate relationships and eight attitudinal
scales reflecting avoidance/distance in maintaining relationships
(e. g. mistrust; attitudinal constraints on closeness; self-reliance;
anger) or anxious/ambivalence (e. g. desire for engagement with others; intolerance of separation; fear of intimacy). The overall classification assessed style (Enmeshed, Fearful, Angry-dismissive, Withdrawn and Clearly Standard) as well as the degree to which
attachment style was insecure or non-standard. All ratings were
agreed at consensus meetings by researchers blind to the presence of
other vulnerability factors or psychiatric symptoms. In the analysis
that follows only styles at marked or moderate levels of insecurity
(non-standard) will be examined, since these are the ones shown to
relate to depression (Bifulco et al. 2002). Those with mildly insecure
styles are included with those secure or standard.
For 5 % of the series a double classification of style was made,
with different primary and subsidiary attachment styles rated. This
occurred when no clear pattern emerged in terms of the recognised
styles. These were when more than one style was evident across relationships (e. g. fear of intimacy and anger both being present) or
when a style evident in one domain was not consistent across other
types of relationships (e. g. clinging with partner and fearful with
other relationships). For most of the analysis the primary attachment
style rating (i. e. the one that reflected the most pervasive style) is
utilised.
Childhood Experience of Care and Abuse (CECA) (Bifulco et al. 1994)
The CECA interview was used to assess a wide range of negative childhood experiences prior to age 17, involving neglect and abuse during
different household arrangements. The scales used in the present
analysis are those previously shown to relate to increased risk of adult
depression and involve severe parental neglect, severe physical abuse
by mother, father or other household member and severe instances of
sexual abuse from any adult. Each of these was rated for severity on
4-point scales and only those with marked or moderate ratings included. Reliability and validity are satisfactory and are reported elsewhere (Bifulco and Brown 1996; Bifulco et al. 1997). The index of
childhood neglect or abuse used is based on the presence of at least
one type of severe neglect, physical or sexual abuse before age 17.
Present State Examination (PSE) (Wing et al. 1974)
Depression was assessed by the PSE over a 12-month period and
judgements of clinical case involved the presence of depressed mood
plus four or more key symptoms (Finlay-Jones et al. 1980). This
threshold has been shown in practice to be virtually identical to major depressive disorder in terms of the DSM-III-R checklist (APA
1987) when both classifications are used (Bifulco et al. 1998a). All
symptom levels and caseness judgements were checked by a psychiatrist experienced in the PSE (Professor T. K. J. Craig) and blind to
other risk factors. Depression was assessed retrospectively over the
prior 12 months in the Childhood Risk and Comparison series. In the
Adult Risk series, where clinical depression was absent at first contact,
onset of symptoms was determined prospectively over the following
12 months. In both series previous episodes of disorder were questioned about and those aged 1319 included as teenage depressions
for this analysis.

63
Data analysis
SPSS9 was used for the data analysis, with corrected chi-squares
used to examine the relationship of attachment style to other vulnerability factors, and logistic regression used to examine which vulnerability factors provided the best model for depression. In order to examine the contribution of each of the attachment styles as categorical
variables, five dichotomous variables representing high levels
(marked or moderate) of Enmeshed, Fearful, Angry-dismissive,
Withdrawn or Standard (mildly insecure/clearly standard) styles
were derived from the overall attachment scale. These comprised the
marked or moderate ratings for each style compared with the remainder.
A separate index identified those with double ratings of style (i. e.
different primary and subsidiary styles rated) in contrast to those
with a single non-standard style (marked or moderate Enmeshed,
Fearful, Angry-dismissive or Withdrawn) and the remainder (mild
degrees of attachment style or clearly standard).

Results
Support
Marital status
Non-standard attachment styles were examined in relation to marital status, single-parenthood and past separation/divorce. Those with Enmeshed and Angry-dismissive styles were more likely to be married or
cohabiting (83 % and 73 %, respectively, compared with
56 % of remainder, p < 0.01). Women with Fearful styles
were least likely to be married/cohabiting (42 %) when
compared with other non-standard styles (69 %) or
those standard (63 %, p < 0.005). Those with Fearful or
Withdrawn styles were more likely to be single parents
(38 % and 28 %, respectively, compared with 15 % of remaining women, p < 0.005). There was no association
between type of style and past partner separation.

Lack of true very close other

Just over a quarter of the women (29 %) had no close
confidant seen at least monthly. This was significantly
related to having a non-standard style (72 % (98/136) vs
13 % (21/166) p < 0.0001, see Table 1). It was significantly
less common among those with Enmeshed style: 38 % vs
60 % of other non-standard and 13 % (21/166) of standard (p < 0.001).

Poor partner support

In the total series 44 % of women who were married/cohabiting had poor partner support. This was highly related to any non-standard attachment style (75 % had
poor partner support compared with 46 % of those with
standard attachment, p < 0.001) (see Table 1, col 3).
When particular styles were examined, Enmeshed
(81 %) and Withdrawn (83 %) had the highest rates.
When such poor support was further examined in terms
of conflict, Enmeshed (56 %), Fearful (45 %) and Angrydismissive (52 %) styles had similarly high rates of con-

Table 1 Attachment style and support

Non-standard
attachment styles
(markedly or
moderately insecure)

Lack true
VCO
N=302
% (n)

Partner*
poor support
N=184
% (n)

Partner
conflict
N=184
% (n)

Partner
indifferent
N=184
% (n)

Enmeshed
Fearful
Angry-dismissive
Withdrawn
Standard
(mildly insecure or
clearly standard).
P<

38 (8/21)
58 (31/53)
55 (18/33)
69 (20/29)
13 (21/166)

81 (13/16)
68 (15/22)
70 (16/23)
83 (15/18)
46 (48/105)

56 (9/16)
45 (10/22)
52 (12/23)
33 (6/18)
23 (24/105)

25 (4/16)
23 (5/22)
17 (4/23)
50 (9/18)
23 (24/105)

0.0001,
4 df

0.002,
4 df

0.007,
4 df

NS

* Those married or cohabiting only

flictful partner relationships when compared with

Withdrawn (33 %) and Standard (23 %) styles (p <
0.001). There was little variation in terms of indifferent
partner relationships, apart from Withdrawn style
which had the highest rate (50 % vs 22 % for remaining
women, p < 0.025) (see Table 1, col 4 and 5). No differences were found for those with a double non-standard
attachment classification and those with single nonstandard style.

Negative Evaluation of Self (NES)

Half the women in the series scored on the index of negative evaluation of self. This was highly related to the
presence of any non-standard style: 71 % (96/136) with
non-standard style had NES vs 36 % (59/166) with standard style (p < 0.001). Highest rates occurred for those
with Fearful (83 % or 44/53) or Angry-dismissive (73 %
or 24/33) styles. This compared with 56 % (28/50) with
other non-standard styles and 36 % (59/166) standard (p
<0.001, 4 df). No differences were found for those with a
double non-standard attachment classification and
those with single non-standard style.

Attachment style, concurrent vulnerability

and depression
A logistic regression was undertaken with each of the dichotomised, derived, non-standard attachment styles
and the concurrent vulnerability factors of NES and
poor support. In order to accommodate both cohabiting
and single women poor support comprised an index of
unsupportive partner for those cohabiting or lack of
close other for those single. Previous analysis has shown
that for cohabiting women conflictful partner relationship is a vulnerability factor even in the presence of
other support outside the home (Brown et al. 1986a).
Those with depression at either case or subclinical level
at point of interview were excluded. Analysis showed
that while Enmeshed, Fearful and Angry-dismissive

64

non-standard styles added significantly to the model,

Withdrawn style failed to add. NES did not add to the
model and poor support was only weakly related, falling
short of the 5 % level of significance (see Table 2).

Table 3 Attachment style and past risk factors

Neglect/abuse < 17

Teenage depression

Non-standard attachment
style (markedly/moderately
insecure)

Present
N=165
% (n)
with style

Absent
N=137
% (n)
with style

Present
N=47
% (n)
with style

Absent
N=255
% (n)
with style

Enmeshed (n=21)
Fearful (n=53)
Angry-dismissive (n=33)
Withdrawn (n=29)
Standard (n=166)
(mildly insecure or
clearly standard)
P<

9 (15)
23 (38)
16 (26)
10 (17)
42 (69)

4 (6)
11 (15)
5 (7)
9 (12)
71 (97)

13 (6)
34 (16)
17 (8)
4 (2)
32 (15)

6 (15)
15 (37)
10 (25)
11 (27)
59 (151)

Early life experience

Just over half the women (55 %, 165/302) in the total series had experienced severe neglect or abuse before the
age of 17. In the comparison series the rate of 20 %
(16/80) was typical of that found in other representative
series in inner-London (Bifulco et al. 1994). The presence of neglect or abuse in childhood was associated
with non-standard attachment styles: 58 % (96/165)
with neglect/abuse had non-standard styles vs 29 %
(40/137) without neglect/abuse (p < 0.001). When type
of non-standard style was examined Angry-dismissive
proved three times as common among those with neglect/abuse as those without (16 % vs 5 %) and Fearful
style proved to be twice as common (23 % vs 11 %, see
Table 3). There was little difference in rates for Enmeshed or Withdrawn styles. No differences were found
for those with a double classification of non-standard
style and those with single.
In the combined series, 15 % (47) of women reported
a depression before age 20 at clinical severity. Teenage
depression was significantly related to non-standard attachment style: 68 % (32/47) of those with teenage depression were rated non-standard compared with 41 %
(104/255) of those with no teenage depression (p <
0.005, see Table 3). Non-standard Enmeshed and Fearful
styles were the most highly related, showing double the
rates among those with such depression when compared
with those without (see Table 3). When those with double classifications of non-standard attachment were examined, these had significantly higher rates of teenage
depression: 56 % (9/16) compared with 19 % (23/120)
with single non-standard style and 9 % (15/166) with
standard style (p < 0.0001, 2 df).
Table 2 Non-standard attachment style, concurrent vulnerability factors and depression. Logistic regression (excluding case or subclinical depression at interview
1, total n=245)
Variable

Odds-ratio

Wald

P<

Non-standard Enmeshed
Non-standard Fearful
Non-standard Angry-dismissive
Non-standard Withdrawn
NES
Poor support

4.88
2.96
3.84
1.54
1.27
1.87

8.20
5.37
7.46
0.55
0.47
3.27

0.004
0.02
0.006
NS
NS
NS (0.07)

In terms of goodness of fit, 78.78 % of subjects correctly classified. Best model is

provided by non-standard (marked or moderate) Enmeshed, Fearful or Angry-dismissive attachment styles. Poor support fails to add to the model at the 5 % level

X2=28.07
P < 0.0001, 4 df

X2=20.62
P < 0.0001, 4 df

Attachment style, past and present vulnerability

and depression
A logistic regression was undertaken to examine the role
of all risk factors in modelling depression in the 12month period.A combined index of non-standard styles
was utilised (markedly or moderately Enmeshed or
Fearful or Angry-dismissive vs the remainder). This was
entered into the model together with concurrent vulnerability factors (NES and poor support) and past risk
factors (childhood neglect/abuse and teenage depression) in relation to depression in the 12-month period.
Women with depression at case or subclinical level at
point of interview were excluded from the analysis in order to control for the possible contaminating effects of
symptomatology. Table 4 shows that non-standard attachment style, poor support and childhood
neglect/abuse provided the best model. Teenage depression did not add to the model at the 5 % level of significance. (When the model was re-run entering a control
for having a sister in the series, the results remained unchanged. Figures available on request.)

Discussion
An overall assessment of attachment style based on an
individuals ability to relate to close others and access
Table 4 Non-standard attachment style, vulnerability and depression in 12
months. Logistic regression (excluding case or subclinical depression at interview,
total n=245)
Variable

Odds-ratio

Wald

P<

Non-standard Enmeshed, Fearful

or Angry-dismissive style
Poor support
NES
Neglect/abuse < 17
Teenage depression

2.34

5.37

0.02

2.09
1.38
2.46
2.10

4.12
0.79
5.85
2.97

0.04
NS
0.01
NS (0.08)

In terms of goodness of fit, 80 % of subjects correctly classified. Best model is provided by non-standard attachment, poor support and neglect/abuse < 17

65

and utilise support in conjunction with attitudes denoting Enmeshment, Fearfulness, Angry-dismissiveness or
Withdrawn-avoidance was utilised in an intensively
studied series of vulnerable and comparison community women. Insecure attachment was significantly related to Negative Evaluation of Self (NES) and poor support (in terms of lack of support from partner or lack of
close confidant). Although each of the non-standard
styles related to such risk, some patterning was evident:
Fearful and Angry-dismissive styles had the highest
rates of NES; all the vulnerable styles (Enmeshed, Angry-dismissive and Fearful) had high partner conflict;
and women with Withdrawn style were the most likely
to lack a close confidant and to experience indifferent
partner relationships.
Although NES was highly related to non-standard attachment and is shown elsewhere to relate to childhood
adversity and poor support (Brown et al. 1986a), it
proved unrelated to depression in the present series
once these other factors were taken into account. Although self-esteem has long shown associations with
depressive disorder, its role has variously been identified
as a concomitant of disorder, a consequence of disorder,
or a vulnerability for disorder (Lewinsohn et al. 1981;
Rohde et al. 1994). It is certainly highly related to ongoing symptomatology, whether depressive or anxious,
and is more highly related to new onsets of depression
when combined with such prior symptomatology
(Brown et al. 1986b). In addition, its role as vulnerability
factor is most effective when combined with negative
close relationships as conjoint psychological and environmental vulnerability (Brown et al. 1990). The present
formulation of non-standard attachment style may have
effectively replaced such conjoint risk, with the advantage of providing a more theoretical framework for such
inter-personal vulnerability.
The association between poor support and attachment style is necessarily a close one in this analysis since
the degree of insecure attachment was predicated on inability to form intimate relationships. However, the two
were not fully overlapping: a number of women had specific poor relationships without any indication of generalising this to a full style of relating. The role of poor
support in relating to disorder once attachment style
was taken into account was somewhat unclear in this series. This was in part because of the nature of the attachment measure which encompassed the extreme
forms of poor support, and in part because of the highrisk nature of the series, with an over-representation of
those with negative adult relationships. When only concurrent risk factors were examined, with each of the
non-standard attachment styles examined independently, poor support fell short of statistical significance
in modelling depression. This was in part due to its high
overlap with Withdrawn style. Once this style was excluded from the vulnerable attachment styles a relationship re-emerged in the series as a whole. Debates on
the role of social support in depression have polarised in
terms of whether lack of support is due to the impove-

rished social environment (low availability of reliable

support) or due to psychological barriers (inability to
perceive existing support as adequate) (Henderson and
Brown 1988). The result of this analysis suggests that
while attachment style (which entails generalised poor
support due to psychological barriers) is central, a role
for environmentally based poor support remains. Further study of these factors on representative and
prospective series may tease apart such associations.
The characteristics of the social environment should
not be underestimated in examining expressions of attachment style. Bowlbys original formulation of attachment theory included environmental trauma and loss
shaping the resulting styles (Bowlby 1977). A dynamic
view of the social environment is frequently neglected in
studies of attachment, although the study of interpersonal life events has been shown to have important interactions with vulnerable attachment style in relating
to psychopathology (Hammen et al. 1995). The series reported here had high rates of social disadvantage in
terms of inter-personal severe life events in the past year
(Bifulco et al. 1998a) as well as over a lifetime (Bifulco et
al. 2000), a high rate of single-parent status and of previous separation/divorce. The non-standard styles exhibited may in part reflect coping strategies activated by
stressful inter-personal situations. Such situations are
likely to involve isolation, rejection and conflict, at least
in part determined by the behaviour of close others, not
only of the individual involved.
Childhood experience in terms of neglect and abuse
has shown consistent relationship to depressive disorder
and to a range of environmental and psychological risks
(Bifulco et al. 1994; Bifulco et al. 1998a). It was confirmed
as additionally relating to attachment style in this analysis, with Angry-dismissive and Fearful styles having
particularly high rates. The relationship appeared not to
be mediated by teenage depression with the latter failing
to add to the model at the 5 % level of significance. However, it needs to be borne in mind that this constituted a
highly selected series with high rates of both vulnerability and prior psychopathology and this may have served
to mask its real contribution. However, it was of interest
to note that teenage depression was highly related to a
double classification of attachment style, suggesting
that this may lead to more disorganised expression of
attachment. Numbers were not sufficient in this analysis
to explore this issue further.
Of interest was the fact that the Angry-dismissive
style, newly differentiated from Withdrawn, was particularly common among those with neglect/abuse. Although the former style is not differentiated by most attachment style categorisations, the relationship between
such anger and childhood experience is consistent with
attachment theory predictions. This dysfunctional
anger is argued to spring from the redirection of submerged childhood anger towards current attachment
figures. This is said to occur either through the projection of rejecting or hostile qualities of early attachment
figures onto close others in adulthood, or through the

66

re-enactment of angry behaviour exhibited by attachment figures during childhood (Bowlby 1973). Further
investigation of the specific relationship of attachment
style to different childhood experiences involving hostile or physically abusive parenting may further illuminate such linkages. Given the relationship of Fearful
style both to childhood neglect/abuse and teenage depression, specificity of style may additionally be influenced by the occurrence of teenage disorder. Experience
of the latter may mould insecure styles deriving from
adverse early experience in more anxious directions.
Childhood experience has been argued to create both
social and psychological disadvantage in later life. Thus,
social impoverishment in terms of unsupported pregnancy and parenthood, deviancy and unreliability of
spouse/partner are all consequences of adverse childhood experience (Harris et al. 1987; Quinton et al. 1993).
In parallel, negative cognitive styles involving low selfesteem, helplessness, poor coping and non-optimal
confiding, all of which have negative impact on relationships, are similarly associated with early adversity (Andrews and Brown 1988; Harris et al. 1990). The present
analysis shows how non-standard attachment style derived from a synthesis of impoverished social environment and negative cognitive-emotional styles may provide a vehicle for examining how such experiences
interact across the life course.
The limitations of the reported analysis revolve
around the limited prospective orientation and the use
of a highly selected series. The advantage of the latter is
that analysis can be undertaken on relatively modest
numbers since psychosocial risks and depression are
maximised in the series. What remains less clear is the
normal prevalence and aggregation of attachment style
with other risk factors in representative series. In addition, further analyses are needed to examine the role of
anxiety, commonly comorbid with depression, in examining the specificity of style to types of disorder and to
add further controls for the influence of symptomatology on non-standard ratings. In addition, examining the
stability of attachment style and the conditions under
which change occurs will give more understanding to
the nature of such styles and the extent to which they are
modified by changes in the environment. These will be
the topic of later reports.
The findings are relevant for intervention in terms of
providing a model for identifying components of vulnerable attachment style, all of which can be tackled
singly or simultaneously during treatment or service
provision. Thus, unsupportive relationships as well as
specific interpersonal cognitive biases and low self-esteem can be the focus of such intervention to enable a
tailoring of help to particular individuals within their
specific social context. Such intervention may aim for
different degrees of change, which can be monitored by
the ASI approach, for example, reducing the severity of
insecurity and, thus, reducing depressive risk, or converting insecure styles to secure and, thus, engendering
better quality of relationships in the longer term. Given

that the likelihood of re-enacting unsupportive relationships in new contexts can be inferred by the presence of
generalised cognitive biases involved in non-standard
attachment, attempts to reduce the likelihood of future
conflictful engagements or isolation can also be tackled
in treatment. Greater precision in monitoring such characteristics will entail a better understanding of an individuals vulnerability in terms of greater psychological
or social-environmental weighting and, thus, aid future
recognition and treatment of depressive risk.
Acknowledgements The research in Islington was supported by a
Medical Research Council programme grant (G702 833 Principle Investigator Professor George Brown). We would like to thank Caroline
Campbell, Hedy Wax and Anne Brackenbridge for data collection,
Tom Craig for advice on psychiatric ratings and Laurence Letchford
for computer analyses. In addition, we would like to thank North London GPs who allowed us access to their patient lists and also to all the
women who gave up hours of their time in being interviewed.

References
1. Ainsworth M, Blehar M, Waters E, Wall S (1978) Patterns of attachment: a psychological study of the strange situation.
Lawrence Erlbaum Associates, Hillsdale, NJ
2. Alexander PC (1993) The differential effects of abuse characteristics and attachment in the prediction of long-term effects of
sexual abuse. J Interpersonal Violence 8: 346362
3. Andrews B, Brown GW (1988) Social support, onset of depression and personality: An exploratory analysis. Soc Psychiat Psychiatr Epidemiol 99108
4. APA (1987) Diagnostic and statistical manual of mental disorders DSM-IIIR (3rd edn revised). American Psychiatric Association, Washington DC
5. Bartholomew K, Cobb RJ, Poole JA (1997) Adult attachment patterns and social support processes. In: Pierce GR, Lakey B (eds)
Sourcebook of social support and personality.The Plenum series
in social/clinical psychology. Plenum Press, New York, pp.
359378
6. Bifulco A, Bernazzani O, Moran PM, Ball C (2000) Lifetime stressors and recurrent depression: preliminary findings of the Adult
Life Phase Interview (ALPHI). Soc Psychiat Psychiatr Epidemiol
35: 264275
7. Bifulco A, Brown GW (1996) Cognitive coping response to crises
and onset of depression. Soc Psychiat Psychiatr Epidemiol 31:
163172
8. Bifulco A, Brown GW, Harris TO (1994) Childhood Experience of
Care and Abuse (CECA): a retrospective interview measure. J
Child Psychol Psychiat 35: 14191435
9. Bifulco A, Brown GW, Lillie A, Jarvis J (1997) Memories of childhood neglect and abuse: corroboration in a series of sisters. J
Child Psychol and Psychiat 38: 365374
10. Bifulco A, Brown GW, Moran P, Ball C, Campbell C (1998a) Predicting depression in women: the role of past and present vulnerability. Psychl Medicine 28: 3950
11. Bifulco A, Lillie A, Ball B, Moran P (1998b) Attachment Style Interview (ASI). Training manual. Royal Holloway, University of
London
12. Bifulco A, Moran P, Ball C, Bernazzani O (2002) Adult Attachment style. I: Its relationship to clinical depression. Soc Psychiat
Psychiatr Epidemiol 37: 5059
13. Bowlby J (1973) Attachment and loss: vol 2. Separation: anxiety
and anger. Basic Books, New York
14. Bowlby J (1977) Aetiology and psychopathology in the light of
attachment theory. B J Psychiat 130: 201210
15. Brown GW, Andrews B, Harris TO, Adler Z, et al. (1986a) Social
support, self-esteem and depression. Psychol Medicine 16:
813831

67
16. Brown GW, Bifulco A, Harris T, Bridge L (1986b) Life stress,
chronic subclinical symptoms and vulnerability to clinical depression. J Affect Disord 11: 119
17. Brown GW, Bifulco AT, Andrews B (1990) Self-esteem and depression: III. Aetiological issues. Soc Psychiat Psychiatr Epidemiol 25: 235243
18. Finlay-Jones R, Brown GW, Duncan-Jones P, Harris TO, Murphy
E, Prudo R (1980) Depression and anxiety in the community.
Psych Medicine 10: 445454
19. Griffin DW, Bartholomew K (1994) Models of the self and other:
Fundamental dimensions underlying measures of adult attachment. J Personality Soc Psychol 67: 430445
20. Hammen CL, Burge D, Daley SE, Davila J, et al. (1995) Interpersonal attachment cognitions and prediction of symptomatic responses to interpersonal stress. J Abnormal Psychol 104: 436443
21. Harris T, Bifulco A (1991) Loss of parent in childhood, attachment style, and depression in adulthood. In: Murray-Parkes C,
Stevenson-Hinde J, Marris P (eds) Attachment across the life-cycle.Routledge, London, New York
22. Harris TO, Brown GW, Bifulco A (1987) Loss of parent in childhood and adult psychiatric disorder: the role of social class position and premarital pregnancy. Psych Medicine 17: 163183
23. Harris TO, Brown GW, Bifulco AT (1990) Depression and situational helplessness/mastery in a sample selected to study childhood parental loss. J Affective Disorders 20: 2741
24. Henderson AS, Brown GW (1988) Social support: the hypothesis
and the evidence. In: Burrows H (ed) Handbook of Social Psychiatry. Elsevier Science Publishers B. V. (Biomedical Division)
25. Holmes J (1993) John Bowlby and Attachment theory. Routledge,
London & New York
26. Kessler RC, Magee WJ (1993) Childhood adversities and adult
depression: basic patterns of association in a US national survey.
Psychol Medicine 23: 679690
27. Kobak RR, Hazan C (1991) Attachment in marriage: effects of security and accuracy of working models. J Personality Soc Psychol 60: 861869

28. Lewinsohn PM, Steinmetz JL, Larson DW, Franklin J (1981) Depression-related cognitions: antecedent or consequent? J Abnormal Psychol 90: 213219
29. McCarthy G (1999) Attachment style and adult love relationships
and friendships: a study of a group of women at risk of experiencing relationship difficulties. B J Medical Psychol 72: 305321
30. Mickelson KD, Kessler RC, Shaver PR (1997) Adult attachment in
a nationally representative sample. J Personality Soc Psychol 73:
10921106
31. Mikulincer M, Nachshon O (1991) Attachment styles and patterns of self-disclosure. J Personality Soc Psychol 61: 321331
32. Murphy B, Bates GW (1997) Adult attachment style and vulnerability to depression. Personality Individual Diffs 22: 835844
33. OConnor P, Brown GW (1984) Supportive relationships: fact or
fancy? J Soc Personal Relations 1: 159175
34. Quinton D, Pickles A, Maughan B, Rutter M (1993) Partners,
peers, and pathways: assortative pairing and continuities in conduct disorder. Develop Psychopath 5: 763783
35. Rohde P, Lewinsohn PM, Seeley JR (1994) Are adolescents
changed by an episode of major depression? J American Acad
Child Adolescent Psychiat 33: 12891298
36. Simpson JA, Rholes WS, Nelligan JS (1992) Support seeking and
support giving within couples in an anxiety-provoking situation:
the role of attachment styles. J Personality Soc Psychol 62:
434446
37. Whiffen VE, Judd ME, Aube JA (1999) Intimate relationships
moderate the association between childhood sexual abuse and
depression. J Interpersonal Violence 14: 940954
38. Wing JK, Cooper JE, Sartorius N (1974) The measurement and
classification of psychiatric symptoms: an instruction manual
for the Present State Examination and CATEGO programme.
Cambridge University Press, Cambridge