Acta Psychiatr Scand 2004: 109: 447–456 Copyright Ó Blackwell Munksgaard 2004

Printed in UK. All rights reserved ACTA PSYCHIATRICA

SCANDINAVICA

Attachment security and parental bonding in

adults with obsessive-compulsive disorder: a
comparison with depressed out-patients and
healthy controls
Myhr G, Sookman D, Pinard G. Attachment security and parental G. Myhr1,2, D. Sookman1,3,
bonding in adults with obsessive-compulsive disorder: a comparison G. Pinard1,2
with depressed out-patients and healthy controls. 1
Department of Psychiatry, McGill University,
Acta Psychiatr Scand 2004: 109: 447–456. Ó Blackwell Munksgaard 2004. 2
Department of Psychiatry, McGill University Health
Centre and 3Department of Psychology, McGill
Objective: This study examines concurrent associations of attachment University Health Centre, Montreal, Quebec, Canada
security, psychopathology and recollections of early parental
interactions, in adults with obsessive-compulsive disorder (OCD),
depression, and in healthy controls.
Method: Thirty-six out-patients with OCD, 16 depressed out-patients
and 26 controls were asked to fill out the Revised Adult Attachment
Scale and the Parental Bonding Instrument (PBI).
Results: OCD and depressed groups were more insecure than controls.
The depressed group recalled less caring mothers than the OCD group,
while the OCD group was indistinguishable from controls on PBI Key words: obsessive-compulsive disorder; depression;
measures. Married status was associated with greater security, but also object attachment; parenting
with recollections of greater parental control, and lower maternal care. Dr Gail Myhr, Department of Psychiatry, McGill Univer-
Conclusion: OCD and depressed groups demonstrated greater sity Health Centre (Royal Victoria Hospital), 1025 Pine
attachment insecurity than controls. No clear relationship emerged Avenue West, Montreal, Quebec H3A 1A1, Canada
between security and PBI recollections. The PBI may not measure E-mail: [email protected]
aspects of early interactions essential for later attachment security, or
recollections may be biased according to diagnosis or attachment style. Accepted for publication November 17, 2003

i.e. the capacity to tolerate negative emotions, and

Introduction
to attenuate them (9, 10). Behaviourally, the secure
Attachment refers to the universal emotional bond, individual embraces attachment relationships –
first arising between infant and primary caregiver, demonstrating comfort with intimacy, the ability
which involves the seeking out of a specific to depend on someone and to be depended on, and
attachment figure in times of stress, and the use the confident use of the attachment relationship to
of this figure as a secure base in the exploration of explore the world, without fear of abandonment
the world (1, 2). Early attachment relationships lay (11).
the ground work for adult attachment relation- A key feature of attachment theory relates
ships, with numerous studies demonstrating that security of attachment in adults to the specific
attachment behaviour demonstrates relative stabil- nature of interactions with early attachment fig-
ity over the life span (3–5), affecting adult roman- ures. Secure attachment behaviour in children has
tic, social and parenting behaviour (3, 6–8). been empirically linked with sensitive, responsive
The concept of attachment – whether secure or caregiving, while insecure attachment behaviour
insecure – has cognitive, emotional and beha- has been linked with inconsistent, neglectful,
vioural components. Cognitively, the hallmark of intrusive or frightening caregiving (1, 12, 13).
secure attachment is the expectation that others Attachment relationships in childhood represent
will be available and responsive in times of need major environmental influences organizing the
(9). From an emotional viewpoint, attachment neurophysiology of the developing brain during
security has been linked with affect modulation its time of maximal growth (9, 10, 14). Inadequate

447
Myhr et al.

relationships may produce stress, which in turn has traumatic experiences that may be expected to
been linked with overpruning and synaptic modi- impact on attachment security. The Italian Group
fication in the areas of the brain responsible with for the Study of Dissociation hypothesized that
emotional regulation – the limbic system and early events impacting on the mother–child dyad
neocortex (14). Long lasting changes in reactivity could lead to disorganized attachment and later
of the hypothalamo-pituitary axis may also occur psychopathology. To test this, they interviewed not
(15). Furthermore, implicit memories of attach- only their subjects, but also their subjectsÕ mothers.
ment experience laid down in the basal ganglia may Both subjectsÕ reports of early trauma and losses
also modify affect regulation and influence future sustained by mothers within 2 years of the subject’s
relationship behaviour in an automatic, reflexive birth were independent risk factors for borderline
manner (10). These factors may explain the cogni- personality (27) and dissociative disorders (28).
tive, emotional and behavioural aspects of insecure Another approach is to try to get a sense of the
attachment: negative expectations about future overall timbre of early attachment relationships.
attachment-related interactions, dysregulation of The Parental Bonding Instrument (PBI), developed
affect, and the tendency to avoid, devalue, or cling by Parker et al. (29), measures recollections of
anxiously to significant relationships. parent–child interactions. The PBI measures two
While insecure attachment is not pathological dimensions for each parent – caring and overpro-
per se, it has been linked to a greater propensity to tection (i.e. controlling). High care scores and low
anxiety and depression in later life (10) (15). overprotection scores are considered to reflect
Studies examining adult attachment and psycho- optimal parenting. The PBI has been used exten-
pathology have found greater insecurity in psychi- sively in community and clinical samples, and has
atric populations compared with healthy controls. been found to be a reliable and valid measure of
One study with a large nationally representative both perceived and actual parenting (30). Parental
sample found childhood adversities of an interper- recollections appear to be independent of mood
sonal nature were strongly associated with insecure state, i.e. PBI scores do not change substantially
adult attachment, which in turn was associated with remission of depression (31, 32).
with various types of psychopathology (16). van Many studies have looked at the relationship
Ijzendoorn and Bakermans-Kranenburg’s (17) between parental bonding and psychopathology.
meta-analysis of 33 studies of clinical and non- The largest of these, using data from 5877 subjects
clinical groups revealed a strong over-represen- of the US National Comorbidity Survey, found
tation of insecure attachment styles in psychiatric that lack of care, especially from one’s mother, was
in-patients, adults with depression, dysthymia, and most consistently associated with adult psychopa-
conduct disorder, and children with oppositional thology. This effect was generally non-specific and
disorders. Fonagy et al.Õs (18) sample of 82 accounted for only 1–5% of the variance in the
in-patients with affective disorders, anxiety disor- occurrence of adult mental disorders (33). A
ders, substance abuse, eating disorders and per- smaller epidemiologic study (n ¼ 386) in a general
sonality disorders had a higher proportion of population revealed no association between PBI
insecure attachment styles than a matched control scores and psychiatric disorder – including anxiety,
group. Patients with obsessive-compulsive disorder psychosis and affective disorders (34). Other com-
(OCD) were included in the heterogeneous anxiety munity-based studies have found recollections of
group. Other studies have replicated the associ- low parental care and maternal overprotection in
ation between insecure attachment and depression individuals with antisocial personality traits (35),
(19, 20), agoraphobia (21), personality disorders and overcontrol characterizing parenting of indi-
(22, 23), eating disorders (24) and post-traumatic viduals with obsessional traits (36).
stress symptoms (25). Burge et al. (26) discovered In clinical samples, low care and/or high over-
not only a concurrent statistical association protection scores have been consistently found
between attachment security and psychiatric in depressed subjects compared with controls
symptoms in young adults, but also found that (31, 37–40). In studies of anxiety disorders, results
insecurity in symptomatic individuals predicted have been inconsistent. One study (41) examined
symptomatology 1 year later. the relationship between parental bonding in
As attachment theory links early experiences to psychiatric out-patients with OCD, comparing
adult attachment security and psychopathology, them to depressed out-patients and healthy con-
the question arises as to how these links can be trols. While the finding of low care and high
studied empirically in adults, when child–caregiver overprotection in the depressed group was repli-
interactions are long over. A common approach is cated, there was no significant difference between
to catalogue a subject’s recollections of early OCD and control groups with respect to parental

448
 Attachment in obsessive-compulsive disorder

bonding scores. A more recent study compared parental overprotection than the other groups,
parental bonding in out-patients with OCD, panic based on the literature reviewed.
disorder with agoraphobia and controls (42).
Anxious subjects recalled their parents as more
protective than controls, though no differences Material and methods
were found between the two anxious groups.
Subjects
Few published studies have explored concurrent
associations of diagnosis, attachment security and The study sample consisted of three groups of
parental bonding. A study of 393 18-year olds (43), subjects: 36 psychiatric out-patients with OCD
revealed that insecure attachment and perception (17 females, 19 males), 16 psychiatric patients with
of lack of maternal care were associated with unipolar depression or dysthymia (12 females and
borderline features. four males) and 26 healthy controls (18 females
Attachment security in adults has been assessed and eight males). Participants were English speak-
using a variety of interviewing and self-report ing, 16–60 years of age.
measures. Those based on a typological model of Subjects in the OCD group were out-patients
attachment describe three or four attachment styles selected from consecutive new referrals to the
and derive either a single predominant attachment Obsessive-Compulsive Disorder Clinic of the
type, or a profile of attachment style scores for an McGill University Health Centre on the basis of
individual (44–47). two semi-structured clinical interviews by the
The dimensional approach postulates that con- Clinic Director (DS) and psychometric criteria.
tinuously distributed constructs underlie attach- Subjects in the depressed group were out-patients
ment behaviour. This is the approach used in this at the same centre, referred by colleagues in
study. The advantages of using dimensional meas- response to a letter soliciting research subjects
urements over typological ones include increased with depression. These subjects underwent one
precision, validity, reliability and statistical power semi-structured clinical interview by the same
(6, 48). Collins and Read’s Revised Adult Attach- author (DS) to confirm the diagnosis of unipolar
ment Scale (RAAS) (6) permits measurement of depression. The comparison group was recruited
three attachment-related dimensions in adult rela- by billboard advertisement from a sample of
tionships: the capacity for closeness, the feeling general hospital employees, ranging in socio-eco-
that one can depend on others and anxiety about nomic status from blue-collar workers to profes-
being unloved or abandoned. sionals. All subjects were asked to fill out several
This paper examines the concurrent associations self-report measures, as well as a questionnaire
of attachment security (as measured by the eliciting demographic and historical data. After
RAAS), recollections of early parental interactions complete description of the study to the subjects,
(as measured by the PBI) and psychopathology in written informed consent was obtained. The study
adults with OCD and depression, compared with was approved by the Hospital Ethics Board.
healthy controls. To our knowledge, there have
been no previously reported studies on attachment
Measures
security specifically in OCD.
The RAAS (6) is an 18-item, self-report measure of
attachment style in romantic relationships. The
Aims of the study
subject is asked to read 18 statements and rate the
Our study had two main hypotheses. First, it was extent to which each describes his/her feelings
predicted that the two clinical groups would toward romantic relationships, on a Likert scale
demonstrate greater attachment insecurity than from 1 Ônot at all characteristic of meÕ to 5 Ôvery
the non-clinical group. Specifically, it was predic- characteristic of meÕ. Six items contribute to a score
ted that individuals with OCD and depression on each of three attachment dimensions: Close,
would report less comfort with closeness and Depend and Anxiety. The ÔCloseÕ subscale meas-
dependency, and greater relationship-centred anxi- ures the extent to which the individual is comfort-
ety than controls. able with closeness and intimacy, and includes such
Secondly, consistent with attachment theory and items as ÔI find it relatively easy to get close to
existing literature, we predicted that attachment peopleÕ. The ÔDependÕ subscale measures the extent
insecurity would be associated with recollections of to which the individual believes others can be
insensitive parenting i.e. recollections of less caring depended on when needed and includes the item:
and/or greater control by one or both parents. We ÔPeople are never there when you need themÕ. The
anticipated the OCD group would score higher on ÔAnxietyÕ subscale measures the extent of anxiety

449
Myhr et al.

about being abandoned or unloved ÔI often worry The psychometric inclusion criterion for normal
that my partner will not want to stay with meÕ. controls was a T score of 58 or lower on the GSI
Secure attachment is defined by comfort with index of the SCL-90-R. Controls had no history
closeness (high ÔCloseÕ scores), capacity to depend of serious psychiatric illness, routine use of
on others (high ÔDependÕ scores) and low relation- psychiatric medications, or psychiatric hospital-
ship anxiety (low ÔAnxietyÕ). The RAAS has ization.
demonstrated good internal consistency and test– Individuals with major physical illness, sub-
retest reliability (6, 26). stance dependence (apart from nicotine), organic
The PBI (29) is a 25-item self-report measure or psychotic symptoms were excluded from all
of a subject’s recollections of his or her parentsÕ groups. Of all subjects assessed for the study, only
rearing practices during his or her first 16 years. four were excluded. One subject in the OCD
It has 12 ÔcareÕ items and 13 ÔprotectionÕ items, group had concurrent bipolar affective disorder,
with each parent being scored on a 4-point one in the depressed group had severe multiple
Likert scale. The Care items reflect the extent to sclerosis. Two controls scored in the pathologic
which the subject found care and nurturance (e.g. range for depressive and OCD symptomatology
Ôspoke to me in a warm and friendly voiceÕ), as respectively, and were invited in for an interview
opposed to neglect and emotional rejection and referred for treatment. This low exclusion
(Ôseemed emotionally cold to meÕ). The Protection rate likely reflects the fact that OCD and
items reflect the dimension of rigid control (Ôtried depressed subjects were part of a referred popu-
to control everything I didÕ) as opposed to lation attending specialty clinics at our tertiary
fostering autonomy (Ôlet me decide things for care centre.
myselfÕ).
The Yale–Brown Obsessive-Compulsive Scale
Statistical analyses
(YBOCS) (49) is a clinician-rated measure of
obsessive and compulsive symptoms. Each of the The Statistical Package for the Social Sciences
10 items are scored from 0 ¼ no symptoms to (SPSS Inc., Chicago, IL, USA), version 10.0 for
4 ¼ extremely severe symptoms, yielding two sub- Windows, was used for data analysis. Between
scale scores – severity of obsessions and severity of group differences in demographic and clinical
compulsions. For this study, the total of the two characteristics were assessed using Chi-square
subscale scores was used. tests, t-tests, or one-way analysis of variance
The Symptoms Checklist-90-Revised (SCL-90- (ANOVA) as appropriate.
R) (50) is a self-report questionnaire that assesses The three Attachment outcomes (Close,
nine symptom dimensions. The anxiety and phobic Depend and Anxiety) and four PBI outcomes
subscales were used as measures of anxiety, and the (maternal care, maternal protection, paternal care
Global Severity Index (GSI) was used as a measure and paternal protection) were entered into a
of general psychopathology. multivariate analysis of variance (MANOVA)
The Beck Depression Inventory (BDI) (51) is an with a three factor design (diagnostic group,
extensively validated self-report measure of depres- gender and marital status). Given a significant
sive symptomatology. main effect of diagnostic group and marital status
in the omnibus MANOVA, and no significant
main effect or interaction involving gender,
Inclusion/exclusion criteria
gender was dropped from subsequent analyses.
Inclusion in the study was independent of past Group differences were assessed for each variable
treatment or duration of illness. The OCD sample with two-way univariate ANOVAs, with diagno-
met DSM-IV criteria. Because of the high rate of sis and marital status as factors. Significant
comorbid depressive symptomatology, OCD interactions were decomposed with simple main
patients with secondary depression (i.e. depressive effects tests using Datasim, Version 1.2. Post hoc
symptoms appearing after the onset of OCD) were pairwise comparisons were conducted using
included. All other patients with mixed Axis I Tukey’s Honestly Significant Difference (HSD)
diagnoses were excluded. test.
The depressed sample met DSM-IV criteria for The relationship of age and BDI scores to
major depressive disorder or dysthymia and had a outcome variables was assessed with linear regres-
minimum BDI score of 12. Depressed individuals sion and calculation of Pearson product–moment
with YBOCS scores >8, or T scores of >63 on correlation coefficient (r).
SCL-90-R anxiety or phobic subscales were exclu- All P-values reported are two-tailed, with signi-
ded. ficance level set at 0.05.

450
 Attachment in obsessive-compulsive disorder

Results 25% of the sample in the moderate or severe range.

To control for the effects of depression, linear
Subject characteristics
regression analysis was performed using BDI as a
Table 1 presents the demographic and clinical predictor of the outcome variables. There was no
characteristics of the subjects in all three groups. significant linear relationship between BDI and
There was no difference in gender between attachment or PBI measures in the OCD or
groups. A higher proportion of OCD patients depressed groups. Thus, depression level, as meas-
were single (75%) than depressed patients (50%) ured by the BDI, was not used as a covariate in
or controls (31%), which may reflect the younger subsequent analyses.
age of the OCD group. There were a small
number of divorced subjects – one OCD subject,
Attachment and PBI outcomes
two depressed subjects and one comparison sub-
ject – and a single widowed depressed subject. Three way multivariate analysis of the seven
These categories of marital status were combined outcome measures revealed a significant main
with the married individuals to form the ÔmarriedÕ effect of diagnostic group (Wilk’s lambda ¼ 0.678,
group for statistical purposes. This choice was F ¼ 1.84, df ¼ 14,120, P ¼ 0.04) and marital
based on the assumption that divorced or wid- status (Wilk’s lambda ¼ 0.688, F ¼ 3.87, df ¼
owed individuals would share attachment charac- 7,60, P ¼ 0.002), but not gender (Wilk’s
teristics with married individuals (for example, lambda ¼ 0.950, F ¼ 0.452, df ¼ 7,60, P ¼ 0.86).
having committed to a permanent relationship at There were no significant two or three way
one point). The married and single groups did not interactions.
differ in severity of psychopathology as measured The results of the univariate ANOVAs with two
by the YBOCS and the BDI in the OCD group factor design (diagnosis, marital status) for attach-
(t ¼ 0.43, df ¼ 34, P ¼ 0.67 and t ¼ 0.44, df ¼ ment outcomes are shown in Fig. 1.
34, P ¼ 0.66 respectively), by the BDI in the On the whole, clinical groups demonstrated
Depressed group (t ¼ 1.16, df ¼ 14, P ¼ 0.27), more attachment insecurity than the comparison
and by SCL-90R/GSI in the controls (t ¼ 0.378, group. The OCD and depression groups had
df ¼ 24, P ¼ 0.71). higher relationship anxiety than the comparison
Regression analyses revealed no significant group (F ¼ 6.34, df ¼ 2,72, P ¼ 0.003), but did
linear relationships between age and any of the not differ significantly from each other. Diagnostic
attachment or PBI outcomes. Thus, age was not group and marital status interacted significantly on
used as a covariate in subsequent analyses. the capacity to depend on others (F ¼ 4.68,
Only OCD patients reported psychopathology df ¼ 2,72, P ¼ 0.012), with depressed married
on the YBOCS. BDI scores were significantly individuals less comfortable with dependence
higher for depressed than controls. There was a than married OCD or comparison individuals,
trend towards higher mean BDI scores for OCD and single OCD individuals less comfortable with
patients compared with controls, reflecting the dependence than singles in the comparison group.
admission of patients with secondary depressive The capacity for closeness in OCD, depressed and
symptoms (P ¼ 0.072). In fact, 69% of the OCD comparison groups was comparable (F ¼ 2.44,
sample had BDI scores in the depressed range, with df ¼ 2,72, P ¼ 0.094).

Table 1. Demographic and clinical characteristics of samples with obsessive-compulsive disorder (OCD), depression and comparison subjects

Depression Comparison
OCD (n ¼ 36) (n ¼ 16) (n ¼ 26) Analysis

Variable n % n % n % Chi-squared df P-value

Female gender 17 47.2 12 75 18 69.2 4.89 2 0.087

Single 27 75 8 50 8 31 12.15 2 0.002
Mean SD Mean SD Mean SD F df P-value Post hoc

Age (years) 29.25 9.12 40.5 9.99 37.35 8.18 10.95 2, 75 <0.001 O < C*, O < D**
YBOCS score 19.28 6.69 1.56 2.63 0.31 0.79 147.72 2,75 <0.001 O > D**, O > C**
BDI score 14.33 8.59 18.69 5.46 2.19 2.3 39.86 2,75 <0.001 D > C**

*P ¼ 0.002, **P < 0.001, Tukey's honestly significantly different test.

Total n ¼ 78.
YBOCS, Yale–Brown Obsessive-Compulsive Scale; BDI, Beck Depression Inventory.

451
Myhr et al.

‘Close’ Dimension position between OCD and depressed groups, but

30
Mean score for ‘close’ did not significantly differ from either. The lowest
25 mean care scores for both parents were found in
20 the depressed group, but the result for fathers did
15
not reach statistical significance. Maternal and
paternal protection scores showed no significant
10
Marital statusa variation with diagnosis.
5 Single The effect of marital status on PBI scores was
0 Married surprising. Married individuals recalled their
Comparison OCD Depression
Diagnosis
mothers to be less caring and both parents to be
a
Significant main effect for marital status (F = 10.61, df = 1,72, P = 0.002) more controlling than singles across diagnostic
groups. Marital status did not affect Ôfather careÕ
‘Depend’ dimension scores.
30
Mean score for ‘depend’

25

20
Discussion
15 This study explored the associations between
10
psychopathology, attachment security and recol-
lections of parental interactions. As predicted,
5 Singleb OCD and depression were associated with insecure
0 Marriedc
Comparison OCDa Depression
attachment: OCD and depressed subjects reported
Diagnosis more relationship anxiety, and depressed subjects
a
Significant effect of marital status for OCD group (F = 8.21, df = 1,72, and single OCD subjects reported more discomfort
P = 0.006). bFor singles, significant difference between OCD and comparison
group (P < 0.01). cFor married, significant difference between depression and OCD
with dependence than did individuals in the
groups (P < 0.01), and between depression and comparison groups (P < 0.01) comparison group. Contrary to our expectations,
few group differences were found in parental
‘Anxiety’ dimension bonding.
20 This is the first study, to our knowledge, that
Mean score for ‘anxiety’

demonstrates greater attachment insecurity in a

15 pure group of OCD subjects – even when con-
trolled for the level of comorbid depression. This is
10 not surprising. A central construct in the cognitive
conceptualization of anxiety disorders involves the
5
notion of personal vulnerability. Anxious individ-
Single
0 Married
uals tend to exaggerate threats from both external
Comparison OCD Depression and internal sources, and underestimate their
Diagnosisa capacity to cope with them (52). The nature of
a
Significant difference between OCD and comparison groups (P < 0.001),
and between depression and comparison groups (P = 0.002).
the threat, and the individual’s cognitive and
Total N = 78, Comparison group (n = 26), OCD group (n = 36), Depression behavioural responses to it, vary from one anxiety
group (n = 16)
disorder to another.
Fig. 1. Adult attachment scale dimensions in obsessive-com- Observing this sense of vulnerability in anxiety
pulsive disorder (OCD), depressed and comparison groups. has lead theorists to consider the influence of
attachment in the genesis of anxiety disorders (53).
Guidano (54) and later Shear (55) have suggested
Being married was associated with greater that insecure attachment may explain difficulties
attachment security. In all groups, married indi- with separation, hypersensitivity to physical sensa-
viduals displayed a greater capacity for closeness tions and attempts to overcontrol interpersonal
and a trend towards lower relationship anxiety interactions found in panic disorder and agora-
(F ¼ 3.67, df ¼ 1,72, P ¼ 0.059). In the OCD phobia. Dugas et al. (56) have postulated that
group, married individuals depended on others insecure attachment may be the root of the intol-
more easily than singles. erance of uncertainty distinguishing generalized
The PBI results are shown in Table 2. The only anxiety disorder. Chorpita and Barlow (57) have
significant effect of diagnostic group was that OCD suggested that uncaring or overcontrolling attach-
subjects reported more caring mothers than did ment relationships could foster a tendency to
depressed subjects. The mean Ômother careÕ score of process events as out of one’s control, leading to
the comparison group occupied an intermediate the ÔhelplessnessÕ of anxiety, and, further along the

452
 Attachment in obsessive-compulsive disorder

Table 2. Parental Bonding Instrument Scale scores in obsessive-compulsive disorder (OCD), depressed and comparison groups

Depression Comparison
OCD (n ¼ 36) (n ¼ 16) n ¼ (26)a Analysis

Variable Mean SD Mean SD Mean SD F df P-value Post hoc

Mother care
Group total 28.8 7.7 20.1 12.4 25.0 9.6 3.51 2,72 0.035 O > D*
Marital status 10.01 1,72 0.002
Single 30.1 6.9 26.2 8.2 27.5 8.6
Ever married 24.8 8.9 13.9 13.3 23.8 10.1
Mother protect
Group total 15.3 8.2 13.8 10.0 12.4 7.8 1.17 2,72 ns
Marital status 6.29 1,72 0.014
Single 14.6 8.6 7.9 5.9 12.0 5.3
Ever married 17.3 7.1 19.8 10.0 12.6 8.9
Father care
Group total 21.5 10.3 17.9 9.4 22.1 10.7 1.13 2,71 ns
Marital status 0.54 1,71 ns
Single 22.2 9.6 16.8 7.5 25.6 10.2
Ever married 19.2 12.5 19.1 11.3 20.4 10.8
Father protect
Group total 11.8 6.4 14.6 8.9 10.6 8.4 1.42 2,71 ns
Marital status 5.06 1,71 0.028
Single 10.9 5.5 10.1 7.5 10.5 9.1
Ever married 14.6 8.5 19.1 8.2 10.6 8.4

* P ¼ 0.005, Tukey's Honestly Significant Difference.

a
n ¼ 25 for father care and protect scores because one subject had no father figure in childhood.
ns, not significant.

Ôlack of controlÕ continuum, the ÔhopelessnessÕ of is consistent with the finding of Vogel et al. (41)
depression. who also found depressed but not OCD subjects
In the area of OCD, Sookman et al. (58) have differed from controls.
shown that greater perception of personal vulner- The only significant group difference was that
ability, difficulty with unpredictability and greater depressed patients remembered their mothers as
need for control distinguishes OCD from other being less caring than OCD patients did. Low
anxiety disorders. They also found that viewing maternal care in the depressed group replicates
strong emotions as dangerous and feeling unable to earlier studies (31, 37–40). Failure to find significant
cope with them was a characteristic of anxiety differences with the control group may reflect the
disorders in general. Insecure attachment may be small size of the depressed sample. This may also
one of the factors that contributes to perceived explain the lower, but statistically non-significant
vulnerability and predisposes to OCD. Negative paternal caring scores found in the depressed group.
expectations about the availability or responsive- One puzzling finding was the contradictory effect
ness of others coupled with intolerance of painful of marital status on attachment and PBI scores.
affects and decreased capacity to soothe them may Married individuals in all groups had significantly
predispose an individual to respond to obsessions greater capacity for closeness and a trend towards
with heightened anxiety, catastrophic appraisals less anxiety in relationships. Married status aug-
and efforts to reduce fear with ritualization and mented comfort with dependence in OCD, but
avoidance. not in depression. However, while married indi-
Parental sensitivity to a child’s ÔsignalsÕ is felt to viduals demonstrate greater attachment security,
be an important factor in the development of they also rate their mothers as less caring, and
attachment security (9). Our second prediction – both parents as more controlling than singles. One
that recollections of child–parent interactions would expect the attachment security found in
would reflect the greater attachment insecurity married individuals to be associated with recollec-
found in clinical groups – was not strongly suppor- tions of better parenting. Truant (59) found an
ted. Attachment insecurity in the OCD group was association between high care scores and marital
not associated with recollections of significantly quality but did not explore the question of marital
less care or more controlling interactions than the status per se.
control group. This finding is opposite to what we The counterintuitive effect of marital status on
expected and to what Turgeon et al. (42) found. It PBI scores is intriguing. Could there be a level of

453
Myhr et al.

ÔcontrollingÕ parental behaviour that contributes to Read (6) reported a Pearson’s r of 0.38 for the
attachment security, which has the effect of indi- Close and Depend subscales, while our study
viduals seeking out committed relationships in found a correlation of r ¼ 0.55. Although the
adulthood? Can it be that some singles have subscales are correlated, they may not measure
memories of greater care from their mothers, interchangeable aspects of ÔAvoidanceÕ, which
which make it difficult for them to engage in the might explain the finding of group differences for
more symmetric and reciprocal attachments of Depend, but not Close scores.
adulthood? Or does the answer lie in the process of We did not attempt to convert dimensional
recall itself? As Main (60) has pointed out, securely scores to discrete attachment styles for individuals.
attached individuals provide more coherent, cred- While this has been performed by Collins and
ible accounts of their early attachment relation- Read (6) using advanced statistical techniques, the
ships – whether these were good or bad. If married loss of precision and power in this process could
individuals are more secure, they may report more have obscured group differences. Studies attempt-
realistic views of their parents than insecure singles. ing to link clinical diagnosis with type of insecurity
Or is it possible that while the PBI reflects have been for the most part inconclusive (17, 18).
something of early attachment-related interactions, Furthermore, we are unable to ascertain from this
it may not measure those aspects of early interac- cross sectional study whether differences noted in
tions essential for later attachment security? attachment dimensions reflect longstanding attach-
The greater security seen in married subjects is ment-related attitudes towards personal rela-
easier to conceptualize. Married status may reflect tionships, or state-dependent factors, such as
the selection of individuals who had more secure generalized discouragement in depression or heigh-
attachment orientations to begin with and who can tened anxiety in OCD. Items such as ÔI often worry
negotiate intimacy with greater comfort and suc- that my partner will not want to stay with meÕ may
cess. Alternatively, there is some evidence that be more salient when one is in the throes of
being in a marriage may augment attachment secur- increased OCD or depressive symptomatology. It
ity over time (61, 62), perhaps by exposing indivi- would be interesting to see if attachment scores
duals to new relational interactions. The remain stable upon reassessment after amelioration
differential effect of marriage in clinical samples of clinical symptoms. A central tenet of attachment
on dependency may reflect the cognitive set of theory suggests that attachment style is a stable
depression where individuals are hopeless about construct. However, it is not known if self-report
themselves, the world and others, and are no more methods of inferring attachment style are more
confident about others being there for them than susceptible to state dependent fluctuation than
are depressed singles. interview methods (64). As previously discussed,
This study has limitations that affect generaliz- PBI scores have more long-term reliability and
ability of findings. As was mentioned, the depressed independence of mood state.
group was small. Furthermore, the mean BDI Self-report measures of current interpersonal
score of the depressed group was only 19 (±5), attitudes and recollections of early parent–child
which suggests only mild-moderate pathology; this relationships are imperfect ways to assess the phe-
may have played a role in underestimating group nomenon of adult attachment – with complex
differences. On the contrary, the inclusion of cognitive, emotional and behavioural components,
dysthymic patients may have augmented overall both conscious and unconscious. Nonetheless, this
mood group insecurity, as Fonagy et al. (18) found study demonstrated an association between inse-
that individuals with dysthymia were more likely to cure attachment and OCD in adults. The observed
be insecure (90%) than those with major depressive lack of a robust relationship between attachment
disorder (60%). security and PBI recollections suggests that the PBI
Dimensional measures were used in this study to may not measure those aspects of early interactions
infer degree of attachment security. In the last essential for later attachment security, or if it does,
decade, attachment researchers have established that attachment security may affect the nature of
that it is likely that two, rather than three, memory itself.
dimensions underlie adult attachment orientations
in intimate relationships – one involving Avoid-
Acknowledgements
ance (discomfort with closeness and dependency)
and the other Anxiety (about abandonment) (48, The authors thank Joseph Rochford PhD, and Lawrence
63). For the RAAS, this means that the ÔCloseÕ and Annable Dip. Stat. for statistical advice, and Jeanne Talbot,
PhD, MD, CM, FRCP(C) and Howard Margolese MD, CM,
ÔDependÕ subscales could be considered aspects of FRCP(C) for their review of the manuscript.
the Avoidance dimension (63). Indeed, Collins and

454
 Attachment in obsessive-compulsive disorder

References 22. Sperling MB, Sharp JL, Fishler PH. On the nature of
attachment in a borderline population: a preliminary
1. Ainsworth MDS, Blehar MC, Waters E, Wall S. Patterns
investigation. Psychol Rep 1991;68:543–546.
of attachment: a psychological study of the Strange Situ-
23. Brennan KA, Shaver PR. Attachment styles and personality
ation. Hillsdale, NJ, USA: Erlbaum, 1978.
disorders: their connections to each other and to parental
2. Bowlby J. Attachment and loss: vol 1. Attachment. New
divorce, parental death, and perceptions of parental care-
York: Basic Books, 1982.
giving. J Pers 1998;66:835–878.
3. Main M, Kaplan N, Cassidy J. Security in infancy, child-
24. Evans L, Wertheim EH. Intimacy patterns and relationship
hood, and adulthood: a move to the level of representa-
satisfaction of women with eating problems and the
tion. In: Bretherton I, Waters E, eds. Growing points in
mediating effects of depression, trait anxiety and social
attachment theory and research. Monographs of the
anxiety. J Psychosom Res 1998;44:355–365.
Society for Research in Child Development, 1985:66–104.
25. Muller RT, Sicoli LA, Lemieux KE. Relationship between
4. Klohnen EC, John OP. Working models of attachment: a
attachment style and posttraumatic stress symptomatology
theory-based prototype approach. In: Simpson JA, Rholes
among adults who report the experience of childhood
WS, eds. Attachment theory and close relationships. New
abuse. J Trauma Stress 2000;13:321–332.
York: Guilford Press, 1998:115–140.
26. Burge D, Hammen C, Davila J et al. The relationship be-
5. Waters E, Merrick S, Treboux D, Crowell J, Albersheim L.
tween attachment cognitions and psychological adjustment
Attachment security in infancy and early adulthood: a
in late adolescent women. Dev Psychopathol 1997;9:151–
twenty-year longitudinal study. Child Dev 2000;71:684–689.
167.
6. Collins NL, Read SJ. Adult attachment, working models,
27. Liotti G, Pasquini P and the Italian group for the study of
and relationship quality in dating couples. J Pers Soc
dissociation. Predictive factors for borderline personality
Psychol 1990;58:644–663.
disorder: patientsÕ early traumatic experiences and losses
7. Ognibene TC, Collins NL. Adult attachment styles, per-
suffered by the attachment figure. Acta Psychiatr Scand
ceived social support and coping strategies. J Soc Pers
2000;102:282–289.
Relat 1998;15:323–345.
28. Pasquini P, Liotti G, Mazzotti E, Fassone G, Picardi A, the
8. Collins NL, Feeney BC. A safe haven: an attachment theory
Italian group for the study of dissociation. Risk factors in
perspective on support seeking and caregiving in intimate
the early family life of patients suffering from dissociative
relationships. J Pers Soc Psychol 2000;78:1053–1073.
disorders. Acta Psychiatr Scand 2002;105:110–116.
9. Seigel DJ. Attachment. The developing mind. New York:
29. Parker G, Tupling H, Brown LB. A parental bonding
Guilford Press, 1999:67–120.
instrument. Br J Med Psychol 1979;52:1–10.
10. Amini F, Lewis T, Lannon R et al. Affect, attachment,
30. Parker G. The parental bonding instrument: psychometric
memory: contributions toward psychobiologic integration
properties reviewed. Psychiatr Dev 1989;7:317–335.
[Review] [113 refs]. Psychiatry 1996;59:213–239.
31. Gotlib IH, Mount JH, Cordy NI, Whiffen VE. Depression
11. Waters E, Cummings EM. A secure base from which to
and perceptions of early parenting: a longitudinal investi-
explore close relationships. Child Dev 2000;71:164–172.
gation. Br J Psychiatry 1988;152:24–27.
12. Crittendon PM. Relationships at risk. In: Belsky J, Nez-
32. Neale MC, Walters E, Health AC et al. Depression and
worski T, eds. Clinical implications of attachment. Hills-
parental bonding: cause, consequence, or genetic covari-
dale, NJ, USA: Erlbaum, 1988.
ance? [see comments]. Genet Epidemiol 1994;11:503–522.
13. Hesse E., Main M. Second-generation effects of unresolved
33. Enns MW, Cox BJ, Clara I. Parental bonding and adult
trauma as observed in non-maltreating parents: dissoci-
psychopathology: results from the US national comor-
ated, frightened and threatening parental behavior. Psy-
bidity survey. Psychol Med 2002;32:997–1008.
choanal Inquiry 1999;19:481–540.
34. Mackinnon AJ, Henderson AS, Scott R, Duncan-Jones P. The
14. Schore AN. Early organization of the nonlinear right brain
Parental Bonding Instrument (PBI): an epidemiological
and development of a predisposition to psychiatric disor-
study in a general population sample. Psychol Med 1989;
ders. Dev Psychopathol 1997;9:595–631.
19:1023–1034.
15. Goldberg D. Vulnerability, destabilization and restitution
35. Reti IM, Samuels JF, Eaton WW, Bienvenu III OJ, Costa Jr
in anxious depression. Acta Psychiatr Scand 2003;108
PT, Nestaldt G. Adult antisocial personality traits are
(Suppl. 418):81–82.
associated with low parental care and maternal overpro-
16. Mickelson KD, Kessler RC, Shaver PR. Adult attachment in
tection. Acta Psychiatr Scand 2002;106:126–133.
a nationally representative sample. J Pers Soc Psychol
36. Cavedo LC, Parker G. Parental Bonding Instrument:
1997;73:1092–1106.
exploring for links between scores and obsessionality. Soc
17. Van Ijzendoorn MH, Bakermans-Kranenburg MJ. Attach-
Psychiatry Psychiatr Epidemiol 1994;29:78–82.
ment representations in mothers, fathers, adolescents, and
37. Parker G, Hadzi-Pavlovic D. Parental representations of
clinical groups: a meta-analytic search for normative data.
melancholic and non-melancholic depressives: examining
J Consult Clin Psychol 1996;64:8–21.
for specificity to depressive type and for evidence of
18. Fonagy P, Leigh T, Steele H et al. The relation of attach-
additive effects. Psychol Med 1992;22:657–665.
ment status, psychiatric classification, and response to
38. Mackinnon A, Henderson AS, Andrews G. Parental ÔÔaffec-
psychotherapy. J Consult Clin Psychol 1996;64:22–31.
tionless controlÕÕ as an antecedent to adult depression: a
19. West M, Rose MS, Verhoef MJ, Spreng S, Bobey M. Anxious
risk factor refined. Psychol Med 1993;23:135–141.
attachment and self-reported depressive symptomatology
39. Duggan C, Sham P, Minne C, Lee A, Murray R. Quality of
in women. Can J Psychiatry 1998;43:294–297.
parenting and vulnerability to depression: results from a
20. Bifulco A, Moran PM, Ball C, Lillie A. Adult attachment
family study. Psychol Med 1998;28:185–191.
style. II: Its relationship to psychosocial depressive-vulner-
40. Enns MW, Cox BJ, Larsen DK. Perceptions of parental
ability. Soc Psychiatry Psychiatr Epidemiol 2002;37:60–67.
bonding and symptom severity in adults with depression:
21. De Ruiter C, Van Ijzendoorn MH. Agoraphobia and
mediation by personality dimensions. Can J Psychiatry
anxious-ambivalent attachment: an integrative review.
2000;45:263–268.
J Anxiety Disord 1992;6:365–381.

455
Myhr et al.

41. Vogel PA, Stiles TC, Nordahl HM. Recollections of par- 54. Guidano V. The agoraphobic cognitive organization.
ent–child relationships in OCD out-patients compared to Complexity of the self. New York: Guilford Press, 1987:
depressed out-patients and healthy controls. Acta Psychi- 139–155.
atr Scand 1997;96:469–474. 55. Shear MK. Factors in the etiology and pathogenesis of
42. Turgeon L, O’Connor KB, Marchand A, Freeston MH. panic disorder: revisiting the attachment-separation para-
Recollections of parent–child relationships in patients with digm [Review] [64 refs]. Am J Psychiatry 1996;153:125–
obsessive-compulsive disorder and panic disorder with 136.
agoraphobia. Acta Psychiatr Scand 2002;105:310–316. 56. Dugas MJ, Buhr K, Ladouceur R. The role of intolerance of
43. Nickell AD, Waudby CJ, Trull TJ. Attachment, parental uncertainty in the etiology and maintenance of generalized
bonding and borderline personality disorder features in anxiety disorder. In: Heimberg RG, Turk CL, Mennin DS,
young adults. J Personal Disord 2002;16:148–159. eds. Generalized anxiety disorder: advances in research
44. Hazan C, Shaver P. Romantic love conceptualized as an and practice. New York: Guilford Press, 2003.
attachment process. J Pers Soc Psychol 1987;52:511–524. 57. Chorpita BF, Barlow DH. The development of anxiety: the
45. Bartholomew K. Avoidance of intimacy: an attachment role of control in the early environment. Psychol Bull
perspective. J Soc Pers Relat 1990;7:147–178. 1998;124:3–21.
46. Bartholomew K, Horowitz LM. Attachment styles among 58. Sookman D, Pinard G, Beck AT. Vulnerability schemas in
young adults: a test of a four-category model. J Pers Soc obsessive-compulsive disorder. J Cogn Psychother 2001;15:
Psychol 1991;61:226–244. 109–130.
47. Griffin DW, Bartholomew K. Models of the self and other: 59. Truant GS. Personality diagnosis and childhood care
fundamental dimensions underlying measures of adult associated with adult marital quality. Can J Psychiatry
attachment. J Pers Soc Psychol 1994;67:430–445. 1994;39:269–276.
48. Fraley RC, Waller NG. Adult attachment patterns: a test 60. Main M. The organized categories of infant, child, and
of the typological model. In: Simpson JA, Rholes WS, eds. adult attachment: flexible vs. inflexible attention under
Attachment theory and close relationships. New York: attachment-related stress. J Am Psychoanal Assoc 2000;48:
Guilford Press, 1998:77–114. 1055–1096.
49. Goodman WK, Price LH, Rasmussen SA et al. The Yale– 61. Grossmann KE, Grossmann K. Attachment quality as an
Brown Obsessive-Compulsive Scale. I. Development, use, organizer of emotional and behavioral responses in a
and reliability. Arch Gen Psychiatry 1989;46:1006–1011. longitudinal perspective. In: Parkes CM, Stevenson-Hinde J,
50. Derogatis LR. SCL-90-R: administration, scoring, and Marris P, eds. Attachment across the life cycle. London:
procedures manual, 3rd edn. Towson, MD, USA: Clinical Tavistock/Routledge, 1991:93–114.
Psychometric Research, 1994. 62. Crowell JA, Treboux D, Waters E. Stability of attachment
51. Beck AT, Ward CH, Mendelson M, Mock J, Erbaugh J. An representations: the transition to marriage. Dev Psychol
inventory for measuring depression. Arch Gen Psychiatry 2002;38:467–479.
1961;4:561–571. 63. Brennan KA, Clark CL, Shaver PR. Self report measure-
52. Beck AT, Emery G, Greenberg R. Anxiety disorders and ment of adult attachment: an integrative overview. In:
phobias: a cognitive perspective. New York: Basic Books, Simpson JA, Rholes WS, eds. Attachment theory and close
1985. relationships. New York: Guilford Press, 1998.
53. De Ruiter C. Anxious attachment in agoraphobia 64. Crowell JA, Treboux D, Waters E. The Adult Attach-
and obsessive-compulsive disorder: a literature review ment Interview and the Relationship Questionnaire: rela-
and treatment implications. In: Perris C, Arrindel WA, tions to reports of mothers and partners. Pers Relatsh
Eisemann M, eds. Parenting and psychopathology. 1999;6:1–18.
Chichester: John Wiley, 1994:281–307.

456