Elsevier

Journal of Affective Disorders

Volume 236, 15 August 2018, Pages 274-290
Journal of Affective Disorders

Review article
Adult attachment representations and depressive symptoms: A meta-analysis

https://doi.org/10.1016/j.jad.2018.04.091Get rights and content

Highlights

  • Attachment theory offers a framework for understanding the etiology of depression.

  • Studies linking attachment and depressive symptoms have shown mixed results.

  • Preoccupied and unresolved attachment were associated with increased depression.

  • Attachment may be a potential target for preventing or treating depression.

Abstract

Background

Attachment theory has long argued that insecure and disorganized attachment representations are associated with vulnerability to psychopathology in general, and depressive symptoms in particular. However, studies assessing the link between insecure and disorganized attachment and depressive symptoms report inconsistent results. The primary aim of this meta-analysis was to examine the extent to which adult attachment representations are associated with depressive symptoms.

Methods

We conducted a literature search for published studies using PsycINFO, MEDLINE, and ProQuest in March 2017, identifying 55 samples (N = 4,386). Eligible studies assessed depression using a well-validated self-report or clinical diagnostic measure of depression and attachment using the Adult Attachment Interview.

Results

Results demonstrated that insecure individuals had higher levels of depression than secure-autonomous individuals (d = 0.21, 95% CI [0.08–0.33]). Specifically, insecure-preoccupied individuals (d = 0.48, 95% CI [0.30–0.65]), but not insecure-dismissing individuals (d = 0.09, 95% CI [−0.03–0.22]), exhibited significantly more depressive symptoms than the secure-autonomous individuals. Additionally, unresolved adults had higher levels of depression than adults with organized attachment classifications (d = 0.29, 95% CI [0.13–0.44]).

Limitations

Due to the nature of the studies reviewed, causal inferences regarding the links found between adult attachment representations and depressive symptoms may not be drawn. Further, the aggregate effect sizes do not solely reflect the magnitude of the association between attachment representations and clinically-diagnosed major depressive disorder.

Conclusions

The results of this meta-analysis have important conceptual and practical implications for attachment-based prevention and intervention efforts, as they suggest that attachment representations, specifically insecure-preoccupied and unresolved, are associated with depressive symptoms in adulthood. Future research using longitudinal and intervention-based designs should examine the extent to which insecure-preoccupied and unresolved attachment lead to increased depressive symptoms, as well as the mechanisms by which these attachment patterns across the life span may increase vulnerability to depressive symptoms in adulthood.

Introduction

The core symptoms of sad mood, diminished interest or joy in daily activities, disrupted sleep routines, disturbed appetite, fatigue, loss of ability to concentrate, and excessive feelings of guilt and worthlessness (American Psychiatric Association, 2013) currently make depression the leading cause of disability in people age 15–44 in the United States (Greenberg et al., 2015). With residual symptoms such as cognitive impairment and social dysfunction, depression is expected to become the leading cause of health loss attributable to disease and injury (i.e., disease burden) worldwide by the year 2030 (Mathers and Loncar, 2006). Given the devastating impact of depression on both the individual and society, it is critical to better understand factors that may be linked to depressive symptoms, as this knowledge may inform clinical assessment, prevention, and intervention efforts.

Individuals with depression often experience a diverse range of social and behavioral difficulties that inhibit their ability to form and maintain relationships (Hames et al., 2013, Tse and Bond, 2004). Given that individual differences in early life attachment patterns are thought to be carried forward and shape our relationships with close others (e.g., peers and romantic partners; Bowlby, 1988, Feeney, 2008, Groh et al., 2014, Holland and Roisman, 2010) and unfamiliar others (Roisman, 2006), attachment theory offers an important framework for understanding the etiology and development of depressive symptoms across the life span.

According to attachment theory (Bowlby, 1969/1982), infants develop secure attachments to their primary caregivers when they are met with consistent sensitive responses during times of distress. Securely attached infants and children are able to maintain a sense of safety derived from physical or perceived proximity to caregivers, which provides them with the necessary confidence to freely explore their environment. However, when caregiving is inconsistent, inattentive, and/or characterized by rejection and neglect, infants may develop insecure attachment patterns (Ainsworth et al., 1978/2015), which entail negative representations of self as unlovable and others as unloving. Negative representations of self and others, in turn, may lead to patterns of interacting with others in overly distancing or demanding ways (Kobak and Sceery, 1988), thereby increasing the probability of experiencing real or perceived interpersonal conflicts and reducing the availability of social support. Thus, insecure attachment may set the stage for processing later relational experiences as stressful and unsatisfying, which may confer risk for depression across the life span (Cummings and Cicchetti, 1990, Hames et al., 2013, Hammen, 2000, Lakey and Orehek, 2011, Williams and Riskind, 2004).

From an attachment theory perspective, psychopathology - including depression - is defined as a deviation from a normal developmental pathway in an attempt toward adaption (Carlson and Sroufe, 1995). In the context of early attachment to caregivers, the most common precursor for such deviation toward adaptation is the “loss of the emotional ties” that infants naturally form with their primary caregivers (Bowlby, 1988). Although the loss of emotional ties to caregivers may include the literal absence of the caregiver (e.g., due to death), it may also include the experience of a caregiver who is unresponsive, rejecting, or overinvolved in the context of the infant's signals of distress. In such rearing environments, where attachment figures are perceived to be or are objectively absent, adaptive strategies to regulate one's own emotional distress are developed (Belsky, 1997a, Main, 1981). Early in life, these adaptive responses result in one of two ‘second best’ relational strategies: insecure-avoidant or insecure-resistant (Main, 1990).

In infancy, attachment strategies are assessed with the Strange Situation Procedure (Ainsworth et al., 1978/2015), which involves a series of brief separations and reunions between the infant and his or her caregiver, serving to activate the attachment system. The infant's behaviors upon reunion are coded along a number of scales that reflect the extent to which infants use their caregiver as a means for regulating distress. The insecure-avoidant strategy, which is thought to either serve to maintain proximity to the caregiver while not driving her away (Main, 1981), or avoid proximity altogether (Mikulincer and Shaver, 2012), is characterized by distant, self-reliant behavior when distressed in the presence of the caregiver (Ainsworth et al., 2015, Main, 2000); in other words, insecure-avoidant infants minimize their displays of distress. In contrast, the insecure-resistant strategy, which serves to attain greater proximity to caregivers in order to ensure more consistent attention and care (Cassidy and Berlin, 1994, Main and Solomon, 1986), is characterized by expressing anger or resistance towards the caregiver while simultaneously seeking proximity (Hesse and Main, 2000). Insecure-resistant infants, thus, maintain proximity by maximizing their displays of distress. Along with secure attachment, these insecure strategies reflect “organized” attachment patterns, given that they are developed and consolidated around the caregiver's consistent availability and responsiveness, or lack of thereof.

In contrast to the organized patterns of attachment (i.e., secure, insecure-avoidant, insecure-resistant), some infants develop disorganized attachment patterns exhibited by the absence or breakdown of a regulatory strategy (Main and Solomon, 1990). Often characterized by odd or anomalous behaviors, freezing or stilling for a period of time, or simultaneous displays of avoidance and proximity-seeking during the Strange Situation Procedure (Hesse and Main, 2006), attachment disorganization typically results from parental abuse or neglect, frightening parental behavior, or elevated risk (Cyr et al., 2010, Schuengel et al., 1999). Accordingly, disorganized attachment behaviors are thought to reflect an “unsolvable dilemma,” where infants perceive their caregiver as both a haven of safety (due to the infant's natural disposition to approach caregivers when threatened) and source of threat (Hesse and Main, 2006). As a result, a breakdown in coping strategies reduces the chance for successful adaptation in non-favorable caregiving environments.

Over time, the attachment system is theorized to integrate and consolidate past affective and behavioral experiences into a generalized relational schema that may have a direct effect on adult functioning (Sroufe, 2005, Sroufe et al., 2005a, Sroufe et al., 2005b). Relational schemas, also referred to as internal working models (Bowlby, 1973, Bowlby, 1982), are thought to guide interpersonal interactions by eliciting emotions, cognitions, behaviors, and expectations (Bretherton, 1991, Bretherton and Munholland, 2008, Main et al., 1985).

A variety of instruments spanning several assessment methods (e.g., self-report, interview) have been developed to measure internal working models in adulthood. In the developmental literature, the semi-structured Adult Attachment Interview (AAI; George et al., 1985) has been routinely administered and is considered the gold standard in adult attachment assessment. The AAI has demonstrated good reliability, stability, and discriminant validity (Bakermans-Kranenburg and van IJzendoorn, 1993, Waters et al., 2000). In addition, the AAI has been linked to the experience of early maternal sensitivity (Haydon et al., 2014), parenting quality with one's own children (Van IJzendoorn, 1995), and romantic relationship functioning in adulthood (Crowell et al., 2002, Holland and Roisman, 2010). The AAI is designed to elicit discourse related to past experiences with one's primary caregivers, including possible trauma or loss events, and how they may have influenced one's development and current personality. Narratives are classified into attachment categories based mostly on the coherence and manner of speech rather than their content (Hesse, 2016), independent of the speaker's verbal ability and general intelligence (Bakermans-Kranenburg and van IJzendoorn, 1993).

Similar to secure infants whose behavior is characterized by free exploration of their environment, secure-autonomous adults exhibit “free” evaluation of past experiences with their caregivers without becoming emotionally entangled with these experiences. Secure-autonomous discourse during the AAI is therefore characterized by a coherent and collaborative manner of speech (Hesse, 2016). Similar to the insecure-avoidant infants, adults who are classified as insecure-dismissing on the AAI exhibit inhibition of emotional arousal and are thought to detach themselves from fully processing potential interpersonal threat or pain (Main, 1981). This representational pattern, commonly referred to as a ‘deactivating strategy’ (Shaver and Mikulincer, 2007), is reflected in the idealization of caregivers, insistence on lack of memory of childhood events, and/or derogation of attachment experiences or relationships (Hesse, 2016). Similar to insecure-resistant infants, adults classified as insecure-preoccupied on the AAI exhibit a tendency to exaggerate expressed emotions and become overinvolved in the availability of others. These individuals present with heightened expression of anger towards one or more caregivers, and/or passivity or vagueness regarding attachment relationships (Hesse, 2016), a behavioral pattern commonly referred to as a ‘hyperactivating strategy’ (Shaver and Mikulincer, 2007). Adult attachment representations may be classified as organized (i.e., secure-autonomous, insecure-dismissing, and insecure-preoccupied) or unresolved. Similar to disorganized attachment in infancy, characterized by disoriented or fear responses when the attachment system is activated, adults who manifest disoriented discourse during discussion about past trauma of loss or abuse are classified as unresolved (Hesse and Main, 2006).

Although extensive research suggests that insecure and disorganized attachment are associated with internalizing symptoms, including depression symptoms, in childhood (Fearon et al., 2010, Groh et al., 2016), empirical findings supporting this link in adulthood are inconsistent (Stovall-McClough and Dozier, 2016). Some studies have found no significant difference in reported depressive symptoms between adults classified as secure-autonomous and insecure (e.g., Dickstein et al., 2004), whereas other studies have found that adults classified as insecure (both dismissing and preoccupied) have higher levels of depressive symptoms or are more likely to be clinically depressed than those classified as secure-autonomous. For example, McMahon et al. (2006) found that mothers who were insecure were more likely to report having chronic postpartum depression than secure-autonomous mothers. As in the case of attachment insecurity, studies assessing the link between attachment disorganization and depression reported mixed results. Although some studies have found an association between unresolved attachment and depression (e.g., Fonagy et al., 1996), others have not (e.g., Patrick et al., 1994, Rosenstein and Horowitz, 1996).

When comparing the types of insecure attachment (i.e., insecure-dismissing, insecure-preoccupied) to secure-autonomous attachment, results have been mixed. Some studies have found that depressive symptoms are elevated specifically among insecure-dismissing individuals. For example, Ivarsson and colleagues (2010) found that, relative to secure-autonomous and insecure-preoccupied individuals, insecure-dismissing individuals were overrepresented in a depressed outpatient sample. However, other studies found depressive symptoms to be elevated specifically among insecure-preoccupied individuals. In a sample of non-psychotic psychiatric patients, those diagnosed with depression (with or without co-morbid conditions) exhibited significantly more involving anger with parents during the AAI, a hallmark of insecure-preoccupied attachment, than those who were not diagnosed with depression (Fonagy et al., 1996). In addition, another study found that low-risk mothers who were diagnosed with postnatal depression were more likely to be classified as insecure-preoccupied than mothers who were not diagnosed with postnatal depression (Murray et al., 2006). Further, insecure-preoccupied mothers who were incarcerated reported higher severity of depressive symptoms than both insecure-dismissing and secure-autonomous incarcerated mothers (Borelli et al., 2010).

Most of the studies that directly compared depressive symptoms between the two subtypes of insecure attachment found that adults with insecure-preoccupied attachment had higher levels of depression than adults with insecure-dismissing attachment. In a group of psychiatrically hospitalized adolescents, females diagnosed with affective disorders were more likely to be classified as insecure-preoccupied than insecure-dismissing (Rosenstein and Horowitz, 1996). Male adolescents with affective disorders, however, were found to be equally likely to have a dismissing or preoccupied attachment. Likewise, in female college students (Cole-Detke and Kobak, 1996) and low risk adolescents (Kobak et al., 1991), individuals classified as insecure-preoccupied tended to report more depression symptoms than individuals classified as prototypically insecure-dismissing.

To our knowledge, there has been only one meta-analysis that has investigated links between adult attachment representations as measured by the AAI and depression, which demonstrated overrepresentation of insecure attachments in clinically depressed samples, compared with non-clinical samples (Bakermans-Kranenburg and van IJzendoorn, 2009). Depressed individuals were found no more likely to be classified as unresolved than individuals without depression. However, this meta-analysis was limited by excluding studies that assessed depression in non-clinical populations and that used self-report measures of depression.

Overall, research assessing the link between adult attachment state of mind and depression has shown inconsistent findings, and a previous meta-analytic study assessing such links only examined distributions of adult attachment classifications in clinically depressed samples compared to non-clinical samples. In addition, although the prevalent notion has been that insecure-preoccupied attachment is more likely to be linked to internalizing disorders, including depression, than insecure-dismissing attachment (Fortuna and Roisman, 2008), no meta-analysis has directly compared depression between the two insecure attachment patterns.

Thus, in the current meta-analysis, we aimed to examine the magnitude of the association between adult attachment representation and depressive symptomatology across clinical and non-clinical populations. Accordingly, we aimed to examine whether there were significant differences in depression between (a) insecure and secure-autonomous individuals, including comparisons between (b) insecure-dismissing and secure-autonomous individuals, and (c) insecure-preoccupied and secure-autonomous individuals; (d) insecure-preoccupied and insecure-dismissing individuals; and (e) unresolved and organized individuals.

Additionally, we examined gender, socioeconomic status and education, ethnicity, and age as potential moderators of the attachment-depression association, given that increased risk for depression has been observed in females (Goodwin and Gotlib, 2004), individuals with low socioeconomic status (Freeman et al., 2016), minority individuals (Dunlop et al., 2003), and older adults (Andrade et al., 2000). In addition, we considered several methodological factors as potential moderators, including whether individuals’ attachment patterns were based on a categorical classifications or a dimensional scores (following debate regarding the underlying structure of adult attachment patterns; Haydon et al., 2012), and whether depression symptoms were measured dimensionally (through self-report) or categorically (through a diagnostic interview).

Section snippets

Search strategy

Peer-reviewed articles and doctoral dissertations investigating the relation between attachment and depression were retrieved in March 2017 by systematically searching PsycINFO, MEDLINE, and ProQuest databases. The keywords attach* or Adult Attachment Interview or AAI were selected and arranged in various combinations with depress*, dysthym*, emotional*, affectiv*, dissociative disorders, psychopatholog*, adolescent psychopathology, mental disorder, mentalization, mood, psychosocial, emotional

Secure versus insecure attachment

Effect sizes for 51 samples with 4074 participants were computed for the meta-analysis examining depression symptoms among secure-autonomous and insecure attachment patterns (see Fig. 2a). There was a wide dispersion of effects, with individual study estimates ranging from −1.79 to 1.65. The aggregate effect size was significant in the predicted direction, such that individuals with insecure attachment had higher depressive symptoms than individuals with secure-autonomous attachment, d = 0.21,

Discussion

The primary aim of this meta-analysis was to examine the extent to which adult attachment representations are associated with depressive symptoms. Individuals with insecure attachment (combining both preoccupied and dismissing subtypes) had higher levels of depression than secure-autonomous individuals and unresolved individuals had higher levels of depression than individuals with organized attachment classifications, with these associations showing small to medium effect sizes. When examining

Conclusion

In conclusion, the current study is the first to meta-analytically examine the association between depressive symptoms and adult attachment patterns as assessed by the AAI across a diverse range of non-clinical and clinical samples. The results suggest that individuals with insecure-preoccupied and unresolved attachment patterns, but not insecure-dismissing patterns, are likely to experience more depressive symptoms than those who are secure-autonomous. Thus, future research should aim to

Conflict of interest

The authors have no conflicts of interest to report.

Role of funding sources

There are no funding sources to report.

Contributors

Dagan, Facompré and Bernard designed the study. Facompré conducted the systematic literature search and Dagan and Facompré completed the coding, under the supervision of Bernard.

Facompré conducted the statistical analyses. Dagan, Facompré and Bernard wrote the first draft of the manuscript and all authors contributed to and have approved the final manuscript.

Acknowledgments

We would like to thank the authors whose data were included in this meta-analytic review.

References (170)

  • L. Andrade et al.

    Cross-national comparisons of the prevalences and correlates of mental disorders

    Bull. World Health Organ.

    (2000)
  • L. *Atkinson et al.

    Attachment and selective attention: disorganization and emotional Stroop reaction time

    Dev. Psychopathol.

    (2009)
  • J.C. *Babcock et al.

    Attachment, emotional regulation, and the function of marital violence: differences between secure, preoccupied, and dismissing violent and nonviolent husbands

    J. Family Violence

    (2000)
  • N.I.G. Bahm et al.

    Unresolved loss, a risk factor for offspring, predicts event-related potential responses to death-related imagery

    Dev. Psychol.

    (2017)
  • M.J. Bakermans-Kranenburg et al.

    The first 10,000 adult attachment interviews: distributions of adult attachment representations in clinical and non-clinical groups

    Attach. Hum. Dev.

    (2009)
  • M.J. Bakermans-Kranenburg et al.

    A psychometric study of the adult attachment interview: reliability and discriminant validity

    Dev. Psychol.

    (1993)
  • M.J. Bakermans-Kranenburg et al.

    Less is more: meta- analyses of sensitivity and attachment interventions in early childhood

    Psychol. Bull.

    (2003)
  • A.G. *Bashe

    Association of Attachment Security With Emotion Regulation and Depressive Symptoms

    (2000)
  • A.T. Beck et al.

    An inventory for measuring depression

    Arch. Geneml Psychiatry

    (1961)
  • J. Belsky

    Attachment, mating, and parenting: An evolutionary interpretation

    Hum. Nat.

    (1997)
  • J. Belsky

    Theory testing, effect-size evaluation, and differential susceptibility to rearing influence: the case of mothering and attachment

    Child Dev.

    (1997)
  • P. *Blumberg

    Adult Attachment and the Organization of Affective Experience

    (1997)
  • J.L. *Borelli et al.

    Language use in the adult attachment interview: Evidence for attachment- specific emotion regulation

    Pers. Relationsh.

    (2013)
  • J.L. *Borelli et al.

    Attachment organization in a sample of incarcerated mothers: distribution of classifications and associations with substance abuse history, depressive symptoms, perceptions of parenting competency and social support

    Attach. Hum. Dev.

    (2010)
  • G. Bosmans

    Cognitive behaviour therapy for children and adolescents: can attachment theory contribute to its efficacy?

    Clin. Child Family Psychol. Rev.

    (2016)
  • J. Bowlby

    Attachment and Loss: Vol. 2. Separation

    (1973)
  • J. Bowlby

    Attachment and loss: Retrospect and prospect

    Am. J. Orthopsychiatr.

    (1982)
  • J. Bowlby

    A Secure Base: Parent-child Attachment and Healthy Human Development

    (1988)
  • K. Brennan et al.

    Self-report measures of adult romantic attachment

  • I. Bretherton

    Pouring new wine into old bottles: The social self as internal working model

  • I. Bretherton et al.

    Internal working models in attachment relationships: elaborating a central construct in attachment theory

  • L.E. Brumariu et al.

    Parent-child attachment and internalizing symptoms in childhood and adolescence: a review of empirical findings and future directions

    Dev. Psychopathol.

    (2010)
  • A.L. *Busch et al.

    Unresolved loss in the adult attachment interview: implications for marital and parenting relationships

    Dev. Psychopathol.

    (2008)
  • A.L. *Busch et al.

    Mothers’ adult attachment interview ratings predict preschool children's IQ following domestic violence exposure

    Attach. Hum. Dev.

    (2010)
  • E.A. Carlson et al.

    The contribution of attachment theory to developmental psychopathology

  • K.M. *Caspers et al.

    Association between the serotonin transporter promoter polymorphism (5-HTTLPR) and adult unresolved attachment

    Dev. Psychol.

    (2009)
  • A. Caspi et al.

    The p factor: One general psychopathology factor in the structure of psychiatric disorders

    Clin. Psychol. Sci.

    (2015)
  • J. Cassidy et al.

    The insecure/ambivalent pattern of attachment: theory and research

    Child Dev.

    (1994)
  • H. *Cole-Detke et al.

    Attachment processes in eating disorder and depression

    J. Consult. Clin. Psychol.

    (1996)
  • R Core Team

    R: A language and Environment For Statistical computing. R Foundation For Statistical Computing

    (2015)
  • J.A. Crowell et al.

    Assessing secure base behavior in adulthood: development of a measure, links to adult attachment representations and relations to couples' communication and reports of relationships

    Dev. Psychol.

    (2002)
  • E.M. Cummings et al.

    Toward a transactional model of relations between attachment and depression

  • C. Cyr et al.

    Attachment security and disorganization in maltreating and high-risk families: A series of meta-analyses

    Dev. Psychopathol.

    (2010)
  • O. *Dagan

    Attachment Security Moderates The Link Between Adverse Childhood Experiences and Cellular Aging

    (2017)
  • C.R. *Davis et al.

    Attachment and the metabolic syndrome in midlife: the role of interview-based discourse patterns

    Psychosom. Med.

    (2014)
  • A.E. *Dawson et al.

    Adolescent insecure attachment as a predictor of maladaptive coping and externalizing behaviors in emerging adulthood

    Attach. Hum. Dev.

    (2014)
  • O.M. *DeKlyen

    Childhood Psychopathology and Intergenerational Relations in the Representation of Attachment: A Comparison of Normal and Clinic-referred Disruptive Preschoolers and Their Mothers

    (1992)
  • C.A. *DeOliveira et al.

    Understanding the link between maternal adult attachment classifications and thoughts and feelings about emotions

    Attach. Hum. Dev.

    (2005)
  • G.S. Diamond et al.

    Attachment-based family therapy for depressed adolescents

    (2014)
  • G. Diamond et al.

    Attachment-based family therapy for depressed adolescents: programmatic treatment development

    Clin. Child Family Psychol. Rev.

    (2003)
  • Cited by (0)

    View full text