Introduction
Attachment theory, as espoused by Bowlby (Reference Bowlby1969; Reference Bowlby1973) is a lifespan developmental model of psychological functioning and affect regulation that emerges from the universal need to form affectional bonds within close relationships, initially with primary caregivers. During infancy, if the caregiver is perceived as available, responsive and sensitive to an individual's proximity seeking attempts, the individual will develop a secure attachment style that is associated with a capacity to manage distress, comfort with autonomy and the ability to form relationships with others (Shaver & Mikulincer, Reference Shaver and Mikulincer2002; Mikulincer & Shaver, Reference Mikulincer and Shaver2007). Conversely, if the caregiver is perceived as unavailable, unresponsive and insensitive to proximity-seeking attempts, an individual will either make use of hyperactivating strategies (e.g., eliciting care from an attachment partner through clinging and controlling responses) which can lead to the development of an insecure–ambivalent attachment style or deactivating strategies (e.g., distancing oneself from an attachment partner to handle a stressful situation alone) that can lead to the development of an insecure–avoidant attachment style (Shaver & Mikulincer, Reference Shaver and Mikulincer2002; Mikulincer & Shaver, Reference Mikulincer and Shaver2007). A fourth disorganised attachment style is said to reflect individuals who make use of both hyperactivating and deactivating strategies which is thought to represent fearful interactions with caregivers (Main & Solomon, Reference Main, Solomon, Brazelton and Yogman1986; Reference Main, Solomon, Greenberg, Cicchetti and Cummings1990).
These early experiences lay the foundations for an individual's internal mental representations or ‘working models’ of the self and others (Mikulincer, Reference Mikulincer1998) and go on to organise cognition, affect and behaviour in adult relationships (Waters & Cummings, Reference Waters and Cummings2000). Research to date has broadly characterised patterns of attachment as either secure or insecure (Ainsworth et al. Reference Ainsworth, Blehar, Waters and Wall1978; Crowell & Treboux, Reference Crowell and Treboux1995; Brennan et al. Reference Brennan, Clark, Shaver, Simpson and Rholes1998) and, although open to revision following significant changes in caregiver interactions (Bowlby, Reference Bowlby1969; Reference Bowlby1973) or adverse life events (Waters and Cummings, Reference Waters and Cummings2000), attachment patterns are considered to be relatively stable across time within the general population (Waters & Cummings, Reference Waters and Cummings2000). Although secure and insecure attachment strategies are functional in their developmental context (Fraley, Reference Fraley2002), difficulties in caregiver bonding and attachment-related adverse childhood experiences (e.g., trauma or loss) have been linked to increased risk of later psychopathology in clinical groups (Greenberg, Reference Greenberg, Cassidy and Shaver1999; Morgan & Fisher, Reference Morgan and Fisher2007; Read et al. Reference Read, Bentall and Fosse2009; Read & Gumley, Reference Read, Gumley and Benamer2010). Given these findings, there has been growing interest in exploring what influence attachment theory may play in furthering our understanding of the development of psychosis.
Two recent comprehensive systematic reviews of the attachment and psychosis literature (Gumley et al. Reference Gumley, Taylor, Schwannauer and Macbeth2014; Korver-Nieberg et al. Reference Korver-Nieberg, Berry, Meijer and Haan2014) have pointed towards the importance of understanding the influence that attachment styles may play in the aetiology, trajectory and recovery of psychosis. These authors concluded that there was evidence supporting the construct validity of attachment measurements in people with psychosis. They found small to moderate associations between attachment styles and outcomes, including positive and negative symptoms, depression and quality of life; and suggested that an insecure–avoidant attachment style may be a risk factor for problematic recovery following psychosis. Furthermore, they found that both insecure–anxious and insecure–avoidant attachment styles are associated with psychotic phenomenology and with an indication that insecurely attached individuals are more vulnerable to developing maladaptive coping strategies in relation to their recovery from psychosis (Gumley et al. Reference Gumley, Taylor, Schwannauer and Macbeth2014; Korver-Nieberg et al. Reference Korver-Nieberg, Berry, Meijer and Haan2014).
The aim of the present paper is to complement these two recent systematic reviews by exploring some of the key findings from the accumulating literature on attachment and psychosis and their relevance to further understanding the critical period of the development of psychosis. More specifically the following questions will be considered:
-
• What are the key attachment measures that have been used within the attachment and psychosis literature?
-
• What are the results of studies that have measured attachment or parental bonding in psychosis and what clinical implications can we derive from it?
-
• What are some of the key questions for future research from these findings in relation to the onset of psychosis research field?
What are the key attachment measures that have been used within the attachment and psychosis literature?
The adult attachment literature has focused on research using semi-structured interviews and self-report measures. The ability of these measures to accurately capture an individual's working model of attachment has raised debate within the literature as to whether they reflect interpersonal dispositions or account more for ways individuals act in close relationships (Pietromonaco & Barrett, Reference Pietromonaco and Barrett2000). More recent reviews of attachment measures (e.g., Fraley & Spieker, Reference Fraley and Spieker2003; Main et al. Reference Main, Hesse, Kaplan, Grossmann, Grossmann and Waters2005) have favoured dimensional approaches to delineating attachment styles and have cautioned that researchers need to consider the assumptions that a specific attachment measure makes in relation to attachment theory, and consider which relationships are under investigation before adopting a particular attachment measure as they may be targeting different constructs (Crowell & Hauser, Reference Crowell, Hauser, Steele and Steele2008; Crowell et al. Reference Crowell, Treboux and Waters1999; Reference Crowell, Fraley, Shaver, Cassidy and Shaver2008).
The most commonly used measures of attachment in psychosis research are reviewed briefly below.
The original Hazan & Shaver (Reference Hazan and Shaver1987) self-report measure of attachment has been used extensively in attachment research. The questionnaire consists of three sets of statements, which delineate the attachment styles of security, insecure–avoidant and insecure–ambivalent. Despite its wide use, the authors have since recommended using more sophisticated measures, which have been developed more recently (http://internal.psychology.illinois.edu/~rcfraley/measures/measures.html).
One such measure is the Attachment Style Questionnaire (ASQ, Feeney et al. Reference Feeney, Noller, Hanrahan, Sperling and Berman1994). The ASQ is a 40-item self-report questionnaire that assesses the individual's internal working model of peer relationships. Participants rate their agreement with the statements on a 6-point Likert-type scale ranging from ‘Strongly Disagree’ to ‘Strongly Agree’. The ASQ has adequate reliability and good convergent validity with other attachment measures, family functioning measures and personality measures.
The most frequently used assessment measure of attachment is the Adult Attachment Interview (AAI, Main & Goldwyn, Reference Main and Goldwyn1984), The AAI is a semi structured interview instrument that classifies adults into secure–autonomous, insecure–dismissing, insecure–preoccupied and unresolved attachment styles whereby the ‘coherence’ of their narrated description of their early attachment relationships are measured. Although the AAI is considered the ‘gold-standard’ measure of attachment it has been noted that when administered to individual's with psychosis, the results can be confounded by the presence of psychotic experiences (Dozier et al. Reference Dozier, Stovall, Albus, Cassidy and Shaver1999).
In recent years, a specific instrument called the Psychosis Attachment Measure (PAM) has been develop to assess attachment in psychosis (Berry et al. Reference Berry, Wearden, Barrowclough and Liversidge2006; Reference Berry, Band, Corcoran, Barrowclough and Wearden2007a , b #50). The PAM is a 16-item self-reported attachment measure where items refer to thoughts, feelings and behaviours in close interpersonal relationships, but do not refer specifically to romantic relationships. Eight of the items assess the construct of avoidance (e.g., ‘I prefer not to let other people know my ‘true’ thoughts and feelings') and eight items assess the construct of anxiety (e.g., ‘I tend to get upset, anxious or angry if other people are not there when I need them’).
The Parental Bonding Instrument (PBI, Parker et al. Reference Parker, Tupling and Brown1979), which assesses perceived levels of parental care and overprotection, has also been widely used within the literature to review the influence of primary caregiver style and its association with the development of psychosis. As such it was also included within the current review. The PBI is a 25-item measure that measures an adult's retrospective account of the parenting they received up to the age of 16 years. Two scales termed ‘care’ and ‘overprotection’ (or ‘control’) measure core parental styles as perceived by the child in respect to care received from the mother and father, respectively. Parker et al. identified four quadrants with different attachment styles and demonstrated that the ‘affectionless control’ style (characterised by low care and high protection) was overexpressed in psychotic participants.
What are the results of studies that have measured attachment or parental bonding in psychosis and what clinical implications can we derive from it?
The details of the papers reviewed in this paper can be found in table 1.
Table 1. Reviewed studies (grouped by assessment instrument and listed in chronological order)
Early onset and longer admissions
Ponizovsky et al. set out to test whether insecure attachment styles were associated with diagnosis and illness course in a sample of male inpatients with a diagnosis of schizophrenia. They used the Hazan & Shaver self-report measure of attachment. Their findings suggested that patients with insecure attachment styles when compared with patients with secure attachments had a significantly earlier age of psychosis onset and longer admissions (Ponizovsky et al. Reference Ponizovsky, Nechamkin and Rosca2007).
A sample of 72 patients with schizophrenia admitted consecutively across four psychiatric units was compared with a sample of controls recruited from General Practitioners using the PBI. Patients' representations of parenting styles impacted on illness onset and course. The results suggested that patients with psychosis were more likely to rate both parents as being less caring and fathers as being more overprotective. Furthermore, patients who rated both parents as being low caring and overprotective also tended to have an earlier age of initial hospitalisation for psychosis and at nine months following discharge, were also more likely to be readmitted (Parker et al. Reference Parker, Fairley, Greenwood, Jurd and Silove1982). These findings were also replicated in study, which assessed 62 patients with a diagnosis of schizophrenia using the PBI. The results indicated that the patients who perceived their parents positively tended to experience fewer relapses (Warner & Atkinson, Reference Warner and Atkinson1988).
Social functioning
The Attachment Style Questionnaire has been used to explore the role of attachment styles, personality characteristics and the implications for social functioning in 96 first episode psychosis service users and controls. Those in the clinical sample were more likely to report greater difficulties in their peer attachments when compared with their matched controls. Results also indicated that first episode psychosis service users reported higher levels of attachment anxiety, discomfort with closeness and a greater need for approval in peer relations compared with controls. Attachment and personality styles both played a role in social functioning (Couture et al. Reference Couture, Lecomte and Leclerc2007).
Psychotic symptoms
Dozier and Lee interviewed 76 patients with psychosis using the AAI. Patients who were more reliant on hyperactivating strategies reported more psychotic symptoms than those using deactivating strategies. However, they found that patients with dismissing attachment styles, who made use of deactivating strategies, experienced more delusions, hallucinations and suspiciousness, and were rated by case workers as presenting as ‘more psychotic’ (Dozier & Lee, Reference Dozier and Lee1995).
MacBeth et al. (Reference MacBeth, Gumley, Schwannauer and Fisher2011) also used the AAI to interview 34 patients with first-episode psychosis. They found that insecure/dismissive attachments were predominant in the sample but no relationship was observed between attachment styles and symptoms of psychosis.
Relationships with others
Researchers using the PAM studied a sample of 58 patients with psychosis on attachment dimensions of avoidance and anxiety in relation to psychiatric staff and parents. Scores on the two attachments dimensions varied depending on the type of relationship and the authors concluded that factors that influence variability in attachment relationships should be considered in treatment plans as it may be possible to support individuals with insecure attachment styles to develop more positive relationships with others (Berry et al. Reference Berry, Wearden and Barrowclough2007b ). In a second study, Berry et al. (Reference Berry, Barrowclough and Wearden2008) used a prospective design to assess attachment in 96 patients with psychosis. They found that higher levels of attachment anxiety and attachment avoidance predicted both symptom severity and difficulties in therapeutic relationships (Berry et al. Reference Berry, Wearden, Barrowclough and Liversidge2006).
A study of parental bonding assessed a sample of 19 patients with schizophrenia, 14 with borderline disorders and 15 control participants with the PBI and similarly observed that patients with schizophrenia spectrum disorders reported less parental care and more overprotection (defined as low care and high control) than their non-clinical counterparts although the difference proved non-significant (Helgeland & Torgersen, Reference Helgeland and Torgersen1997).
In a sample of 36 patients with schizophrenia or schizoaffective disorders compared with their siblings, Willinger et al. (Reference Willinger, Heiden, Meszaros, Formann and Aschauer2002) found that patients had a greater tendency to describe their mothers as being less caring and more overprotective towards them compared with descriptions from their healthy siblings. The perceptions of higher maternal overprotection remained a key factor even after controlling for the influence of premorbid personality.
Interestingly, a study examining 12 monozygotic and 19 same-sex zygotic twin pairs discordant for DSM-III-R schizophrenia found that patients reported higher levels of parental overprotection than their probands, raising the question of whether the differences in parental bonding could be explained by the presence of a psychotic disorder (Onstad et al. Reference Onstad, Skre, Torgersen and Kringlen1994).
Clinical implications
In combination, these studies indicate a relationship between childhood attachment, patient reflections on parenting style and psychosis. Significant relationships were found between recollections of early attachment relationships and attachment style (Berry et al. Reference Berry, Wearden and Barrowclough2007b ). Individuals rated as having insecure attachment styles in adulthood were also more likely to describe early caregiving relationships as being characterised by rejecting or inconsistent parenting (Fonagy et al. Reference Fonagy, Steele, Steele, Higgitt and Target1994). In line with Bowlby's attachment theory (Bowlby, Reference Bowlby1969; Reference Bowlby1973), it seems that recollections of adverse early experiences with primary caregivers are likely to limit the capacity to form secure attachments in adulthood as these studies suggest that individuals who perceived their caregivers as insensitive or indifferent to their distress were more likely to experience difficulties in relating to others.
Within the field of psychosis an individual's attachment style has been suggested as a clinically relevant construct in relation to the development, course and treatment of psychosis. More specifically it has been suggested that the attachment experience of individuals with psychosis is an important construct for understanding how social information is processed and how mentalisation skills are developed within this population (Korver-Nieberg et al. Reference Korver-Nieberg, Berry, Meijer and Haan2014). Gumley's systematic review of attachment and psychosis found that individuals with psychosis who had a secure attachment had better engagement and greater treatment adherence, whereas insecure attachment was found to be related to disengagement with treatment services and avoidant attachment was related to help-seeking difficulties, poorer use of treatment, longer hospital admissions and lower-rated therapeutic alliance. Their findings also suggested that the attachment system is activated in the relationships that individuals develop with their service providers and as such, highlight attachment theory as a useful framework in which to consider recovery within individuals with psychosis. These authors further note the importance of services being aware of how these systems may be activated within individuals as a means to provide ‘an attuned response to the needs of individuals’ and establish ‘a safe haven and secure base for recovery’ (Gumley et al. Reference Gumley, Taylor, Schwannauer and Macbeth2014).
What are some of the key questions for future research from these findings in relation to the onset of psychosis research field?
Attachment studies in psychosis have mainly focused on multiple episode samples where the results have invariably been confounded by the presence of secondary disabilities impacting upon quality of social relationships. The extent to which insecure/avoidant representations predominate in first episode samples requires further study. Establishing the distribution of secure and insecure attachment representations in first onset psychosis groups, and implicit within this the cognitive–affective–interpersonal model that each attachment classification represents, could provide a basis for tailoring treatment models towards the specific needs of the individual. Future research in attachment in psychosis should also consider that various attachment styles are assessed by different instruments. Instruments such as the Hazan and Shaver measure were developed within the social psychology research tradition, others came from a developmental psychology tradition (e.g., AAI; Roisman et al. Reference Roisman, Holland, Fortuna, Fraley, Clausell and Clarke2007; Crowell et al. Reference Crowell, Fraley, Shaver, Cassidy and Shaver2008). To date only the PAM was developed specifically to measure attachment styles in individuals with psychosis (Berry et al. Reference Berry, Barrowclough and Wearden2008). Finally, future research should evaluate the strength and weaknesses of self-report measures versus semi-structured interview measures of attachment in psychosis research.
Limitations
The studies reviewed in this paper suggest that insecure attachment representations are evident at different illness phases including during the at-risk mental state (Couture et al. Reference Couture, Lecomte and Leclerc2007; Gajwani et al. Reference Gajwani, Patterson and Birchwood2013). They are linked to a poorer quality of interpersonal relationships and less integrated recovery styles (Berry et al. Reference Berry, Wearden and Barrowclough2007b ; Gumley et al. Reference Gumley, Taylor, Schwannauer and Macbeth2014). Thus, insecure attachment styles may serve as a vulnerability factor for both the development and persistence of psychosis. Recent data from at-risk psychosis populations (e.g., O'Brien et al. Reference O'Brien, Gordon, Bearden, Lopez, Kopelowicz and Cannon2006; Tienari et al. Reference Tienari, Wynne, Sorri, Lahti, Laksy, Moring, Naarala, Nieminen and Wahlberg2004; McFarlane & Cook, Reference McFarlane and Cook2007) attests to the important role played by family relationships in the expression and course of psychosis symptomatology. However, as most attachment studies are cross-sectional, it is equally possible that having a more severe course of illness can render individuals more likely to develop and/or recall difficulties in attachment relationships.
Concluding remarks
In conclusion, reports of caregiver attachment difficulties may play an important role in the development of psychosis. The narrative review speaks to the importance of greater attention being given to the assessment and understanding of attachment needs and difficulties that are experienced and reported during the early years, and it emphasises the need to develop interventions that seek to compensate for these difficulties. Secure attachment may confer advantages in facilitating how individuals make sense of their experiences and their readiness to engage with therapeutic interventions and seek help. In contrast, individuals presenting with insecure attachment representations may find the reciprocal process of engagement with clinicians threatening or overwhelming, and consequently may disengage from services to regulate affect. An understanding of attachment representations may therefore be relevant to understanding differences in recovery trajectories in the first few years after treatment for psychosis is initiated, particularly in understanding the role help seeking may play in accelerating or forestalling relapse (Gumley et al. Reference Gumley, Braehler and Laithwite2010; Onwumere et al. Reference Onwumere, Bebbington and Kuipers2011).
Acknowledgements
We acknowledge the Fondazione CariVerona for their support (please see also below Financial Support).
Financial support
This study was supported by the Fondazione CariVerona, who provided a 3-year grant to the WHO Collaborating Centre for Research and Training in Mental Health and Service Organization at the University of Verona, directed by Professor Michele Tansella. The grant ‘Promoting research to improve quality of care. The Verona WHO Centre for mental health research’, supports the main research projects of the following Units of the Verona WHO Centre: Psychiatric Case Register, Geographical Epidemiology & Mental Health Economics (Head: Professor Francesco Amaddeo); Clinical Psychopharmacology & Drug Epidemiology (Head: Professor Corrado Barbui); Environmental, Clinical and Genetic Determinants of Outcome of Mental Disorders (Head: Professor Mirella Ruggeri).
Conflict of Interest
None.
