Carer Attachment Moderates Resilience for Young People Experiencing Psychosis or Depression

This paper explored the transactional nature of carer attachment, proposing that carer insecure attachment may influence the development of psychopathology and resilience in Young People (YP) experiencing psychosis or depression. Seventy-five matched carer and YP dyads were grouped according to the YP diagnosis, psychosis forty-three and depression thirty-two. YP and carers completed an attachment protocol, Attachment Style Questionnaire and the YP’s scores on the Ego-Resiliency Scale measured resilience. Diagnosis of Depression was associated with low YP resilience and high carer insecure (avoidant) attachment. Diagnosis of Psychosis was associated with high YP resilience and insecure (anxious) carer attachment. The results are plausibly explained by the type of carer insecure attachment combining with the YP's level of resilience to influence development of psychopathology in YP. If so, the assessment of carer attachment is likely to provide insights into the interpersonal transactions that influence resilience in YP experiencing Mental Health problems.


Introduction
Rutter [1] suggested that emotional and behavioural responses in children and young people may be grounded in the quality of their family relationships, and that family processes may influence the development and maintenance of psychopathology. According to Hill, Fonagy, Safier, & Sargent, [2] there is a deficit in theory and measurement of the link between family processes and individual family members. They suggest that attachment theory [3] can accommodate these issues by exploring the attachment processes that serve as interpersonal templates for carer and family relationships. It may be argued therefore that the carer's attachment style may influences their capacity to respond appropriately, and as such, contribute to the severity of illness experienced by the young person [4].
The present study investigates the relationship between carer's attachment style, and the closely linked concept of resilience in young people, to diagnosis of psychosis and depression in Young People. If the suggestion of Hill and colleagues [2] is valid then measured patterns of resilience in youths diagnosed with depression and the attachment characteristics of their carers should differ clearly from patterns measured in youths diagnosed with psychosis, and their carers.
Bowlby [5][6][7] has identified the antecedents of attachment as biologically determined 'proximity seeking' from which the individual develops a particular attachment style He proposed that the quality of interaction and experience between a child and its caregiver contributes to the formation of Internal Working Models (IWMs), which function as interpersonal cognitive schemata across the life span. IWMs, in this case attachment styles, serve as templates for affect regulation, family adjustment and the caregiving system, which tend to be consistent across the life span. In terms of the success of attachment styles in regulating affect and adjustment, Ainsworth and collegues, [8] identified attachment styles as either secure or insecure. According to Feeney, Noller and Hanrahan, [9] a secure attachment style was measured by higher scores on the Confidence dimension and lower scores on measures of Anxious and Avoidant attachment. Insecure attachment was represented by lower scores on the measured Confidence dimension in the presence of higher scores on either Anxious attachment (Preoccupation with Relationships or Need for Approval) or Avoidant attachment (Discomfort with Closeness or Relationships as Secondary) or higher scores on both Anxious and Avoidant attachment.
Anxious attachment relationships are marked by interpersonal intrusion, anxiety, and the fear of attachment dissolution. Avoidant attachment is insecure because the attachment bonds are weak, often because the avoidantly attached person is rejecting and has not learned how to form stable and rewarding attachments. Mikulincer & Shaver [10] found insecure attachment to be associated with early onset depression and psychosis; specifically, depression has been associated with an Anxious attachment style and psychosis with Avoidant attachment. On the other hand Mickelson and colleagues [11] found that the strength of association did not differ, nevertheless, attachment styles provide a way of assessing individual differences in the expression of interpersonal responses to the Caregiver-Young Person dyad.
A key factor in understanding the dynamic nature of attachment relates to what attachment styles represent. Mikulincer and Shaver [12] make the point that although attachment styles, Secure, Anxious and Avoidant, provide a convenient categorical nomenclature, the dimensional aspects of attachment require that research subjects are distributed continuously rather that categorically.

Resilience
Attachment styles influence the manner in which individuals form constructs and attributions about their psychosocial experiences, including defective organisation of attention and distortions of essential schemata [13]. Schemata influence emotional responses and affect the manner in which the individual appraises a social situation in order to maintain a coherent image of self (14,15]. Such schemata, developed from the caregiver-self interaction [16], serve to predict and maintain self-other relationships and provide the basis for coping strategies used to regulate attachment-sensitive social and emotional distress [17] and thus form the foundation for what is generally called Resilience [10], defined here as the ability to be competent and adaptive despite the experience of significant stressors. Implicit in this definition is the understanding that resilience contributes to, but does not guarantee, good outcomes and that resilience is inferred by adaptive behaviour in the context of threats to normative development [18] Rutter, [19], points out that the diversity of responses to environmental risk factors, including psychosocial and physical difficulties, and individual variations in response and rates of recovery from adverse conditions, have fostered investigations into the construct of resilience.
Bowlby's attachment theory [3] may contribute to understanding the development of resilience and vulnerability. This model proposes that an infant develops the active cognitive schemata of his/her caregiver and the surrounding environment, employing them to predict the caregiver's behaviour and to regulate his/her own behaviour, ultimately internalising these schemata as attachment experiences and employing them as prototypes for later relationships. According to Egeland and colleagues [20] resilience can be understood as a developmental process that becomes organised by way of transactions between and within individuals. From this perspective, individual characteristics and environmental dynamics influence behaviours that may shape vulnerability and risk or protective factors [21].
According to Lazarus and Folkman [22] stress is the perception that demands exceed the individual's resources, which jeopardises a positive sense of self. Stress may interact with attachment styles and stimulate resilience-associated responses, for example, by activating attachment related proximity seeking as a means to buffer adverse conditions. However, protracted negative caregiver interactions may hypersensitize stress responses, increasing the likelihood of attachment insecurity and a disabled resilience response and the likelihood for expressions of psychopathology. Partial support for this perspective has been suggested by Gumley, and colleagues [23] who suggest that the attachment system may provide a meaningful construct to explore the developmental and transactional sources of emotional regulation and resilience.
It is proposed that an individual's attachment experiences influence their developmental outcomes because the level of attachment security they encounter affects how they categorise and organise their life experience, which in turn influence their resilience [24]. However, it is essential that the connection between attachment and psychopathology not be understood as being deterministic, but rather that attachment and psychopathology are understood to be associated with a coalescence of risk and protective dynamics influencing the probability of individual's degree of resilience [21].
Following Sroufe and colleagues [21] we predicted that YP resilience, irrespective of a diagnosis of depression or psychosis, was positively associated with their carer's secure attachment. However, it was also expected that where Carers revealed insecure attachment, those young people with psychosis would reveal a stronger negative relationship between resilience and Carer discomfort with closeness (avoidant attachment style) than depressed young people.
Conversely, for young people with depression, we expected attachment styles indicating Carer anxious attachment styles, (such as Preoccupation with Relationships), to negatively mediate the relationship between diagnosis and resilience.
Ethical permission for this research was granted by the Human Research Ethics Committees of Sydney West Area Heath Service (Approval No. 06/004).

Attachment
Attachment was measured using the Attachment Style Questionnaire (ASQ) [25]. The ASQ is a continuous measure containing 40 items in five subscales. Evidence indicates support for the psychometric properties of the ASQ. Secure attachment (Confidence, 8 items) showed acceptable reliability, with alpha = 0.80. Reliability for Insecure attachment was more variable, but within acceptable limits for research purposes: Discomfort with Closeness (10 items) alpha = 0.84; Need for Approval, (7 items) alpha = 0.79; Preoccupation with Relationships (10 items) alpha = 0.76; Relationships as Secondary (7 items) alpha =0 .76. Test re-test reliability measured at 10 weeks ranging from .67 to .78 indicated that the measure had acceptable levels of stability [25].

Resilience
Psychological resilience is characterised by a person's ability to bounce back from negative emotional experiences and to respond flexibly to the changing demands of stressful experiences [26]. In this study, the Ego-Resiliency Scale (ER89) [27] was used to assesses Young People's psychological resilience. The scale consists of 14 items requiring responses on a 4-point Likert scale. High scores correspond with high levels of resilience. The scale has been shown to have a coefficient alpha reliability of .76. [27] Across a five-year assessment period the test-retest reliabilities were .67 and .51 (adjusted for attenuation) for females and males respectively.

Severity of Illness
New South Wales Health implemented the Mental Health Outcome and Assessment Tools (MH-OAT [28]) to ensure the collection of minimum standard data on consumers, including young people accessing mental health services who have experienced early-onset mental illness. Included in these tools are the Health of a Nation Outcome Scales for Children and Adolescents (HoNOSCA) [29], the standard measure used for Young People up to 17 years; and the Health of a Nation Outcome Scales (HoNOS) [30], the equivalent measure used for people above the age of 18. Adequate levels of validity and reliability are associated with these measures [31]. The HoNOS or HoNOSCA were used depending on the age of the YP at the time of assessment in accordance with the NSW Mental Health clinical guidelines.
The HoNOSCA includes the four domains: Behaviour, including disruptive behaviour, overactivity, self-injury and substance abuse; impairment, reflecting, academic impairment, physical illness and disability; symptoms, including, psychotic, somatic and emotional indicators; and social measures, including problems with family and peers, poor self-care and school attendance [32]. The HoNOS is comprised of four domains: Behaviour, overactive, aggressive, disruptive or agitated behaviours, self-harm and problems with substance abuse; impairment, reflecting cognitive and physical disability; symptoms, indicating positive symptoms, depressed mood and other mental or behavioural problems; Social and vocational factors reflecting relationship problems, living conditions and difficulties with daily living [33]. Possible scores on the HoNOSCA range from 0-52 while those on the HoNOS range from 0-48, higher scores reflecting greater severity of illness. In accordance with the MH-OAT, the criteria for illness severity are measured by the baseline total on the HoNOS (18+years) or HoNOSCA (12-17years).
Item variation between the HoNOSCA and HoNOS scales poses a problem for data comparability. To make the scores from these measures compatible, a severity metric was developed by assuming conceptual equivalence between the four subscales of each measure and converting these to percentages. Thus the formula for standardising the two scales (to a common metric) is as follows: (HoNOS/48)*100 and (HoNOSCA/52)*100. In this study the common metric was identified as HoNOS/CA.

Procedure
The groups were recruited from three Mental Health Services; Adult MH servicing the needs of people over the age of 18 years, Child and Adolescent Mental Health Services (CaAMS) and Headspace, a Mental Health service dedicated to the provision of Mental Health services to YP 15-25. The primary criterion for participation was a diagnosis of first episode psychosis or depression, the diagnosis having been made by a psychiatrist working at the Area Mental Health Service. All YP and their carers were approached at the time of their first psychiatric assessment and diagnosis of psychosis or depression. Only complete dyads of carer and YP were included in the study.

Data Analysis
To explore the relationship between the dependent variable of diagnosis and the predictor variables of attachment style and YP resilience, the analytic strategy was to use univariate and multivariate discriminant function analyses (DFA), with diagnosis (psychosis vs depression) as the outcome variable. This approach allows relationships between diagnosis, resilience and Carers' attachment style to be explored while controlling for the variables of young person's attachment style and symptom severity as measured by the HoNOS/CA. No group mean comparisons are reported here because the object of the study was to uncover relationships within the set of continuous predictors, related to diagnosis. This is fundamentally a correlational procedure.
The number of predictors permitted for multivariate DFA = n-2, although Poulson and French [34] recommend a five-to-one ratio of cases-to-predictors for DFA. In this analysis, 12 continuous predictor variables were examined. They were the young person's five attachment style scores, their resilience score, and the five attachment style scores of their carer. The measure of severity of illness at examination was also included in the predictor variable set, to control for possible confounding of diagnosis with severity.
The continuous variables were screened for normality by calculating skewness and kurtosis. Variables were either normal, or showed non-normality too minor to distort the planned analysis. Predictor variables were screened for multicolinearity with Pearson correlations. All intercorrelations were < 0.6. The primary dichotomous predicted variable, diagnosis (psychotic/depressed), was dummy-coded, with psychosis coded 1 and depression coded 2.
Point biserial correlations, also known as Simple Discriminant Function Analyses, are the correlational form used to compare dummy-coded group membership to a continuous predictor variable. They are equivalent to a between-groups t-test, producing identical p values when t-test assumptions are met, as they were here, with the advantage over t-tests of including a strength of effect measure in the form of a correlation coefficient. They are directly comparable to the multivariate DFA used for the multivariate stage of the analysis. The point-biserial correlations identified simple bivariate relationships between group (psychotic/depressed) and all 12 predictor variables. Predictor variable patterns which distinguished diagnoses were further examined using multivariate discriminant function analysis (DFA). As with the point-biserial correlations, the variable predicted was diagnosis; psychosis or depression. For ease of interpretation, results are presented here in the more familiar regression format, rather than as discriminant function equations.
Because of the direction of dummy coding, positive point biserial correlations or DFA beta weights indicate young people diagnosed with depression tended to have a higher score on the continuous predictor than young people diagnosed with psychosis. Negative correlations or beta weights indicated the opposite pattern. The analyses were conducted using SPSS v18 and v19., and Instat+ v3.036.
Within group correlations between YP and Carers' attachment were not significant, however they are available from the author. Table 1 reports the means, standard deviation and range for the Carers and Young People's attachment styles by the Young Person's group membership. Table 2 reports the means, standard deviations and ranges for the resiliency variable for the combined sample and the group of Young People with depression or psychosis separately. When correlating group with resilience an r pb = -0.317, p = 0.006 indicated that psychotic YP were found to experience significantly greater resilience than depressed YP.  Table 3 shows the bivariate relationships between predictor variables and diagnosis. Only one variable, Carer Discomfort with Closeness, was strongly correlated with diagnosis. Two variables, YP Preoccupation with Relationships and YP Resilience, were moderately correlated. Two other YP variables were weakly related to diagnosis. An initial discriminant function analysis (DFA), with all 12 predictor variables, was highly statistically significant (F 12,62 = 21.97, p < 0.001) and accounted for 81% of the variation in diagnoses. Examination of coefficients suggested only three variables was responsible for the prediction, so the DFA was re-run using only these variables.

Results
The second DFA was also highly significant (F 3,71 = 88.28, p < 0.001) and accounted for 79% of variation in diagnoses.
Interestingly, symptom severity and YP's attachment style were not related to diagnosis, therefore there was no further exploration of these variables. Table 4 shows that the three significant predictors were YP Resilience, Carer Discomfort with Closeness, and Carer Preoccupation with Relationships. Beta weights suggested a diagnosis of psychosis is associated with lower Carer Discomfort with Closeness, higher Carer Preoccupation with Relationships, and (paradoxically) higher YP Resilience. A diagnosis of depression is associated with higher Carer Discomfort with Closeness, lower Carer Preoccupation with Relationships, and lower YP Resilience.

Discussion
In this study, a diagnosis of psychosis was found to be associated with higher resilience, compared to a diagnosis of depression. Prior research [35] has suggested that secure attachment, indicating a positive sense of self and the capacity to seek others for emotional support during stressful conditions, are precursors for resilience, however our results indicate that psychotic YP whose carer exhibited insecure attachment, low attachment avoidance (Carer Discomfort with Closeness) and high anxiety (Preoccupation with Relationships) were more resilient than the depressed YP who experience high levels of carer avoidant attachment (Carer Discomfort with Closeness) and low anxious attachment (Preoccupation with Relationships). It should be noted that the YPs attachment per se was not related to their diagnosis.
The attachment system may be understood as a mix of attachment dimensions that interact with attachment styles in which coherence is the key. Thus, when things cohere, attachment is protective, but when there is a lack of coherence, greater vulnerability is to be expected. For example, when stress is experienced, the individual's attachment system, which reflects the developmental relationship between the YP and their carer, is activated setting the occasion for the expression of the attachment system (secure or insecure). Where the attachment system is insecure, there is a potential for the individual to develop depression. Compared to secure individuals insecure Preoccupied YP are more highly prone to depression because they perceive themselves as lacking self-confidence and experience feelings of unworthiness [36,37].
Preoccupied individuals attempt to validate themselves by actively seeking proximity with a caregiver and, in so doing, become overly dependent, increasing the likelihood of developing a dysthymic perspective in the context of their need for intimacy not being requited at a level commensurate with their emotional needs. This situation may be exacerbated when the Anxious Preoccupied individual experiences caregiving from an Avoidantly attached carer. In this context the interpersonal interactions may be represented by overt proximity seeking from the Anxious YP and rejection from the Avoidant carer.
This study found support for the negatively mediating relationship between carer insecure attachment and diagnosis. YP who experienced carers high on Avoidance were found to experience depression, while the experience of an Anxious carer was found to be associated with YP experiencing psychosis. These results suggest that the influence of carer attachment insecurity may be understood to set the occasion for the development for type of psychopathology developed.
At times of significant stress, the caregiver's attachment system becomes activated eliciting attachment related behaviours reflecting the IWM associated with their attachment dispositions. For example, an Avoidant carer may be characterized as having a positive view of self and a negative perspective on others. The Young Person who has an Avoidant carer may experience a carer who lacks genuine affection and expresses emotion in a cold or derisive response to support or comfort seeking from the care receiver. This contrasts with the Anxious carer who may express real albeit over concern and intrusive anxiety in response to a negative view of themselves and yet a positive view of others.
The Anxious carer expresses over-caring and over-controlling dispositions however compared to avoidant carers, anxious carers are not unresponsive or unavailable [38]. Rather they are motivated by a need to support care receivers because they are strongly affectedly connected with them and feel obligated and hope that their altruism may be reciprocated because they are concerned that they will be valued and accepted. Thus their need for attachment is great but their expression can be overly involved and obtrusive. In contrast to Avoidant cares, Anxiously attached persons have been found not to be neglectful nor lacking in empathy, rather they are over-involved and compulsive with their caregiving, but not without evidence of affection [39].
Feeney, [40] investigated the attachment characteristics of caregivers and found that secure carers were more sensitive, responsive and cooperative than insecure carers. Avoidant carers were inconsistent in terms of their responsiveness and low on compulsive caring. The Anxious carers tended to fall between the Avoidant and securely attached cohorts, indicating that they experience greater attachment security relative to avoidant carers. This aspect of relative attachment security may contrast with the avoidant carer's lack of availability towards the YP and be reflected in the anxious carer's heightened level of availability to the care receiver.
Mikulincer and colleagues [41] have proposed that attachment figure availability strengthens reliance on the attachment figure and encourages the development of self-regulation. These authors also found that psychological unavailability predicted low resiliency. This position was supported by the results in this study where the YP who experienced avoidant carers were found to be low in resilience, on the other hand YP assessed as being resilient experienced anxious caregivers. It may be argued that due to the transactional nature the carer/care recipient relationship, the resilient YP experiencing psychosis positively influenced their Anxious carers and moderated the expected negative association related to carer anxiety.
The transactional perspective may be supported by the characteristic responses of resilient YP who exhibit a broader repertoire of coping and behavioural responses that are likely to promote attunement between Young People and their cares. However, these characteristics are unlikely to be available to those low on resilience whose experience of avoidant caregiving is likely to be associated with rejection and a dismissing disposition towards proximity seeking.
Previous studies displayed an association between the Avoidant style and psychosis. This association has been systematically reviewed [23,42]. These studies reflect a trend towards Avoidant attachment being the paramount indicator of a diagnosis of psychosis. However, research also indicates that attachment styles vary and that Secure attachment [43] Anxious attachment [44] and Disorganised attachment [45] have also been found to occur in samples of psychotic subjects.
It is worth noting however that the overrepresentation of relative insecure attachment characteristics in people experiencing psychosis cannot discretely account for the expression of psychopathology. Rather it is the underlying constituents of attachment styles for example, impaired interpersonal sensitivity, dysregulated emotional expression, attentional control and related psychological processes that may mediate and moderate the development and course of psychopathology [46].

Strengths and Limitations
The strength of this study resides in the fact that the YP in the sample were representative of the clinical characteristics often associated with presentation to Early Intervention services: the age of the carers was similar between the two groups, and there was a preponderance of long term female carers, indicating that the attachment systems between the Young People and their carers were well established and therefore likely to be an accurate representation of the attachment relationship between carers and YP generally.
Ideally a larger sample would have improved the study since the number of subjects classified into the five attachment styles rendered the analysis under-powered, increasing the likelihood of Type 2 errors while the correlational nature of the study precluded assumptions of causality and the size of the sample curtailed the possibility of running a more sophisticated analysis.

Clinical Implications
This research indicates that the carer-care seeker attachment relationship plays an important part in the development and maintenance of mental health. From this perspective an assessment of the attachment relationship at presentation to mental health services may provide insights into positive and negative dispositions that may enable treatment and aid the process of recovery. In this research much has been made of the quality of carer's attachment relationship. Bowlby [5] has made it clear that blaming the carer for the onset and maintenance of mental illness is unhelpful. He argues that "it has long been recognized that the misguided behaviours of parents is more often than not the product of their own difficult and unhappy childhood" [5 p. 145]. This study was not about the causes of psychosis or depression, or a comparison of differences between well and unwell YP. The sample was from a population already identified as symptomatic. Rather, we sought to identify factors that might influence the type of symptomatology developed. Further research that explores the role of carer's Expressed Emotion [47] in the development of resilience is likely to provide further insights into the influences of attachment transactions of YP experiencing mental health problems.
The expected negative influence of carer insecure attachment on the level of resilience experienced by the YP was moderated by the level of the YP's resilience. The Depressed YP who experienced Avoidant caring were found to express lower levels of resilience than YP whose carers' were Anxiously attached. These results provide a degree of support for the notion that carer attachment moderates the predictive role of resilience, nevertheless, while carer attachment insecurity may contribute to the development of psychopathology in Young People, attachment needs to be understood as a possible contributor rather than the defining characteristic of developmental competencies and vulnerabilities.

Highlights
 Investigates the relationship between carer's attachment style, and resilience, to recovery from psychosis and depression in Young People.  resilience was found to be higher in YP experiencing psychosis when compared to the depressed group.  YP people who experienced carers high on Avoidance were found to experience depression,  The experience of an Anxious carer was found to be associated with YP experiencing psychosis.