Relatives’ Current Behavior Impacts on the Service User
Relationships with family, friends and peers all play a significant role in psychosis. Firstly the mere presence of close relationships and friends seems to be important. For example, there is evidence that social isolation, particularly in minority immigrant populations is associated with increased risk of psychosis (Cantor-Graae and Selten
2005), whereas living with a relative is associated with significantly better outcome. Social support from family and close friends during the early stages of psychosis predicts better functioning 5 years later, even controlling for other variables such as education, duration of untreated psychosis, symptoms and baseline functioning (Norman et al.
2012). Using more real world momentary methods of assessment, being in the presence of familiar people, rather than alone or with strangers decreases risk of experiencing delusions in people with chronic psychosis (Myin-Germeys et al.
2001), and in those at risk of psychosis, the presence of familiar friends or family reduces paranoid thinking (Collip et al.
2011) and reporting of unusual experiences (Verdoux et al.
2003). A large multisite RCT testing the effectiveness of CBT and FI for people who had recently relapsed with non-affective psychosis, found no effect of either treatment on outcome, but people with an identified close relative had a significantly better outcome than those without, and the presence of a relative was associated with a more positive response to either treatment (Garety et al.
2008). There are many potential confounds that could account for these findings, but the positive impact of social support is fairly robust.
A number of theories have been put forward as to how relatives’ support may improve outcome. Unsurprisingly, where relatives are present, the quality of relationship is crucial, and most research in this area has focussed on the concept of Expressed Emotion (EE). EE is a measure of the emotional response of relatives towards the service user, rated from relatives’ reports during the Camberwell Family Interview (CFI; Leff and Vaughn
1985; Vaughn and Leff
1976). Relatives’ are rated along five scales: hostility, criticism, over-involvement, warmth and positive remarks, and those who score six or more on critical comments, any hostility, or a rating of three or more on emotional over involvement (EOI), based on overprotective, excessively devoted or self-sacrificing style towards the service user are described as high EE, compared to low EE relatives who do not meet this criteria. Interestingly, ratings of warmth or positive remarks do not contribute to the EE rating. Early studies in the 1960s first measured the importance of the family environment for people with schizophrenia (Brown and Rutter
1966) and a meta-analysis of 26 studies in this area concluded that living in a high EE critical or hostile home environment more than doubles the risk of relapse over 9–12 months for people with psychosis (Butzlaff and Hooley
1998). Further, interventions that reduce high EE can significantly improve outcome for service users (Hooley
2007), supporting a causal role for relationship quality in relapse.
The exact mechanism by which EE predicts relapse is not yet clear. Attempts have been made to observe differences in behavior towards the service user between high and low EE relatives to see if specific behaviors can be identified that could play a role in the relapse process and which could be targeted in interventions with relatives. Using methods of coding relatives’ behavior during an interaction with the service user, such as the Kategoriensystem für Parnerschaftliche Interaktion (KPI; Interaction Coding System; Hahlweg and Conrad
1985), relatives categorised as high EE on the CFI, or rated as having a higher level of criticism considered alone, demonstrate higher levels of negative verbal or nonverbal behavior when compared with low EE or less critical relatives (Hahlweg et al.
1989; Hooley
1986; Mueser et al.
1993; Simoneau et al.
1998). Using an alternative rating of behavioral control based on coding statements from the CFI interviews, Hooley and Campbell (
2002) found that high EE relatives behaved in a more controlling manner than low EE relatives. Furthermore, behavioral control was a significant predictor of relapse at 9 months. The association between high EE relatives and the use of more controlling behaviors has been replicated in a sample of people with recent onset psychosis, and further developed by distinguishing behavioral styles between high EE-critical relatives and high-EE over-involved (Vasconcelos e Sa et al.
2013). Critical relatives tended to describe using more “direct influencing” in which they attempt to change the service users’ behavior using mild behaviors such as a polite request, or gentle reminder, through to extreme behaviors such as intimidation or ultimatums. Alternatively, relatives rated as high-EE–EOI used more “buffering” ways to take control, or do things for the service user, ranging from mild supervising or joint planning, to more intrusive actions like taking control of finances, or dealing with personal mail. Despite not finding a direct relationship between behavior and relapse in this sample, and a number of methodological limitations (including rating behavior and EE from the same interview transcripts), this study does support the idea that there are direct behaviors associated with EE.
Several potential processes have been suggested to explain how the relatives’ behavioural style impacts on psychosis in the service user (Garety et al.
2001). Firstly, relatives’ behavior could act to reinforce negative core beliefs about self, world and others that in turn impacts on information processing biases (link 5). For example, relatives behaving in a very critical way could reinforce beliefs about being useless or unlovable, and that others are critical or dangerous, leading to a bias towards negative interpretations of the behavior of others, and behavioral responses of withdrawal and avoidance that are likely to follow from this. There is some evidence to support this. In a cross sectional model, Barrowclough et al. (
2003) showed the positive association between criticism from relatives and scores on the Positive and Negative Symptom Scale (PANSS; Kay et al.
1987) positive symptom scale was mediated by negative self-evaluation (interview based assessment). In a 5 year follow-up of the same sample, negative self-evaluation also predicted time to relapse even when controlling for baseline symptoms and duration of illness (Holding et al.
2013). Secondly, relatives’ behavior could act as a direct triggering event (link 6) which is then misinterpreted within a delusional framework. Once such a framework has been established, then even benign behaviors from relatives can be misinterpreted as malevolent, fuelling psychotic symptoms. Thirdly, relatives’ behavior could impact on psychosis by increasing negative emotion and arousal levels in the service user (link 7), which in turn impacts on information processing, including reasoning skills, generating further misinterpretation of triggering events (link 8). Brown et al. (
1972) suggested that in high EE families, the environment was too over-stimulating, and that, consistent with the stress vulnerability model of psychosis (Zubin and Spring
1977), this acted as a direct trigger for psychosis. Support for this hypothesis comes from psychophysiological studies that show elevated autonomic arousal levels in service users with high EE relatives, compared to those with low EE relatives (Tarrier and Turpin
1992), and from self-reported elevated stress levels from service users in the presence of high EE relatives, compared to those with low EE relatives (Cutting et al.
2006). Further support comes from studies using functional magnetic resonance imaging (fMRI) which show enhanced activation of brain regions concerned with processing of aversive social information in response to hearing relatives’ critical comments compared to neutral comments, suggesting a potential neural basis to the impact of high EE environments on outcome (Rylands et al.
2011).
A major limitation of the EE research is that it has tended to study a dichotomy of high versus low EE focusing heavily on the characteristics of high EE including criticism, hostility and emotional over-involvement, with far less investigation of the characteristics and impact of low EE relatives, or the specific impact of warmth and positive remarks. What little has been done, has shown that this is potentially a very important area of investigation that could provide valuable insights into how to develop more effective solution focused treatments for psychosis that involve relatives. Cross sectional associations between warmth in relatives and satisfaction with life in people with psychosis have been shown (Greenberg et al.
2006), but more significantly, prospective studies show a predictive relationship which strengthens the argument for a causal link between positive family environments and outcome (López et al.
2004). In an attempt to replicate the original EE studies of Brown and Birley (
1968), the link between high EE and subsequent relapse was reproduced (Bertrando et al.
1992), but in addition, the authors found that high levels of warmth reduced the risk of relapse over 9 months, and led to lower admission rates, even within families that were also rated as high EE. The protective impact of positive family environments has also been demonstrated in adolescents at risk of psychosis (O’Brien et al.
2006; Schlosser et al.
2010) and following first episode of psychosis (Lee et al.
2013), and highlighted as an important moderator of the negative impact of EOI in some cultures (Singh et al.
2013). These studies support the independence of the negative and positive ratings within EE and suggest that the tendency to categorise relatives as high or low EE is too simplistic to capture the multidimensional complexity of family relationships and how they impact on outcome.
From a more positive perspective, supportive behaviors from the relative could reduce vulnerability to psychosis by firstly challenging negative core beliefs about self/world/others and confirming more positive beliefs. Warmth and supportive behavior, would lend support to positive beliefs about the self and others and increase drive towards positive social interactions (link 5). Positive self-evaluation is an even stronger predictor of time to relapse than negative self-evaluation (Holding et al.
2013) so relatives who can build this, even in the presence of continuing negative self-evaluation may be able to increase resilience. Secondly, relatives can provide an alternative perspective to the misinterpretation of triggering events. Several groups have pointed out that social isolation reduces access to alternative and normalising explanations for anomalous experiences, and that the failure to be part of a normalising social network is one factor distinguishing those who develop psychosis from those who do not (Hodges et al.
1999; Van Os et al.
2000).
Relatives’ behavior, if supportive and calming could reduce arousal levels, increasing information processing capacity, and directly trigger positive emotions (link 7). These emotions may in turn initiate “upward spirals” of positive affect which have several potential beneficial effects (Garland et al.
2010). Firstly, the immediate cognitive and emotional benefits of positive emotion are likely to directly impact on common experiences associated with psychosis. The misinterpretation of ambiguous information has been identified as an important underlying cause of both hallucinations, in the form of misinterpretation of anomalous experiences (Morrison
2001) and delusions, in the form of cognitive biases towards jumping to conclusions (Garety et al.
2001). The broaden and build theory of positive emotion (Fredrickson
1998,
2001,
2003) postulates that, in the same way that negative emotion has been shown to narrow cognition and focus behavior to specific survival responses, positive emotion leads to a broadening of cognition and an increase in behavioral flexibility (link 8). Effects include broadening the scope of visual attention (Fredrickson and Branigan
2005; Rowe et al.
2007), expanding people’s repertoires of desired actions (Fredrickson and Branigan
2005), and their openness to new experiences (Kahn and Isen
1993), and critical feedback (Raghunathan and Trope
2002). Further effects at the interpersonal level, include an increase in people’s sense of “oneness” with close others (Waugh and Fredrickson
2006), and their trust in acquaintances (Dunn and Schweitzer
2005). Linking this to psychosis, we can see how positive emotions triggered by warm supportive relative’s behavior could reduce cognitive biases by triggering this more broaden and build perspective and consequently reduce vulnerability to psychotic experiences (link 8). A further common, and often equally debilitating experience in psychosis, is the loss of anticipatory pleasure for activities of life (Kring
1999), leading to lack of motivation to engage and general withdrawal, often referred to as negative symptoms. It is easy to see how a vicious cycle is created in which the loss of anticipatory reward and greater social withdrawal become entwined. Attempts to break this cycle often involve exposure to situations which may trigger positive affect to a level that can ignite anticipatory pleasure in future exposure (Tarrier
2010). Relatives who behave in warm supportive ways with a degree of consistency which can ignite anticipatory reward may therefore generate both immediate positive affect which in turn also increases the likelihood that the person with psychosis will expose themselves to other potentially rewarding social interactions (link 3), reducing the risk of withdrawal and isolation associated with long term mental health problems and breaking the vicious cycle thought to underlie negative symptoms.
Relatives’ Responses are Partly Determined by Their Appraisals
Consistent with the cognitive model, there is now good evidence that relatives’ appraisal of the service user’s behavior is an important determinant of their emotional and behavioral response (link 10). First to explore this was Brewin et al. (
1991) who found that carers rated as high EE on the basis of criticism or hostility were more likely to make controllable and personal attributions than over-involved or low EE carers (Brewin et al.
1991). Since then, there have been many studies exploring how relatives’ attributions impact on their responses (see Barrowclough and Hooley
2003 for a review). In summary, the attribution style studies suggest that relatives who behave in highly critical ways are more likely than those expressing low criticism to believe that service users are substantially in control of the negative events that relatives experience. They are also more likely to ascribe them greater personal responsibility for these negative events. Underlying personal responsibility attributions are judgements that the behavior of the service user is a result of factors that are internal and personal to that individual—but also could be controlled by them if they wished. Responsibility appraisals are even more apparent in relatives rated as hostile as well as critical. This helps explain why behaviors such as substance misuse, negative symptoms and violence are more likely to lead to critical or hostile responses in relatives. These behaviors are less obviously “symptoms” of an illness, appearing in the non-psychosis population and generally construed as under active control.
Most of the evidence for links between underlying beliefs and relatives’ behavioral responses (link 11) has come from coding of CFI transcripts. Hooley and Campbell (
2002) used this methodology to demonstrate that, making more attributions of control is associated with behaving in more controlling ways, suggesting that the attributions may be driving behavioral responses which may in turn be linked to relapse in the service users. In contrast, relatives rated as high EOI, who tend to behave in ways that “buffer” the service user from the demands of life, tend to make very few attributions of responsibility to service user for any of their behaviors. This pattern has been described as “victim appraisals” in which the service user is seen as a victim of psychosis (Barrowclough and Hooley
2003). As a consequence, high EOI relatives often take a lot of responsibility for both the development of the psychosis, and the process of recovery. These associations were initially identified in a sample of relatives of people with chronic psychosis (Barrowclough et al.
1995; Hooley et al.
1987), but have recently been replicated in a recent onset group (Vasconcelos e Sa et al.
2013). Despite far less exploration of attributions associated with low EE, there is evidence to suggest that low EE relatives make what have been described as “survivor appraisals”. They tend to see the service user as less responsible for negative events than high EE critical relatives, but more responsible for positive events (Grice et al.
2009).
Relatives’ attributions also determine their emotional responses (link 13), in particular distress levels. Unsurprisingly, relatives who blame themselves for the mental health problems of their family member show higher levels of distress (Barrowclough et al.
1996; Boye et al.
2001; Fortune et al.
2005). This association has also been replicated in recent onset families, in which the most common self-blaming attribution was a perceived failure to recognise and respond to early signs of illness (Vasconcelos E Sa
2014).
Research exploring attributions underlying relatives’ responses has been immensely useful in guiding the development of our understanding of interpersonal dynamics in families of people with psychosis and in developing effective interventions which try to identify and modify attributions (e.g. Barrowclough and Tarrier
1992; Kuipers et al.
2002). This area of work is still developing, and recent advances include the wider exploration of beliefs about psychosis, beyond focussing on attributions about specific behaviors, and the insight that the interpersonal dynamic may be better understood as a function of the discrepancy between beliefs held by service users and relatives, rather than understanding just one perspective.
The Self-Regulation Model (Leventhal et al.
1984), applied to psychosis (Lobban et al.
2003) proposes that relatives develop working models of psychosis (as they would with any illness) which helps them to make sense of their experiences and guides their coping strategies. Specifically, they will hold beliefs along a number of dimensions including the identity of the illness, likely consequences, the controllability, the cause and the likely timeline. Barrowclough et al. (
2001) found that the number of critical comments made by relatives was associated with a perceived greater frequency of symptoms, even when controlling for an objective measure of illness severity. The greater the criticism, the less sense there was of the illness being amenable to control/cure and the less able relatives felt to control the illness themselves. Finally, relatives rated as high EE perceived a more chronic timeline for the illness. This work suggests that a wider exploration of relatives’ beliefs that goes beyond attributions of control and responsibility may highlight other key beliefs that underlie distress in relatives, and or interpersonal difficulties with the service user, and which may provide fruitful targets for therapy. However, it is unlikely that understanding the relatives’ model of psychosis in isolation will provide the whole picture. EE reflects the quality of the relationship between the service user and relative from the relative’s perspective. Relationships are by definition between two or more people. Therefore, it is likely that the impact of beliefs held by the relative about the illness will depend upon how much they are in (dis)agreement with the beliefs held by the service user. Lobban et al. (
2006) were the first to test this in psychosis and found that a comparison between models held by high and low EE relatives showed no significant differences between the groups—but when discrepancy scores were compared which showed the difference between the service user and relatives beliefs within each dyad, high EE dyads showed greater levels of discrepancy than was seen in low EE dyads, with the relatives tending to hold a more negative overall model of illness that then service user. Kuipers et al. (
2007) used the same SRM framework with a larger sample of dyads. Although they found no direct link between illness beliefs and EE, they did show that the discrepant views were related to greater distress, depression and lower self-esteem in both service users and relatives. Taken together, these studies support the idea of more dynamic interpersonal application of the CBT framework in which the impact of beliefs is recognised as being dependent on the degree of discrepancy with those of significant others.
Consistent with the SRM, underlying attributions, appraisals, and illness beliefs are all important because they impact on the coping styles of the relatives, and on their emotional responses (links 13 and 11). As with much of the work in this area, there has been too little focus on understanding the underlying beliefs, appraisals, and working models of relatives who are able to manage psychosis without high levels of distress and who are able to successfully support the service user through the process of recovery. In addition to the identification of the “survivor appraisal style” characteristic of low EE relatives and described above, some interesting qualitative work (Treanor et al.
2013) has highlighted other key factors which may help us understand what we need to be working towards in supporting relatives. In a small study in which eight relatives rated as low EE on the CFI were interviewed in depth about their experiences of supporting a close relative with psychosis, the authors identified key themes underlying the relatives’ responses. The relatives shared an acceptance that they were unable to change what the service user was experiencing or doing—but an ongoing commitment to support them with managing these experiences. They demonstrated a deep emotional understanding of how the service user was feeling, and had complex working models of the cause and maintenance of the problems. Coping styles focussed around humour, distraction and time out, and downward social comparison—recognising that things could be (and often are) a lot worse for others. Characteristic of the relatives interviewed was the presence of realistic optimism for the future, characterised by an acceptance of a change in life course for the service user rather than perceiving a failure to achieve previously identified goals. This preliminary work highlights how much more we can learn from in-depth interviews with relatives who have already nurtured the relationships we aspire to achieve through clinical interventions.
Finally, the SRM (and other cognitive models) would suggest that behavior is not determined by cognitive representations alone—but also directly by emotional representations, and there is some evidence to support this assertion in relation to relatives’ responses to psychosis (link 12). Anxiety, fear and grief have all been explored in this context. Relatives’ coping strategies and behavioral responses to the service user may reflect their attempts to manage their emotional responses to psychosis and allowing relatives to express and work through these emotions may facilitate behavior change. For example, Greenley (
1986) found that high EE was associated with relatives being more fearful and anxious and suggested that their controlling behavior was a way of trying to manage these emotions. Patterson et al. (
2005), showed strong links between high EOI and loss, and suggested that use of buffering behaviors were attempts by the relative to deal with loss, and that over time this loss would either reduce, with an associated switch towards low EE and more supportive behavior, or remain and this could lead to critical and controlling behavior in an attempt to change the situation and get the service user to return to pre-morbid functioning. This has interesting implications for working therapeutically with relatives to facilitate the grieving process, and also highlights the dynamic nature of relatives’ responses and factors that may influence change over time.