In order to tackle the widely reported stigma around mental health [1], the Time to Change anti-stigma campaign was launched in England in 2007, followed by the 2011 White Paper Strategy No Health Without Mental Health. In 2012, the Compassion in Practice 6C initiative focused on putting the person at the centre of National Health Service (NHS) practice [2], which was subsequently reflected in with the NHS Constitution [3]. Since the introduction of anti-stigma campaigns, the level of public discrimination has reportedly reduced [4,5], but a third of mental health service users continue to report stigma and discrimination when accessing whilst the level of discrimination in mental health services and other parts of the NHS remains static [6]. More recently, the Care Quality Commission [CQC] [7] identified that services for those experiencing a mental health crisis often lack basic respect, warmth and compassion.

      However, crisis and acute inpatient services are a crucial part of mental health care, providing support and care when people are most vulnerable and unwell [8]. Working in acute wards is demanding, with staff having to manage difficult interactions and challenging behaviour. In addition to other stressors associated with nursing, psychiatric nurses experience increased stress given the risk of violence and suicidal behaviour in service users [9] whilst having to be contain their emotional responses and attitudes in order to develop and maintain positive therapeutic relationships [10].

      Research on staff attitudes towards individuals experiencing mental health difficulties is mixed as although positive attitudes are found, there is evidence of negative attitudes and expectations, particularly with respect to social acceptance of people with mental illness [11]. Differences relating to gender [12] have been reported but not universally [13] as well as age with younger nurses holding more favourable views regarding mental health difficulties. Differences between qualified and nonqualified staff are also reported, with the latter holding both more positive and negative compared to qualified staff [12].

      To date, most studies have used questionnaires together with case vignettes and/or interviews to explore attitudes towards service users, methods that are prone to bias due to social desirability [14]. In addition, interpretation of such can be problematic with researchers making assumptions about participants’ responses and intended meaning. In contrast, the repertory grid technique [15] allows exploration of staff perceptions about actual clients without making interpretations about participant responses, [15]. Repertory grid technique is based in [16] grounded in Personal Construct Theory (PCT), which postulates that people actively form representations (constructs) of others (elements) to understand the world around them and to make predictions about likely outcomes and patterns of behaviour through the exploration of similarities, differences and themes based upon previous experience. This technique facilitates elucidation of people’s views through their own idiosyncratic language, thus minimising the impact of social desirability bias [15]. Repertory grids have been used to explore staff attitudes towards clients with mental health difficulties [17-19]. Two recent repertory grid studies exploring staff construal of clients with a dual diagnosis [20] and of mothers with mental health problems [21] identified that all staff made critical judgements about some clients, whilst clients with a personality disorder and those considered a ‘bad’ mother were viewed as most different to the self [21] and clients with a dual diagnosis and substance misuse were negatively construed, with staff having less judgemental views towards non-clients who used substances [20].

      To date, no studies have used repertory grids to explore staff views of service users admitted to an inpatient psychiatric ward. Therefore, the current study explored whether the cognitive complexity of staff influences construing of clients (including those with first time admissions compared to those with two or more admissions) and how inpatient ward staff construe clients and non-clients in comparison to the self, as well examining the relationship between construal of individual client elements.

Design

      A series of repertory grids were elicited for each participant, providing an idiographic representation of each person’s construal of clients, non-clients and themselves. The study was granted full ethical approval by the University of Manchester’s Research Ethics Committee and the local Research and Development department.

Participants

      All staff working on an inpatient ward with 31 beds at a North West England NHS Trust were invited to participate if they had a minimum of 12 months experience working on the ward, were permanent members of staff (part- or full-time) and were proficient in English in order to provide informed consent and participate in the interview. Nursing students and ‘bank staff’ were excluded from the study.

Data collection

      The 31-bed-ward provided inpatient care, including the assessment, development and implementation of individualised care programmes to both male and female adults of working age experiencing significant mental health difficulties. Clients were ‘informal’ voluntary admissions or had admitted under the MHA [22]. The ward had a team of medical staff, occupational therapists, nurses and nursing assistants.

      Demographic information relating to the participants (e.g., age, gender, ethnicity, current job role, time (in years) working on the ward) was obtained.

      The brief 21-item Depression, Anxiety and Stress Scale (DASS-21) [23], designed to measure the severity of a range of symptoms common to stress, anxiety and depression, was used to provide an overview of participants’ wellbeing and to provide additional contextual information. Items were rated on a 4-point-likert scale indicating the presence of a symptom over the previous week. Each item is scored from 0 (did not apply to me at all over the last week) to 3 (applied to me very much or most of the time over the past week).

Data analysis

      An audio-recorded semi-structured repertory grid interview [15], lasting approximately one hour was completed with each participant. Participants were presented with the seven elements and asked to think of clients (either current or those who had left the ward) that fitted each category element: 1) an individual client who has a first time admission to XX Ward, 2) an individual client who has a second or more admission to XX Ward, 3) an individual client with a dual diagnosis (client with substance misuse), 4) An individual client who you (i.e., staff member) find easier to care for, 5) an individual client who has been compulsorily detained, 6) a client with whom you had a difficult professional relationship and 7) your hypothetical ideal client. Four further elements relating to the participant and their non-working life were included: 8) a family friend or relative with mental health difficulties, 9) a person you care about/care for, 10) yourself (now) and 11) your ideal self.

      The triadic opposite method was used to generate constructs [15] with participants presented with three randomly selected elements and asked in what way two were similar to each other (emergent construct) and different from the third. Participants then generated the opposite end of the construct (implicit construct), resulting in bipolar constructs with corresponding behavioural descriptions for each end of the construct. Participants were then asked to rate each element on a 5-point-scale along the construct poles that had been elicited, with the emergent construct rated as 1 and the implicit construct as 5. Those elements that had been used to elicit the construct were rated first, the remaining elements were then presented in a random order and rated along the scale. This process was repeated with random element combinations until the participant could not generate any further constructs. During a second interview, participants were presented with a visual representation (biplot) of their grid and a list of elements that were highly correlated. Participants were asked to discuss the findings and whether the analysis was a reasonable explanation of their views [15]. At the end of the study participants received £10 as a thank you for volunteering their views and time.

      Idiogrid version 2.4 [24] was used to analyse the repertory grid data, with standardised Euclidean distances identifying elements most similar / dissimilar to each other. Elements have an expected inter-elemental distance of 1.00 and were considered significantly ‘similar’ if the inter-element Euclidean distance was less than 0.50 and significantly ‘different’ if greater than 1.50 [25]. Principal Components Analysis of each repertory grid plotted the relationships between constructs and between elements (biplot) which enabled the most relevant constructs that defined each element to be identified, i.e. the element that was closest to the construct on the biplot. Eigenvalues (percentage variance accounted for by each principle component) were calculated with the eigenvalue for the first principal component measuring the complexity and ‘tightness’ of an person’s construing, with higher eigenvalues indicating less complex and more tight construing [25].

      A content analysis of individual participants constructs using the Classification System for Personal Constructs [26] (CSPC) was undertaken with each construct being categorised in one of seven CSPC domains (moral, emotional, relational, personal, intellectual and operational, values and interests and other).

Participant characteristics

      Twelve staff (2 male; 10 female) from a staff team of 48 participated and completed the repertory grid interview with 10 participants attending individual feedback sessions. A sample of 12 participants is a sufficient size for a repertory grid study [20, 21]. Mean age of participants was 39.2 years (23 to 67 years), with nursing staff having a mean of 4.8 years of experience working on the ward and nursing assistants a mean of 7.3 years of experience (1-16 years) [Table 1]. All staff indicated that nonwork related factors were the principal cause for their reported mood difficulties.

      All twelve participants identified individuals who fitted the client elements, eight participants provided nine constructs, one provided twelve constructs and three participants provided seven constructs.

Cognitive complexity

      Principal Components Analysis was used as a measure of ‘cognitive complexity’ of the construal of clients and non-client elements. Repertory grids with no principal components are fragmented; one principal component indicates monolithic structure and two or more principal components are indicative of cognitive complexity [27]. Additionally, if the first principal component accounts for a high percentage of the variance, an individual is using fewer dimensions to construe the behaviour of others, which is a further indication of the level of a person’s cognitive complexity [27], indicating that a more varied and differentiated way of construing other people’s actions [28].

      Two participants (P2 & P7) had less cognitively complex repertory grids with over 80% of the variance being accounted for by the first principal component (82.48% and 91.67%, respectively). In the bi-plot for P7 (Figure 1), the construct lines were close together indicating they were highly correlated. Element ratings are represented by the proximity of the elements to each construct and for grids that are less cognitively complex there is less distance between construct lines. In contrast the construing of clients by P1, P3 and P6 were very complex (i.e., both had three principal components). The remaining eight participants showed moderately complex construct systems (i.e., two principal components), also indicating differentiated construing of others. When the cognitive complexity of participants’ bi-plots were taken into account, a pattern emerged which suggested that staff members whose bi-plots demonstrated less cognitive complexity tended to make a greater distinction between themselves and clients in comparison to staff members whose grids demonstrated greater cognitive complexity.

Construal of clients, non-clients and the self

      Biplots were computed for each repertory grid and Figures 1 and 2 show the biplots for P7 and P10 respectively, exemplifying the diversity of construing among individual participants in relation to clients, non-clients and the self. P7 had 12 tightly construed constructs (Eigenvalue= 91.6%), in contrast P10 had eight constructs which were more complicatedly construed (Eigenvalue = 62.12%). The elements of the client with a) two or more admissions to the ward, dual diagnosis, b) who was compulsorily detained and c) with whom you had a difficult professional relationship were construed as most dissimilar to the self by P7 (Euclidian distance = 1.45, 1.48, 1.39, 1.45 respectively) compared to P10 who only identified the client with two or more admissions to the ward as most dissimilar to the self (Euclidean distance = 1.65) 0 (Table 2). P7 construed family friend or relative with mental health difficulties as similar to a client who you find easier to care for (Euclidian distance = 0.30); however, P10 also construed family friend or relative with mental health difficulties as being similar to a client with a first time admission to the ward, dual diagnosis and client who was compulsorily detained (Euclidian distance = 0.46, 0.53 & 0.46, respectively).

      As per Table 2, participants rated themselves as most similar to the non-client elements of someone you care for/care about (mean Euclidean distance = .56) and the ideal self (mean Euclidean distance = .58) as well as to the hypothetical ideal client (mean Euclidean distance = .67). Only P6 rated the non-client element family friend or relative with mental health difficulties as similar to themselves. Five participants (P1, P4, P5, P8 & P12) identified themselves as similar to any of the client elements (excluding the hypothetical ideal client), with two participants (P1 & P4) identifying themselves as most similar to a client who has a first time admission to the ward (Euclidean distance = .51 and .55 respectively). P5 identified themselves as most similar to a client with two or more admissions to the ward (Euclidean distance = .56), and P8 and P12 identified themselves as most similar to the client whom you find easier to care for (Euclidean distance = .46 and .52 respectively). Five participants (P2, P5, P7, P8 & P12) rated client with whom you’ve had a difficult professional relationship (mean Euclidean distance = 1.31) as significantly different from the self, with three different client elements being rated as significantly different by three participants and one element (client with two or more admissions to the ward) being seen as significantly different from two participants (P7 & P10).

Relationship between construal of the individual client elements

      Client who has a first time admission to the ward. The construal of clients who have a first time admission to the ward was varied, with P2 and P7 construing them in a more negative way, putting them closest to the constructs of “not thinking rationally”, “being impulsive”, “self-centred”, “impatient”, “passive” and “aggressive” . This was in contrast to other participants, who viewed this client group as similar to other client element, including Client who find easier to care for and hypothetical ideal client and to non-client elements, especially a family friend or relative with mental health difficulties. Content analysis using the CSPC [26] revealed constructs relating to this client group included moral values, in particular staff construed this client group as being concerned about others (‘altruist’ ) and ‘sincere’, as well as describing clients as emotionally ‘warm’. Staff also construed the relational style of clients with first time admission as more ‘introverted’ and, ‘peaceable’, as well as being more ‘conformist’ in relation to suggestions and support.

      Client with two or more admissions to the ward. Staff construal of clients with two or more admissions to the ward was also varied but only P11 viewed such clients as similar to their hypothetical ideal client or client with whom they found easier to care for. Two participants (P1 & P10) identified that they were significantly different from these client groups. Three staff (P9, P11 & P12) identified this client group as significantly similar to a family friend or relative with mental health difficulties. Although construed in a varied way by staff, content analysis [26] revealed that for this client group participants identified constructs that can be identified within the category ‘rigid’ which relates to their personal way of being as well as the categories ‘rebel’ and ‘aggressive’ describing their styles of relating to others. Constructs can be identified within the ‘moral’ category of the CSPC which is concerned with the moral value of the person or element and is based on a judgement around the person’s moral character (e.g., good, altruistic, proud). Some participants viewed clients with two or more admissions as ‘insincere’.

      Client with dual diagnosis. All participants construed the client with a dual diagnosis towards the negative end of the pole, with the element being aligned with moral constructs relating to only thinking of oneself (‘egoist’) and being untrustworthy (‘insincere’). Relationally clients with a dual diagnosis were construed as ‘aggressive’ and ‘unsympathetic’. Additionally, using the categories identified by the CSPC some participant constructs relating to the area of personality and way of being identified this group of clients as ‘lazy’ but also unsure of self/insecure (‘self-criticism’). Clients with a dual diagnosis were construed as most similar to clients who were compulsorily detained and those with whom staff had a difficult professional relationship, with four participants identifying significant similarities between these elements (P7, P9, P10 & P11). Although only one participant viewed clients with a dual diagnosis as distinct from themselves (P7), the data suggest that participants did construe themselves and non-client elements of someone care for/care about in a different way to this client group, but less different from a family friend or relative with mental health difficulties. Only P9 viewed this client group to be significantly similar to a friend/relative with mental health difficulties.

Client whom you find easier to care for and hypothetical ideal client.

      The elements a client who you find easier to care for and hypothetical ideal client were considered to be most similar to the self and to the non-client element of someone you care for/ care about, with four participants (P1, P8, P10, P12) construing either one or both of these as significantly similar to the client whom it is easier to care for. Seven participants (P5, P7, P8, P9, P10, P11 & P12) also construed either one or both of these elements as similar to the hypothetical ideal client element. Three participants (P5, P8 & P12) identified the client who you find easier to care for and the hypothetical ideal client as significantly different from the client with whom you have a difficult professional relationship. Additionally the hypothetical ideal client was construed as significantly different from the client who has been compulsory detained (P8, P9 & P12). Only P2 identified the client who you find easier to care for as significantly different from the self and someone you care for/care about and most similar to client with whom you have a difficult professional relationship and a client with a dual diagnosis. P2 explained that: “these clients engage in the same behaviours on the ward but this client [client you find easier to care for] it is due to their mental health difficulties and so can’t help it, whereas this person [client with whom had difficult relationship] it’s who they are.”

      Content analysis revealed that these two client groups were mainly aligned to personal qualities of being ‘hard working’, ‘flexible’, as well as having a relational style that was ‘peaceable’ and ‘independent’, and were considerate of others (‘altruist’).

      Client who has been compulsorily detained and client with whom you’ve had a difficult professional relationship. Clients who have been compulsorily detained and those with whom staff had a difficult professional relationship with were considered to be most similar with four participants (P6, P7, P8 & P9), identifying significant similarities between these client groups. Only P5 identified a significant difference between these clients. Two participants (P5 & P10) identified that clients who had been compulsorily detained were significantly similar to non-client elements of a family friend or relative with mental health difficulties, someone you care for/care about and the self; however, no participants identified any similarities between non-client elements and the client with whom you’ve had a difficult professional relationship. Eight participants identified significant differences between either all or some of the non-client elements and the client with whom you’ve had a difficult professional relationship, with P5 identifying the most significant differences. Least differences were construed between a family friend or relative with mental health difficulties and this client group. Content analysis revealed that both client groups were construed towards the negative end of constructs with participants providing constructs that related to relational styles that were categorised as ‘rebel’ and ‘aggressive’. Furthermore, constructs were categorised as ‘visceral’ in regards to their emotional attitude towards life, suggesting that participants saw clients as ‘impulsive’, ‘emotional’ and ‘reactive or quick to temper’. In terms of moral judgements made about these two client groups, participants identified constructs under the category ‘egoist’ which indicated these client groups tended to think more about themselves than others. Some differences were observed with some participants construing a client who had been compulsorily detained as more ‘dependent’ and ‘introverted’ in their relational style.

Differences between participant demographics and construal of clients

      Staff nurses and nursing assistants. Independent samples Mann-Whitney U test revealed no significant difference between Euclidian distances for employment type (staff nurses and nursing assistants) and any of the client and non-client elements ((Mdn = 8.93), U = 22.0, z = 1.02, p = .368, r = .29), suggesting that both staff nurses and nursing assistants construed the self to clients and non-clients in the same way. Although no significant difference was found between the two groups, visual scrutiny of the Euclidian distances suggested that in this group staff nurses viewed themselves to be more similar to clients with two or more admissions to the ward than nursing assistants (mean Euclidian distance = 0.89 and 1.25 respectively). Both staff nurses and nursing assistants construed the hypothetical ideal client as most similar to themselves (mean Euclidian distance = 0.67 and 0.68 respectively) and client with whom had a difficult professional relationship as most different from the self (mean Euclidian distance = 1.36 and 1.22).

      Gender, years working on ward and age of participants. Only two male staff participated in the study and visual scrutiny of the Euclidean distances indicated that they appeared to construe clients with two or more admissions to the ward and clients with a dual diagnosis as more similar to themselves than female participants. It was not possible to compare staff constructs based on age or years working on the ward given that the groups were heavily weighted by employment type.

Content analysis

      Of the 103 constructs elicited, the majority were relational (n= 42), with remaining constructs mostly related to moral (n=23), personal (n=20) or emotional (N=13). The intellectual and operational category, essentially relating to personal insight, and the supplemental existential category accounted for four and one constructs respectively. The individuality of construing was apparent during the content analysis. For example, a number of participants used the word “caring” to describe the positive end of a construct to describe someone who was not demanding and who was thoughtful around the demand for staff and them not being able to meet their needs immediately, others used the construct “accepting” or “patient”, to describe similar moral attributes. This highlights that the detailed descriptions of the constructs were important, because individual words held different meanings for each participant.

      The present study explored the meanings constructed by staff in the course of working on an acute inpatient ward and identified that very few staff members construed clients in a similar manner to their construal of themselves and highlighted the differences between the construal of different client groups, with clients with first time admission to the ward being construed more positively than other client groups.

      Less than half of participants construed clients (excluding hypothetical ideal client) as similar to themselves, with only those participants whose grids showed greater cognitive complexity indicating less distinction between the construal of the clients, themselves and non-clients. Clients with more complex presentations, in particular those who were admitted to the ward on two or more occasions and those with whom staff had a difficult professional relationship, were construed as most dissimilar to the self. Previous research has noted that staff expressed increased negative attitudes and less empathy towards clients with a diagnosis of borderline personality compared to those with a diagnosis of major depressive disorder or generalised anxiety disorder [29]. Blundell et al [21] reported that mothers with psychosis or personality disorder were construed towards the negative end of the pole and that attributions about the clients’ behaviour as well as the clients’ interactional style negatively influencing the ability of staff members to develop positive relationships. In the current study, the interactional styles of clients with more complex difficulties were generally perceived to be more demanding and aggressive.

Limitations

      Only 25% of the staff team and were a self-selected sample. Whilst representative in terms of age and years working on the ward, it was not representative of the employment type or gender. As the findings suggest that the two male staff members construed clients somewhat differently, it would have been beneficial to see if this was representative on a wider scale. A further limitation is that participants were free to select any individuals who they felt matched the provided elements. As a result, it is possible that participants selected different individuals to represent the same element. Furthermore, for the element of a family friend or relative with mental health difficulties it was apparent during the interview that there was significant variation in the degree of mental health difficulties of the individuals selected for this element. Participants were not explicitly asked to identify what difficulties this person experienced and if they thought of individuals with less complex difficulties, this may account for differences in construal between clients and family friend/relative with mental health difficulties.

Clinical implications

      A number of recommendations arose from the findings of this study. Some of which are outlined in Table 3.

      According to Edwards [30], new staff members can feel stifled by ward culture in particular around negative stereotypes attached to service users with conditions such as personality disorders. For staff who have past experiences of individuals there is also the risk that negative evaluations of individuals based upon prior interactions will reduce the potential of optimum care when clients return to the ward. Thus, supporting staff to formulate and understand an individual’s distress, behaviour and needs may help staff to focus away from anticipated constructs and biases. In the current study, some participants expressed that having the time to identify similarities between clients was helpful, because it was something they had not previously done and allowed them to think of clients whom they had originally perceived in a particular way and to consider them in a different manner. Additionally, following the feedback of the pingrids, one participant identified that the process had enabled them to consider how they worked with clients, in particular they reflected on how they could ensure there was not as great a distance between them self and the client whom they had a difficult professional relationship with.

      Compassionate care from nursing staff has been aligned with actions which can often take time as well as fleeting actions [31]; however, it has been highlighted that the administrative duties required in nursing impacted on the time staff were able to spend with clients, in particular staff reported that these duties were the dominant culture on the wards and was seen as having greater importance than valuing human contact [30]. Edwards [30] also highlighted that some nurses saw their role as patient management and containment, especially as the focus on risk is now central in the provision of mental health care [32]. Staff constraints on time and completing demands to support clients and undertake other duties was discussed by participants with staff construing some clients who wanted immediate support (in no risk situations) as demanding or selfish.

      In the current study staff described the context under which their perceptions were reached, with all staff describing personal experiences they had faced on the ward. Staff described challenging situations they had been placed in by clients, especially when clients were hostile or aggressive. The NHS constitution [3] identifies that staff themselves should receive compassion as well as clients, therefore in situations when staff experience aggression it is understandable that cognitive and emotional reactions are elicited. It is therefore important for staff to have support mechanisms to manage such stressful situations, especially given that individuals are more prone to socially biased decisions and behaviours when highly stressed or tired and when decisions need to be made quickly [33]. Support mechanisms, such as clinical supervision with a trained professional [34] are important. Similarly, providing access to therapeutic support such as mindfulness has also been found to reduce burnout [35] and reduce stigmatizing attitudes [36].

Further research

      Previous research used repertory grid technique to explore clients’ perspectives of nursing staff and compassionate care highlighting that clients found staff who were flexible in their approach to be more supportive which facilitated recovery [37]. Understanding how clients admitted to larger wards construe staff should identify if staff resources impact on the relationship between compassion experiences and recovery. Future research could examine staff perceptions of their clients by recruiting participants from different locations and it could also explore if there are differences in the construal of clients between acute inpatient ward staff and community mental health nurses. Staff working in these areas hold more positive attitudes towards clients because they are most likely to see service users recover and return to independent living compared to staff who work within residential settings [38].

      In this study, nursing staff whose grids showed greater cognitive complexity demonstrated more perceived similarities between themselves and their clients admitted to an acute inpatient ward. Yet clients with more complex presentations were considered to be most different from the self by all staff members. These results underlie the importance of nursing staff having support mechanisms, including clinical supervision with a trained professional for staff to voice difficult reactions towards clients in order to contain anxieties and to formulate and understand an individual’s distress, behaviour and needs. This will not only improve the well-being of staff but ultimately contribute to improved care for service users.

      The authors would like to thank the staff who participated in this study and shared their views with us and also the Ward Manager, Amanda Jackson, for their strong support of this study. We would also like to express our gratitude to Brigid Corrigan and Adam Morris for their support with this study.