Perspectives of crisis intervention for people diagnosed with 'borderline personality disorder': an integrative review

WHAT IS KNOWN ON THE SUBJECT?: People diagnosed with "BPD" often experience crisis and use services "BPD" is a controversial diagnosis, and the experience of crisis and crisis intervention is not well understood WHAT THIS PAPER ADDS TO EXISTING KNOWLEDGE?: People diagnosed with "BPD" have different experiences of crisis, and using the diagnosis alone as a basis for deciding care and treatment is not appropriate There are many human factors which can influence how professionals deliver care to people diagnosed with "BPD" WHAT ARE THE IMPLICATIONS FOR PRACTICE?: The education of staff, views on responsibility, team conflicts and access to clinical supervision can have an impact on how care is delivered, and should be addressed by organizations providing crisis care. Access to care often occurs when a person is self-harming or suicidal, but does not address underlying distress. Crisis care should go beyond managing behaviour and address any underlying needs. ABSTRACT: Introduction "Borderline personality disorder" ("BPD") is associated with frequent use of crisis intervention services. However, no robust evidence base supports specific interventions, and people's experiences are not well understood. Aim To explore the experiences of stakeholders involved in the crisis care of people diagnosed with "BPD." Method Integrative review with nine databases searched January 2000 to November 2017. The search filtered 3,169 titles and abstracts with 46 full-text articles appraised and included. Results Four themes were constructed from thematic analysis: crisis as a recurrent multidimensional cycle, variations and dynamics impacting on crisis intervention, impact of interpersonal dynamics and communication on crisis, and balancing decision-making and responsibility in managing crisis. Discussion Crisis is a multidimensional subjective experience, which also contributes to distress for family carers and professionals. Crisis interventions had limited and subjective benefit. They are influenced by accessibility of services, different understandings of "BPD" and human dynamics in complex decision-making, and can be experienced as helpful or harmful. Implications for practice Subjectivity of crisis experiences shows limitations of the diagnostic model of "BPD," emphasizing that interventions should remain person-centred. While thresholds for intervention are often met after self-harm or suicidality, professionals should review approaches to care and support people with underlying distress.

Papers from January 2000 to November 2017 were accessed through several databases, ensuring the search was comprehensive. The nine databases were; Cochrane Library, CINAHL, Medline, SocIndex, PsycINFO, PsychARTICLES, Web of Science, Knowledge network and ProQuest. To ensure completeness, further articles were identified through the reference lists of included papers (Aveyard, Payne and Preston, 2016). The SPICE (setting/perspective/intervention/comparison/evaluation) framework (Booth, 2004(Booth, , 2006 was used to develop an effective search strategy and refine the questions being asked (See table 1).  (Booth 2002(Booth , 2004 Setting All settings in UK and Ireland, Continental Europe, Europe, USA, Canada, Australasia and New Zealand

Perspective
People with a diagnosis of 'borderline personality disorder'/'emotionally unstable personality disorder', their family carers and professionals involved in their care.

Intervention
Crisis Intervention for people diagnosed with 'borderline personality disorder'/'emotionally unstable personality disorder'

Comparison
Comparison may be drawn between: • The perceptions and experiences of people diagnosed with 'borderline personality disorder'/'emotionally unstable personality disorder', their family carers, and professionals involved in their care. • Variations of peoples experience of crisis and clinical outcomes from intervention. Evaluation

Clinical outcomes, views and experiences of crisis intervention
Comprehensive search terms were developed (see table 2) by identifying relevant terminology, identifying synonyms and using terms already found in relevant publications (Aveyard, Payne and Preston, 2016). The wildcard symbol '*' was used to capture variations of root words (Hewitt-Taylor, 2017). ('Borderline personality disorder' OR 'emotionally unstable personality disorder' OR 'BPD' OR 'EUPD') AND Search 2: (Cris* OR emergenc* OR urgent OR risk* OR acute* OR critical* OR intensive* OR respon* OR Self-Injurious Behav* OR self harm* OR self injur* OR self mutilat* OR self poison* OR overdos* OR self burn* OR self cut* OR suicid*) AND Search 3: (experien* OR prefer* OR belie* OR perce* OR attitud* OR opinion* OR view* OR judg* OR reaction* OR impression* OR feel* OR satisf*) *Wildcard -utilised to capture variations of root word.
The question was; 'what are the experiences and perceptions of PdxBPD, their family carers, and professionals around crisis intervention for 'BPD'? Specific aims should be a logical continuation of the research question (Hewitt-Taylor, 2017), and the following sub-questions were explored (see table 3). 3.
In which contexts do these crisis interventions take place, and does the context impact on experience?

4.
What are the barriers and facilitators to people feeling a crisis intervention has been beneficial? *Note: "People" refers to the multiple perspectives of: • People diagnosed with 'BPD' (PdxBPD) • Families and carers (Family carers) • Health, social care and emergency services staff (Professionals).
Inclusion and exclusion criteria were developed with reference to the University of Melbourne guidelines (2019)(see table 4).

Exposure of interest
Primary research studies evaluating crisis intervention, or including experiences of crisis and/or crisis intervention Interventions that go beyond one month Crisis intervention defined as an action to "ensure safety and recovery and lasts no longer than one month" (Borschmann et al 2012

Primary research required for integrative review
The protocol for this study was developed in collaboration between all authors, and registered with the International Prospective Register of Systematic Reviews (PROSPERO) (Warrender et al, 2017). The full protocol can be accessed at: http://www.crd.york.ac.uk/PROSPERO/display_record.php?ID=CRD42017075123.

Data Collection
The review captured papers January 2000 to November 2017, with details reported through the PRISMA flow diagram (Moher et al, 2009) (See Figure 1). To increase quality assurance, two reviewers from the team were involved at each stage of the process; reviewing titles and abstracts, agreeing inclusion of papers, data extraction, data analysis and synthesis. Papers were appraised using the mixed methods appraisal tool (MMAT) (Hong et al, 2018). This tool has specific questions for five categories of empirical study, qualitative research, randomized controlled trials, non-randomized studies, quantitative descriptive studies, and mixed methods studies. Each study was reviewed against relevant criteria. Scoring is discouraged (Higgins andGreen 2008, Crowe andShepard 2011), and excluding studies on the basis of quality is not recommended (Hong et al, 2018). The MMAT was therefore used to aid description. Percentages of affirmative MMAT responses were recorded, with negative responses requiring comment in the data extraction table (see appendix).
Thematic Analysis Braun and Clarke's (2006) thematic analysis framework guided analysis and organisation of data. Data was analysed by all authors, identifying patterns within the dataset which were relevant to the research questions. This was initially a theoretical thematic analysis, providing a detailed analysis of dataset content relevant to research questions (Braun and Clarke, 2006). However, themes evolved through interpretation, identifying the significance of patterns in relation to not only research questions but relevant literature presented in the introduction to this paper. Themes were presented to provide a "concise, coherent, logical, nonrepetitive, and interesting account of the story the data tell" (Braun and Clarke, 2006, p.23). These themes give readers an in-depth narrative of the human experiences, influences and variables associated with crisis and crisis intervention for PdxBPD.
Twenty-nine papers collected data on PdxBPD, eleven on professionals, with family carers the least represented with six. Gender was overwhelmingly female in PdxBPD participants, who all met or had met DSM diagnostic criteria. Professionals included practitioners from mental health nursing, psychiatry, psychotherapy, psychology, counselling, case management, social work, art therapy, police and occupational therapy.
Many papers were qualitative with small sample size, though agreement about quality in qualitative research is elusive (Aveyard, Payne and Preston, 2016). It is argued that weaker studies would simply contribute less, rather than distort findings (Thomas and Harden 2008).
Some papers were represented once, and others across themes. Findings illuminate a variety of perspectives which may reflect the experience of 'crisis' and intervention for PdxBPD. Four themes emerged; crisis as a recurrent multidimensional cycle, variations and dynamics impacting on crisis intervention, impact of interpersonal dynamics and communication on crisis, and balancing decision making and responsibility in managing crisis.

Crisis as a recurrent multidimensional cycle
Twenty-eight papers characterised crisis as a recurrent, unpredictable, subjective, multidimensional and overwhelming experience. Internal or external triggers precipitated self-harm, which was a selfmanagement or help-seeking strategy. Distress was paralleled in experiences of family carers and professionals. The word 'crisis' did not feature in all papers, though all included experiences of feeling out of control. Two additional terms were identified; 'aversive tension' (Stiglmayr et al, 2005(Stiglmayr et al, , 2008 describing extreme emotional dysregulation which often precedes self-harm, and 'agitation', with measures including tension, uncooperativeness, hostility, and poor impulse control (Damsa et al, 2007).
Crisis is complex with subjective precipitating factors. Brooke and Horn's (2010) interviews (n = 4) identified distal and proximal factors, while Black, Murray and Thornicroft's (2014) (n = 9) described internal and external dynamics. Distal factors included histories of trauma (Brooke and Horn 2010, Holm and Severinsson, 2011 and proximal factors/external dynamics related to interactions with others , Brooke and Horn, 2010, Black, Murray and Thornicroft, 2014. Internal dynamics also saw crisis arise within the self (Black, Murray and Thornicroft, 2014). Quantitative studies identified PdxBPD self-reporting triggers as feeling rejected, being alone, failure (39%, n = 63) (Stiglmayr et al, 2005) and inner helplessness (Stiglmayr et al, 2008). 'BPD' was distinguished from other diagnoses in that tension arises from a negative view of the self (Stiglmayr et al, 2008).
PdxBPD felt crisis could arise suddenly, sometimes without warning and impact on emotional and perceptual states. Henderson et al's (2013) qualitative case series (n = 4) described crisis as having a quick onset, sometimes without warning signs (Helleman et al, 2014). PdxBPD felt on edge, overwhelmed by emotions (Perseius et al, 2005) as if they were going to explode (Brooke and Horn, 2010), with a desperate need to gain peace or escape (Holm and Severinsson, 2011). Perceptual changes included dissociative experiences (Brooke and Horn, 2010, Black et al, 2014, while Slotema et al's (2017) quantitative study (n = 89) found PdxBPD with auditory verbal hallucinations correlated with a higher frequency of suicidal plans and attempts. Reasons for referral to emergency services included depression, anxiety, psychosis, drug abuse/dependence and disruptive behaviour (Pascual et al, 2007).
PdxBPD described difficulty articulating their experiences (Black et al, 2014, Helleman et al, 2014. This was confirmed by family carers (Dunne and Rogers, 2012). Some attempted to hide their experience of crisis (Perseius et al 2005, Holm andSeverinsson 2011), at times to protect family (Black et al, 2014). Conversely, sometimes PdxBPD had difficulty thinking of others (Holm and Severinsson, 2011).
PdxBPD may have enduring negative emotional states which could impact experiences of crisis. A quantitative study of self-reported emotional responses found emotional reactivity similar whether PdxBPD were in crisis or not, indicating that negative thinking may be enduring (Staebler et al, 2009). Furthermore, hourly self-reporting over 48 hours saw aversive tension found to be more frequent, more intense and longer lasting in PdxBPD than in healthy controls (Stiglmayr et al, 2005). Unsuccessful suicide attempts could become a cycle of crisis, where feelings of failure reinforce suicidal thoughts (Black et al, 2014).
PdxBPD self-managed crisis through self-harm, or seeking help from professionals (Black, Murray and Thornicroft, 2014). Self-harm, predominantly cutting and burning, was a self-managed personal crisis intervention. This was used as a response to dissociation , Black et al, 2014 or emotional dysregulation Horn, 2010, Henderson et al, 2013), with people reporting subjective analgesia (Philipsen, Schmahl and Lieb, 2004). Some people described progressive systems of coping, moving through self-harm to suicidality (Brooke and Horn 2010). Brown et al (2002) found self-harm had a dual role, for reducing or expressing the feeling of crisis. Difficulties articulating distress could lead to self-harm as a means of communication (Brooke and Horn, 2010). A quantitative online survey found that feeling suicidal, feelings of self-harm and feeling unsafe were the most common reasons for PdxBPD seeking hospital admission (Lawn and McMahon, 2015a). Decisions to hospitalize were often associated with risk of suicide (Pascual et al, 2007).
Family carers could experience a parallel crisis to PdxBPD. Dunne and Rogers (2012) focus groups found family carers experiencing their own distress, with unstructured interviews finding distress related to PdxBPD self-harming and attempting suicide (Giffin, 2008). Free text questionnaires and group interviews (n = 19) saw distress described as a permanent crisis and 24-hour duty of constant worry, which included powerlessness, frustration, and mirrored the experience of PdxBPD (Ekdahl et al, 2011). Over one third of family carers in a quantitative study (n = 32) knew little about 'BPD' (Hoffman et al, 2004), while focus groups in a grounded theory retrospective study found family carers lacked skills for helping PdxBPD's issues, and wished for more information on how to handle situations Rogers, 2012, Lohman et al, 2017). However, greater knowledge about 'BPD' was associated with higher levels of family burden, distress, depression and greater hostility towards PdxBPD (Hoffman et al 2004). This may link to guilt felt by families due to preconceived ideas that parents are responsible for development of 'BPD' (Ekdahl et al 2011) Furthermore, professionals experienced distress. Interviews (n = 29) identified a need for emotional support (Bergman and Eckerdal, 2000) with the threat of suicide considered the most distressing (McGrath andDowling, 2012, Hughes et al, 2017). Professionals further described feeling frustrated, inadequate, challenged (Commons Treloar, 2009), confused, uncertain, drained and personally distressed (Warrender, 2015).
In summary, crisis has multiple triggers, is subjective and people manage their distress in different ways. Crisis was recurrent and could have a quick onset, which may be linked to enduring negative thinking. The constant nature of crisis could contribute to distress in family carers and professionals, who may mirror PdxBPD's crisis. The subjectivity of crisis experience may make it a challenge to treat, though threat of self-harm and suicide is often the reason for crisis intervention.

Variations and dynamics impacting on crisis intervention
Twenty-five papers explored access to care during crisis. The impact of professional interventions, resources available, treatment options and preferences and variables impacting on team approaches emerged.
PdxBPD experienced challenges accessing care. Self-referral could be difficult to arrange in the midst of crisis, and while PdxBPD could self-refer to a psychiatric emergency service, ambulance was the most common means of arrival (Pascual et al, 2007). Helleman et al's (2014) qualitative study (n = 17) found PdxBPD using preventative hospital admissions felt security and reassurance knowing admission was available. To the contrary, some PdxBPD and carers were refused hospital admission and reported significant distress McMahon, 2015a, 2015b). Morris, Smith and Alwin's (2014) qualitative study (n = 9) saw PdxBPD describe services as reactive rather than proactive regarding risk, feeling thresholds for intervention were only met in immediate risk of suicide. Pasqual et al's (2007) quantitative retrospective examination of patient records (n = 540) reported that though PdxBPD did self-refer, professionals' decisions to hospitalize were often based on suicide risk.
Access to and continuation of care can be influenced by the 'BPD' diagnosis. PdxBPD in a qualitative study (n = 5) felt they were excluded from services or had care withdrawn based on their diagnosis (Horn, Johnstone and Brooke, 2007), and professionals confirmed they had witnessed colleagues refusing to treat PdxBPD (Commons Treloar, 2009). In contrast, for mild to moderate suicidal ideation, PdxBPD were more likely to be discharged from toxicology services and admitted to psychiatric hospital than people diagnosed with depression (Carter et al, 2005).
Professional interventions were often standard care, though took place in a variety of contexts including inpatient settings (Philipsen, Schmahl and Lieb, 2004, Koekkoek et al, 2010, Helleman et al, 2014, Warrender, 2015, a crisis intervention unit (Berrino et al 2011), emergency departments (Damsa et al, 2007, Pascual et al, 2007, toxicology service (Carter et al, 2005), a 'BPD' resource centre (Lohman et al, 2017), intensive home treatment (Turhan and Taylor, 2016) and an intensive outpatient therapy (McQuillan et al, 2005). Professionally influenced interventions included joint crisis plans (Borshmann et al, 2013) and a smartphone application (Prada et al, 2017). PdxBPD accessed their general practitioners for support and referral to other services, and they were rated by family carers as the most responsive professionals (Lawn and McMahon, 2015b).
Hospitalisation was common, though had subjective value. Preventative hospital admission saw a slight decrease in services used in terms of inpatient days recorded, and was evaluated positively by PdxBPD (n = 8) (Koekkoek et al, 2010). However, PdxBPD experiences of a 3-night hospital admission with support were both positive and negative (Helleman et al, 2014). Focus groups of inpatient mental health nurses (n = 9) using mentalization based therapy (MBT) skills felt more empowered and able to facilitate positive changes for PdxBPD, though no patient outcomes were recorded (Warrender, 2015). Some professionals considered hospitals too busy and not conducive (Warrender, 2015), feeling that PdxBPD were best managed as outpatients without medication, receiving consistent support (Little et al, 2010).
There were positive impacts of services specifically purposed to manage crisis intervention. Admission to a crisis intervention unit (n = 100) saw reduced rates of self-harm (8%) and hospitalisation (8%) compared to treatment as usual (TAU) (n = 100, 17% and 56%) (Berrino et al, 2011), and IHTT (n = 27) noted improvement in most PdxBPD (Turhan and Taylor, 2016). However, these interventions showed limited benefit for suicidality, with IHTT not enough to manage suicide risk in 34% of cases where PdxBPD were hospitalised (Turhan and Taylor, 2016). Although improved compared to TAU, the crisis intervention unit still recorded treatment failure through suicidal crisis (Berrino et al, 2011). McQuillan et al's (2005) quantitative study on intensive outpatient dialectical behavioural therapy showed acceptability with high treatment completion and retention rates, and improvements on depression and hopelessness scales. A quantitative study on joint crisis plans co-developed by PdxBPD and mental health teams showed no reduction in instances of self-harm, though was used by 73.5% (n = 25/34) during a crisis, contributing to a greater feeling of control for 47.1% (n = 16/34) of participants followed up (Borshmann et al, 2013). A smartphone application using mindfulnessbased exercises was evaluated as user friendly, and though mechanisms of change were unclear, it contributed to reduction in aversive tension (Prada et al, 2017).
Crisis intervention using specific medications was reported in two papers. Damsa et al (2007) found intramuscular Olanzapine (10mg) reduced agitation, however this was after refusal of oral medication and included 80% (n = 20/25) of participants being physically restrained. Naloxone (0.4mg) administered intravenously showed improvement in dissociative symptoms, though was not better than placebo (Philipsen, Schmahl and Lieb, 2004). Medications were often prescribed at a psychiatric emergency service (Pascual et al, 2007), though PdxBPD in a qualitative study (n = 7) felt that medication was often used due to a lack of appropriate resources (Rogers and Acton, 2012).
Outcomes of crisis interventions are influenced by several factors. Resources available did not always meet demand (Lohman et al, 2017), and family carers (n = 121) described a lack of choice in services for PdxBPD (Lawn and McMahon, 2015b) and identified the need for an appropriate base and crisis accommodation (Dunne and Rogers, 2012). Commons Treloar's (2009) qualitative study (n = 140) saw professionals across emergency medicine and mental health services perceive current services as unsuitable for PdxBPD's needs. Once crisis and imminent risk of suicide was over, PdxBPD felt professionals were not interested in their underlying distress (Morris, Smith and Alwin, 2014). Family carers further identified a lack of long-term consistent support as contributing to anxiety in PdxBPD (Lawn and McMahon, 2015b).
PdxBPD's preferences for care included therapeutic relationships giving emotional and practical support, while specific treatment refusals included particular medications and use of involuntary treatment (Borschmann et al, 2014). PdxBPD were mixed in perception of the usefulness of identifying early warning signs, developing crisis plans and hospital admission (Lawn and McMahon, 2015a). These were respectively found to be very unhelpful for around a quarter of carers in the study (25.4%, n = 18; 28.6%, n = 20; and 23.9%, n = 17).
Professionals identified conflict in teams regarding approaches to working with PdxBPD (Commons Treloar 2009), describing lack of collaboration negatively impacting care (Bergman and Eckerdal, 2000). Family carers echoed this and at times heard contradictory advice (Giffin, 2008). An interagency quantitative study (n = 378) found that health and welfare, mental health, and police responded to PdxBPD in different ways (Little et al, 2010).
Conflict could be due to different levels of education on 'BPD' which varied between professionals (Bergman and Eckerdal, 2000). Family carers experienced staff with little knowledge (Ekdahl et al, 2011), some telling them "it's just behaviour" (Dunne and Rogers 2012). Professionals identified the need for specific education on 'BPD' (Commons Treloar, 2009), and whilst they utilised clinical supervision (Berrino et al, 2011) and emphasised its importance (Commons Treloar, 2009), it was not always accessible (Warrender, 2015). Focus groups (n = 9) saw teams using MBT skills describe increased consistency in their approach (Warrender, 2015).
To summarise, PdxBPD had varying experiences of accessing care which could be influenced by diagnosis. Professional interventions took place in a variety of contexts but were most often nonspecialist inpatient units and emergency departments, and showed limited or subjective benefit. Outcomes may be influenced by resources available, thresholds for intervention, conflict in teams, differing levels of professional education and access to clinical supervision.

Impact of Interpersonal dynamics and communication on crisis care
This theme was illuminated by 22 papers, highlighting interpersonal dynamics as a trigger to crisis and relationships holding contradictory roles in relieving or adding to suffering. Reputations for selfharm and the 'BPD' diagnosis itself could contribute to discriminatory experiences.
Interpersonal issues could precipitate crisis (Black et al, 2014) and be a catalyst to self-harm , with rejection self-reported as a precipitating factor to aversive tension (Stiglmayr et al, 2005). Brooke and Horn (2010) found PdxBPD used self-harm as a means of regaining self-control and inhibiting interpersonal behaviour which may be deemed inappropriate. A quantitative study (n = 75) using clinical history interviews recorded instances of parasuicide (suicide without supposed intent to die) and found 20% had an interpersonal influence (Brown Comtois and Linehan, 2002). However, this study did not define parasuicide, and acknowledged limitations in that self-reporting of intent may not be known or remembered.
Social relationships had a subjective role, with PdxBPD's preferences in crisis contrasting between connecting with others, and the desire to be left alone (Borshmann et al, 2014). Black, Murray and Thornicroft (2014) found relationships with family could be protective against suicide, as a purposeful family role and responsibility engendered self-preservation. However, the same study found this responsibility to protect loved ones could lead PdxBPD to hide their distress.
PdxBPD (n =17) valued contact with professionals (Helleman et al, 2014), and particularly those who invested in them and offered hope (n = 8) (Veysey, 2014). PdxBPD valued being treated like a person (Morris, Smith and Alwin 2014), shown dignity and respect, and receiving emotional and practical support (Borshmann et al, 2014). Collaboration was valued, as 47.1% (n = 16/34) of PdxBPD self-reported that developing joint crisis plans with professionals had improved their relationships (Borshmann et al, 2013). An aspect of crisis is a difficulty communicating and articulating experiences, and a qualitative study interviewing professionals (n = 9) described their role as slowing things down and helping PdxBPD to think (Bowen, 2013).
Relationships with professionals had a duel role. Qualitative studies found they could relieve or add to suffering (Perseius et al, 2005)(n = 10) as PdxBPD experienced both helpful and discriminatory experiences (Veysey, 2014)(n = 8). Through interviews, professionals (n = 5) perceived that PdxBPD have high expectations of them and are sensitive to interpersonal disappointment due to adverse childhood experiences, further considering therapeutic relationships potentially re-traumatising patients when ending (Rizq, 2012). Professionals perceived PdxBPD's difficulties with them as a parallel process and a repetition of experiences outside of care, though also valuable opportunities for learning (Bowen, 2013).
PdxBPD described 'non-caring care', with some professionals perceived to be reluctant, unwilling or unable to work with them or dedicate time to therapeutic relationships (Morris, Smith and Alwin, 2014), and lack of contact in an inpatient context contributing to negative emotions (Helleman et al, 2014). PdxBPD experienced professionals being dismissive of their distress (Rogers and Acton, 2012)(n = 7), describing being 'dumped' or left in wards following frequent admissions (Rogers and Dunne, 2011)(n = 10). Dismissiveness was confirmed by family carers, with focus groups describing some professionals as unprofessional and unhelpful (Dunne and Rogers, 2012)(n = 8), and an online survey identifying the most challenging issue for PdxBPD as not being taken seriously (Lawn and McMahon, 2015b)(n = 121). Self-reporting emotional reactions of health and non-health related agencies (n = 378) found the police as more likely to see PdxBPD as a nuisance, as in contrast to mental health professionals, police felt they needed to be available all the time (Little et al, 2010).
PdxBPD had perceived discrimination from professionals. An online survey found that 65.4% (n = 78/96) of PdxBPD who had accessed care for ten years or more had experienced discrimination, particularly as inpatients (Lawn and McMahon, 2015a). Some PdxBPD felt they were not seen as a person (Walker, 2009, Holm andSeverinsson, 2011), and Walker's (2009) narrative interviews (n = 4) found PdxBPD perceive their reputations as a 'self-harmer' as overshadowing other issues. Veysey's (2014) qualitative study found through semi-structured interviews that PdxBPD (n = 8) with selfharm histories had increased experiences of discrimination, which impacted on self-image.
Stigma attached to the 'BPD' diagnosis had further impact. Interviews saw mental health nurses acknowledge the stigma attached to PdxBPD as they arrived at their service (McGrath and Dowling, 2012)(n = 17) and a questionnaire found reduced sympathy for people with the diagnosis (Markham and Trower 2003)(n = 48). PdxBPD felt their distress was often viewed in terms of 'BPD', and they could be misunderstood as being deliberately difficult (Morris, Smith and Alwin, 2014). PdxBPD described professionals indicating they were selfish (Holm and Severinsson, 2011), and family carers acknowledged a stigma from professionals who described distress as 'just behaviour' (Dunne and Rogers, 2012).
In summary, this theme showed the complex nature of social and professional relationships. Interpersonal issues were often a trigger to crisis, with social relationships of varying benefit. PdxBPD emphasised the value of the therapeutic relationship with professionals, though also described its double role through experiences of 'non-caring care', often experiencing discrimination which was sometimes related to diagnostic stigma.

Balancing Decision Making and Responsibility in managing crisis
Nineteen papers contributed to balancing decision making and responsibility in managing crisis. Shared decision making was identified as important, though experiences of this varied with complexity in power dynamics. There were often differing views on where responsibility lay for the management of crises, and this created difficulties for professionals and family carers.
PdxBPD welcomed choice and joint decision making, though decisions were not always collaborative. PdxBPD (n = 17) with choice of hospital admission reported an improved sense of autonomy and responsibility (Helleman et al, 2014). Koekkoek et al (2010)(n = 8) identified that preventative hospital admission contributed to feelings of control over crisis, with PdxBPD feeling that having access to admission if needed and having control over their own treatment promoted their ability to self-manage their own difficulties. Focus groups of PdxBPD with experiences of inpatient settings identified good joint decision making (Rogers and Dunne, 2011) and professionals also emphasised its importance (Bowen, 2013). Although noted in under half of participants (47.1%, n = 16/34), using a joint crisis plan had contributed to greater feelings of control over problems (Borshmann et al, 2013). Professionals (n = 9) described their being placed in the expert role as unhelpful, shared decision making encouraging shared responsibility, and that PdxBPD becoming disillusioned with them could lead to looking inward to their own resources (Bowen, 2013).
The removal of responsibility and choice was not welcomed by PdxBPD. Holm and Severinsson's (2011) qualitative interviews (n = 13) saw PdxBPD describe having responsibility removed and lacking the power to make decisions as a barrier to effective intervention. Involuntary treatment was a specific treatment refusal in joint crisis plans (Borshmann et al 2014), though this was used at times with 9/13 PdxBPD reporting difficulty accepting this and feeling violated (Holm and Severinsson, 2011). Particular psychotropic medications were a specific treatment refusal in some joint crisis plans (Borshmann et al, 2014), though PdxBPD described little choice regarding the use of medication in inpatient settings (Rogers and Acton, 2012). Furthermore, 80% of all participants (n = 20) in one study were physically restrained prior to medication administration (Damsa et al, 2007).
Power dynamics appeared to play a role in treatment, with the act of refusing the advice or guidance of professionals interpreted as pathology. 'Uncooperativeness' was a measure of agitation (Damsa et al, 2007) while 'noncompliance with treatment' was a reason for hospitalization (Pasqual et al, 2007). The illusion of choice was noted by Rogers and Dunne (2011, p.229) through de-facto detention, with PdxBPD describing experiences as inpatients where professionals told them that they could be voluntary patients, "or we can section you" using mental health legislation. Some PdxBPD identified powerlessness and the paradox of being told to use their strengths, yet simultaneously having decisions made on their behalf (Holm and Severinsson, 2011).
Uncertainty emerged regarding who should hold responsibility for PdxBPD. Three qualitative studies using interviews found professionals felt responsibility for the safety of PdxBPD (Rizq, 2012) and sometimes felt this was transferred to them by patients (Nehls, 2000, Hughes et al, 2017. However, some PdxBPD countered that suicidality could actually be through their desire to take responsibility for themselves (Holm and Severinsson, 2011). Hughes (2017) interviews found some community mental health teams (n = 4) feared being blamed in event of patient suicide. Furthermore, Krawitz and Batcheler's (2006) quantitative self-report questionnaire found that decisions are sometimes made out-with PdxBPD's best interests to protect professionals from legal repercussions. Defensive practice was influenced by the PdxBPD's family and friends, though the biggest influence was cited as the media (Krawitz and Batcheler 2006). Nonetheless this was contradicted by a self-report questionnaire (n = 378) across professional agencies, which found that though the police felt they needed to be constantly available, there were no concerns in any group regarding damage to professional credibility nor legal consequences if suicide were to occur (Little et al, 2010).
Family carers described an all or nothing responsibility transaction between them and professionals. Qualitative studies using focus groups, interviews and questionnaires found family carers held full responsibility until their significant other was in hospital, then felt overlooked and had no responsibility (Dunne and Rogers, 2012)(n = 8), and were sometimes told by professionals that they were not needed (Ekdahl et al 2011)(n = 19). Giffin's (2008) unstructured interviews (n = 4) saw family carers perceive that responsibility was often left with them, with their support used as a reason to avoid professional intervention. Family carers involvement in care was often limited, though Lohman et al (2017) randomly reviewed resource requests (n = 500) to find that they desired more communication with professionals. Family carers also had mixed experiences of care plans being shared (Dunne and Rogers, 2012) and felt there were no discharge plans (Giffin, 2008).
Professionals struggled with suicide risk and felt uncertain whether intervention was required or not (Rizq, 2012). Nehls (2000) interviews with professionals (n = 17) described this as balancing over and under concern. Hughes et al (2017) found professionals from community mental health teams (n = 4) describe balancing patient responsibility with professional responsibility, and found considerable variation in professional views regarding risk. This variation in views corresponds with carers being given contradictory advice by professionals (Giffin, 2008). Family carers experience paralleled that of professionals, describing the challenge of balancing support and enablement between themselves and PdxBPD (Dunne and Rogers, 2011).
Mental health professionals appeared the most comfortable with handing responsibility back to PdxBPD, and were more understanding than police or health and welfare of why a person may be discharged or not admitted to hospital for ongoing suicidality (Little et al, 2010). PdxBPD felt that they were compared to people with other diagnoses, and seen as having more control than patients diagnosed with schizophrenia (Rogers and Dunne, 2011). This was confirmed in a study of mental health nurse attitudes, which viewed PdxBPD as being in control of their behaviour (Markham and Trower, 2003).
This theme saw PdxBPD welcome choice and joint decision making. However, decisions were not always collaborative and the removal of responsibility was perceived as a barrier to effective intervention, particularly recognising power dynamics between PdxBPD and professionals. There was uncertainty between professionals, family carers and PdxBPD as to who held responsibility, with family carers describing their responsibility as all or nothing. Both professionals and family carers described difficulty in balancing the level of responsibility they shared with PdxBPD for their safety. Mental health professionals appear to be the most comfortable in handing responsibility back to PdxBPD.

Discussion
This integrative review will inform evidence-based practice around crisis intervention for PdxBPD with RCT's lacking (Borschmann et al, 2012). Crisis is a subjective term and crisis intervention is not well understood. This justified an integrative review, including a broad and diverse range of literature (Aveyard, Payne and Preston 2016). This approach is appropriate to defining concepts and reviewing theories (Whittemore and Knafl, 2005), and can provide foundations for future knowledge and research.
The contribution of this review to existing knowledge comes through the synthesis of 46 papers which highlight key themes on this complex topic. The overall quality of research was good, with affirmative MMAT responses ranging between 60% and 100%. The vast majority (29/46) achieved all affirmative responses, while negative responses often related to a lack of clarity rather than poor research practice. Some studies did not acknowledge the variables which may influence their outcomes, and this review contributes to understanding of these factors.
A conceptual map of the potential journey from crisis to crisis intervention (see figure 2) provides a visual representation of themes discussed. Sansone (2004) described crisis as being precipitated by an event, and this review identified events as internal or external, triggered from within the self or interpersonally. The influence of both self and others may be understood through the concept of mentalization. Mentalizing is "the process by which we make sense of each other and ourselves, implicitly and explicitly, in terms of subjective states and mental processes" (Bateman and Fonagy, 2010, p.11). Difficulties mentalizing are influenced by childhood trauma or neglect, and can lead to difficulties in the experience of oneself, and a vulnerability to interpersonal interactions (Bateman and Fonagy, 2010). It would however be unfair to suggest that all difficulties for PdxBPD in interpersonal relationships were due to their failure to mentalize, given family carers and professionals descriptions of stigma and discrimination.
PdxBPD experienced crisis in different ways, not surprising given the heterogenous diagnosis. Sudden and recurrent onsets may relate to the consistent availability of triggers which could come from self or others, and recurrent crises may relate to enduring negative thinking. Emotional dysregulation and perceptual changes were features of crisis, and these would further impact mentalizing ability. The experience of feeling overwhelmed was consistent with general definitions of crisis (James and Gilligand, 2005), though the subjectivity of experiences indicates the need for sufficient flexibility in any intervention, remaining person centred rather than diagnosis centred.
A prominent self-management strategy for PdxBPD was self-harm. Felitti et al's (1998) adverse childhood experiences (ACEs) study proposed that health risk behaviours such as smoking and obesity are viewed as societal problems, yet are solutions from the perspective of individuals. There is a high prevalence of ACEs in the histories of PdxBPD (Herman, Perry, Van Der Kolk, 1989, McFetridge et al 2015 and people who self-harm in general (Everett and Gallop, 2000;Vivekananda, 2000). This review found the 'problem' of self-harm was often a solution for PdxBPD. Professional responses sometimes did not see beyond self-harm, treating personal solutions as problems, and not exploring the underlying distress. Self-management also contrasted between hiding distress and help-seeking, with hiding distress emphasising the subjective value of social relationships and complex relationships with family carers.
Family carers experience distress, which paralleled crisis for PdxBPD, yet often had limited involvement with care and all or nothing responsibility. This emphasises the importance of the 'triangle of care' (Carers Trust 2016) with carers involved in care planning and treatment, in true partnership working between people experiencing mental distress, family carers and professionals. However, this experience may not be unique to crisis intervention for PdxBPD, with a literature review across diagnoses finding that collaborative decision making was not a regular experience, and that there was an 'us and them' divide between family carers and professionals (Doody et al, 2017).
Though some interventions contributed to reduced self-harm and hospitalisation, improvement on depression and hopelessness scales, and improvement in dissociative symptoms, largely interventions were of subjective and limited benefit. This review identified factors which may influence the quality of any crisis intervention. Professionals described deficits in resources and knowledge, and their own need for support. Targeted education on 'BPD' can impact staff attitudes (Miller and Davenport, 1996, Commons Treloar and Lewis, 2008, Shanks et al, 2011, and it may be prudent to target professionals basic training (Warrender and Macpherson, 2018) and have education co-produced with experts by experience (Dickens et al, 2019). Given the prevalence of trauma histories in PdxBPD, trauma informed care should also inform therapeutic relationships (Sweeney et al, 2018). Clinical supervision has been specifically recommended to professionals working with PdxBPD (Bland and Rossen, 2005) and may be particularly valuable given complexities in decision making and potential for team conflicts. Professional decision-making regarding risk has been described as an ethical dilemma, with well-intended decisions having the potential for iatrogenic consequences (Warrender, 2018).
PdxBPD had positive and negative experiences of care. These were polarised between feeling professionals were person centred or diagnosis centred, having access to care or finding it difficult, being included in joint decision making or having responsibility removed, and feeling a therapeutic relationship had been established or experiencing stigma and discrimination. Regardless of any interventions design, these factors influence the experience. Furthermore, given interpersonal relationships as a potential trigger to crisis, professional stigma and discrimination can have an iatrogenic and counterproductive impact, as PdxBPD may be triggered back into crisis and feel worse in care (see figure 3). The lack of hope has been described as a self-fulfilling prophecy, where the attitudes of professionals contribute to poor outcomes (Warrender and Macpherson 2018). The subjectivity of crisis experience shows the limitations of diagnosis, emphasising that any intervention should remain person centred. Whilst thresholds for intervention were often met after self-harm or suicidal behaviour, professionals should review the ease of access to their services and ensure care goes beyond behaviour management and supports PdxBPD with underlying distress. PdxBPD preferences for care were not surprising or unrealistic in having access to care, joint decision making and valuing therapeutic relationships. These findings highlight that PdxBPD may have poor experiences of care, and that limited resources, deficits in knowledge, uncertainty, team conflict, distress, and a lack of clinical supervision are potential factors which influence how professionals respond. Family carers should have access to appropriate support to manage their own distress, and the opportunity to be involved in care planning as per the triangle of care.

Limitations
Limitations of this review include the exclusion of groups including under 18's, and people with comorbid 'BPD' though not as their primary diagnosis. Furthermore, the exclusion of other personality disorders was necessary to aim for validity around a common experience, and thus the difficulties and complexity of all personality disorder diagnoses have not been captured.

Crisis
Professional intervention

Stigma and Discrimination Conclusion
The experience of crisis for PdxBPD is complex, with subjectivity in precipitating factors, experience and ways of coping. Family carers experience their own distress and require support, and should be given more opportunities for involvement by professionals. Interventions are available, though often standard care, and despite showing some benefit to PdxBPD, there is inconsistency in that people have positive and negative experiences of care. Several factors influence professional interventions, and implications for practice suggest a review is required of crisis intervention services. This review will inform future research by highlighting the complexity and array of human factors in the delivery of crisis intervention, which may have an influence on recorded outcomes. Future research may be wise to focus on perspectives within single cases, comparing PdxBPD, family carer and professional perspectives on shared experiences, to provide in-depth exploration of interpersonal factors.  (2000) To broaden understanding of what it means for professionals to manage PdxBPD Inpatient and outpatient service 29 professionals (Sweden)

Qualitative Grounded theory Individual interviews
Need identified for emotional support and education on BPD.

Differing levels of education on BPD and this influenced approaches to working with PdxBPD
Organisations/teams that fail to work together and collaborate effectively perceived to have a negative impact on patient care.

100%
Berrino et al (2011) To assess whether crisis intervention at a general hospital is a suitable management strategy for PdxBPD referred to the emergency room for self-harm.

Quantitative
Prospective 3 month follow up using patient records Crisis intervention unit had 8 beds, max 5 night stay, intensive interdisciplinary care and daily clinical supervision After 3 months CI group had reduced rates of self-harm and hospitalization (8% + 8%) compared to TAU (17% + 56%) Treatment failure was defined as suicidal crisis with supplementary inpatient treatment, and was observed in both groups (CI=14 / TAU=56).
CI was more cost effective than TAU

80%
Crisis intervention group also received unplanned cointerventions Black, Murray and Thornicroft (2014) To understand the phenomenology of BPD from the patients perspective.

Qualitative Interviews
PdxBPD experienced dramatic perceptual and psychological changes, impacts on ability to communicate, experience of pain, memory loss and hallucinations.
Responses to crisis were help seeking and selfharm.

80%
Unclear why interpretive and not descriptive phenomenology Families were perceived as either protective or burdensome.
People felt a cycle as recovery from suicide attempt could generate new feelings and further suicidal thoughts.
Crisis is multidimensional, with a complex relationship between internal and external factors in the experience of crisis.
Crisis can rise from within the person (internal factors) as well as through experiences with others (external factors).

Borschmann et al (2013)
To examine the feasibility of recruiting and retaining adults with a diagnosis of BPD to a pilot RCT investigating the potential efficacy and cost effectiveness of using a joint crisis plan.

Qualitative
Open discussion (using crisis plan subheadings as a basis) on joint crisis plans created by participants.
Variation in peoples preferences regarding crisis intervention, emphasising the importance of individually tailored crisis plans Being treated with dignity and respect and receiving emotional and practical support is important to PdxBPD Some PdxBPD identified the importance of connecting with others during crisis, but several indicated the desire to be left alone during a future crisis.

100%
Specific treatment refusals during crises included particular types of psychotropic medication and involuntary treatment Bowen (2013) To explore the experiences of good practice among mental health professionals working in a service that provided specialist treatment for PdxBPD 9 clinicians (4 nurses, 3 social therapists, 1 art therapist, 1 psychiatrist) (UK) Qualitative, Semi-structure interviews Professional role felt to be to slow things down, to help PdxBPD to think Shared decision making and shared responsibility felt to be important Interpersonal issues between PdxBPD and professionals seen as a repetition of experiences outside their care, though this was seen as an opportunity for learning Professionals felt that when PdxBPD placed staff in the expert role it was unhelpful.
Professional felt when PdxBPD become disillusioned with staff, they look to their own resources 100% Brooke and Horn (2010) To explore the meanings of self-injury and overdosing and the relationship of each to the other for women who have fulfilled the diagnostic criteria for BPD Psychotherapy service 4 PdxBPD (UK)

Qualitative
Interpretive phenomenological analysis Interviews Both distal and proximal factors perceived as potential antecedents to crisis Crisis symptoms included feeling "like a pressure cooker", "about to burst" and dissociative experiences Self-harm identified as private form of selfhelp for regaining control of emotional dysregulation, or public form of communicating distress People have progressive systems of coping with distress, ranging from cutting to burning and overdosing.

100%
Brown, Comtois and Linehan (2002) To better understand reasons for suicide attempts and non-suicidal selfinjury in women diagnosed with BPD 75 PdxBPD (USA) Quantitative, 'Parasuicide history interview' Recorded a comprehensive 47item semistructured interview measuring details of single parasuicide episodes.
People may feel that crisis is something to be reduced, or expressed.
Motives for suicidality are complex, and people may have multiple reasons.

80%
PdxBPD confirmed reasons from a prepared list. New reasons not collected as data.
20% of participants cited interpersonal triggers to suicidality. Carter et al (2005) To compare the initial clinical management of hospital-treated deliberate self-poisoning patients with major depressive disorder (MDD) or borderline personality disorder (BPD) Diagnostic group had no effect on length of stay in the HATS unit, or psychiatric follow up.
PdxBPD less likely to have a GP follow up arranged.
For mild to moderate suicidal ideation, PdxBPD were more likely to be discharged to a psychiatric hospital than people dx MDD. Families feel traumatic stress as they experience their child self-harming, attempting suicide and being near death.
Families feel the burden of care was put onto them as professional would use family support as a reason to avoid clinical crisis intervention.
Families did not feel supported by professionals, and heard contradictory advice regarding how much to support their loved one.
Families felt there were no discharge plans.

80%
Approach to data analysis not clear Suicidality could be PdxBPD's desire to take responsibility for themselves PdxBPD had difficulty thinking of others in times of crisis, and how suicidality may impact on loved ones PdxBPD felt the need to hide their experiences of wanting to kill themselves PdxBPD saw barriers to feeling crisis intervention was beneficial were having responsibility removed, not having the power to make their own decisions, nurses indicating that they may be selfish and not being seen as a person.

100%
Horn, Johnstone and Brooke (2007) To explore user experiences and understandings of being given the diagnosis of BPD. PdxBPD evaluated preventative admission positively PdxBPD using preventative admission felt more control over crisis and felt encouraged to self-manage symptoms until next admission Preventative admissions sees a slight decrease in services used in terms of inpatient days

90%
Quantitative measures of quality of therapeutic alliance only obtained from professional, not PdxBPD Krawitz and Batcheler (2006) To conduct a pilot survey about clinician views on defensive practice when working with adults with borderline personality disorder Community mental health, acute inpatient and crisis teams 29 Professionals (New Zealand) Quantitative Self-report survey questionaire Professionals admitted to making decisions which are not in PdxBPD's best interests but protect professional from legal repercussions.
PdxBPD's families/friends were cited as an influence on defensive practice Media cited as biggest influence on defensive practice 60% Sampling strategy not clear. Inclusion and exclusion criteria not given.
Approach to data analysis not clear Lawn and McMahon (2015a) To explore experiences of care from the perspective of Australians dx with BPD.
153 PdxBPD (Australia) Quantitative Online survey Feeling suicidal, feelings of self-harm and feeling unsafe most common reasons for PdxBPD when seeking hospital admission.
PdxBPD had difficulty accessing services.
PdxBPD reported high levels of distress when refused hospital admission. 65.4% (n = 78) reported experiencing discrimination, particularly as inpatients Variation in preferences regarding care: Usefulness of identifying early warning signs, developing crisis plans and hospital admission 100% was mixed, some finding these helpful and some unhelpful. Lawn and McMahon (2015b) To explore carers experiences of being carers, their attempts to seek help for PdxBPD, and their own needs 121 family carers of PdxBPD (Australia) Quantitative Online survey Carers perceived lack of choice of support services, difficulty accessing support when needed, and lack of long term consistent support all contributing to anxiety in PdxBPD Carers perceived most challenging issue for the PdxBPD was not being taken seriously Carers had often asked for PdxBPD to be admitted to hospital but been refused.
GPs were rated as the most responsive health professionals.

100%
Little et al (2010) To explore the emotional reactions, concerns and beliefs related to working with PdxBPD by health and non-health related agencies Police, health and welfare, and mental health services 378 Professionals (Australia) Quantitative Self-report questionnaire Different agencies respond in different ways to PdxBPD Police saw PdxBPD as a nuisance Mental health staff believed PdxBPD were best managed out of hospital and without medication Mental health staff more likely to understand why a person either wasn't admitted or was being discharged despite ongoing suicidality than police or health and welfare professionals 80% Variation in sample sizes of professional groups. Police (n = 210) Health and welfare (n = 120) Mental health (n = 51) Lohman et al (2017) To identify key resources for and barriers to obtaining supportive and treatment services for BPD from the perspective of individuals seeking information or services related to BPD ("BPD care seekers").
BPD resource centre 500 randomly selected subscripts from 6253 resource requests (USA)  (2014) To explore the experiences of individuals with a diagnosis of BPD in accessing adult mental health services and to better understand which aspects of contact with services can be helpful or unhelpful.