Crisis and acute mental health care for people who have been given a diagnosis of a ‘personality disorder’: a systematic review

People with complex emotional needs (CEN) who may have been given a diagnosis of a ‘personality disorder’ require access to good quality services when they experience a crisis. For this population, crises can be a time of extreme emotional distress with increased risk of suicide, suicide attempts, non-suicidal self-harm (NSSH) and loss of social functioning. Some people with a ‘personality disorder’ diagnosis are frequent users of mental health crisis services [1] and can present to a spectrum of other services such as Emergency Departments and the police in a crisis [2]. A crisis can be difficult to define due its subjective and personal nature, but common themes in a range of definitions are that crises represent a time of heightened vulnerability, threaten the person’s sense of equilibrium, involve a period of overwhelming emotions and loss of control, can result in increased risk of self-harm or harm to others and/or might result in a person presenting to mental health or health services for urgent help [3]. Crisis and acute mental health services that a person may access include crisis or home treatment teams, acute psychiatric hospital admission, acute day units, crisis houses, crisis-focused psychological therapies or crisis houses/sanctuaries. The availability of different service models tends to vary considerably not only between countries, but between areas within the same country [4].

The term ‘personality disorder’ is controversial and historically the diagnosis has excluded people from services [5] as it was wrongly perceived as untreatable [6]. There is significant stigma associated with the label personality disorder, both amongst the wider public but also among clinicians in health services [7]. Some service user groups prefer the term ‘Complex Emotional Needs’ (CEN) and services in the United Kingdom are increasingly adopting this [8]. Some services use the term CEN to avoid the therapeutic pessimism and stigma associated with the diagnosis of “personality disorder”, often including in this term a broader population of people, including people who might have some needs associated with the diagnosis such as recurrent self-harm, but might not meet the full diagnostic criteria for a ‘personality disorder’[8]. However, there is presently no consensus about what terms and language should be used in the academic literature, mental health services and society at large to describe the difficulties that are currently labelled as a ‘personality disorder’. We advocate future work to develop less stigmatising and more valid ways of describing these difficulties. However, in this review we have used the ‘personality disorder’ term and construct to determine the inclusion criteria and search strategy, as this is the term used in the primary literature that we are reviewing.

Previous reviews examining crisis interventions and services for people with a ‘personality disorder’ diagnosis have limited inclusion criteria to randomised controlled trials (RCTs) or have not included studies reporting outcomes from commonly used crisis service models. A Cochrane review performed in 2022 of crisis interventions designed for people with a ‘personality disorder’ diagnosis found only two studies: a RCT of a brief admission intervention [9] and a pilot RCT of crisis planning [10, 11], both reporting no effect of the study intervention compared to treatment as usual. A systematic review conducted in May 2022 searched for randomised and observational studies of crisis-focused psychosocial interventions for ‘borderline personality disorder’ that can be delivered within acute care and found five studies [12]. In four of these studies the intervention was a form of psychological therapy and in one study the intervention was joint crisis planning. This review found that crisis psychosocial interventions are feasible but there was insufficient evidence to recommend a particular intervention. Both of these reviews focused only on crisis interventions designed specifically for people with a ‘personality disorder’ diagnosis, rather than more commonly delivered transdiagnostic models of crisis care. Neither review included studies that reported outcomes for people with a ‘personality disorder’ diagnosis following use of the crisis services that are currently available in the UK and many other countries, such as crisis teams, hospital admission, acute day services or crisis houses.

Other reviews have examined crisis interventions in a wider range of clinical other populations. For example, a 2015 Cochrane review of crisis interventions for people with severe mental illness (which included people with a ‘personality disorder’ diagnosis) did investigate the effects of crisis interventions, including models that a person with a ‘personality disorder’ diagnosis may be able to access such as crisis teams [13]. However, this review did not investigate effects or outcomes by diagnostic group. Given that it is thought that people with a ‘personality disorder’ diagnosis might experience particularly poor outcomes from crisis care and that recommendations are made in national guidelines to take particular care in avoiding over-use of inpatient crisis care for this diagnostic group [14], there is justification for investigating outcomes specifically for people with this diagnosis. A recent review of the qualitative literature has been performed by our research group synthesising experiences of non-inpatient crisis services, and found only a small number of published studies focusing on experiences of emergency departments [15].

Some commentators have argued that certain types of acute care, especially inpatient admissions, are thought to have potential to cause harm for people with a ‘personality disorder’ diagnosis, through encouraging regression, loss of control and coercion [11]. This is reflected in national guidelines for care of people with a ‘borderline personality disorder’ diagnosis, advising against admission to hospital in a crisis if possible, or advising that when admissions are used they are brief in duration [9, 16, 17]. These recommendations reflect the approach used in an evidence-based psychological intervention for people diagnosed with a ‘personality disorder’, Dialectical Behavioural Therapy (DBT), in which steps are taken to avoid admission to hospital where possible, with the aim of not reinforcing behaviours such as self-harm and to prevent the loss of coping skills [18]. However, clinical practice recommendations often cite a small number of studies relying on observational data [19] or expert opinion [16] and the strength of the evidence examining the benefits or harms of hospitalisation in this group appears to be low. Poor experiences of acute care may be due to a lack of the application of therapeutic models or interventions specifically aimed at supporting people with difficulties associated with a ‘personality disorder’ diagnosis in crisis services, and some suggest that a well-functioning treating team can provide an opportunity for useful care in acute inpatient settings [20]. Additionally, if it is the case that hospital admission is to be avoided, then there is little information about what alternative crisis care options might better meet the needs of this group. It is therefore important to establish what is known about the outcomes experienced following use of crisis care models for people with a ‘personality disorder’ diagnosis from existing evidence.

No previous review has focused on the outcomes experienced by people with a ‘personality disorder’ diagnosis following use of the currently available models of crisis and acute care. The inclusion criteria for the current review have been kept broad given that this field of research is in its early stages. It is therefore of benefit to describe all preliminary data that is available. This will add to the previous more focused reviews which have only found a small handful of eligible studies.

We aimed to summarise current knowledge of the clinical and social outcomes experienced by people with a ‘personality disorder’ diagnosis following use of acute/crisis services and interventions. We also aimed, subject to sufficient evidence being available, to compare the effectiveness of the different available models of crisis care for people with a diagnosis of a ‘personality disorder’.

We performed a systematic review following PRISMA guidelines [21]. The protocol was pre-registered on Prospero prior to commencement of screening (PROSPERO 2022 CRD42022313720).

Searches of bibliographic databases (including grey literature databases) returned 16953 results, of which 4585 were duplicates. We screened 12368 titles and abstracts, with 11763 results excluded. We assessed 605 full texts for eligibility resulting in 35 studies included in the review. Reference lists of reviews that had been identified were searched and forward and backward citation searching was performed but this did not identify any new studies meeting inclusion criteria (see Fig. 1, PRISMA flow chart). No grey literature that met the inclusion criteria was found.

Based on the 10% of records abstract screened by a second reviewer, there was initially 85% agreement. After discussion all discrepancies were resolved. For full text screening, there was 83% agreement. Discrepancies were resolved after discussion, with two discrepancies resolved by a third reviewer. At both stages there was revisiting of the inclusion criteria with both reviewers.

The number of eligible studies identified in this review was low considering the wide inclusion criteria, demonstrating the lack of evidence in this field. Overall, the quality of evidence available was low, with only six randomised studies found across all models and a large number of uncontrolled studies. The majority of studies were judged to be at high risk of bias and the certainty of evidence was judged to be low or very low for all models except for brief admission and outpatient-based crisis-focused psychosocial and psychological crisis interventions. The literature for acute care is smaller than for community care for people with this diagnosis, which is also under researched [67]. The review included studies that were found in the previous more focused reviews [10, 12], but also included a wider range of literature.

There were a large number of studies found reporting only changes in pre-post measures over time without a comparison group, particularly for the outcome of symptomatic improvement. The majority of these studies reported improvements in symptom scores after using crisis care, whether that was hospital admission, crisis teams, acute day units or psychotherapies. Result of these uncontrolled studies have very limited implications however, as it would be expected that participant will show an improvement over time, particularly when baseline measures are performed during an acute crisis. The lack of comparisons between different crisis models means that there is little information about the degree of improvement when using one service over another or compared to not accessing crisis services.

The literature on crisis teams is particularly sparse and low quality, with only one study identified, and this reported only changes in pre-post outcomes over time. This is an important evidence gap given that crisis teams are recommended in the most recent UK NICE guidelines as a first line intervention in a crisis for people with a ‘borderline personality disorder’ diagnosis who may need hospitalisation [16]. A previous systematic review of qualitative literature also found no published qualitative studies exploring experiences of people with a ‘personality disorder’ diagnosis using crisis teams [68].

Two outpatient-based crisis-focused psychotherapy models were the only models of care identified that were supported by RCT evidence of a statistically and clinically significant benefit compared to treatment as usual. These were a 4-session crisis focused psychosocial intervention from Australia [39] and a 3-month intensive manualised therapy called ‘abandonment psychotherapy’ [28], designed to be used following presentation in crisis and focusing on relationship losses. Both interventions were delivered within a system of acute care that provided concurrent risk assessment and the option of brief hospitalisation, medication and more intensive input if the individual needed this. The Australian intervention was also offered within a stepped care model that offered referral to longer term treatments. Crisis-focused psychological and psychosocial outpatient interventions that can be delivered as part of an acute care system may therefore be a promising area for future research. That these services do not operate as standalone interventions but require interaction with the rest of an acute care system may limit generalisability to contexts where acute care options are not extensively developed.

The evidence we identified that investigates acute hospitalisation for people with a ‘personality disorder’ diagnosis in crisis was of low quality for all outcomes. The lack of certainty in the evidence is important, as many clinicians and policy makers believe that hospitalisation for people with a ‘personality disorder’ diagnosis can cause iatrogenic harm and can cause a deterioration in symptoms and in risk [19, 69], although much of the academic literature is more equivocal on this question [11, 16, 70, 71]. Evidence that demonstrates clearly whether and when acute psychiatric admissions might be helpful or harmful was not found in the studies identified by this review. Four included studies reported an improvement in symptom scores during hospital admission [37, 38, 53, 54] and no studies reported a deterioration, although most studies did not capture adverse events or other possible negative outcomes such as self-harm. Improvements in symptoms between admission and discharge tended to be reported in studies, but whether hospitalisation had contributed positively to theseimprovements was unclear, as improvements over time may well have occurred in other settings and people are more likely to be discharged at a point when improvement has been observed. Negative experiences that may result from hospitalisation, such as violence, coercion and other traumatising experiences [72], tended not to be reported in the studies we retrieved. Based on the quality of evidence available, conclusions cannot be drawn on the effectiveness of acute hospital admission in crisis for people with a ‘personality disorder’ diagnosis. The current lack of evidence supports neither offering acute hospital admissions nor refraining from doing so wherever possible. The basis of beliefs that hospitalisation is harmful is questionable given the limitations of the evidence regarding harms. We would suggest that clinical recommendations should reflect this uncertainty and care should be taken to reduce the impact of stigma against this group in influencing treatment choices. Given that people with this diagnosis continue to be offered acute admissions, there is a need to develop and implement co-produced strategies to improve experiences and outcomes of inpatient admissions when they occur. This review did not compare outcomes for people with a ‘personality disorder’ diagnosis to other diagnostic groups using acute care, which could be useful further work.

The evidence for brief admission was considered separately from general acute psychiatric hospital admission, as brief admission in the majority of studies referred to a distinct model tailored to people with a diagnosis of a ‘personality disorder’ in which individuals at high risk of admission to hospital entered into a pre-agreed treatment contract and modifications were made to the inpatient experience, for example the person managed their own medications. Although the quality of evidence for brief admission was of higher quality than acute psychiatric hospital admission, it is not possible to draw generalisable conclusions about hospitalisation from the randomised controlled trial of brief admission, as both the intervention and treatment as usual groups also experienced general psychiatric hospital admissions during the follow-up period.

It was not possible to perform a meta-analysis in this review due to the wide range of outcome measures used in different studies. This was particularly the case for measures used to quantify symptomatic improvement, where there were very few studies with any overlap in the measures used. Studies tended to use a large battery of different measures, covering domains such as anxiety, depression, self-harm, hopelessness, and social participation. The suitability of measures that were developed to focus on relapse and recovery in depression and anxiety when applied to people with a ‘personality disorder’ diagnosis is not known. There is debate about the suitability of traditional concepts of recovery for people with a diagnosis of a ‘personality disorder’, for whom the course of a person’s difficulties might not be characterised by episodic relapses and linear, sustained improvements in symptoms, but living with more longstanding difficulties and more frequent fluctuations that may require crisis support, but with gradual improvements in the development of relationships, social functioning and sense of self alongside this [73‐75]. The need for a wide battery of measures to capture different aspects of symptoms and social functioning may place a larger burden on participants, potentially reducing recruitment and retention in studies, particularly of those who are most unwell. In contrast services have previously been found to focus too narrowly on improvements in self-harm and emotional dysregulation [75]. The development of a core set of outcome measures should be considered for future studies that are meaningful for people with a ‘personality disorder’ diagnosis and can reflect the fluctuations in symptom severity that can occur. The most recent UK national guidelines (NICE) for the care of people with a ‘borderline personality disorder’ diagnosis recommend the development of an agreed set of outcome measures as a priority for future research [16]. The International Consortium for Health Outcomes Measurement (ICHOM) conducted a Delphi Study published in 2021 aimed to develop a standard set of patient reported outcomes for ‘personality disorder’ but only one measure recommended in this paper was found in the literature for this review (WHODAS) and it is unclear the extent to which this has been adopted in ongoing research [76].

Other than measures of symptomatic severity, the next most commonly reported groups of outcomes were service use, including hospitalisation and days in hospital. There was a lack of information across studies about experiences of care of both staff and people using services. The subjective experience of care seems particularly important in a crisis where the aim of services is to provide support during a period of distress. A meta-synthesis we performed of qualitative studies describing experiences of crisis care for this group found a paucity of literature, but with participants emphasising the importance of the perceived attitudes of staff, skills in communication and the quality of relationships formed between crisis service staff and service users as being central to the experience of crisis services, regardless on the model of care used [15]. The lived experience working group for this review expressed scepticism for the expectation that a person should improve on quantitative symptomatic measures over a short space of time in a crisis and instead said they hoped crisis services could offer the feeling of being held and known about by a supportive team that can offer contact and containment whilst they experienced a crisis and began recovery. Efforts should be made in future research to capture outcomes measures more suitable for this group of people.

The literature in this field is sparse and of overall low quality. There were no high-quality studies that investigated outcomes following crisis teams or hospital admissions for this group despite these being the core models of care available in most UK and European mental health services. There were few comparisons available that provide any information about which types of models of care might be more or less effective for this group in crisis. Studies of crisis focused psychosocial and psychological interventions were small in number but showed some promising results. Future work should aim to provide better information on which to base clinical decisions for this group in a crisis, through higher quality studies and considering the use of core outcomes measures that are meaningful for people with a ‘personality disorder’ diagnosis that would allow comparison of different models.

However, while it is certainly validating to know that the evidence base does not support exclusion on the basis of this stigmatising label, that doesn’t change the harsh reality that too many people with CEN continue to suffer the consequences of such discriminatory practices, which are not only limited to hospitalisation, but extend into other forms of crisis care as well.

As the studies included in this review confirm that symptomatic improvements can in fact be achieved with a variety of interventions, there really is no justification for a blanket exclusion of service users with CEN from desperately needed support in crisis.

The lack of direct comparisons between different models of crisis care seems significantly less important, especially considering the highly subjective nature of a ‘personality disorder’ diagnosis and large heterogeneity amongst service users with CEN. As there simply is no ‘one-size-fits-all’, crisis care – like any other mental health service – should always be flexible and person-centred, providing options tailored to meet individual needs and empower service users instead of relying on professionals to make decisions about us, without us.

Maybe it is no coincidence that the interventions that showed the most significant benefit were outpatient-based crisis-focused psychotherapy models that operated within integrated systems which were adaptable and provided opportunities for more collaborative care.

Ultimately, instead of more RCTs with outcome measures of questionable suitability for people with CEN, future research in this area should rather focus on qualitative studies of service user experiences of crisis care as well as addressing the underlying stigma and pejorative attitudes that continue to result in malignant alienation of service users on the basis of a ‘personality disorder’ diagnosis.

After all, when someone seeks help in a crisis, any offer of support is better than none."