The woodland is a magical place full of elves who are very clever and technically brilliant. They create new knowledge, they understand how things work, they can look and see whether the emperor is exquisitely tailored or whether his bum is showing. These elves have not written this blog. These elves are off doing important elf things and instead a shouty, ranty elf has wandered in.
Today we’re looking at a paper that examines the evidence for interventions offered at crisis time to those who are given the diagnosis of “personality disorder”. I’m putting “personality disorder” in quotes to recognise the contentious nature of this diagnosis. Some feel that saying someone has a personality disorder is akin to saying that that they have a disordered personality. Others feel that if you abandon the rules of the English language this is not the case.
“Personality disorder” was used in this study though, purely for the practical reason of wanting to find out what was offered to those who get the label when they are in crisis. The study took the idea that people with this diagnosis get offered some help when in crisis (not always the most helpful help) so tried to discover the different types of help offered and what the evidence base for the different types were.
How did they do it?
The authors wandered into the library and started pulling texts from the shelves with wild abandon. They looked for studies where:
- The participants were adults.
- They met the criteria for a personality disorder diagnosis.
- At least one outcome measure was used.
- At least one crisis service was used.
The studies had to include a population where at least 50% met the criteria for ‘personality disorder’ or where those with that diagnosis were evaluated as a a subgroup.
You counted as a crisis/acute service if…
- You said you were
- The service was less than 3 months long and you could be referred from ER/A&E/the community
- You described yourself as an alternative to admission.
They did a lot of things to make sure they found the right papers and that the papers were relevant. My research background isn’t strong enough to tell you whether the search strategy was inclusive enough, whether they might have missed some key things or whether too much was included. They’ve written lots on it, they’ve described a method to ensure quality and there’s some searchers of stature among the authors so they’ve convinced me.
[Editor: they followed PRISMA guidelines, pre-registered their protocol, and included “quantitative study designs, including RCTs, quasi-experimental studies and observational studies reporting pre-post outcomes”. More detail here].
They found 12,368 original papers which they finessed and refined down to 35 relevant papers. You might think that given they were including papers published over the last 70 years, 35 is a pretty low number of studies. They all took place is western, wealthy countries. There were five RCTs, one cluster RCT, two non-randomised studies with contemporaneous controls, six non-randomised studies with historical controls and 21 cohort studies reporting only changes in pre-post outcome measures over time. Two of the six controlled trials had a high risk of bias, and the other four had a medium risk of bias, which leaves us with the sense that the evidence being presented shouldn’t be relied on too much.
So, what works?
Acute Psychiatric Hospital Admission
The studies showed that people CAN improve, but the absence of control groups means we don’t know if the improvements would have happened anyway.
This is where “a treatment contract was negotiated in advance, and it was a novel approach designed specifically for people with a ‘personality disorder’ diagnosis that aimed to develop a more positive therapeutic alliance than with general psychiatric admission”. Admissions tended to last about 3-5 days and people generally had more autonomy.
People who received this intervention had less bed days in all the studies, but not compared to groups in controlled studies who received treatment as usual. We can be moderately certain that brief admissions do not reduce bed days.
There was only one eligible paper for this with a team where 18% of patients made up 51% of the crisis team referrals. The evidence that this intervention had a benefit wasn’t really there.
Acute day units/day hospitals/acute partial hospitalisation
Once again, there were no controlled studies, so it can be argued that “people improved” but there’s no counter to the statement “they might have improved anyway”.
Outpatient‑based crisis‑focused psychotherapies or psychosocial interventions
These looked at a range of different approaches from Dialectical Behaviour Therapy inspired interventions to the chillingly monikered ‘abandonment therapy’. Nevertheless, the results for these approaches were promising and the evidence suggests that we can be moderately certain that they reduce hospitalisation and suicidality. Interestingly Joel Paris is often vilified for being the champion of not admitting those with a Borderline Personality Disorder diagnosis. What he actually champions is not admitting people, but doing this instead.
Psychological or psychosocial therapies based in Emergency Departments or Psychiatric Emergency Services
“Studies found promising improvements in outcomes following introduction of crisis focused psychosocial therapies in the emergency department for people with a diagnosis of a ‘personality disorder’. A lack of randomised studies means that firm conclusions cannot be drawn about the effectiveness of these interventions”. The certainty we can have about them being effective is very low.
Psychotherapy groups on inpatient wards
The certainty of the evidence was low.
Mother and Baby Units
There was no evidence of benefit.
Joint Crisis Plans
There was no difference between subject and comparison groups.
Community Early Intervention Services
There was no real evidence for these services.
Well… over the past 70 years very little has been written. There is little reliable evidence.
One of the areas with least evidence, in proportion to how much it’s used, is the crisis team. We can’t say that it helps people diagnosed with a personality disorder. We can’t say that it doesn’t. If I were designing a service for people who had experienced high levels of abuse, had poor experiences of care where the evidence is for sustained and containing relationships, I would probably avoid designing a service based on brief contacts with multiple members of staff with timings that are variable and unpredictable.
The two interventions with the greatest success sound interesting. One was the abandonment therapy and another part of a stepped care model in Australia. The latter is something both NICE and the Royal College of Psychiatrists suggest rather than services working in silos.
The headline finding of this paper or at least, the bit that I see most people talking about is the idea that there isn’t evidence that hospital causes harm. The National Collaborating Centre for Mental Health are clear about avoiding admission, the NICE guidelines say:
- ensure the decision is based on an explicit, joint understanding of the potential benefits and likely harm that may result from admission
- agree the length and purpose of the admission in advance
- ensure that when, in extreme circumstances, compulsory treatment is used, management on a voluntary basis is resumed at the earliest opportunity.
For those I work with, I struggle to recommend environments with huge power imbalances, massively reduced autonomy, arbitrary rules, treatment delivered under compulsion (or implied threat of compulsion) where staff often don’t know how to help. That’s not a reason to blanket ban people with a particular diagnosis, more a reason to see how people react to a ward environment then learn whether that way of supporting people helps.
Is the idea about keeping those with a ‘personality disorder’ diagnosis off wards a product of stigma or a way of protecting them from it? Our wards should be better equipped to help this group of people but until they are, I’m helping people avoid them. My opinions are heavily influenced by the clear harms I see admission having on those I work with. These are just opinions though and if people were motivated to build the evidence around the harms of admission, there is lots of work to do to capture the experiences of those who have been detained. This is unlikely to come from an RCT and as such, these qualitative options will take a lot of time to influence practice. How long? Well, it has been 20 years since No Longer A Diagnosis of Exclusion (NIMHE, 2003) told us:
“The DH [Department of Health] will engage in dialogue with the Royal Colleges, regulatory bodies and curriculum setting bodies
- to address the gap in training provided at pre-registration and pre-qualification for key disciplines
- to influence the content of undergraduate syllabuses
- to influence the mechanisms determining the selection of CPD educational opportunities”
And “The DH expects to pump prime the development of new training opportunities, inviting tenders from recognised sites of good practice and from training providers to offer a range of inputs to trusts delivering personality disorder services, and to expand the pool and range of personality disorder courses available nationally.”
Has this happened?
Not really. If ‘personality disorder’ gets mentioned at all in training then it is generally along the lines of people having a disordered personality. Less the impact of trauma, abandonment, and abuse. None of our crisis services will help if we don’t prepare the staff better and with the vacancies the way they are it would be a luxury to be able to select new staff.
Strengths and limitations
Briefly, there were broad search criteria. Some studies included those without a personality disorder diagnosis if it was a mixed cohort. Some participants weren’t formally diagnosed but then we have the real situation of people “seen as having a personality disorder”, whether diagnosed or not. It didn’t include things like crisis houses, assessment and planning in A&E… just because the evidence isn’t there. All the measures used in these studies were around symptoms, self-harm, and other things that users of mental health services are less likely to prioritise. It would be useful to have some agreed measures, but equally I wonder what a crisis service (community or hospital) would look like if designed by the people who used it? Nothing like the current models I suspect.
The chasm of missing evidence is largely due to the paucity of trials with a control group. Why is this important? This is where the idea of “return to baseline” comes along. People in crisis are, by definition, in crisis. The worst things could be. The peak of distress, suffering and pain. People don’t stay at that level forever, so we have to assume that most people will improve regardless. As such, all the studies showing people in crisis got better, without being able to say “better than those who had something else” or “better than those who had nothing” aren’t really of much value.
Conclusion (mine, not the authors)
- We have very little evidence for what helps those who end up being given a diagnosis of ‘personality disorder’ when they are in crisis.
- The studies with the best evidence were those most related to what we know helps people with this diagnosis (or what helps people with recurrent self-harm and suicidality with histories of abandonment and abuse).
- It could be argued that the areas without the evidence are the ones that move furthest from what is known to help.
- It would be useful if services for this group of people were designed with them in mind, rather than creating environments that include what is known to be unhelpful, while a failure to thrive is attributed to their diagnosis.
Maconick, L., Ikhtabi, S., Broeckelmann, E. et al. Crisis and acute mental health care for people who have been given a diagnosis of a ‘personality disorder’: a systematic review.BMC Psychiatry 23, 720 (2023). https://doi.org/10.1186/s12888-023-05119-7
National Institute for Mental Health in England (2003) https://www.nimhe.org.uk/downloads/PDFinal.pdf Accessed 10th November 2023