Our aim was to describe the typical cases presented to the panel, the discussions that took place and the advice given. To best produce this narrative from the data available, we felt a qualitative approach was appropriate. We required a method of qualitatively analysing transcribed data, such as the meeting transcripts that formed our data-set, and decided on thematic analysis. This provided a framework with which to identify and organise the discrete ideas within the transcripts, outlining patterns in the data but also indicating their relative frequency within the meetings as a whole.
Although qualitative research uses many specific, prescribed techniques to evaluate non-empirical data, there has historically been some discrepancy as to what strictly constitutes thematic analysis and its methodology. For our purposes, we have followed the model proposed by Braun and Clarke,4 which describes six discrete stages of thematic analysis.
The stages of the above method as applied to our analysis are described in the Appendix.
Discrete ideas (referred to in the process as ‘codes’) and broader groupings of these ideas into key patterns (‘subthemes’) were identified. The subthemes can then be grouped together to identify the main themes within the data.
The Risk Reference Panel meetings are attended by a secretary who records what is said by typing verbatim as far as is possible; there is no audio recording of the meetings. These transcripts are then used to produce letters to the referring team, outlining what was discussed and the recommendations of the panel.
A pilot analysis was conducted whereby authors A.T. and G.J. independently applied the method described below to a sample of original transcripts for 8 of the 48 cases that had been presented to the Risk Reference Panel between May 2010 and April 2016. Ethical approval was not required as our study is a retrospective service evaluation where no change to the service was implemented and no service users were interviewed. All reported data are anonymised. Anonymised samples of the transcribed data, as well as examples of the codes generated from them and how they were recorded, are given in Boxes 1 and 2.
Box 1 Sample transcription and coding process
Box 2 Box 2 Sample of code database
The pilot analysis generated an initial list of codes that were used as the basis for coding the rest of the transcripts, as well as demonstrating an acceptable consensus on what constituted individual codes. The remaining 40 transcripts were then coded, 20 each by authors A.T. and G.J., and the frequency of codes were recorded. The resultant database allowed the rest of the thematic analysis to be completed. A thematic map provided a diagrammatic representation of the various themes, subthemes and relationships between them (Fig. 1).
Fig. 1 Thematic map of Risk Reference Panel meeting transcripts. CJS, criminal justice system.
A feedback form allowed participants to indicate, via a Likert scale, the degree to which they agreed with a number of statements (outlined under ‘Feedback’ within Results) and to provide free-text feedback.
Using this method, 16 subthemes were identified to group similar codes. The subthemes were further grouped into four principal themes: childhood risk factors, current presenting difficulties, assessment and management of risk to self or others and team or wider organisational issues. The first two themes encapsulate information presented to the panel during presentations by referrers; the latter describe recommendations made or issues identified by the panel.
The subthemes are organised by frequency of codes, referred to hereafter as occurrences, with absolute frequencies (i.e. the number of times a code occurred in the whole of the transcribed data, given as n) and the proportion of occurrences within that theme (given as a percentage). Percentages are rounded to the nearest 0.5%. There were a total of 522 occurrences of codes throughout the 48 transcripts analysed.
Childhood risk factors
The first theme contains codes pertaining to the history of the case, outlining risk factors pertinent to the current problem. There were four individual codes that accounted for 31 of the 522 occurrences (i.e. 6% of all occurrences). These were primarily adverse experiences that the person had endured in childhood, but also included childhood traits that are associated with subsequent antisocial behaviour.
The most common childhood adverse experience was being a victim of physical abuse (n = 14, 45% within theme) or sexual abuse (n = 11, 35%), followed by having been in care (n = 3, 10%). A history of aggressive behaviour as a child was also identified in a number of individuals (n = 3, 10%).
Current presenting difficulties
This theme accounted for the greatest proportion of individual codes, with 245 occurrences (47%). Five subthemes were identified within current presenting difficulties, namely nature of offences, problematic behaviour, factors affecting current presentation, police/criminal justice system involvement and established diagnoses.
The nature of offences subtheme explored the various types of offences committed, if applicable (only some of the cases presented had criminal convictions, hence their low frequencies). The offences captured in the transcripts included arson (n = 3, 1%), non-assaultive sexual offences (n = 3, 1% within theme), threats to kill (n = 1, 0.5%) and sexual assault (1, 0.5%).
The most common problematic behaviour was suicide attempts/self-harm (n = 32, 13%), which was often seen to be chaotic and instrumental. However, there were several cases where severe self-harm without suicidal behaviour was the primary problem. The next most frequent was a lack of engagement with services (n = 19, 8%). The remaining codes within the problematic behaviour subtheme were sexually inappropriate behaviour (n = 7, 3%), antisocial behaviour (n = 7, 3%), disruptive behaviour (n = 6, 2.5%), harassment of care-giving staff (n = 6, 2.5%), multiple contacts by phone (n = 6, 2.5%), absconding risk (n = 6, 2.5%), complaints to health board/Welsh Assembly Members/counsellors (n = 5, 2%), false allegations (n = 4, 1.5%) and litigation (n = 1, 0.5%). Although the most prevalent individual code was risk of self-harm or suicide, the combined frequency of codes relating to risk to others was around 1.5-times as high.
Another subtheme was factors affecting current presentation, which captured issues in the person's life at present that were affecting their level of risk. Substance misuse was the most prominent among these (n = 27, 11%). Other codes within this subtheme were chronic pain/illness/physical issues (n = 7, 3%), social isolation (n = 7, 3%), multiple admissions to psychiatric hospital (n = 7, 3%), homelessness/moving area frequently (n = 6, 2.5%), family history of mental health issues (n = 5, 2%), admissions under the Mental Health Act (n = 5, 2%) and lack of insight (n = 5, 2%).
Less commonly, people had problems with anger (n = 3, 1%), bereavement (n = 3, 1%), sexuality (n = 3, 1%) and personality traits without a diagnosis of personality disorder (n = 2, 1%). There was police involvement in nine cases (3.5%) and the person was subject to the criminal justice system in 15 cases (6%).
Most of the cases presented to the panel had an established diagnosis, although within the 60 code occurrences in this theme, ‘conflicting/unclear diagnosis’ occurred 13 times (21.5%). These subthemes grouped the diagnoses by psychopathology.
‘Developmental disorders’ were the most common presentation and included autism spectrum disorder (n = 8, 13%), intellectual disability (n = 5, 8%), attachment disorder (n = 5, 8%), ‘Special needs’ (n = 3, 5%) and attention-deficit hyperactivity disorder (n = 3, 5%). Emotionally unstable personality disorder (EUPD) was the only specified personality disorder diagnosis (n = 9, 15%) and the most common individual pathology. This was followed by psychotic illnesses, with ‘treatment-resistant psychosis’ accounting for seven cases (11.5%). Schizophrenia was only given as a named diagnosis in one case (1.5%). Lastly, affective/anxiety disorders included depression (n = 3, 5%), eating disorders (n = 2, 3%) and post-traumatic stress disorder (n = 1, 1.5%).
Risk assessment and management
There were 31 occurrences of codes (6%) related to an assessment of the level of risk posed to others by the individual. Three of these (9.5% within theme) were an observation that there was a poor understanding of the risks posed and that specific further information was required.
Much more of the output from panel meetings was regarding the management of specific risks to the person or others, with 95 code occurrences. Within this theme, four subthemes were identified: need for further assessments, specialist care plans, capacity/placement factors and responsibility of other services.
Most commonly, the panel advised that specific assessments would be beneficial to manage risk. In disorders that modulated risk, such as autism spectrum disorder, the panel would often suggest ‘expert/specialist opinion required’ (n = 17, 18%). If the risk was more general, then suggestions of which avenues to pursue included a need for HCR-20 (Historical Clinical Risk Management-20; an established actuarial tool for assessment of violence risk) (n = 3, 3%); legal advice (n = 4, 4%) and a need for more background information (n = 11, 12%) or updated formal risk assessment (n = 8, 8%).
Issues regarding the assessment of capacity, or the established lack of capacity, were also commonly cited. There were 14 occurrences (15%) of recommendations that a Mental Health Act assessment had not been considered and could be of use in containing the risks posed. Further to this, there were 13 recommendations (13.5%) that the acute problems should be managed as an in-patient.
Several of the cases indicated longer-term in-patient or other compulsory interventions were necessary, such as the need for low secure placement (n = 5, 5%) and management under guardianship (n = 2, 2%) or Protection Of Vulnerable Adults scheme (n = 2, 2%).
Team issues within the context of risk management were need to improve engagement (n = 1, 1%), need to take positive risks (n = 1, 1%) and need for child protection training for team members (n = 1, 1%).
In eight cases (8%) the panel identified that responsibility for the patient's actions did not lie with health services as they were independent of mental health issues. These were divided into a recommendation that their behaviour be dealt with under the criminal justice system (n = 4, 4%), or an acknowledgement that although there is no criminal element to pursue in some cases, all appropriate actions to reduce risk have been considered and there is no more that the team can realistically do (n = 4, 4%).
Occasionally, the panel recommended specific management plans to manage particular disorders. Most frequently this was stated as a need for a specialist care plan (n = 24, 77%), incorporating advice from specialist assessments. The need for structure in the patient's life was identified (n = 3, 10%), as was the need for family work/family meeting (n = 1, 3%). More specifically, clozapine for treatment-resistant psychosis (n = 2, 6%) and dialectic behaviour therapy for EUPD (n = 3, 10%) were suggested.
Team/wider organisational issues
The final theme generated by the analysis contained 29 occurrences (0.5%) between five codes, and was subdivided into team working issues (n = 8, 27.5%), funding issues (n = 7, 24%), need for special staff/staff rotation (n = 6, 20.5%), need extra support for the staff involved (n = 5, 17%) and trust management issues/needs escalation (n = 3, 10%).
Team working issues captured differences within the team that may have hindered progress in the management of particular behaviours. For example, the panel acknowledged that because of splitting in the context of EUPD, it was necessary to establish a clear and consistent boundary narrative. However, outright disagreement within teams, over significant issues such as primary diagnosis and level of risk posed, was identified as an issue.
Funding issues identified that responsibility for funding was not always clear and had delayed the provision of appropriate placement. This was particularly relevant where the person had moved frequently between areas or was of no fixed abode.
The panel stressed that the staff who were dealing with particularly demanding individuals needed extra support themselves, which can often be neglected, and suggested rotation of staff to prevent burnout. In a small number of cases it was identified that the particular service or treatment indicated for a person's condition, such as dialectic behaviour therapy, was not available, and it was suggested that this was escalated on a trust level on the basis that this would reduce risk of harm.
Feedback was overwhelmingly positive, with 20 out of 22 respondents (91%) indicating agreement or strong agreement with the following statements: ‘New ideas were generated in the session’, ‘The case feels safer as a result’, ‘I learned a great deal during the session’, ‘The questions I brought were answered’ and ‘I would recommend the process to a colleague.’
The Royal College of Psychiatrists has highlighted a need for improved, tiered risk assessment and interventions on both a local and national level.5 A qualitative analysis of the content of Risk Reference Panel meetings has illustrated how it provides a potential model for services to address this need. The panel is an innovative service model that was set up to provide a further level of support for teams who had been looking for support in managing very complex individuals with high-risk behaviour and has provided new direction and specialist advice in the management of these cases.
Most cases described physical or sexual abuse, with many experiencing multiple adverse childhood events (ACEs). These findings support the existing body of work on the relationship between multiple ACEs and subsequent morbidity and risk behaviours, such as that by Bellis et al,6 which showed that individuals with four or more ACEs were several times more likely to smoke, drink heavily, have low mental well-being and chronic health conditions. This also correlates with well-documented risk factors for self-harm,7 suicide8 and interpersonal violence.9 These data have more general implications on history-taking in risk assessment, with the inclusion of this background information crucial to the working of the panel. The potential for referrers to exaggerate the risk, both consciously and unconsciously, to encourage acceptance by the panel should be acknowledged.
Although a person's risk of harm cannot be entirely removed,5 particularly in such complex cases, a thorough understanding of risks posed and their severity is crucial. The panel often identified areas where risk was not fully understood and suggested specific assessments or specialist opinion that would be beneficial. Having said this, the responsibility for risky behaviour, particularly in a capacitous patient, does not lie entirely with mental health teams and it is prudent to identify when all reasonable steps have been taken, or if involvement of the criminal justice system is more appropriate. Taking the case to such a panel allows an independent debate and makes this decision more defensible.
For patients who remain in the community, it is particularly difficult for teams to provide care and ameliorate risk if engagement is poor. Surprisingly, the meetings often highlighted that capacity assessments and use of the Mental Health Act beyond admission for brief assessment had not been considered. This may reflect a lack of appropriate placements available to community teams, particularly long-term in-patient services when patients cannot be managed in the community.
This model provides a valuable multidisciplinary source of special expertise in the management of complex and exceptional cases, where risk of harm is significant and clinical teams have ongoing concerns. The format may not be appropriate for every service and there is also a place for improved training in formulation and management of patients with complex personality difficulties, but model could certainly be reproduced in other areas where clinical teams feel it would have a beneficial role, particularly to inform Care Programme Approach unmet needs, training needs within the staff workforce and service development needs.
Lastly, it is pertinent to consider alternatives to such a panel. As the heaviest burden appears to be personality pathology, these would primarily consist of improving training in managing patients with personality disorders. In some cases, it may be that this is the better and more economical option, although it has been acknowledged that the panel approach uses relatively few resources in terms of clinician sessions while providing an additional tier of support for community teams to reduce the risk of harm associated with mental illness.
This study is only able to describe the function and output of the panel in question, rather than give a rigorous analysis of this model in comparison with other models, and there were no available data on matched people who were not referred. We were also unable to comment on the make-up and training of the teams referring into the panel; it is possible that the referrals represent professional anxiety that stems from sources other than the patient in question, such as gaps in senior management or reluctance to engage with complex individuals with personality pathology, although the panel meetings analysed cover a period of 6 years where it is likely that there would be some fluctuation in team make-up.
All referrals to the panel are made via a written referral letter, which details the history of the case, the assessments that have previously been conducted, a risk assessment and concerns that the referrers wish to address. As the panel will already be familiar with this information, some of it may not have been verbalised at the meetings and thus not captured by the transcription. However, it was noted that there was a verbal presentation of each case and that the relevant history was given.