It is estimated that 4% of people have a personality disorder. ^{Reference Coid, Yang, Tyrer, Roberts and Ullrich1} Personality disorder has a significant impact on individuals, their families and society at large. Despite this, people with personality disorder have historically been overlooked by mainstream mental health services. Many people with personality disorder are dissatisfied with the care they receive and some clinicians are reluctant to work with this group. ^{Reference Lewis and Appleby2,Reference Ramon, Castillo and Morant3} Concern about the quality of personality disorder services prompted the Department of Health to recommend the development of dedicated community-based services in England and fund 11 new ‘pilot’ personality disorder services. ⁴ Although evidence about the effectiveness of specific treatments for people with personality disorder is beginning to emerge, ^{Reference Binks, Fenton, McCarthy, Lee, Adams and Duggan5} there is little information to guide the development of such services. We therefore conducted a Delphi study to develop a consensus on how dedicated community-based services for people with personality disorder might best be delivered.

Method

Delphi studies involve sequential rounds of questionnaires in which participants are given feedback summarising the views of others and asked whether they would modify their views in the light of this feedback. ^{Reference Murphy, Black, Lamping, McKee, Sanderson, Askham and Marteau6} We developed a 49-item questionnaire on the organisation and delivery of dedicated personality disorder services based on in-depth interviews with service users and providers, and recommendations from the study team. ^{Reference Crawford, Rutter, Price, Weaver, Josson, Tyrer, Gibson, Gillespie, Faulkner, Ryrie, Dhillon, Bateman, Fonagy, Taylor and Moran7} Each question comprised a statement and an accompanying Likert scale ranging from 1 (disagree) to 9 (agree). Two additional items asked panellists to rank 11 priorities for service development and 6 measures of service outcome. Panel members were also encouraged to provide suggestions for additional items to be included in the subsequent round.

The panel comprised service providers (from the 11 national pilot services), service users (recruited via the Department of Health's national personality disorder programme, a mental health charity (the Mental Health Foundation) and a national service user group (‘Borderline UK’)) and academic experts. Academic experts were identified from an electronic search of bibliographic databases of all those who had published at least one peer-reviewed paper on personality disorder services in the UK in the past 10 years. We aimed to recruit equal numbers from each group; the final panel comprised 34 service providers, 34 service users and 31 academic experts. Three rounds were carried out. In the first round, eight additional statements were suggested, which led to five new statements in the second round of the survey. This meant that a total of 54 items were rated, plus the two ranking items.

For analysis, the 9-point scale was divided into bands of disagree (1–3), neutral (4–6) and agree (7–9). Consensus was judged to have been reached when 75% of panellists responded in the same 3-point band. Feedback on the results of the previous round was given to participants in rounds two and three, including the person's previous response, the median rating and information about differences between stakeholder groups where applicable.

Results

Eighty-eight people (89%) responded to the first-round questionnaire, 85% in the second round and 82% in round three. Expert authors and service providers included 21 (34%) psychiatrists, 13 (21%) psychologists, 12 (20%) psychotherapists, 8 (13%) nurses and 2 (3%) social workers.

Consensus was reached on 21 items (39%), which are listed in the online Table DS1. Ten items were endorsed in round one, five in round two, and a further six in round three. Consensus was not reached on most items, including: whether services should work with people with a history of violent offending; the role of assertive outreach; whether people should be assessed in their own homes; use of medication; whether teams needed to have medical input; use of compulsory treatment orders; and whether dedicated out-of-hours services should be provided.

Details of items endorsed by each of the three groups of stakeholders are shown in the online Table DS2. Major differences between stakeholders were not found. However one item, ‘It does not matter if personality disorder services do not have a clear treatment model, as long as there are positive outcomes for service users’, was rejected by expert authors and service providers, but it was endorsed by over 40% of service users.

Priorities for service development were rated by 88 (89%) panellists. Four types of service scored consistently highly: those aimed at reducing stigma and discrimination associated with personality disorder and community-based services providing psychological treatments (median score=8), and dedicated day services and consultation services providing expert guidance to colleagues working with people with personality disorder (median score=7). Two types of service, therapeutic communities and in-patient units for people with severe personality disorder, were rated lower (median score=5). All measures of service outcome were scored highly. Quality of life was ranked highest by all three groups. Expert authors and service providers placed social function as the next most important outcome, with service users opting for reduction in symptoms of mental distress. All three stakeholder groups rated user satisfaction with quality of care the least important outcome measure.

Discussion

Consensus was reached on only 39% of items, which is lower than that achieved in previous Delphi studies examining general and dedicated mental health services. ^{Reference Fiander and Burns8,Reference Heather, Dallolio, Hutchings, Kaner and White9} Although we involved panellists with a wider range of backgrounds than most previous studies, it is unlikely that this resulted in the low level of consensus as the more homogeneous subgroups of service providers and expert authors agreed on even fewer (35%) items. Instead, we think that the lack of consensus reflects the rudimentary experience of service provision for people with personality disorder and the limited evidence base that exists to support their development. The differences between groups that we did find may reflect a greater interest in the outcome rather than the process of care among service users, and a tendency to place greater importance on user involvement among service users and providers than among the expert authors that participated in the study.

Despite low level of consensus, agreement was reached on several statements, some of which apply equally well to generic mental health services, although others are more specific. For instance, general mental health services often need unilaterally to take steps to reduce risk of harm to self or others: in contrast, panellists agreed that the reduction of risk to people with personality disorder involves placing a high degree of choice and personal responsibility with the patient. Panellists stated that most people with personality disorder should have access to dedicated services. In 2002, only 17% of trusts in England provided such services. ⁴ Although dedicated services may reduce the need for in-patient and emergency medical services, ^{Reference Bateman and Fonagy10} the financial implications of providing them for all those with personality disorder would be significant.

Strengths of the study include the range of perspectives that we accessed and the high response rate. However, all panellists came from Great Britain and although several factors would seem important in other countries, others are more closely related to the way that services are currently configured in Britain. A more fundamental limitation of this, and other Delphi studies, is that findings are based solely on expert opinion. A consensus based on expert opinion can be of value, especially when other forms of evidence do not exist, but expert opinion may still be flawed. So, for instance, we do not know whether services that adhered to the principles that we identified would be any more effective than those that do not.

In England, mental health service providers have been asked to ensure that people with personality disorder have access to dedicated community-based services. Members of the Delphi panel recommend that such services should provide psychological treatments via out-patient and day-patient units, and should work to reduce stigma associated with personality disorder. They stated that interventions need to be delivered for years rather than months and that service outcomes should be assessed in terms of improved quality of life and social functioning, and reductions in mental distress. Findings of this Delphi study highlight good practice and indicate the need for further research in this area.