Treatment models for those with severe mental illness and comorbid personality disorder

Background Dual diagnosis of personality disorder and severemental illness is an important clinical association that hasbeen under-researched with regard to clinical management.

Aims To compare the outcomes of different treatment models.

Method The outcome of patients with this combined diagnosiswas compared in a systematic review of three randomised controlledtrials in which different forms of community outreach treatmentor intensive case management were compared with standard care.

Results The results from the three studies showed that the outcome of comorbid diagnoses was worse than that of single diagnoses.Although assertive approaches reduced in-patient care, theysometimes did so at the expense of increasing social dysfunctionand behavioural disturbance.

Conclusions For those with comorbid severe mental illness and personality disorder, the policy of assertive outreach and carein community settings may be inappropriate for both publicand patients unless modified to take account of the specialneeds of this group.

INTRODUCTION

TOP

INTRODUCTION

The comorbidity of severe mental illnesses, such as schizophreniaand bipolar disorder, with personality disorder is common andis one of the most frequent dual diagnoses found in clinicalpractice. Between 30 and 60% of those with psychotic disordershave a personality disorder also (Casey, 2000) and the proportiontends to be higher in in-patient populations (Cutting et al,1986). The outcome of those with comorbid personality disorderis generally less good than the outcome of those with singlemental state disorders, with less improvement in symptoms,poorer quality of life and greater dissatisfaction with treatmentafter 2 years (Tyrer & Seivewright, 2000). Despite theseimportant clinical correlates, these comorbid conditions arenot often recognised in ordinary practice, mainly because theidentification of personality disorder is often difficult inpatients who have widespread abnormalities. In the case of schizophrenia, these may overflow into the personality domain and lead to poorreliability of assessment (Tyrer et al, 1983).

There is now a general belief that most patients with severemental illness can be treated largely in the community, withonly brief periods of admission. However, for those with grosspersonality disorder who are treated at special hospitals avery much longer period of treatment is common. We felt thatit might be valuable to examine the outcome of those with comorbidsevere mental illness and personality disorder, to determinewhether there were important differences between the effectsof different service intervention policies.

METHOD

TOP

METHOD

The relevant data from three studies of different models ofcare, carried out in west London (Paddington and Brent), aresummarised in Table 1. In each of the studies (all randomisedcontrolled trials) a community-focused service was compared with a more hospital-focused or standard service, which hadfewer community resources. The time span of the three studieswas 8 years and during this period the standard service steadilyimproved in its community orientation, so there was an expectationthat differences would become fewer over time. The overallfindings supported this view, with the early study showing the strongest differences in favour of the community service (Merson et al, 1992).

View this table:
[in this window]
[in a new window]

Table 1 Three randomised controlled studies of models of community management in severe mental illness

In this current set of investigations, however, all effectswere examined between those with and without personality disorder;and the influence of personality status on response to eachservice intervention (i.e. the interaction between personalityand service type) was recorded. This was part of a Cochranesystematic review first established in 1997 (Tyrer et al, 1999). The only specific hypothesis tested was that those with comorbidpersonality disorder and severe mental illness would have abetter response to focused community treatment, as they aregenerally considered to be ill-placed in hospital.

Patients, assessments and procedures
All patients in two of the studies had a psychotic illness withfrequent hospital admissions, and in the third (Merson etal, 1992) the patients were emergency presentations to thepsychiatric services; 70% of these had schizophrenia or affectivedisorders. Assessments of clinical symptoms in all studieswas with the Comprehensive Psychopathological Rating Scale(CPRS) (Åsberg et al, 1978). Social function was recordedwith the Social Functioning Questionnaire (SFQ) (Tyrer, 1990)in two of the three studies reported here.

The procedures for randomisation and assessment intervals differed.In one study, randomisation took place at the point of presentationas an emergency (Merson et al, 1992), with assessments atbaseline and after 2, 4 and 12 weeks. In the second study, randomisation took place at the time when in-patients were assessedas fit for discharge (Tyrer et al, 1998). In the UK700 study,randomisation took place at the point of discharge and a follow-up(Creed et al, 1999).

In all three studies personality was assessed using the Personality Assessment Schedule (PAS) (Tyrer & Alexander, 1979) althoughin the UK700 study a shorter version, the Rapid PersonalityAssessment Schedule (PAS—R) (Van Horn et al, 2000) wasused. In all analyses a simple distinction was made betweenpersonality disorder and no personality disorder. The thresholdfor the diagnosis of personality disorder using the PAS isa little higher than that for ICD-10 personality disorders (Tyrer et al, 1994), which equates to the level of personalitydifficulty in a dimensional scale (Tyrer & Johnson, 1996).

RESULTS

TOP

RESULTS

The findings from the first two studies are summarised in Tables 2 and 3. In the first study, the admission of all patientswas reduced markedly in the community service, but this wasachieved at the expense of a poorer outcome in those with personality disorder. This was most marked for social function and depressivesymptoms. In the second study, there were no important differencesin symptomatic outcome, but those with personality disorderwere not kept out of hospital to a significantly greater extent.However, in this study there was greater improvement in thosewith personality disorder treated in the community-orientedservice. Meta-analysis of the proportion of patients making a significant improvement (to a CPRS score of 10 or less) showed54% of personality disorders improving in the community groupcompared with 19% in the hospital group (P < 0.002) (Table 4).

View this table:
[in this window]
[in a new window]

Table 2 Comparison of outcomes of comorbid and single diagnoses in community- and hospital-based teams (Merson et al, 1992): trial of patients presenting as emergencies (n=100)

View this table:
[in this window]
[in a new window]

Table 3 Summary of results of trial of community-orientated and hospital-orientated care programmes separated by comorbid personality disorder with assessments at baseline and after 1 year (Tyrer et al, 1998; Gandhi et al, 2001) (n=138)

View this table:
[in this window]
[in a new window]

Table 4 Meta-analysis of studies 1 and 2 showing greater improvement in those with personality disorder in community-oriented service, measured by proportion of patients whose symptoms had been significantly relieved

In the second study there was a major shortage of beds in theBrent area, and for much of the period of the study a significantproportion of admissions were extra-contractual referrals todifferent hospitals (Tyrer et al, 1998). This probably accountedfor the greater length of admission of Brent patients; thosewith personality disorder, in particular, had very long periodsof in-patient treatment in the year after recruitment to thestudy. The interaction between personality status and siteof service was significant.

In a separate part of the study, the number of contacts withpolice were recorded in the year of the study. Of 26 incidentsinvolving 16 patients with the police, most were found in thosewith personality disorder within the flamboyant (cluster B)grouping. These were significantly more common in patientsallocated to community-oriented management (Gandhi et al, 2001). All but two of these incidents were in the Paddington componentof the project, significantly more than one would expect bychance ( ${chi}$ ²=4.7 (after Yates' correction), d.f. 1, P=0.03) and this is unlikely to be explained by demographic differencesalone. It seems likely that the long period of in-patient treatmentof those with the comorbid diagnosis in the Brent area reducedproblems in the community and could be perceived as givingsome protection to the public.

In the third study, there was an imbalance between the allocationof patients with personality disorder to intensive or standardcase management, so that approximately only 1 in 4 of thosewith personality disorder were allocated to intensive casemanagement. The small numbers made the interpretation of datadifficult and there were no clear differences in any of themain outcome variables, with the exception of duration of in-patientcare, the main outcome measure of the study. Those with comorbidpersonality disorder and psychosis had a shorter duration ofin-patient treatment in the 2 years of the study (Fig. 1).

View larger version (11K):
[in this window]
[in a new window]
[as a PowerPoint slide]
Fig. 1 Mean days in hospital over 2 years for 145 patients (114 with no personality disorder ( ${square}$ ) and 31 with personality disorder ( ${blacksquare}$ ). Shorter duration of hospital care for those with personality disorder treated by intensive case management (ICM) (n=8) compared with those treated with standard case management (SCM) (n=23, P=0.06 (for interaction)). Of those with no personality disorder, 59 and 55 were treated with ICM and SCM, respectively. (Data from Tyrer et al, 2000.)

DISCUSSION

TOP

DISCUSSION

The evidence base for interventions in comorbid personalitydisorder and severe mental illness is limited, and in a recentCochrane review of the subject the only randomised controlledtrials were those reported here. The three studies demonstratefour features:

Aggressive community care may be successful in keeping thesepatients with dual diagnoses out of hospital.
The price ofthis success is often more impaired social functionin those who remain in community care.
There may be greater risks tothe public if community treatmentfor this group is pursued.
When a community-oriented approach is pursued with less emphasisplaced on keeping patients out of hospital, there are betteroutcomes, both for patients and in protecting the public.

The findings go some way in supporting the notion that personality disturbance is more important than mental illness state in determining disturbed and antisocial behaviour, and perhaps should be assessedmore commonly in ordinary practice. Whereas violence in severemental illness is the same for those treated by intensive andby standard case management (Walsh et al, 2001) the additionalmeasurement of personality status adds an extra dimension. In the UK700 study, violent episodes were found to be more frequentin those with personality disorder (P. Moran, personal communication,2002). What is abundantly clear is that treatment policiesof those with comorbid personality disorder and severe mentalillness should not be assumed to be the same as for those withsevere mental illness alone, and that further work is neededon specific interventions for this group.

Clinical Implications and Limitations

TOP

Clinical Implications and...

CLINICAL IMPLICATIONS

Patients with comorbid personality disorder and severe mentalillness have worse outcomes if an aggressive community treatmentpolicy is adopted that reduces admission to hospital to a minimum.
The assessment of personality status in those with severementalillness would help clinical management.
The nuisanceand distress created for the public by patientswith comorbiddiagnoses needs greater attention to reduce thestigma of mental illness.

LIMITATIONS

The studies described, although using randomised controlleddesigns, were not hypothesis-driven and most analyses werepost hoc.
The studies were carried out in London at a timewhen bed pressureswere very great, and this may have distortedgeneralisationof the findings.
No acknowledgement was madeof the implications of comorbidpersonality disorder in treatmentpolicies, and the resultsmight have been different if treatmenthad been adjusted accordingly.

ACKNOWLEDGMENTS

TOP

ACKNOWLEDGMENTS

These studies were supported by the North Thames Regional Research Committee and Central Research & Development at the Departmentof Health. We thank Dr Phil Harrison-Read and Dr Tony Johnsonfor assistance and advice.

REFERENCES

TOP