Abstract
Borderline personality disorder (BPD) is a serious mental illness, with a mortality which approaches 10% [1–4]. An American study [5] suggests that over 20% of psychiatric inpatients and about 10% of psychiatric outpatients can be given this diagnosis. Although once considered ‘unanalysable’ and so beyond the therapeutic pale, cautious optimism has arisen following recent reports of treatment outcome in this condition [6–9]. However, although the morbidity associated with BPD generates heavy expenditure and is a considerable burden on health services, scepticism about treatment effects persists within health bureaucracies [10],[11] since adequately designed studies in this area are few.
An adequate design is prospective; involves control measures; has a clearly described therapeutic approach; uses outcome measures which are clearly related to the condition rather than non-specific; and includes outcome evaluation not only at the cessation of treatment but also at a later time. Two studies of BPD meet most criteria for adequate design [6–7]. However, as Lambert and Bergin authoritatively remark, controls in psychotherapy outcome studies are never entirely satisfactory [12, p. 152]. In our study [6], patients were their own controls. Events related to the BPD condition (time in hospital, episodes of violence and self-harm, drug use, unemployment, medical attendance) were collated for the year before treatment and compared with a collation of the same events for the year after treatment. For all indices, there was a significant improvement after therapy.
The design of our study is appropriate when considering treatment in a chronic condition which remains much the same and shows little change over, say, a 5-year period [13]. Recent evidence suggests that this is the case for BPD [14],[15]. Patients of the average age of our cohort (29.4 years) changed very little over 5 years in the Chestnut Lodge study [2]. Nevertheless, although the design is acceptable, it is not ideal [13]. There is value in a comparison group. We present here an additional control measure, comparing the treated cohort with a group of BPD sufferers who had ‘treatment as usual’.
Method
Subjects
The majority of patients were originally referred to Westmead Hospital by psychiatrists and trainee psychiatrists who had been working unsuccessfully with them using a variety of treatments which included, for example, drugs, behaviour modification, inpatient care and psychotherapy or combinations thereof. The remainder came from general practitioners, community and self-referral. The subjects were consecutive referrals.
These patients were, in the main, severely disabled people. Some had had multiple hospital admissions, or had been misdiagnosed as schizophrenic (e.g. case description [16]).
All patients were diagnosed according to DSM-III criteria for BPD. Forty-eight people fulfilled criteria and accepted treatment. Eight dropped out, typically in the first 3 months. Forty patients were treated. At the end of 1 year it was considered unethical, in seven cases, to cease therapy, since it seemed likely that termination would jeopardise the gains made in treatment. Three patients could not be contacted. This report concerns the remaining 30 who completed 12 months' therapy.
Controls
When this project began in the 1980s, it was considered neither ethical nor feasible to allocate some BPD patients to a wait list control. However, since the clinic is the only one of its kind in the State, a wait list inevitably grew. Those patients who had to wait a year or more did so because of individual circumstances (e.g. they applied to the clinic at a time when no therapists were available who could be allocated to them). In this way, entry into either of the treated or wait list group was a matter of chance. Nevertheless, the cohorts cannot, in the technical sense, be seen as randomly chosen.
The first 30 patients on the wait list who had been waiting 12 months or more made up the comparison group. During the waiting period, they had their usual treatments, which were various (e.g. supportive psychotherapy, crisis intervention only, cognitive therapy, pharmacotherapy). Some patients were hospitalised. There was no typical course of treatment. It seems likely, at least to us, that the somewhat lower DSM scores in this group is explained by the most unstable people being unable to tolerate the wait list. As a consequence, they moved on.
Treatment method
The treatment model [17] was consistent with, and an elaboration of the Conversational Model of Hobson [18],[19]. The Conversational Model has been manualised as ‘interpersonal-psychodynamic’ psychotherapy (IP) [20].
The DSM catalogue of BPD criteria is made up of three factors [21]. They concern affect, impulse and self. It is not known which, if any, of these constellations of experience and behaviour is primary. We have taken the view that the third factor, which includes the experience of emptiness, is the most fundamental. The model is based on the idea that borderline personality disorder is a consequence of a disruption in the development of the self. The principal assumption is that a certain kind of mental activity, found in reverie and underlying symbolic play, is necessary to the generation of the self. This kind of mental activity is non-linear, associative and affect laden. In early life, its presence depends on a relationship with caregivers in which their responsiveness is resonant with, or shows a re-knowing of re-presentation, of the child's core experience. Piaget likened this relationship to a ‘life of union’ [17, p.39].
Development is disrupted where this environmental provision is lacking or where the requisite form of relatedness is repeatedly broken up. These breaks impact upon the child rather like a loud noise, causing the child to orient outwards. This effect also arises through actual stimuli (abuse of various kinds is common in the early lives of borderline patients [22–24]), through high anxiety, and through responses that do not connect with the child's immediate reality and so seem to come from ‘outside’.
The aim of therapy is maturational. Specifically, it is to help the patient discover, elaborate, and represent a personal reality (i.e. a reality that relates to an inner life and which has the form and feeling of what William James called the ‘stream of consciousness’) [25].
The first task is to establish the enabling atmosphere in which the generative mental activity can arise. In order to do so, the therapist must imaginatively immerse himself or herself in the embryonic inner life of the patient. This, however, is no easy task since the pathogenic environment has caused the developing individual persistently to orientate towards the outer world. This orientation allows the subject little chance of establishing the experience of the stream of consciousness. The BPD sufferer is as if ‘stimulus entrapped’ [26]. The therapeutic conversation is determined by a linear form of mental activity leading to the production of ‘chronicles’ [27] concerning events of the immediate past, such as problems with family or at work, or accounts of the sensations of the body.
The work towards getting the sense of self going depends upon a growing feeling of ‘connectedness’ between patient and therapist. Attempts are constantly made to amplify those elements of the personal and inner world which appear in the conversation, particularly as feelings or implicit metaphors [18],[28],[29].
The developing conversation, however, is repeatedly interrupted by the intrusion of memories of the traumatic impacts of the environment upon his or her psychic life which occurred in the individual's childhood. These memories are recorded in memory systems which are not those underpinning the stream of consciousness. In this way, they are ‘unconscious’ [30]. As a consequence, they are sensed as occurring entirely in the present. In this state, the dyad is given the characteristics of the original traumatising situation. The patient experiences himself or herself as bad, worthless, weak and so forth, in the face of someone who is cold, critical, controlling and so forth.
These traumatic memories are triggered, often in very minor or muted form, by contextual cues, particularly the circumstances of the therapeutic conversation, for example, by a chance word, or a particular tone of the therapist's voice. The second main therapeutic aim is to identify these moments when traumatic memory breaks into ordinary consciousness, in order to work towards the integration of the traumatic memory system into the system of self. Such ‘disjunctions’ are indicated by: (i) negative affect (e.g. deadness, anxiety); (ii) linear thinking; (iii) an orientation toward events and the outer world; (iv) a change in self state (e.g. devaluing, grandiose); and (v) emergence of transference phenomena [8]. The experiences during these moments are the starting points of experiential explorations necessary to the integrating process.
Further aspects of the treatment approach are set out elsewhere [31–34].
Adherence to the model was achieved by use of audio tapes of the therapeutic encounter. The therapists presented their tapes each week to their supervisors. In order to further ensure a coherence of treatment approach, supervisors, at times, supervised together.
Treatment arrangements
All subjects received psychotherapy for 1 h twice a week over a 12-month period. Before entering the program, its purpose was explained to them and they gave written consent to the procedures, including audio taping of sessions. The patients signed a contract outlining the arrangements, committing themselves to them. The inpatient unit at Westmead Hospital provided back-up to the outpatient program. Patients could be admitted to this facility when in crisis.
Most patients, on entering the trial, were on some form of medication. These medications, in the typical case, were slowly withdrawn.
The therapists
The 20 therapists were mainly psychiatrists in training. There were also two senior psychiatric nurses and a psychologist. They were young (average age = 30.6 years), 12 were single, eight married. Eleven were male, nine female; six had postgraduate qualifications. They were enthusiastic, optimistic and formed a cohesive group. It seems likely, at least to us, that this general atmosphere contributed to the success of the program.
The therapists received weekly didactic lectures about the method. However, their most important training came through the weekly supervision by means of audiotapes of their sessions. The audiotapes were essential to the process since they allowed the therapists to become aware of the shifts taking place in the therapeutic conversation. They were helpful also in fostering the therapist's imaginative and empathic understanding of their patients. Finally, as previously remarked, the tapes ensured adherence to the therapeutic model.
Outcome measures
On entry into the program, demographic data were collected on all patients. The number of DSM-III criteria for borderline personality disorder fulfilled by the subject was determined. In addition, a 27-point scale made up of all the items in each criterion category, was constructed. Each patient was scored according to this scale in a clinical interview conducted by two people, a psychiatrist and research psychologist. The interview was of a semistructured kind, designed to elicit information necessary to give a ‘yes/no’ response to each of the 27 items. Such clinical interview ratings of DSM-III criteria, and DSM-III derived ratings have been shown to be reliable [21],[35].
Methods
The psychotherapeutically treated group was similar to the wait list group in demographic terms: in the treatment group 25 were single and five married/de facto; in the wait list group 21 were single and nine married/de facto. Eight of the treatment were employed; five of the comparison group had a job. Twenty-one of the treatment group and 19 of the comparison group had School Certificate/HSC. Six in each group had received higher education. Three of the treatment group and five of the wait list patients left school before Year 10. The mean age of the treatment group was 29.4 years (SD = 7.9); of the comparison group 32.9 years (SD = 7.8).
The wait list group was assessed 12 months after the initial assessment in 23 cases; 18 months after in two cases; 31 months in two; 43 months in one; 48 months after in two. The average length of follow-up was 17.1 months.
Of the 30 treated patients, 30% no longer met criteria for the diagnosis of BPD after a year's psychotherapy. The 30 patients on the wait list for 1 year or more showed no change in diagnosis.
Table 1 shows mean DSM scores at time 0 and at follow-up for treatment and control group.
Mean and SD scores at initial assessment and follow-up for treatment and wait-list control group
Multiple regression analysis using SPSS 6.1 (SPSS Inc., Chicago, IL, USA) was used to perform an analysis of covariance. Group membership and DSM score at the first assessment time (time 0) were entered as independent variables. The treatment group was coded as 1 and the control group was coded as 2. The dependent variable was change in DSM scores between time 0 and at follow-up (second assessment time).
The results are shown in Table 2. The change in DSM score depends on both the initial DSM score (p = 0.03) and on treatment (p = 0.0007). Individuals with higher DSM scores at time 0, irrespective of group membership have a greater reduction in DSM scores at follow-up. After adjusting for DSM at time 0, the DSM scores of individuals in the treatment group decreased by an average of 4.78 more than subjects in the control group, over the 12-month period.
Multiple regression analysis of change in DSM score, with group and DSM score at initial assessment as covariates
Discussion
This study suggests that psychotherapy based on psychodynamic principles provides benefit for patients with borderline personality disorder. After 1 year's treatment, the experimental group showed a significant reduction in DSM scores. The wait list group showed no significant changes. The improvements in the group treated psychotherapeutically have been maintained both at the 1 year [6] and the 5-year follow-up [36].
Our findings can be compared with the only other study of the effect of treatment on borderline patients which was controlled, prospective and involved a follow-up beyond the cessation of treatment. Linehan et al. [37] found that 6 months after therapy the improvement was largely maintained. However, at 12 months the improvement had begun to dissipate. It was sustained for Global Assessment Scale scores, but there were no longer significant differences between the experimental group and the treatment-as-usual group in terms of parasuicidal episodes, anger, and self-reported social adjustment.
Are there possible alternative explanations for our findings in a group of people who characteristically ‘do not fare well at follow-up’ [38]? One such explanation might be that the condition was not severe in these 30 patients. This was not the case. Fifty-seven percent of the cohort showed all eight DSM-III criteria compared with 7% in Stone's large series [1].
A second possible explanation concerns those seven patients whose treatment had to be continued beyond 12 months because it was considered improper to discontinue therapy for the sake of the trial. Would their outcome status have diminished or cancelled the positive results achieved by the remaining 30? Again this was not the case. Their outcome was good. Two of them have been the subject of individual reports [16],[39].
A third possible explanation might be that, since entry into either the wait list or the treatment groups was not randomised, the less ill patients may have been taken into therapy so enhancing the chances of improvement. However, the opposite was the case. Those on the wait list showed somewhat lower levels of disorder than the treatment group. In demographic terms, the two groups were similar.
This study, which is prospective, involving control measures, and concerned with outpatient work, is not directly comparable with the two major investigations in the field of borderline personality disorder which come from McGlashan [2] and Stone [1]. Both studies are large and necessarily based on retrospective diagnoses. In both cases, a large majority of the cohort was located for follow-up (Stone had located 95% of his subjects). These two studies concern the long-term outcome of patients who received prolonged inpatient psychotherapy. Since the modest improvements shown by these patients were temporally distant to the point of termination of treatment, it has sometimes been inferred that these outcome data reflect the natural history of the condition. However, a recent study from Fairburn et al. [40], admittedly of a different disorder, suggests that psychotherapy, relative to cognitive behaviour therapy, may have temporally distal effects. The average patient at Chestnut Lodge had 2 years' treatment, while those at the New York State Psychiatric Institute received 12.5 months of inpatient care. It is hard to discount the effect of such prolonged therapy.
Our own study and that of Linehan et al. [7],[37] suggest that outpatient treatment, of a specific kind, is at least as effective as these more extensive treatments. In both these studies, the drop-out rate was low, 16%, suggesting a greater tolerance by patients of the treatment methods used in these programs compared with earlier methods [41],[42]. The possibility arises that Linehan's multimodal therapeutic approach, which is directed towards problems of affect disregulation and impulsivity, is complementary to our own, which focuses on self.
The possibility that treatment based on principles related to self theory may be particularly helpful in severe personality disorder is compounded by the recent findings of Monsen et al. [43]. Their study, although not controlled and not specifically focused on BPD, was prospective. They treated 25 outpatients with personality disorders and psychosis for a mean period of 25.4 months with a mean follow-up of 5.2 years using a therapeutic approach focusing on ‘affect consciousness’ [8]. A significant and substantial change was found in symptoms. There was a 72% reduction in axis II diagnoses. There was also significant improvement in psychosocial function [9].
These studies, together with our own data, suggest that psychotherapy of a specific kind can lead to sustained improvement in severe personality disorder. The precise characteristics of the beneficial aspects of this therapy will need to be determined by detailed investigations of the therapeutic encounter.
Footnotes
Acknowledgements
The authors wish to thank Dr Karen Byth, Consultant Statistician to the Western Institute of Health Research, Westmead Hospital, for statistical guidance.