Statistics from Altmetric.com
QUESTION: In patients with personality disorders, does psychotherapy improve symptoms, social role functioning, and core psychopathology?
Studies were identified by searching Medline and PsycINFO (1974–98).
Studies were selected if the investigators used systematic methods to diagnose personality disorder, used validated outcome measures, and reported data that allowed calculation of effect sizes or assessment of recovery from borderline personality disorder.
Data were extracted on patient characteristics, type and duration of treatment, length of follow up, severity of disorder, attrition, and outcomes. Outcomes rated by patients (Symptom Checklist-90-R, target complaints, Inventory of Interpersonal Problems, and Beck Depression Inventory) and observers (Health-Sickness Rating Scale, Global Assessment Scale, and Social Adjustment Scale) were analysed separately. Within-condition effect sizes were calculated for each study.
15 studies met the selection criteria. 3 studies were randomised controlled trials (RCTs) that compared (1) short term anxiety provoking psychotherapy and brief adaptive psychotherapy with waiting list conditions, (2) dialectical behaviour therapy with standard care for patients with borderline personality disorder, and (3) 3 types of short term behavioural therapy with a waiting list condition. Studies involved patients with borderline personality disorder (4 studies), borderline personality disorder and schizotypal personality disorder (1 study), avoidant personality disorder (1 study), antisocial personality disorder (1 study), and mixed types of personality disorder (8 studies). Treatment lasted from 10 days to 25.4 months (mean 35.5 wks), and follow up in 14 studies ranged from 1–62 months (mean 14.9 mo). In 12 studies, participants who received active psychotherapy showed improvement at follow up (mean effect size 1.1, 95% CI 0.9 to 1.3 for self report measures and 1.3, CI 0.8 to 1.8 for observer rated measures). Participants in the control groups of RCTs did not show an improvement at follow up (mean effect size –0.3, CI –0.4 to 0.9 for self report measures [3 RCTs] and 0.5, CI –2.3 to 3.3 for observer rated measures [2 RCTs]). Data from the RCTs showed that active psychotherapy was more effective than no psychotherapy or non-specific treatment (mean difference in effect size 0.8, p=0.006). This difference was not statistically significant when only the 2 RCTs that used observer ratings were pooled (mean difference in effect size 0.5, p=0.14).
Psychotherapy may be of benefit in people with personality disorders.
There is a paradox in a high quality meta-analytic review of a sparse and methodologically flawed field of studies. By my reckoning, this review by Perry et al meets 7 out of the 8 quality criteria of Oxman and Guyatt,1 making it the best yet in this field. The authors have laboured hard and used some ingenious methods to reveal underlying truth. Yet what a hotchpotch the 15 studies are! Combining results is not easy from such differing lengths, types, and settings of treatments with different case identification methods, referral criteria, and outcome measures. Crucially, to obtain even this small dataset, the authors have included reports of the effect of personality disorder on outcome of therapy for other mental health problems, such as depression and eating disorder. These are not strictly studies of psychotherapy for personality disorders, which makes it more difficult to generalise the results to referred populations.
Effect sizes for combining results are within treatment and therefore inflated by sources of bias (eg, type I error of diagnosis, regression to the mean, and improvements not from therapy). Even so, the striking finding is that these patients do improve their mental health and social functioning, whether measured by self report or observer ratings. Although methodological problems prevent us accepting effectiveness as proven, this is an important finding, given the urgent need to develop better ways of managing these disorders. The conclusions stand to be confirmed or amended by subsequent randomised controlled trials and large N field studies of outcomes. For example, an RCT of psychotherapeutic day hospital treatment too recent to be included also found evidence of effectiveness.2
Given the methodological problems, is this attempt at a meta-analytic review premature? Possibly, but it is still of value. This review serves its function in combating therapeutic pessimism, showing the wide range of approaches that are of potential value, and spelling out how to improve psychotherapy and health services research in personality disorder.
Source of funding: no external funding.
For correspondence: Dr J C Perry, Institute of Community and Family Psychiatry, Sir Mortimer B Davis-Jewish General Hospital, 4333 Chemin de la côte Ste-Catherine, Montreal, Quebec H3T 1E4, Canada. Fax +1 514 340 7507.
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.