Background Borderline personality disorder (BPD) is a relatively common personality disorder with a major impact on health services as those affected often present in crisis, often self-harming. Objectives To evaluate the effects of psychological interventions for people with borderline personality disorder. Search strategy We conducted a systematic search of 26 specialist and general bibliographic databases (December 2002) and searched relevant reference lists for further trials. Selection criteria All relevant clinical randomised controlled trials involving psychological treatments for people with BPD. The definition of psychological treatments included behavioural, cognitive-behavioural, psychodynamic and psychoanalytic. Data collection and analysis We independently selected, quality assessed and data extracted studies. For binary outcomes we calculated a standard estimation of the risk ratio (RR), its 95% confidence interval (CI), and where possible the number need to help/harm (NNT/H). For continuous outcomes, endpoint data were preferred to change data. Non-skewed data from valid scales were summated using a weighted mean difference (WMD). Main results We identified seven studies involving 262 people, and five separate comparisons. Comparing dialectical behaviour therapy (DBT) with treatment as usual studies found no difference for the outcome of still meeting SCID-II criteria for the diagnosis of BPD by six months (n=28, 1 RCT, RR 0.69 CI 0.35 to 1.38) or admission to hospital in previous three months (n=28, 1 RCT, RR 0.77 CI 0.28 to 2.14). Self harm or parasuicide may decrease at 6 to 12 months (n=63, 1 RCT, RR 0.81 CI 0.66 to 0.98, NNT 12 CI 7 to 108). One study detected statistical difference in favour of people receiving DBT compared with those allocated to treatment as usual for average scores of suicidal ideation at 6 months (n=20, MD -15.30 CI -25.46 to -5.14). There was no difference for the outcome of leaving the study early (n=155, 3 RCTs, RR 0.74 CI 0.52 to 1.04). For the outcome of interviewer-assessed alcohol free days, skewed data are reported and tend to favour DBT. When a substance abuse focused DBT was compared with comprehensive validation therapy plus 12-step substance misuse programme no clear differences were found for service outcomes (n=23, 1 RCT, RR imprisoned 1.09 CI 0.64 to 1.87) or leaving the study early (n=23, 1 RCT, RR 7.58 CI 0.44 to 132.08). When dialectical behaviour therapy-oriented treatment is compared with client centred therapy no differences were found for service outcomes (n=24, 1 RCT, RR admitted 0.33 CI 0.08 to 1.33). However, fewer people in the DBT group displayed indicators of parasuicidal behaviour (n=24, RR 0.13 CI 0.02 to 0.85, NNT 2 CI 2 to 11). There were no differences for outcomes of anxiety and depression (n=24, 1 RCT, RR anxiety BAI >/=10 0.60 CI 0.32 to 1.12; RR depression HDRS >/=10 0.43 CI 0.14 to 1.28) but people who received DBT had less general psychiatric severity than those in the control (MD BPRS at 6 months -7.41 CI -13.72 to -1.10). Finally this one relevant study reports skewed data for suicidal ideation with considerably lower scores for people allocated to DBT. When psychoanalytically oriented partial hospitalization was compared with general psychiatric care the former tended to come off best. People who received treatment in a psychoanalytic orientated day hospital were less likely to be admitted into inpatient care when measured at different time points (e.g. n=44, RR admitted to inpatient 24 hour care > 18 to 24 months 0.05 CI 0.00 to 0.77, NNT 3 CI 3 to 10) Fewer people in psychoanalytically oriented partial hospitalization needed day hospital intervention in the 18 months after discharge (n=44, 1 RCT, RR 0.04 CI 0.00 to 0.59, NNT 2 CI 2 to 8). More people in the control group took psychotropic medication by the 30 to 36 month follow-up, than those receiving psychoanalytic treatment (n=44, 1 RCT, RR 0.44 CI 0.25 to 0.80, NNT 3 CI 2 to 7). Anxiety and depression scores were generally lower in the psychoanalytically oriented partial hospitalization group (n=44, 1 RCT, RR >/= 14 on BDI 0.52 CI 0.34 to 0.80, NNT 3 CI 3 to 6), as are global severity scores. People receiving psychoanalytic care in a day hospital had better social improvement in social adjustment using the SAS-SR at 6 to 12 months compared with people in general psychiatric care (MD -0.70 CI -1.08 to -0.32). Rates of attrition were the same (n=44, 1 RCT, RR leaving the study early 1.00 CI 0.23 to 4.42). Authors' conclusions This review suggests that some of the problems frequently encountered by people with borderline personality disorder may be amenable to talking/behavioural treatments but all therapies remain experimental and the studies are too few and small to inspire full confidence in their results. These findings require replication in larger 'real-world' studies.
