Pharmacological interventions for people with borderline personality disorder

Borderline Personality Disorder (BPD) is prevalent (2% in the general population, 20% among psychiatry in-patients) and has a major impact on health facilities as those affected often present in crisis but then make poor use of further attempts to help them. To evaluate the effects of pharmacological interventions for people with borderline personality disorder. We conducted a systematic search of 26 specialist and general bibliographic databases (October 2002) and searched relevant reference lists for further trials. We included all randomised clinical trials comparing any psychoactive drugs with any other treatment for people with borderline personality disorder. 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 synthesised using a weighted mean difference (WMD). We found ten small (total n=554), short, randomised studies involving eight comparisons from which we could extract usable data. Studies comparing antidepressants with placebo were small (total n=79, 2 RCTs) but for ratings of anger fluoxetine may offer some improvement for those on antidepressant therapy over placebo (n=22, 1 RCT, RR anger not improved 0.30 CI 0.10 to 0.85, NNT 2 CI 2 to 9). The one small study investigating the important outcome of attempted suicide found no difference between mianserin and placebo (n=38, 1 RCT, RR 0.82 CI 0.44 to 1.54). Haloperidol may be better than antidepressants for symptoms of hostility and psychotism. There were few differences between MAOIs and placebo except that people given MAOIs were less hostile (n=62, 1 RCT, MD -9.19 CI -16.12 to -2.26). Although some ratings were statistically significant the comparison of MAOIs with antipsychotics did not show convincing differences. Antipsychotics may effect some mental state symptoms more effectively than placebo but results are difficult to interpret clinically and there is little evidence of advantage of one antipsychotic over another. Finally mood stabilisers such as divalporex may help mental state (n=16, 1 RCT, RR no improvement in mental state 0.58 CI 0.36 to 0.94, NNT 3 CI 2 to 17) but data are far from conclusive. Pharmacological treatment of people with BPD is not based on good evidence from trials and i