Randomised, double-blind placebo controlled trials that evaluated mood stabilisers, antidepressants or antipsychotic agents in patients where all, or the majority of participants, were diagnosed with borderline personality disorder were eligible for inclusion. Trials had to report parametric data (numeric observations with a mean and standard deviation) on anger or depressive symptoms to be included.
All trials used structured diagnostic assessment to make the diagnosis of borderline personality disorder, but specific tools used varied across trials. Most trials only included patients with borderline personality disorder. Some trials also included patients with comorbid disorders (range 0 to 100%). Specific interventions evaluated in the included trials were mood stabilisers (carbamazepine, divalproic acid, lamotrigine and topiramate), antidepressants (amitriptyline, fluoxetine, fluvoxamine, phenelzine and tranylcypromine) and antipsychotics (aripiprazole, haloperidol, olanzapine and trifluoperazine). Treatment duration ranged from five to 24 weeks, drop-out rates ranged from 5 to 65%, the proportion of patients in psychotherapy ranged from 0 to 100% and 73% of patients were women. Most trials excluded patients with alcohol and substance abuse, suicidality and self-harm behaviours. In included trials, measures of depression included the Hamilton Rating Scale for Depression, Symptom Checklist-90, Beck Depression Inventory and measures developed by the study authors; measures of anger included the Symptom Checklist-90 Hostility, Overt Aggression Scale-Modified, State-Trait Anger Expression Inventory, Anger Out, and the Profile of Mood States Anger. Most trials lasted 12 weeks or less.
Two reviewers independently assessed studies for inclusion. Disagreements were resolved through consensus.