Bogenschutz 2004.

Study characteristics
Methods	12‐week trial with 2 arms: olanzapine (2.5‐20 mg/d), most 5‐10 mg/d, mean at endpoint 6.9 mg/d (SD 3.2) placebo Duration of trial: 12 weeks (patients had to be free of mood stabilisers, antipsychotics, benzodiazepines, and antidepressants for at least 2 weeks prior to participation) Country: no information Setting: outpatient
Participants	Methods of recruitment of patients: Patients were recruited from the community and from outpatient clinics at a university psychiatric hospital. Overall sample size: 40 Diagnosis of borderline personality disorder: DSM‐IV Means of assessment: SCID‐II Mean age: 32.6 years (SD 10.3; range = 18‐54) Sex: 62.50% women Comorbidity: "Patients met criteria for a mean of 2.9 SCID‐II personality disorders (including BPD) and a mean of 2.2 Axis I diagnoses from the MINI" (Bogenschutz 2004, p. 106). Inclusion criteria: no information Exclusion criteria Schizophrenia Schizoaffective disorder Bipolar affective disorder Current major depressive episode Psychotic disorder due to substance or general medical condition Substance dependence not in full or partial remission Suicide attempts in past 6 months Having current suicidal intent or definite plan Pregnancy Neurologic impairment
Interventions	Experimental group Treatment name: olanzapine Number randomised to group: 20 Duration: no information Control/comparison group Comparison name: placebo Number randomised to group: 20 Duration: no information Both groups Concomitant psychotherapy: "Patients were allowed to continue ongoing psychotherapy (if initiated more than 3 months prior to randomisation)" (Bogenschutz 2004, p. 105). Concomitant pharmacotherapy: medication for stable, chronic medical conditions such as hypertension Proportions of participants taking standing medication during trial observation period: no information
Outcomes	Primary outcomes Suicidal ideation, measured by CGI‐BPD‐recurrent suicidal ideation. Assessed at baseline and at 2, 4, 8 and 12 weeks (EOT) Mental health status (functioning), measured by CGI. Assessed at baseline and at 2, 4, 8 and 12 weeks (EOT) Secondary outcomes Anger, measured by CGI‐BPD (inappropriate anger) and AIAQ. Assessed at baseline and at 2, 4, 8 and 12 weeks (EOT) Affective instability measured by CGI‐BPD (affective instability). Assessed at baseline and at 2, 4, 8 and 12 weeks (EOT) Feelings of emptiness, measured by CGI‐BPD (chronic feelings of emptiness). Assessed at baseline and at 2, 4, 8 and 12 weeks (EOT) Impulsivity, measured by CGI‐BPD (impulsivity) and OAS‐M (aggression). Assessed at baseline and at 2, 4, 8 and 12 weeks (EOT) Interpersonal problems, measured by CGI‐BPD (unstable interpersonal relationship). Assessed at baseline and at 2, 4, 8 and 12 weeks (EOT) Avoidance of abandonment, measured by CGI‐BPD (abandonment). Assessed at baseline and at 2, 4, 8 and 12 weeks (EOT) Identity disturbance, measured by CGI‐BPD (identity disturbance). Assessed at baseline and at 2, 4, 8 and 12 weeks (EOT) Dissociative symptoms, measured by CGI‐BPD (transient paranoia or dissociation). Assessed at baseline and at 2, 4, 8 and 12 weeks (EOT) Attrition Adverse effects, measured by weight. Recorded at each visit
Notes	Sample calculation: no information Ethics approval: no information Funding source: funded or partially funded by pharmaceutical industry Conflicts of interest: No other conflicts of interest were reported besides funding from the pharmaceutical industry. Comments from trial authors (limitations) Based on the relatively small sample size and the large variance for the scales (OAS‐M, ASI, AIMS, Simson‐Angus Scale), the trial may not have had adequate power to consistently detect differences on these scales. There was a discrepancy between the SCL‐90 scores in this trial and a previous trial. There was no clear explanation for this; however, it could be due to demographic factors. This trial sample included men, was more ethnically diverse and possibly lower in socioeconomic status and education than the sample in the referred previous trial (Zanarini 2001).
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Comment: The article referred to the trial as being randomised, however inadequate information on random sequence generation was provided to permit a judgement of low or high risk of bias.
Allocation concealment (selection bias)	Unclear risk	Comment: no information on allocation concealment to permit a judgement of low or high risk of bias
Blinding of participants and personnel (performance bias) All outcomes	Low risk	Quote: "double‐blind trial", "pseudo‐dose [...] for patients receiving placebo" (Bogenschutz 2004, p. 105).
Blinding of outcome assessment (detection bias) All outcomes	Unclear risk	Comment: The article referred to the trial being double‐blind, however, there was insufficient information on how blinding of outcome assessors was carried out and maintained to permit a judgement of low or high risk of bias.
Incomplete outcome data (attrition bias) All outcomes	Unclear risk	Comment: "evaluable patients" refers to all patients remaining at least 2 weeks in the study, i.e. who attended both baseline and preliminary assessment after two weeks; reasons for early termination specified (Bogenschutz 2004, p. 106). However, 2 patients dropped out of the olanzapine group due to "violation of protocol" (Bogenschutz 2004, p. 106). Of the 40 patients enrolled, 23 completed the full 12 weeks of the trial (10 in olanzapine group, 13 in placebo group). Reasons for early termination: Lost to follow‐up: 2 in the olanzapine group, 5 in the placebo group Lack of efficacy: 2/2 Weight gain: 2/0 Sedation: 2/0 Violation of protocol: 2/0 Continuous data based on LOCF of patients remaining at least 2 weeks in the trial Dichotomous data based on ITT sample
Selective reporting (reporting bias)	Unclear risk	Comment: No protocol available
Vested Interest (funding and/or author affiliations)	High risk	Quote: "Supported by a grant from Eli Lilly and Co, Indianalpolis, Ind." (Bogenschutz 2004, p. 104)
Other bias	Low risk	Comment: No other apparent sources of bias were found.