Markovitz 1995a.

Study characteristics
Methods	A 14‐week trial with 2 arms: fluoxetine placebo Duration of trial: 14 weeks Country: USA Setting: inpatient
Participants	Methods of recruitment of patients: no information Overall sample size: 17 Diagnosis of borderline personality disorder: DSM‐III‐R Means of assessment: SCID‐II and Gunderson's DIB20 Mean age: no information Sex: no information Comorbidity: Axis I: Each patient had on average 3.0 current Axis I diagnoses and 4.7 lifetime diagnoses at the time of the study. Axis II: 100% borderline, 82% self‐defeating, 82% paranoid, 71% compulsive, 65% avoidant, 65% dependent, 59% histrionic, 59% passive‐aggressive, 53% schizotypal, 35% narcissistic, 35% antisocial. Axis III: 92% premenstrual syndrome, 47% headaches/migraines, 41% IBS, 35% fibrocytis, 29% neurodermatitis, 29% sleep apnoea Inclusion criteria: no information Exclusion criteria: no information
Interventions	Experimental group Treatment name: fluoxetine Number randomised to group: 9 Duration: 14 weeks Control/comparison group Comparison name: placebo Number randomised to group: 8 Duration: 14 weeks Both groups Concomitant psychotherapy: no information Concomitant pharmacotherapy: no information Proportions of participants taking standing medication during trial observation period: no information
Outcomes	Primary Psychosocial functioning, assessed by the Global Assesment Scale at baseline and every other week until week 14 (EOT) Secondary Depression, assessed by Becks Depression Inventory and Hamilton’s Depression Scale at baseline and every other week until week 14 (EOT) Attrition in terms of patients lost after randomisation in each group at EOT
Notes	Sample calculation: no information Ethics approval: no information Funding source: no information Conflicts of interest: no information Comments from trial authors (limitations): "All of the rating instruments were continuing to show increasing improvement at 14 weeks in patients on fluoxetine, suggesting the trial may not have been long enough". (Markovitz 1995a, p. 271) Comment from review authors: This trial was not included in quantitative analyses due to data unavailability for effect size calculations.
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Comment: The study was referred to as randomised; however, the method used to generate the allocation sequence was not described in sufficient detail to allow an assessment of whether it produced comparable groups.
Allocation concealment (selection bias)	Unclear risk	Comment: Allocation concealment and the method used to conceal the allocation sequence was not described.
Blinding of participants and personnel (performance bias) All outcomes	Unclear risk	Comment: The study was referred to as being double‐blind but there was no information on method or if the blinding was successful.
Blinding of outcome assessment (detection bias) All outcomes	Unclear risk	Comment: The study was referred to as being double‐blind but there was no information on method or if the blinding was successful.
Incomplete outcome data (attrition bias) All outcomes	High risk	Quote: “Dropout after 3 weeks but prior to 14 weeks and their final rating scores carried through for each subsequent time point. Seven of nine patients on fluoxetine completed the entire study, and seven of eight patients on placebo completed all 14 weeks of the trial”. (Markovitz 1995a, p. 271) Comment: last‐observation‐carried‐forward
Selective reporting (reporting bias)	Unclear risk	Comment: no protocol available. Insufficient information to permit judgement of high or low risk of bias
Vested Interest (funding and/or author affiliations)	Unclear risk	Comment: insufficient information on funding and conflict of interest to permit judgement of high or low risk of bias
Other bias	Low risk	Comment: The study appeared to be free of other sources of bias.