Objective: To compare the efficacy of a neuroleptic (haloperidol) to a
monoamine oxidase inhibitor antidepressant (phenelzine sulfate) again
st the affective, cognitive, and impulsive-aggressive symptoms of crit
eria-defined borderline inpatients in an effort to dissect apart affec
tive and schizotypal symptom patterns or subtypes using medication res
ponse. Design: Randomized, double-blind, placebo-controlled trial. Set
ting: Inpatient unit of a tertiary care university psychiatric hospita
l serving a large public catchment area. Patients: One hundred eight c
onsecutively admitted borderline inpatients defined by Gunderson's Dia
gnostic Interview for Borderline Patients and DSM-III-R criteria, rand
omly assigned to 38 phenelzine, 36 haloperidol, and 34 placebo trials.
Interventions: Following 1 week free of medication, haloperidol (aver
age dose, 4 mg/d), phenelzine sulfate (average dose, 60 mg/d), or plac
ebo were given for 5 weeks with weekly symptom ratings and plasma drug
level determinations. Main Outcome Measures: Efficacy was measured on
depression (Hamilton Rating Scale, Beck Depression Inventory), global
severity (Global Assessment Scale, Symptom Checklist-90 items [SCL-90
]), anxiety, anger-hostility (SCL-90, Inpatient Multidimensional Psych
iatric Scale [IMPS], Buss-Durkee Hostility Inventory), psychoticism (S
chizotypal Symptom Inventory, SCL-90, IMPS), impulsivity (Ward Scale,
Barratt Impulsiveness Scale, Self-Report Test of Impulse Control), and
borderline psychotherapy (Borderline Syndrome Index). Results: Three-
way comparisons between groups indicated superior efficacy for phenelz
ine, followed by placebo and haloperidol on measures of depression, bo
rderline psychopathologic symptoms, and anxiety. Pairwise comparisons
between medication and placebo revealed significant efficacy for phene
lzine against anger and hostility but no efficacy against atypical dep
ression or hysteroid dysphoria. We were unable to replicate prior repo
rts of efficacy for the neuroleptic. Conclusions: Pharmacologic dissec
tion of borderline personality disorder patients into affective and sc
hizotypal subtypes could not be demonstrated.