Background: To determine the predictors of desipramine-refractory depr
ession, the authors examined the outcome in patients with major depres
sion who were admitted to a general hospital and treated with desipram
ine adjusted to an adequate blood level. Method: Sixty-eight consecuti
ve inpatients with DSM-III nonpsychotic unipolar major depression who
had failed to respond to 1 week of hospitalization without drug treatm
ent were studied. Outcome was assessed with the Yale Depression Invent
ory after a 4-week desipramine trial in which 24-hour plasma concentra
tions were used to rapidly achieve a therapeutic desipramine level. Re
sults: Poor response to a therapeutic desipramine trial, which occurre
d in 15 of 50 patients, was significantly associated with definite per
sonality disorder, prior treatment failure, near delusional status, ag
e less than or equal to 35 years, duration of depressive episode, recu
rrence of depression, dysthymia, and secondary depression. The first f
our items remained significantly correlated with poor response when th
e presence of the other items was accounted for using multiple regress
ion. Drug response was not predicted by the diagnosis of melancholia (
DSM-III and DSM-III-R) or initial severity of the depressive episode.
Conclusion: The four strongest correlates of outcome were highly predi
ctive of drug response. In patients with two or more predictors, only
25% (4 of 16) responded, while in those with or no predictors, 91% (31
of 34) responded.