The primary outcome of this study is to determine if predictors of response can select a population of patients with MDD that is effectively treatable by augmentation with ziprasidone.

Major depressive disorder (MDD) is a broad category, including many forms of depressive illness, including those with only a single major depressive episode, those with episodic recurrence with intervening well states, those with chronic depressive/anxious states without intervening euthymia, and those with manic symptoms that do not meet threshold definitions of full mania/hypomania.

In this heterogenous, large diagnostic definition, important groups of patients do not appear to respond well to antidepressants, and, conversely, based on observational studies, may respond well to neuroleptics. These predictors of response have begun to be identified and may serve to better design studies of neuroleptics in depressive illnesses.

Among these predictors of response in MDD are clinical features that are more similar to bipolar illness than unipolar depression. These include a family history of bipolar disorder, antidepressant-induced mania, highly recurrent depressive episodes (>5), atypical depression, early age of onset of depression (< age 20), failure to respond to antidepressants, and antidepressant tolerance (initial response followed by later loss of response).

The investigators propose to use these predictors to pick out patients that are more likely to respond to Geodon for MDD. This will be the first RCT of these predictors of depressive response applied to neuroleptics.

This will be a three-site, block randomized (1:1 ratio) double-blind, placebo-controlled prospective cross-over study with 50 subjects. Patients will be randomized to receiving ziprasidone-washout-placebo or placebo- washout-ziprasidone for 13-weeks.

Primary and Secondary and safety outcomes: The primary outcome measure will be change from baseline Montgomery-Asberg Depression Rating Scale (MADRS) score to end of treatment. Safety outcomes will be determined by spontaneously reported adverse events on the case report form.

Inclusion Criteria:

1. Age 18-70 years.
2. If female, nonpregnant/nonlactating
3. If a sexually active female of reproductive potential, must be using adequate contraception (i.e., oral contraceptives, barrier protection, or prior tubal ligation)
4. Currently meets DSM-IV criteria for a major depressive episode, non-psychotic.
5. Having at least 3 of the following criteria listed for predictors of depressive response to neuroleptics: a family history of bipolar disorder, antidepressant-induced mania, highly recurrent depressive episodes (>5), atypical depression, early age of onset of depression (< age 20), failure to respond to antidepressants, and antidepressant tolerance (initial response followed by later loss of response). Inadequate response to antidepressants is identified as follows: having a score of ≥14 on the 17-item HAMD or a CGI-S score of ≥ 3 after a retrospective confirmation of an adequate trial of a single antidepressant (defined as a ≥ 6-week trial of acceptable therapeutic dose [≥ 40 mg of fluoxetine, paroxetine or citalopram, 20 mg of escitalopram, 60 mg of duloxetine, 37.5 mg of paroxetine CR, 150 mg of sertraline, 100 mg of fluvoxamine, 225 mg of venlafaxine XR, 30 mg of mirtazapine, 300 mg of bupropion, 75 mg of nortriptyline, 20 mg of protriptyline, 100 mg of amitriptyline or imipramine)

Exclusion Criteria:

1. Bipolar depression
2. Sensitivity to or failure to respond to ziprasidone by history or ziprasidone use in previous 3 months
3. Active substance abuse or dependence in the previous 3 month
4. Psychotic disorders
5. Serious suicidality as evidenced by score of 3 or greater on suicide item of MADRS
6. Medically unstable as judged by study investigators
7. Lack of capacity to provide informed, written, consent to investigators
8. Previous diagnosed cardiac arrhythmias](streamdown:incomplete-link)

13 weeks

Adverse events are reported as ziprasidone arm events over placebo given the crossover nature of the study.

All crossover studies have the limitation of order effects: the order in which treatments were given could impact the results. The definition of bipolarity predictors also may not have effectively identified groups correctly.