To compare the efficacy and safety of a select serotonin re-uptake inhibitor (SSRI, sertraline) and a tricyclic antidepressant (TCA, nortriptyline) in outpatients over the age of 60 who meet DSM-IV criteria for unipolar major depression, excluding patients who meet criteria for psychotic or atypical subtype. To test the hypothesis that medication condition interacts with diagnostic subtype (melancholic vs non-melancholic) in determining antidepressant response. To examine the roles of symptom severity and alternative diagnostic subtyping in contributing to this pattern.

SSRIs are effective in the treatment of major depression. However, there is also evidence that SSRIs may be significantly less effective than TCAs for depressed patients with melancholia. This issue is of particular concern in late-life major depression. SSRIs have important safety advantages with respect to overdose and a benign cardiovascular profile. Furthermore, the SSRIs do not have significant anticholinergic effects, and appear to be better tolerated than the TCAs. Perhaps most important, the SSRIs currently are prescribed widely as the medication treatment of first choice for major depression in late life. Therefore, if it were determined that SSRIs are considerably less effective than TCAs in the treatment of melancholia in the elderly, there would be significant ramifications for clinical practice.

Randomization to sertraline (a SSRI) or nortriptyline (a TCA) is stratified by the presence or absence of melancholia. Outcome measures for the 12-week acute phase include clinician and patient ratings of symptoms, side effects, and an evaluation of the health-related quality of life (HRQOL). At the end of the acute treatment phase, patients who meet criteria for clinical response participate in a 6-month continuation phase.