Improvement in the methodology of longitudinal investigations and increasing research interest in depressive disorders led to findings of clinical and heuristic importance. Outcomes, such as chronicity of depression, relapse, recurrence, and development of dementia, appear to be predicted by different clinical and laboratory findings. Chronicity of depression may be predicted by long duration of the current or previous episodes, coexisting medical illness, high severity of depression, nonmelancholic presentation, delusions, and perhaps cognitive impairment and neuroradiologic abnormalities. Predictors of relapse and recurrence of geriatric depression include multiple previous depressive episodes, high severity of illness, "double depression," presence of "exit" events, and intercurrent medical illnesses. Development of dementia may be predicted by a transient dementia syndrome during a depressive episode ("pseudodementia"), onset of the first depressive episode in the senium, and neuroradiologic abnormalities such as cortical atrophy and rapidly evolving ventricular enlargement. Long-term antidepressant treatment, if not controlled by a research protocol, usually is of low intensity and has a questionable effect on the outcome of depression over a long period of time. For this reason, naturalistic treatment studies are useful for identifying subgroups of depressives and time periods of high risk for specific adverse outcomes. This knowledge is particularly important in frail elderly populations who are vulnerable to side effects of antidepressant treatments. The next step is to conduct controlled-treatment studies and examine the capability of antidepressant treatments to prevent adverse outcomes in the high-risk populations identified through naturalistic treatment studies. Controlled-treatment studies can provide findings that clinicians can use to assess the risk-benefit ratio of continuation and maintenance treatments of geriatric depression. The heuristic importance of knowing the outcome of geriatric depression is that it permits identification of clinically and, to some extent, biologically-homogeneous groups. Given the absence of specific and sensitive laboratory tests, outcome is perhaps the "next best thing" to brain autopsy for subclassifying geriatric depression. Biologic measures of structural and functional abnormalities can then be used in homogeneous subgroups for the pursuit of pathophysiologic or etiologic studies.
