studies have been mixed, with most studies reporting lower prevalence of depression in the elderly (Regier et al., 1988), and some studies showing higher prevalence (Blazer, Burchett, Service, and George, 1991). Similar trends in low prevalence of mood disorders among ethnic elders also have been reported (Weissman, Bruce, Leaf, Florio, and Holzer, 1991), with no significant overall differences between groups in the Epidemiological Catchment Area (ECA) study (George, Blazer, Winfield-Laird, Leaf, and Fischbach, 1988). It would seem that healthy, functioning older adults are at no greater risk for becoming depressed. Instead, age-related effects may be attributable to physical health problems, functional and cognitive disability, chronic illness, low social support, and financial difficulties (Blazer et al., 1991; Roberts, Kaplan, Shema, and Strawbridge, 1997a).

Nevertheless, a number of methodological limitations may lead to underestimates in rates of depression in community studies. Karel (1997) noted that low prevalence rates of major depression among older adults may reflect (1) invalid measurement of depression in older adults (e.g., diagnostic difficulties, symptoms being misattributed to medical causes, symptom recall, and older adults being viewed as less likely to be functionally impaired by depressive symptoms), (2) sampling bias (e.g., older adults may have died, may be unable to participate due to illness and disability, and may be institutionalized or residing in community dwellings for the elderly), or (3) cohort effects due to sociocultural changes (i.e., rates of depression increase in cohorts born after World War II). Several studies have confirmed that rates of depression are increasing in the United States and worldwide (Cross-National Collaborative Group, 1992).

In examining depression in the elderly, it is also important to distinguish between recurrent illness that began earlier in life and first onset illness that manifests itself in late life. Early onset depression has been associated with a more malignant course (Klein et al., 1999; Lewinsohn, Fenn, Stanton, and Franklin, 1986; Sorenson, Rutter, and Aneshensel, 1991), greater vulnerability to chronic life stress (Hammen, Davila, Brown, Gitlin, and Ellicott, 1992), greater psychiatric comorbidity (Kasch and Klein, 1996; Lewinsohn, Rohde, Seeley, and Fischer, 1991), and greater genetic liability (Klein, Taylor, Harding, and Dickstein, 1988; Lyons et al., 1998). Early disorder onset is also associated with greater family psychiatric burden, neuroticism, and dysfunctional past maternal relationships (Brodaty et al., 2001; Van den Berg et al., 2001). On the other hand, first episode of depression in later life is associated with greater medical disability, and decreased neuropsychological and psychophysiological functioning (Lewinsohn et al., 1991). Therefore, it is important to distinguish between depressive subgroups that may possess different etiological pathways: (1) those with early onset and longstanding psychobiological and familial vulnerability, who carry this risk for recurrence into old age; (2) those who