Standardized interviews to detect depression have been available since the 1980s. In order to determine the prevalence of major depression, most studies have adminstered these interviews to samples of people in the community. There is, however, a potentially serious problem with this approach. Not all episodes of depression are necessarily remembered. Furthermore, the instruments attempt to identifiy episodes by asking about specific symptoms, but people may not remember, for example, a series of weeks during which their sleep or appetite patterns were altered. Furthermore, of course, peoples' memories of events can change over time. The end result is that the prevalence of major depression in the population may be seriously underestimated. To explore this possibility, I recently examined the accumulation of depressive episodes during 14 years of follow-up in the National Population Health Survey. Since the participants in this cohort were interviewed every two years, they were probably much more likely to have episodes detected than other studies where they have been asked to recall episodes occuring much earlier in their life.
The hypothesis was confimed: the prevalence of depression as measured by the prospective approach was twice the usually cited figures for Canada. Major depression is probably much more common than is currently beleived. Details (in preliminary form) may be found here in the journal BMC Psychiatry.