Saturday, September 5, 2009
The biggest project in psychiatric epidemiology lately is the World Mental Health Survey initiative, which is conducting standardized surveys in more than 30 different countries. A remarkable thing about these surveys is that almost all of the authors from each participating country have concluded that there is a cohort effect for major depression. The standardized interviews (the CIDI, version 3.0) all inquire about lifetime history of specific disorders. One expects that as people get older and have a longer time at risk, the frequency of a positive lifetime history would increase - but this doesn't happen. Instead, the lifetime history starts to decrease in middle age. The idea of a cohort effect is an interpretation of this finding - with the interpretation being that people born more recently (younger at the time of the survey) are at higher risk of depression and this means that there will be an increase in depression in older age groups as these recent birth cohorts age. However, there are other possible explanations - differential mortality (people with depression may die more often as they get older) and recall bias (people may forget about past episodes, or reframe those experiences, so they come out as lifetime negative later in life when they would have come out as positive had they been interviewed earlier). I have been looking at these issues in simulation models, and have produced a brief video showing the results, click here and follow the link to "view." As it turns out, expected levels of forgetting/reframing of past episodes can easily explain the observed pattern. There is no reason to suppose that a cohort effect will lead to higher rates of depression in the future.