Minor change in the diagnostic threshold leads into major alteration in the prevalence estimate of depression

BackgroundAlthough highly structured diagnostic interview instruments are reportedly reliable, it has been suggested that even small changes in the diagnostic threshold or wording of the questions may substantially affect the results. General population data on this topic are scarce.MethodsA random sample of 15–75-year-old Finnish men and women was interviewed in 1996 (N = 5993). The diagnosis of DSM-III-R major depressive episode (MDE) was made by using the Composite International Diagnostic Interview Short Form (CIDI-SF). Prevalence estimates derived by using three different thresholds for the intensity of the depressed mood (“depressed mood all day” vs. “most of the day” vs. “half of the day”) are compared.ResultsThe use of the threshold “depressed mood all day”, yielded the prevalence estimate of 4.67% [95% CI 4.12, 5.22] for MDE, while according to the thresholds of “depressed mood most time of the day” and “at least half of the day” the prevalences were 9.23% [95% CI 8.47, 9.99] and 11.9% [95% CI 11.0, 12.8], respectively. A consistent female to male ratio was observed across the categories, while an age effect was noted so that younger age associated with less frequent depressed mood. The diagnostic thresholds associated with impairment, episode duration, treatment need and use.ConclusionsMinor changes in case definition within the same measuring instrument may produce major differences in prevalence estimates. The categories defined for the purposes of this study were on a continuum where the frequency of depressed mood associated with other measures of the depressive episode.