The sensitivity of the K6 as a screen for any disorder in community mental health surveys: A cautionary note

Abstract

Objective: Short screening instruments, which exclude respondents unlikely to have psychiatric disorders, can make epidemiologic surveys shorter and more cost-effective. The Kessler 6-Item Psychological Distress Scale (K6), a measure of generalized distress, has been proposed for this role and has shown good agreement with the Composite International Diagnostic Interview (CIDI). However, performance of the K6 may vary for individual disorders or combinations of disorders. In this report, we examine the ability of the K6 to detect disorders among respondents in different diagnostic categories. Method: We used data from Cycle 1.2 of the Canadian Community Health Survey to assemble 5 groups of respondents with different 12-month psychiatric disorders (n = 4481). A sixth group comprised those with 2 or more disorders. We examined the sensitivity of the K6 among respondents with an individual disorder as well as those with multiple disorders. Results: The sensitivity of the K6 varies significantly by disorder; it is highest among respondents with multiple disorders and lowest among those with agoraphobia only. Conclusions: Use of the K6 as a screen for the CIDI is likely to result in biased prevalence estimates. However, both instruments should be compared with a third standard to fully assess the benefits and drawbacks of their combination.