Abstract
BACKGROUND: Cultural biases may affect the individual responses to questionnaires for depression and thus confound the international or multiethnic researches on depression. OBJECTIVE: We compared the diagnostic accuracy of the Center for Epidemiologic Studies Depression Scale (CES-D) for major depressive disorder (MDD) in late life between Korean and US Caucasian elderly. METHODS: This study included 332 US Caucasian MDD patients, 116 Korean MDD patients, 125 US Caucasian nondepressed subjects and 700 Korean nondepressed subjects. Differential item functioning and factor analyses were conducted to examine the differences in the response patterns to the CES-D between the US Caucasian and Korean elderly. Diagnostic accuracy of the CES-D for MDD was compared using the area under the receiver-operating characteristic curves (AUC). RESULTS: The Korean elderly were more likely to endorse 6 items compared to the US Caucasians, and the US Caucasian elderly were more likely to endorse 5 items compared to the Koreans. The factor solutions from both ethnic groups were not comparable since the congruence coefficient for the second factor was below 0.46 and that for the first factor did not reach 0.90. The AUC of the CES-D for MDD in Koreans (AUC = 0.850, 95% CI = 0.801-0.899) was significantly smaller than that in US Caucasians (AUC = 0.973, 95% CI = 0.960-0.987), and the optimal cutoff score of the CES-D in the Korean elderly (21/22) was 2 times higher than that in the US Caucasian elderly (10/11). CONCLUSION: Cross-cultural issues may significantly influence the diagnostic accuracy of depression questionnaires and thus should be considered more carefully than before in both clinical and research settings on multiethnic populations.