items loaded on four subscales, which she labeled well-being, depressed mood, somatic symptoms, and optimism. However, the subscales were not consistent across different samples and she raised concerns that for elderly people the somatic items may falsely identify physical problems common with aging, thus reducing their validity as indicators of depression (18). This hypothesis was subsequently tested by Steuer et al., who found that the overall SDS scores did not correlate with physicians’ ratings of overall health, although one somatic item (fatigue) did correlate significantly (rho = 0.35) (17, p686). A principal components analysis of the Japanese version of the SDS identified three factors reflecting cognitive, affective and somatic symptoms (19).