Background. Previous research has reported inconsistent evidence of the trajectory of depressive symptoms across the adult lifespan. We investigated how the distributions of each item score change with age and determined whether the trajectory of depressive symptoms varied with the scoring methods of the questionnaire. Methods. We analyzed data collected from 21,040 subjects who participated in the national survey in Japan. Depressive symptoms were assessed using the Center for Epidemiologic Studies Depression Scale (CES-D). The CES-D has 20 items, each of which is scored in four grades of "rarely," "some," "much," and "most of the time." We used the exponential distribution model which fits the distributions of 16 negative symptom items of CES-D, with the probabilities of "some," "much," "most," and "rarely" expressed as P, Pr, Pr (2), and 1 - P × (r (2) + r + 1). Results. The distributions of the responses to 16 negative symptom items followed the common exponential model across all age groups. The mean of the estimated parameter r of 16 negative items showed a U-shape pattern, being high during 12-29 years, remaining low during 30-50 years, and then increasing again over 60 years. The trajectory of depressive symptom scores simulating the binary method was different from that of the empirical scores using the Likert method. Conclusions. Our findings show that the increase in the depressive symptoms score during older age is based on the increase of the parameter r. The differences in the scoring method may contribute to the different age-related patterns across the adult lifespan.
Keywords: CES-D; CIS-R; Depression; Depressive symptoms; Exponential distribution; Item response; Latent trait; Mathematical model; Questionnaire.