Accurate measurement of depressive symptoms in the cancer setting is critical for ensuring optimal quality of life and patient outcomes. The present study compared the one-factor, correlated two-factor, correlated four-factor, and second-order factor models of the Center for Epidemiologic Studies Depression Scale (CES-D), a commonly used measure in oncology settings. Given the importance of adequate psychometric performance of the CES-D across age groups, a second aim was to examine measurement invariance between younger and older adults with cancer. Participants (N = 663) were recruited from outpatient clinics at a large cancer center. Over one-fourth of the sample endorsed clinically significant depressive symptoms (25.9%, n = 165). Confirmatory factor analysis of the CES-D supported the hypothesized correlated four-factor model as the best fit. The second-order factor also demonstrated good fit, but interpretations of the factors were more complex. Factors were highly correlated (range = .38-.91). There was also support for full scalar invariance between age groups, suggesting that regardless of age, respondents endorse the same response category for the same level of the latent trait (i.e., depression) on the CES-D. Taken together, the results suggest that the CES-D is a viable depression screening option for oncology settings and does not require scoring adjustments for respondent age.
Keywords: CFA; aging; cancer; depression; screening.