Item non-response on self-reported depression screening questionnaire among community-dwelling elderly

Background: In responses to self-reported depression screening questionnaires, item non-response, which refers to the absence of answers to specific questions, is problematic. The objective of this study were (1) to clarify the features of respondents with item non-response on a self-reported elderly depression screening questionnaire (15-item geriatric depression scale; GDS-15) as compared to respondents with full responses, and (2) to compare positive depression screening rates calculated using two methods: excluding respondents with item non-response (complete case analysis; CCA) and estimating by multiplying mean scores from valid responses by the total number of GDS-15 items for respondents with item non-response. Methods: This was a cross-sectional study conducted from 2010 to 2012. Of 4794 elderly subjects (65 years and older) living in one town in Japan 2836 community-dwelling elderly people (59.2%) were included in the analysis. Results: Item non-response was observed in 25.0% of respondents. Respondents with item non-response had a higher rate of depression and mental and physical problems. Respondents with depression (estimated GDS-15 score >= 6) and suicidal ideation both had a 1.6-times higher risk of item non-response on the GDS-15. The positive depression screening rate on GDS-15 by CCA was 16.5%, compared with 18.9% when calculated by the estimated GDS-15 score. Limitations: Our survey was conducted in one rural area and targeted only elderly people. Conclusion: The incidence of item non-response among community-dwelling elderly people was associated with depression of the respondent. Excluding subjects with item non-response when calculating positive depression screening rates in elderly individuals causes the rate to be underestimated. (C) 2014 Elsevier B.V. All rights reserved.