RÉSUMÉ
OBJECTIVE: The issue of non-response in dementia epidemiological studies, which may result in the underestimation of the prevalence of dementia, has attracted little attention. We aimed to explore the causes and related factors of non-response in a dementia survey among Chinese veterans. METHODS: A two-phase, cross-sectional study investigated the prevalence of dementia and mild cognitive impairment in Chinese veterans aged ≥ 60 years. We collected the socio-demographic data and prior medical history, evaluated the health status of veterans and their caregivers, assessed the cognitive status of veterans, and evaluated the care burden of caregivers by Caregiver Burden Inventory (CBI). RESULTS: Of 9676 eligible participants, 525 (5.4%) veterans in phase 1 and 1706 (35.0%) veterans among 4875 veterans in phase 2 did not respond. Illness, hospitalization and death accounted for 63.0% and 75.5% non-response in phases 1 and 2, respectively. Non-participation in social activities, self-perceived poor health status, worsened health changes, self-reported need for life care, and history of hearing loss or glaucoma independently predicted non-response in phase 1 or 2. The heavy care burden, suggested by the higher CBI scores and self-reported health deterioration of the primary caregivers, predicted non-response in phase 1 or 2. CONCLUSIONS: The negative factors from both the participants and their caregivers independently predicted the non-response in the dementia study in an older population. Preventative strategies from the perspectives of the participants and caregivers should be developed to improve the response rates in both phases in a cross-sectional study.