Challenges with interviewer administration of EQ-5D questionnaires in a large-scale population survey in China: a qualitative analysis

INTRODUCTION: While most adult measures of health-related quality of life (HRQoL) are designed for self-completion, their use in low- and middle-income countries involves interviewer administration. Large-scale population surveys in China that collect EQ-5D data via face-to-face interviews typically show high ceiling effects. We aimed to identify non-standard interviewing approaches in the collection of EQ-5D data in a large-scale population health survey in China and to explore reasons for these non-standard approaches and plausible mechanisms through which these deviations may influence EQ-5D response patterns. METHODS: A random sample of routinely recorded interviews from population health surveys in Nanjing, China were thematically analysed to identify interviewing approaches to collecting EQ-5D-3L. Focus group discussions with ten interviewers were used to explore the reasons for those approaches. RESULTS: Six non-standard interviewing approaches were identified: skipping questions; altering question wording; combining multiple dimensions into one question; altering response options; selecting responses based on interviewer interpretation of respondent's narrative; and not asking the interviewee any questions. Three main reasons motivated these non-standard approaches: (a) comprehension and response challenges among low-literacy respondents; (b) survey structure, including overall length, positioning of EQ-5D questions and respondent burden leading to time-saving interview strategies; (c) limited interviewer knowledge of self-reported health measures, including a tendency to view EQ-5D as an observer-assessed instrument. CONCLUSION: Our findings highlight challenges in interviewer administration of the EQ-5D in large-scale population surveys. The non-standard approaches observed may affect data quality, potentially contribute to ceiling effects, and complicate the interpretation of these data being 'self-reported'. Providing training on EQ-5D data collection to interviewers may be a first step to strengthening data collection processes.