Presentation (CTI)

OBJECTIVES: This study examined how modifications to item order, item polarity, and response options affected the psychometric performance of the EQ-HWB in an older Chinese population.
METHODS: A 2-week longitudinal survey was conducted among Chinese older adults (≥60 years) using social care services. Participants completed both the original and modified EQ-HWB versions alongside the EQ-5D-5L. Psychometric performance was evaluated through analyses of missing data, response distributions, convergent validity, structural validity, known-groups validity, test-retest reliability, and cross-instrument change consistency.
RESULTS: Seventy-eight participants completed the baseline survey and 73 completed follow-up. The modified version showed more missing responses at baseline, fewer at follow-up, and more ceiling effects. Correlations with the EQ-5D-5L were similar across versions. EFA supported a four-factor structure for the original version and a three-factor structure for the modified version, with the additional factor in the original defined by three positively worded items. Both versions demonstrated known-groups validity, although the original version showed greater discriminatory ability and higher test-retest reliability (ICC=0.893 vs 0.640). Cross-instrument change consistency was moderate for both versions.
CONCLUSIONS: Changes in item polarity, item order, and response options influenced psychometric performance. The findings suggest that converting positively worded items into negatively worded items may alter item meaning rather than represent a neutral wording change. The original version may capture broader quality-of-life aspects, whereas the modified version appears more closely aligned with health-related outcomes.