Agreement and psychometric properties among interviewer administration, self-complete, and proxy-report of EQ-5D-Y-3L Tigrinya version in Ethiopia

Abstract

Objective: To assess agreement and psychometric properties of self-complete (SC), interviewer-administration (IA), and proxy-report of the EQ‐5D‐Y-3L Tigrinya version in Ethiopia. Methods: Children and adolescents aged 8 to 18 years completed EQ‐5D‐Y-3L in SC, IA and proxy-report. The modes of administration (MoA) were assessed for feasibility, agreement between each pair of MoA, test-retest reliability, known-group validity and responsiveness. Repeated measures were conducted to assess test-retest reliability, after 10 days among those who reported unchanged health status, and responsiveness after one month among those who reported improved or worsened health status, based on the global rating of change scale. Responsiveness was evaluated using standardized response mean (SRM) to determine the magnitude of change of mean level sum score (LSS) for each MoA. Results: A total of 636 pairs of children and adolescents and proxy were recruited. Missing data were lowest in the IA (< 3.00%) across all dimensions. Agreement was highest between SC and IA across all dimensions (Gwet’s AC ≥ 0.83), and lowest between SC and proxy-report (Gwet’s AC ≥ 0.60). Test-retest reliability was substantial to almost perfect across MoA dimensions, with Gwet’s AC ranging from 0.75 to 0.96. The intraclass correlation coefficients (ICCs) for the LSS and EQ-VAS indicated excellent test-retest reliability (≥ 0.90). Known-group validity was demonstrated, with significant differences in LSS across health conditions (χ² = 291.60 for SC; χ² = 305.70 for IA; p < 0.001). Responsiveness was large in SC and IA (SRM > 0.80) for both improved and worsened groups, but insignificant in proxy-repot for worsened group (SRM = 0.18). Conclusion: The SC and IA versions of the EQ-5D-Y-3L Tigrinya version demonstrated strong agreement and better psychometric properties than proxy-report. The SC and IA can be used interchangeably, while the IA is recommended for children and adolescents with limited literacy.