Psychometric evaluation of the EQ-5D-Y-3L in Ethiopian pediatric inpatients: comparing self and proxy reports

Limited evidence exists regarding the measurement properties of the EQ-5D-Y-3L across different modes of administration. This study aimed to examine changes in parent/caregiver-child/adolescent dyad agreement concerning health-related quality of life (HRQoL) over time, assess variations in health status according to socio-demographic factors, and evaluate the responsiveness of the EQ-5D-Y-3L within a pediatric population in Ethiopia.

The study was conducted at the University of Gondar Comprehensive Specialized Hospital, involving children/adolescents aged 4-18 years admitted to the pediatric inpatient unit. Children/adolescents completed the EQ-5D-Y-3L self-complete version at admission and discharge, while parents/caregivers completed the proxy version. Health status was analyzed utilizing the EQ-5D-Y-3L descriptive profiles, utility values, and the EuroQol Visual Analogue Scale (EQ VAS) scores, categorized by age, gender, and residence. Agreement between parent/caregiver and child/adolescent reports was evaluated using weighted Cohen's kappa for dimension levels and the intraclass correlation coefficient (ICC) for utility and EQ VAS scores. Responsiveness was assessed through paired t-tests and the Paretian Classification of Health Change (PCHC) analysis, which classifies health status changes as improved, worsened, mixed, or unchanged based on changes across EQ-5D dimensions.

A total of 957 children/adolescents, with a mean age of 10.7 ± 4.3 years, along with their parents/caregivers, participated in the study. The predominant diagnoses included pneumonia, meningitis, malaria, malnutrition, and glomerulonephritis. Both child/adolescent and parent/caregiver reports indicated poorer health status among older adolescents (13-18 years), boys from rural areas. Agreement on the EQ-5D-Y-3L dimension levels was fair to moderate at admission (weighted kappa ranging from 0.28 to 0.38) and was poorer at discharge for the 'worried, sad or unhappy' dimension (weighted kappa of 0.15). Agreement on utility and EQ VAS scores was acceptable at both admission and discharge (ICC: 0.498-0.676), with moderate to good agreement observed among children/adolescents aged 7-16 years. However, agreement on utility scores decreased at discharge for older boys (13-18 years old) and urban residents, while it increased for the younger age group (4-6 years old). Responsiveness analysis demonstrated significant improvements in the dimensions of 'Looking After Myself,' 'Mobility,' and 'Worried, Sad or Unhappy,' with most children/adolescents exhibiting health improvements according to PCHC criteria.

Parent/caregiver-child/adolescent dyad agreement concerning HRQoL was low to moderate and varied according to socio-demographic factors. The EQ-5D-Y-3L instrument demonstrated responsiveness to changes in health status, supporting its utility in pediatric populations. These findings underscore the importance of employing age-appropriate and context-sensitive HRQoL assessment tools in pediatric care and health policy. Incorporating both child/adolescent and parent/caregiver perspectives can inform clinical decisions and resource allocation, especially in low-resource settings. Further research is warranted to explore factors influencing these variations and to enhance understanding of their underlying causes.