OBJECTIVES: The three-level EQ-5D-Y (Y-3L) and five-level version (Y-5L) are health-related quality of life instruments for children and adolescents. With the introduction of the Y-5L, it is essential to compare their performance. An ongoing review showed moderate construct validity and test-retest reliability for both versions. This review explores factors influencing the construct validity and test-retest reliability of Y-3L and Y-5L using published head-to-head studies.

METHODS: Eight databases were searched for validation papers comparing the Y-3L and Y-5L, published in English up to February 14, 2024. All full texts were assessed for eligibility by two independent reviewers. Measurement properties were rated as “sufficient” (good), “inconsistent” (moderate), or “insufficient” (poor) per COSMIN guidelines. Seven factors were examined: (1) population type, (2) geographical region, (3) age, (4) survey language, (5) respondent type, and (6) interval period [only for reliability]. Additionally, we examined the impact of these factors on construct validity using meta-regression.

RESULTS: The review included 17 studies from Southeast/East Asia (N=9) and Africa (N=5), mostly cross-sectional (N=12) with consecutive sampling (N=16). High-certainty evidence supports moderate construct validity for both versions. Meta-regression of 788 tests (Y-3L) and 866 tests (Y-5L) from 14 articles showed that general populations (Y-3L: OR 1.83, p<0.001), adolescents aged 12-18 (Y-3L: OR 1.83; Y-5L: OR 3.07, p<0.05), and the English (Y-3L: OR 1.91; 5L: OR 1.72, p<0.05) and Chinese versions (Y-3L: OR 3.29; Y-5L: OR 5.52, p<0.001) were associated with positive construct validity. Low-certainty evidence supports moderate test-retest reliability for Y-3L and Y-5L (N=11, 59 tests), with Y-3L showing better reliability in studies conducted in Africa and Oceania, as well as among child respondents.

CONCLUSIONS: Both EQ-5D-Y versions showed similar construct validity, while the Y-3L outperformed Y-5L in test-retest reliability. The review identified high heterogeneity in results and highlighted the need for more high-quality head-to-head comparative validation studies.