OBJECTIVES: Critical questions remain as to differences in adult and adolescent preferences for health states described by the EQ-5D-Y. There is no existing research comparing the test-retest reliability of reported preferences.

METHODS: A Best Worst Scaling (BWS) experiment was conducted in Spain using health states constructed from the EQ-5D-Y. A general population sample of 1006 adults, and 1000 adolescents (aged 11 to 17 years) completed the BWS survey online. We compare the test-retest reliability of ‘best’ and ‘worst’ ratings.

RESULTS: 470 (47%) adults and 323 (32%) adolescents completed the identical repeat survey (average 2.92 days later for adolescents, 3.35 days for adults). Adolescents and adults consistently rated ’no pain and discomfort’ most often as best in original and repeat surveys (Adolescents 51%, 54%, Adults 51%, 53%). Adolescents and adults consistently rated ‘a lot of pain and discomfort’ most often as worst in original and repeat surveys (Adolescents 43%, 48%, Adults 44%, 46%). Chance corrected agreement and intra-class correlation in utility score will be reported with regression analyses illustrating the relationship between reliability and socio-demographic variables.

CONCLUSIONS: Concerns have been raised regarding the stability of utility values generated using BWS methods. In addition, little is known about how adolescent and adult preferences differ in reliability. This research indicates that repeat responses for the EQ-5D-Y for both adults and adolescents show re-test reliability. This will make critical contributions to gaps in current knowledge and will guide future decisions about how to best elicit preferences for child health utility instruments.