Assessing Outcomes for Cost-Utility Analysis in Children and Adolescents With Mental Health Problems: Are Multiattribute Utility Instruments Fit for Purpose?

Abstract

Objectives: The objective of this study was to compare the concurrent and construct validity, as well as the sensitivity of 5 multiattribute utility instruments (MAUIs), including the Assessment of Quality of Life-6D (AQoL-6D), EQ-5D-Y, Health Utilities Index (HUI)-2 and HUI-3, and the Child Health Utility 9D, 1 generic pediatric quality of life instrument, with 3 routinely collected outcome measures in Australian mental health services (Strengths and Difficulties Questionnaire, Clinical Global Assessment Scale [CGAS] and the Health of the Nation Outcome Scale for Children and Adolescents) in children and adolescents diagnosed of internalizing (eg, anxiety/depression), externalizing (eg, attention deficit hyperactivity disorder/conduct disorders), and trauma/stress related mental disorders. Methods: A cross-sectional survey of measures, including demographic and basic treatment information, in children/adolescents recruited via 5 child and youth mental health services in Queensland and Victoria, Australia. Measures were either proxy or self-report completed, the CGAS and the Health of the Nation Outcome Scale for Children and Adolescents were clinician completed. Results: The sample included 426 participants and had a mean age of 13.7 years (range 7-18 years). Utilities (as calculated from MAUIs) were generally lower in older adolescents and those with internalizing disorders. All MAUIs and self-reported clinical measures significantly correlated with each other (absolute correlation range 0.40-0.90), with the AQoL-6D showing generally higher levels of correlations. Correlations between the MAUIs and clinician/proxy-reported measures were weak, regardless of diagnosis (absolute correlation range 0.09-0.47). Generally, EQ-5D-Y, HUI-2, and AQoL-6D were more sensitive than Child Health Utility 9D and HUI-3 when distinguishing between different severities according to clinician-assessed CGAS (effect size range 0.17-0.84). Conclusions: The study showed that the commonly used MAUIs had good concurrent and construct validity compared with routinely used self-complete measures but poor validity when compared with clinician/proxy-completed measures. These findings generally held across different diagnoses.