Mapping PedsQLTM scores onto CHU9D utility scores: estimation, validation and a comparison of alternative instrument versions

Background The Paediatric Quality of Life Inventory(TM) 4.0 Generic Core Scales (PedsQL) is a non-preference based instrument for assessing health related quality of life (HRQoL) in children. Recent papers presented algorithms of parental proxy and short-form versions of the PedsQL onto the validated preference-based Child Health Utility 9D (CHU9D) instrument, to enable conversion of PedsQL scores to quality adjusted life years for use in economic evaluation. However, further research was needed to both validate these algorithms, and assess if use of the full 23-item PedsQL self-report instrument is preferable to other PedsQL versions for mapping onto child self-report CHU9D utilities. Objective To develop a mapping algorithm for converting the 23-item PedsQL instrument onto the CHU9D instrument and provide an external validation of two recently published algorithms that might be considered alternatives. Methods Data from children in the Longitudinal Study of Australian Children (LSAC) were used (N = 1801). Six econometric methods were compared to identify the best algorithms, assessed against a series of goodness-of-fit criteria. The same data and goodness-of-fit criteria were used in the external validation exercise for previously published mapping algorithms. Results The optimal mapping algorithm was identified, which used PedsQL dimension scores to predict the CHU9D utilities. It performed well against standard goodness-of-fit tests. The external validation exercise revealed the recently published alternative algorithms also performed relatively well. Conclusion The identified mapping algorithms can be used to facilitate cost-utility analysis in comparable populations when only the PedsQL instrument is available. Results from this population indicate the algorithms identified in this paper are well suited for estimating CHU9D self-report utilities when the full 23-item self-report PedsQL instrument has been used.