OBJECTIVES: To explore the quality of data collected in the paediatric multi-instrument comparison (P-MIC) study, and assess instrument feasibility, acceptability, and psychometric performance.

METHODS: P-MIC data for children aged 5-18 years included the EQ-5D-Y-3L, EQ-5D-Y-5L, CHU9D, PedsQL, HUI, PROMIS-25 and AQoL-6D. Data quality, participant demographics and dropout rates were analysed descriptively. Feasibility and acceptability, self-reported difficulty completing instrument, time to complete, and instrument response patterns were analysed descriptively. Known group (children with a chronic health condition) and convergent validity were assessed. Sub-group analyses were undertaken by child age, proxy versus self-report, child health status and recruitment method.

RESULTS: More participants recruited via the online panel sample were removed for not meeting quality criteria (35%) compared to those recruited via hospital (2%). After their removal, the quality of the data was similarly good across samples. The initial survey was completed by 5,444 children and their caregivers. Higher follow-up rates were found in the hospital sample (79%) compared to online panel (24%). Over 70% of participants reported finding each instrument either easy or very easy to complete with a median time to complete of < 3 minutes. Proxies and online panel participants reported finding instruments easier to complete and completed them more quickly compared to their counterparts. Moderate to high ceiling effects were found in all instruments apart from PedsQL. All instruments except PROMIS-25 demonstrated known group validity with large effect sizes. There were moderate correlations largely in the expected direction between most instruments, except for the HUI.

CONCLUSIONS: This study is the largest of its kind internationally. Results suggest none of the instruments are clearly superior to others: each has different properties that might confer advantages (or disadvantages) in particular settings. The preference-weighted instruments all suffered from ceiling effects – a possible concern for their use in detecting improvements in child health.