Expanding the experimental EuroQol toddler and infant populations (EQ-TIPS) descriptive system: comparing 3-level and 5-level performance for infants and toddlers with chronic conditions

OBJECTIVES: The EuroQol Infant and Toddlers Populations (EQ-TIPS) instrument may support both 3-level and 5-level versions, mirroring the structure of the EuroQol Adult and Youth descriptive systems. This study compared the performance of the experimental version 2.0 EQ-TIPS-3L (3 L) with EQ-TIPS-5L (5L) in children 0-4 years living with a health condition. METHODS: Caregivers of children were recruited from specialist outpatient clinics in South Africa. The feasibility of EQ-TIPS-5L and EQ-TIPS-3L was compared using the absolute reduction in the ceiling effect (111111). Discriminatory power was evaluated with Shannon's H' and J' index for absolute and relative informativity. The redistribution of dimension responses between the 5 L and 3L versions was evaluated for inconsistency, defined as a 5 L response that was two or more levels removed from a 3 L response. Convergent validity of EQ-TIPS dimensions was calculated with Kendall Tau B and Gamma correlations. Known-group severity was calculated for the EQ-TIPS-5 L and EQ-TIPS-3 L LSS scores using severity groupings based on PedsQL total score (≤ 74.2) and EQ VAS (≤ 80). RESULTS: Data from 176 children with a median age of 29 months, and more males (54%) were included. Most respondents were mothers (78%). Children were grouped by disease category: epilepsy (9%), neuromuscular disease (7%), chronic gastrointestinal disease (16%), renal disease (27%) and oncological or haematological diseases (41%). The ceiling effect (111111) decreased minimally by 2% points from the 3L to 5L. Absolute informativity (H') of dimensions improved by 0.096 on the 5L, with retention of the spread of responses for all dimensions. Convergent validity was similar, with strong correlations on paired 3L and 5L dimensions. The 3L and 5L were similarly able to significantly differentiate between known groups based on PedsQL and EQ VAS severity cut-offs. CONCLUSION: These results indicate that a 5L version of the EQ-TIPS does not influence the ceiling effect in the same way as the Adult and Youth descriptive system. The modest increase in informativity of the EQ-TIPS-5L suggests the potential advantages of expanded levels of reporting, but this requires testing in samples with more frequent reporting of severe health states.