Methods: Fit to the Rasch model was examined in 6648 8-to-15-year-olds from Australia, New Zealand, Brunei, Cambodia, Hong Kong, Malaysia, Thailand, Germany, United Kingdom, Brazil and Mexico.
Results: In all but two items, the initial five answer options were reduced to three or four, to increase precision of the children’s selection. Items 10 (Shy/embarrassed) and 11 (Concerned what others think) showed an ‘extra’ dependency between item scores beyond the relationship related to the underlying latent construct represented by the instrument, and so were deleted. Without these two items, the CPQ was unidimensional. The three oral symptoms items (4 Food stuck in teeth, 3 Bad breath and 1 Pain) were required for a sufficient person-item coverage. In three out of 14 items (21%), Europe and South America showed regional differences in the patterns of how the answer options were selected. No differential item functioning was detected for age.
Conclusion: Except for a few modifications, the present analysis supports the combination of items, the cross-cultural validity of the CPQ with 14 items and the extension of the age range from 8 to 15 years.
- Patient-reported outcome measures
- Measurement accuracy
- Cross-border care
- Psychometric properties
- Item response theory