Cross-cultural analyses of the EORTC QLQ-C30

This dissertation aims to explore whether linguistic or cultural factors affected responses to the EORTC QLQ-C30 cancer quality of life instrument. Questionnaire data from 125 separate studies and representing over 39,000 respondents from 49 countries were received and recoded into a standard format. Differential item functioning (DIF) analyses using ordinal logistic regression were conducted in order to determine whether there were different response patterns to each item of the EORTC QLQ-C30 according to (1) the translation used, and 92) the geographic/cultural grouping, after controlling for the overall level of quality of life in that domain. Although most response patterns were similar, statistically significant DIF was identified in at least one item for nearly all translations and geographic groupings examined. interviews with bilingual people were used to help decide whether the DIF was due to the translation or to cultural factors, or whether this was caused by DIF in another item in the same scale (pseudo-DIF). Overall, the results suggest that most of the DIF was caused by translation issues, but there was a suggestion of cultural response patterns for East Asian versus Western countries. A literature review of differential item functioning (DIF) methods found a large number of available methodologies, which are described in this thesis, but no clear consensus as to the recommended method. Computer simulations were used to examine the performance of the ordinal logistic regression DIF methodology when the scale had a small number of items. Separate scale-level regression analyse were also conducted to determine whether there were cultural differences in the emphasis placed on each aspect of quality of life.