OBJECTIVES: To assess, by item response theory (IRT) and confirmatory factor analysis (CFA), the psychometric characteristics of the EORTC QLQ-C30.

METHODS: We analyzed responses to QLQ-C30 from 1107 patients with advanced lung cancer who were included in five clinical trials. We used nonparametric Mokken analysis in a training sample (n=332) to preliminary assess the dimensional structure of the QLQ-C30. We then used CFA in the test sample (n=775) to confirm the measurement dimensions. We also used a parametric IRT (Samejima's graded response model) to evaluate the difficulty and discrimination of the ítems.

RESULTS: The Mokken analysis showed QLQ-C30 represents a unidimensional scale. All items showed adequate scalability indexes (Loevinger’s index > 0.30), and the overall scalability of QLQ-C30 was 0.47, defining a medium scale. The unidimensionality of the QLQ-C30 was confirmed by CFA (CFI = 0.98; RMSEA = 0.055) with all items presenting salient factorial loadings (>0.40). The Samejima model showed most QLQ-C30 items presented adequate difficulty and discrimination.

CONCLUSIONS: The EORTC QLQ-C30 fits an underlying unidimensional latent construct of perceived quality of life in advanced lung cancer patients. Subsequent studies on differential item functioning across cancer populations are warranted, as well as its sensitivity to change and minimum important difference.