Gepubliceerd in:
01-09-2015
Psychometric evaluation of the EORTC QLQ-PR25 questionnaire in assessing health-related quality of life in prostate cancer survivors: a curate’s egg
Gepubliceerd in: Quality of Life Research | Uitgave 9/2015
Log in om toegang te krijgenAbstract
Purpose
The EORTC QLQ-PR25 was primarily developed to measure disease-specific health-related quality of life (HRQoL) of prostate cancer (PCa) patients undergoing active treatment. The growing number of cancer survivors has focussed interest on assessing survivors’ HRQoL. We evaluated psychometric properties of the EORTC QLQ-PR25 questionnaire amongst PCa survivors.
Methods
A postal questionnaire, including the QLQ-PR25, was administered to 6559 PCa survivors 2–18 years post-diagnosis, identified through population-based cancer registries in Ireland; 3348 participated. The QLQ-PR25 has been suggested to have five multi-item subscales measuring urinary (US), bowel (BS) and hormone treatment-related symptoms (TS), sexual activity (SA) and sexual functioning (SF), and a single item measuring bother due to using incontinence aids (IA). Reliability analysis, divergent validity, discriminant validity (compared to EORTC QLQ-C30, DASS-21 and EuroQoL EQ-5D-5L), and exploratory and confirmatory factor analysis (EFA, CFA) were undertaken.
Results
The percentage of survivors completing QLQ-PR25 subscales was: US-79 %; IA-20 %; BS-83 %; TS-86 %; SA-87 %; and SF-26 %. Reliability was acceptable (Cronbach’s α > 0.7) for three subscales (US, SA, SF). The instrument discriminated well between clinically distinct groups, especially those defined by primary treatment(s) and stage at diagnosis. The SA and SF subscales showed good discriminant validity compared to QLQ-C30, DASS-21 and EQ-5D-5L; other subscales did not. EFA reproduced a near fit to the proposed factor structure. CFA confirmed this.
Conclusion
This is the largest ever QLQ-PR25 validation study. When used in PCa survivors, although the proposed factor structure was confirmed, some of the subscales (e.g. BS and TS) showed poor reliability, a lack of discriminant validity and moderate levels of item non-response.