Psychometric properties of a prostate cancer radiation late toxicity questionnaire

BACKGROUND To construct a short prostate cancer radiation late toxicity (PCRT) questionnaire with health-related quality-of-life (HRQoL) domains. METHODS The PCRT was developed by item generation, questionnaire construction (n = 7 experts, n = 8 focus group patients), pilot testing (n = 37), item reduction (n = 100), reliability testing (n = 237), and validity testing (n = 274). RESULTS Reliability of the three item-reduced subscales demonstrated intraclass correlation coefficients (CC) of 0.811 (GU), 0.842 (GI), and 0.740 (sexual). Discriminant validity demonstrated Pearson CC of 0.449 (GU-GI), 0.200 (sexual-GU), and 0.09 (sexual-GI). Content validity correlations between PCRT-PCQoL were 0.35-0.78, PCRT-FACT-G(c) were 0.19-0.39, and PCRT-SF-36(R) were 0.03-0.34. CONCLUSION We successfully generated a PCRT HRQoL questionnaire including subscales with very good psychometric properties.