Reproducibility of self-reported melanoma risk factors in melanoma patients

As melanoma researchers continue to investigate environmental and lifestyle-related risk factors, questionnaire data remain important. The reproducibility of a questionnaire on melanoma risk factors was investigated using a test-retest approach in 389 Dutch melanoma patients. In 2011, 389 melanoma patients filled out a questionnaire on melanoma risk factors twice. Test-retest reproducibility was assessed by calculating kappas (κ), weighted kappas (κw), and intraclass correlation coefficients (ICCs) for categorical, ordinal, and continuous variables, respectively. Stratified analyses were carried out by sex, age group, education level, and time since diagnosis. The median time between the questionnaires was 31 days. The reproducibility was substantial for questions on phenotypic characteristics (κ/κw/ICC=0.62-0.77), fair-to-substantial for sun exposure and sun protection behavior (κ/κw/ICC=0.38-0.79), and moderate for sunburn history (κ/κw=0.42-0.51). No clear differences were observed between men and women. Younger patients showed a better reproducibility in nine of the 29 questions compared with older patients and higher educated patients showed a better reproducibility in four of the 29 questions. Patients with a diagnosis shorter than 1.5 years ago had a better reproducibility in four out of 29 items compared with patients with a diagnosis 1.5-3.0 years ago. Our study showed that self-reported information on melanoma risk factors is fairly well reproducible. Although this does not guarantee validity, this type of questionnaire seems to be useful in research settings. The reproducibility is slightly better in young patients and patients with a higher education level, which can be taken into account when interpreting results from epidemiological studies.