Exposure to ultraviolet radiation has consistently been linked to an increased risk of melanoma. Most studies rely upon retrospective self-report of exposure, which is susceptible to measurement error, which can distort the observed effect sizes. The reliability of self-report of many sun exposure factors has been previously described, but the self-report of use of tanning beds and self-tanning creams has been less well characterized. The purpose of this reliability study was to determine the reproducibility of a survey used in a case–control study of melanoma and to quantify the effect that measurement error could have had on the overall study results. A mailed survey was readministered after completion of the initial survey to 76 randomly selected participants in a case–control study of melanoma. Cases and controls were diagnosed in 1999 and 2000 and ascertained from the Iowa Cancer Registry in 2002. Colorectal cancer patients served as controls. We assessed the consistency of self-reported use of sunlamps and self-tanning creams, sun sensitivity, and history of sunburns using Cohen's kappa (κ). There was substantial reliability in reporting the use of sunlamps or self-tanning creams (κ = 0.83 and 0.94, respectively). Therefore, adjusting the observed odds ratios for use of sunlamps or self-tanning creams to account for measurement error had little effect. The reliability of other factors associated with melanoma was also substantial, with κ estimates of 0.62 to 0.78 for several sun-sensitivity factors. Overall, the survey instrument demonstrated substantial reproducibility for factors related to use of tanning beds and self-tanning creams and for sun sensitivity factors. This indicates that this survey instrument is reasonably reliable and that measurement error was unlikely to have biased the overall study results.
Read full abstract