Résumé :
|
[BDSP. Notice produite par INIST 1R0xDMek. Diffusion soumise à autorisation]. The authors linked interview data drawn from Utah participants in the Diet, Activity, and Reproduction in Colon Cancer (DARCC) Study (1992-1995) to genealogic and cancer information contained in the Utah Population Database (UPDB). They evaluated the sensitivity of subjects'reports of familial cancers and measured the overall agreement between reported and database records with the kappa (K) statistic. They calculated odds ratios from logistic regression to compare the relative risk estimates that would result from use of either data set (or both data sets). Overall, 37.6% (331 of 881) of the Utah DARCC subjects were linked to the UPDB genealogy. High sensitivities were observed for subjects'reports of breast (83%), colorectal (73%), and prostate (70%) cancers, while ovarian (60%) and uterine (30%) cancers were not reported as well. Results for K were similar, with values of 0.63 for breast cancer and 0.56 for colorectal cancer. Although the observed Ks of 0.36 and 0.25 for ovarian and uterine cancers, respectively, exceeded chance expectations, the agreement between subjects'reports and database records was unimpressive. No consistent difference was observed between cases and controls in the accuracy of self-reports. In general, higher sensitivities were observed among younger subjects than older subjects ; females reported family histories of cancer only slightly better than males. (...)
|