Self-reported family history of cancer: The utility of probing questions

Background: Many studies of familial aggregation of cancer have relied on self-reported cancer family history data because verification through medical records was not feasible. Self-reported data might be nonspecific and prone to errors, although of high quality for certain cancers. We evaluated the utility of a brief follow-up survey to elicit further details regarding uncertain cancer conditions. Methods: We reviewed all of the cancer family histories reported by 13,972 survivors participating in the Childhood Cancer Survivor Study. There were 533 affected relatives (obtained from 500 cases) identified as having uncertain cancer histories. Of these 500 cases, 386 (77%) were successfully recontacted. They completed a 5- to 7-minute telephone survey containing focused, probing questions regarding their 406 relatives with uncertain cancer conditions. Two investigators independently reviewed the survey responses. Results: A disease site and malignancy status could be determined for 81% (331 of 406) of the relatives. Among the 312 relatives with a single uncertain cancer condition, the additional survey identified 58% as having a malignant condition with a known primary site, 7% as having a malignant condition with an undetermined primary site, 22% as having a benign condition, and 13% for whom no further clarification was possible. Among the 94 relatives with multiple reported cancers, 64% were found to represent a single diagnosis. Conclusions: Incorporation of additional probing questions into the initial family history interview would improve the ability to classify cancer reports by site and malignancy status. The utility of these additional questions could vary by the site and specificity of the initial cancer reports.