Does this patient have a family history of cancer? An evidence-based analysis of the accuracy of family cancer history.

CONTEXT: A family history of certain cancers is associated with an increased risk of developing cancer. Both cancer screening and genetic services referral decisions are often based on self-reported pedigree information.
OBJECTIVE: To determine the accuracy of self-reported family cancer history information. DATA SOURCES: English-language articles were retrieved by searching MEDLINE (1966-June 2004) using Medical Subject Headings family, genetic predisposition to disease, medical history taking, neoplasm, and reproducibility of results. Additional articles were identified through bibliography searches. STUDY SELECTION: Original studies in which investigators validated self-reported family history by reviewing the identified relatives' medical records, death certificate, or cancer registry information were included, as well as studies that evaluated breast, colon, ovarian, endometrial, and prostate cancers. DATA EXTRACTION: Two of the 3 investigators independently reviewed and abstracted data for estimating the likelihood ratios (LRs) of self-reported family cancer history information. Only data from studies that evaluated both positive and negative family cancer histories were included within the analyses. A total of 14 studies met the search criteria and were included in the review. DATA SYNTHESIS: For patients without a personal history of cancer, the positive and negative LRs of a family history of the following cancers in a first-degree relative were 23.0 (95% confidence interval [CI], 6.4-81.0) and 0.25 (95% CI, 0.10-0.63) for colon cancer; 8.9 (95% CI, 5.4-15.0) and 0.20 (95% CI, 0.08-0.49) for breast cancer; 14.0 (95% CI, 2.2-83.4) and 0.68 (95% CI, 0.31-1.52) for endometrial cancer; 34.0 (95% CI, 5.7-202.0) and 0.51 (95% CI, 0.13-2.10) for ovarian cancer; and 12.3 (95% CI, 6.5-24.0) and 0.32 (95% CI, 0.18-0.55) for prostate cancer, respectively. Positive predictive values tended to be better in articles concerning first-degree relatives compared with second-degree relatives.
CONCLUSIONS: Patient-reported family cancer histories for first-degree relatives are accurate and valuable for breast and colon cancer risk assessments. Negative family history reports for ovarian and endometrial cancers are less useful, although the prevalence of these malignancies within families is low.