Validity of self-reported cancer: Comparison between self-report versus cancer registry records in the Geelong Osteoporosis Study.

BACKGROUND:Determining the validity of self-reported data is important. The aim of this study was to assess the validity of self-reported cancer and investigate factors associated with accurate reporting in men and women. METHODS:Study participants (n = 1727) from the Geelong Osteoporosis Study, located in south-eastern Australia, were utilised. Self-reported cancer data were compared to Victorian Cancer Registry records. Age, socioeconomic status (SES), education and time between cancer diagnosis and study appointment were investigated as factors associated with accuracy of self-report. RESULTS:There were 142 participants who self-reported a cancer and 135 with a VCR record. Comparing self-report to any registry record, sensitivity was 63.7 %, specificity 96.5 %, PPV 60.6 %, NPV 96.9 %, and overall agreement ĸ0.588. Comparing exact-match records, sensitivity was 58.8 %, specificity 95.5 %, PPV 49.3 %, NPV 96.9 % and overall agreement ĸ0.499. In logistic regression models, post-secondary education was independently associated with accuracy of any (OR 1.72, 95 % CI 1.10-2.70) and exact-match (OR 1.59, 95 % CI 1.05-2.42) self-report, compared to cancer registry record. For any cancer, being aged >70 years was inversely associated with accuracy (OR 0.24, 95 % CI 0.15-0.38). Likewise, for matched cancer reporting, those aged 60-70 years (OR 0.51, 95 %CI 0.30-0.88) and >70 years (OR 0.23, 95 % CI 0.15-0.35) were less accurate. No other significant associations were detected. CONCLUSION:Results suggest moderate agreement between self-report and registry data for any cancer among men and women. However, when comparing self-report to registry data for exact-match cancer type, level of overall agreement deteriorated. Self-report cancer data may be acceptable for determining a history of cancer, although, is less accurate in identifying history of specific cancer types documented in registry-based data.