Subjects were selected from WEB study participants born in Western New York State between 1920 and 1964 for whom we were able to obtain a New York State birth certificate. Addresses were geocoded and mean distance and SD between self-reported and certificate addresses were calculated.
Our findings indicate good agreement between self-reported and birth certificate addresses (77 % are within 0.25 mile difference), and no difference in recall for cases compared with controls.
This study provides evidence that self-reported and birth certificate-based addresses may be used together to achieve increased accuracy of historical records in early life when place of birth is used as a proxy for early life environment in relation to the development of breast cancer or other chronic diseases.