Misclassification bias related to definition of menopausal status in case-control studies of breast cancer.

It is often assumed, but has not been consistently observed, that some characteristics of reproductive history are specifically related to breast cancer of pre- or postmenopausal onset. To determine whether inconsistent reports may be due to differences in definition of menopause, we computed the relative odds (RO) of breast cancer for nulliparity, age at first live birth, family history of breast cancer and prior history of benign breast disease, separately in pre- and postmenopausal women, using seven different definitions of menopause. Results show that (i) relative odds of breast cancer and their confidence intervals may vary according to definitions of menopause; (ii) age-based definitions of menopause are associated with moderate differential misclassification bias between cases and controls; (iii) nulliparity, late age at first birth and family history of breast cancer seem to be specific risk factors for pre- but not postmenopausal breast cancer when cutoff for menopausal status is 10 years or more after last menses; and (iv) when information on menstrual history is not available, 50 years of age may be the best proxy for all menses-based definitions of menopause. We conclude that inconsistent findings on the effect of menopausal status in the association of breast cancer with some reproductive factors are partly due to statistical imprecision and differential misclassification bias associated with different age-based or menses-based definitions of menopause. Researchers should either test whether their conclusions hold using several definitions of menopause or give a biological rationale for the choice of a given definition of menopause.