Age at menopause and breast cancer: estimation of floating absolute risks

Late menopause is consistently related to breast cancer risk, but direct comparison of the modifying effects of type of menopause and of time factors across studies has been made difficult by the heterogeneity in the choice of the reference category, and in the definition of postmenopausal status. These effects were investigated in the absence of an arbitrary baseline category by computing floating absolute risks (FARs). Pooled data from two case-control studies conducted between 1983 and 1994 in major teaching and general hospitals of six Italian centres were used. Participants were 3576 postmenopausal women with incident, histologically confirmed breast cancer, and 3578 postmenopausal controls admitted to hospital for acute, non-neoplastic, non-hormonal, non-gynaecological conditions. When all types of menopause were considered together, FARs were 0.49 for <35 years, 0.81 for 35–39 years, 0.82 for 40–44 years, 0.88 for 45–47 years, 1.02 for 48–50 years, 1.23 for 51–53 years and 1.24 for 54–56 years, with a significant linear trend. The same pattern was seen in women reporting a natural menopause, with FAR of 0.14 for women with menopause when aged <35 years versus 1.20 for those with menopause at 54–56 years (ratio between extreme FAR ESTIMATES = 8.6). No trend emerged in the overall surgical menopause group (including hysterectomy alone or with monolateral oophorectomy, and bilateral oophorectomy). However, when women reporting a bilateral oophorectomy were selected, a significant linear trend was observed. No heterogeneity emerged when risks were evaluated in strata of age at diagnosis/interview. Thus, using a unique large dataset and an innovative method of analysis, this study documented an over 8-fold ratio of risks between extreme categories of age at menopause, but no evidence of a latency period before the establishment of an effect of age at menopause.