The curvilinear relationships between grand parity and incidence of hormone-dependent cancers; follow-up of postmenopausal women in the Norwegian 1960 Census

Breast, ovarian and endometrial cancers are named hormone dependent cancers (HDC) because of the effects of endogenous hormones, including parity, on the incidence rates. Here, we will test the hypothesis that each additional child has the same relative preventive effect on the risk of each of the three cancer sites and with similar shapes of the incidence curves over the extended exposure range, 1 - 18 children in postmenopausal women. The study is based on parity information from the Norwegian 1960 Census for women aged 45-89 years. A total of 385 816 married women answered the question of number of children in present marriage to a civil servant. Follow-up continued until the first of any of the HDC diagnosis, death, or end of follow-up 2005 through linkages based on the unique Norwegian birth number. Included were 16 905 breast cancers, 3 827 ovarian cancers and 3 834 endometrial cancers. Based on person-years (PY), the percentage change in incidence rates of the three cancer sites for each additional child was calculated using a logit regression model including models with higher order terms. A new statistical method for analyses of collinearity between parity and age at first birth has been developed. Age at marriage was used as a proxy for age at first birth. Parity had strong linear effects on the incidence rates for all three cancer sites (p