All previous studies estimating risk of parity in relation to breast, ovarian and endometrial cancers have used nulliparous or uniparous as reference group with less than ten as the high parity group. Based on epidemiological findings in a large national Norwegian follow-up study covering the entire fertility range of 1-15 children the aim is to invert the risk estimation by introducing high parity as the reference group.
Material: The Norwegian 1960 Census was vital for the implementation of a national unique birth number for identification of all Norwegians. The information was collected by civil servants that visited all households. All married women were asked about number of children in the current marriage. This information was used in a linkage study with the Cancer Registry of Norway and information on deaths for women aged 45 years or more (postmenopausal) with follow-up from 1960 and out 2005. The linkage was performed within Statistics Norway. Analyses are based on logit modelling using Poisson distribution and Cox proportional hazard models.
Results: Number of women were 319 708 with 16,905 breast cancers, 3,827 ovarian and 3,834 endometrial cancers. High parous women, ten or more pregnancies, counted 5 049 women. Relative risk (RR) was estimated over the entire fertility range of 1-15 children separately for each cancer site and for each parity. The three cancer sites had almost identical risk reduction per child; breast RR=10.5% (95%CI; 9.6-11.4), ovarian cancer RR= 13.2% (95%CI; 11.2 –15.3) and endometrial cancer 10.9% (95%CI; 8.9-12.8) with the same curvilinear relationship.
Inverting the risk estimation with high parous women, 15 children, as reference group gave a RR of 4.9 for breast cancer, 7.1 for ovarian cancer and 5.0 for endometrial cancer. Age at first birth had only minor influence on the estimates, less than one child.
Discussion: We found consistent and similar reduction in risk of all three cancer sites in the order of 11% over the entire fertility range with uniparous as reference group. Inverting the risk estimation using the extended fertility range showed very strong relative risks for decreasing parity. This explains why low parity dominates the high incidence rates in low fertility countries. In Norway more than ninety-five percent of the children born in 2024 had birth number three or less. There is no selection bias in this national program. No people were lost to follow-up except for a small proportion emigrating. Parity is not associated with sex hormone levels. The relationship for each site is too strong to be explained by confounding, and the strong similarity between the curves makes confounding even less plausible.
Conclusion: Inverting the risk function with high parous women (15 children) as reference group gives novel understanding of the importance of parity as cause of breast, ovarian and endometrial cancers with particular relevance for low fertility countries.