Grant Number:	5R01CA073703-02 Interpret this number
Primary Investigator:	Hsieh, Chung-Cheng
Organization:	Univ Of Massachusetts Med Sch Worcester
Project Title:	Validating a Model for Postpartum Breast Cancer Risk
Fiscal Year:	1999

A full-term pregnancy may have two opposing effects on breast cancer risk: an adverse one shortly after delivery and a beneficial one in the long run. We propose to study the postpartum time-varying breast cancer risk by comparing two adjacent parities at a time and examining current age as a modifier of the effect of age at delivery. Effect modification can be assessed by stratified analysis or modelling with interaction terms. For a case-control study with a moderate sample size, the modelling approach treats age and age at delivery as continuous variables under the assumption of linearity in logit of risk. Applying the model with interaction terms, a transient increase in breast concern risk lasting for about 15 years after a childbirth has recently been reported in a large population- based study. A linear model would fail to reveal any short-term excess in postpartum breast cancer risk that might peak several years after the delivery. A large database which links together the Swedish Fertility Registry and national Cancer Registry will be utilized to determine, using single-year indicator variables for age at delivery and stratifying by current age in one-year intervals, whether there is a postpartum period when the risk of breast cancer is the highest. We then evaluate whether the model with the linearity assumption yields effect estimates similar to those derived from the procedure using categorical variables. Other questions to be studied include whether the transient increase in risk varies significantly with age at the pregnancy, and whether any short- or long-term effect of a multiple birth on breast cancer risk differ from that of a singleton birth. Included in the analysis will be an estimated number of 34,000 breast concern cases that occurred in Sweden between 1958 and 1994 and 170,000 age-matched population controls, randomly selected from among those in the source population, alive at least to the date of diagnosis of the index case, and who had not been diagnosed with breast concern by that date.





None