Abstract:
|
Family history is an important risk factor for breast cancer. In statistical modeling, family history is usually treated as a yes/no binary variable or a categorical variable of 0/1/1+, based on first-degree relatives. However, simply using family history as a two or three-level categorical variable may reduce its value as a breast cancer risk predictor. We proposed a family history score that takes number of diseased first-degree relatives, family composition, and both their current ages and ages at diagnosis into consideration through a Bayesian approach.We have successfully shown a high agreement between the observed and the predicted 5-year risk in the Sister Study data. When using the proposed score in the Gail model, the model has a higher predictive power. Our score has the potential to both help identify women at high risk who may warrant more frequent screening and help women with low risk to avoid un-necessary mammogram expos
|