Population attributable risk of a competing-risk model for breast cancer and non-breast cancer death among women ≥ 65 years

Purpose To inform decision making around mammography-screening frequency and cessation, we previously used Fine-Gray competing-risk regression to develop and validate a model to estimate older women's 10-year risk of breast cancer and their competing risk of non-breast cancer (non-BC) death. Here, we aimed to understand the amount of incident breast cancer and non-BC death risk explained by our model among women >= 65y. Methods We included women >= 65y who completed the 2004 Nurses' Health Study questionnaire (NHS, n = 59,662) or who participated in the Women's Health Initiative-Extension Study (WHI-ES, n = 82,528). We calculated our model's full and risk factor-specific population attributable risk (PAR%) for incident breast cancer and non-BC death. Results Mean age of the NHS participants was 73.5y (SD 5.2); 3.1% were diagnosed with breast cancer and 26.1% experienced non-BC death within 10 years. Mean age of WHI-ES participants was 73.6y (SD 5.4); 4.2% were diagnosed with breast cancer and 17.7% experienced non-BC death within 10 years. The full-model PAR% for breast cancer was 58.8% (22.7-80.6) in NHS and 54.8% (24.8-75.2%) in WHI-ES. Modifiable risk factors explained approximately 1/3 of breast cancer risk; BMI >= 30 had a PAR% of 6.5% (3.1-9.9%) in NHS and 12.2% (8.5-16.0%) in WHI-ES. For non-BC death, the full-model PAR% was 94.2% (91.4-96.1%) in NHS and 86.2% (80.9-90.0%) in WHI-ES. Conclusions Our competing-risk model explained the majority of breast cancers and non-BC deaths in women >= 65y, and we identified risk factors (e.g., elevated BMI) that may be targeted to reduce the burden of breast cancer in older women.