Background
Evidence of mammography screening primarily came from Western countries, while there remains an unmet need for a more efficient and tailored screening strategy.
Purpose
Risk factors for breast cancer development were identified from big data analyses.
Methods
Using a unique ID, we identified subjects attending screening at least once between Jan 2007 and Sep 2014, matching the cancer registry concurrently with a two-year’s extension of the screening interval to Aug 2016.
Results
Multi-variate analyses identified family history of cancer, the number of affected sisters, years of hormone replacement, breast symptoms, breast examinations within two years, previous breast surgery, educational level, and breast composition as risk factors for breast cancer diagnosis, while menopausal status, breast feeding, sonography within two years, compared with previous mammography, times of screening mammography, and served with a mobile mammography van were protective. The model showed an area under the receiver operating characteristic curve of 0.6766. Screening-detected cases were associated with an earlier disease stage, while clinically detected breast cancer remained an independent risk factor for relapse-free and overall survival.
Conclusion
Using big data analysis for risk model construction, several risk factors for Taiwanese breast cancer development were identified, and the efficacy of mammography screening was ascertained for Taiwanese women.
Impact:
Further studies incorporating genetic data may augment the predictive power substantially and pave the way for personalized screening.