Analyses of factors associated with breast cancer and construction of a risk stratification model among adult women in suburban Shanghai

Objective To examine factors associated with breast cancer among adult women in suburban Shanghai and develop a risk stratification model, providing a basis for the identification of high-risk populations. Methods A cross-sectional study was conducted using baseline data from the Shanghai Suburban Adult Cohort and Biobank. A total of 39683 women aged 20-74 years old were included, comprising 249 breast cancer cases and 39434 nonbreast cancer cases. Information for demographic characteristics, reproductive health, lifestyle, chronic diseases, and psychological status was collected. Missing data were handled using multiple imputation. Multivariable Firth logistic regression analyses were used to identify factors associated with breast cancer, and restricted cubic spline analyses were performed to assess the dose-response relationship between hemoglobin A1c and breast cancer. With breast cancer as the outcome, the data were randomly split into a training set and a test set at a ratio of 7∶ 3. A logistic regression model was developed for risk stratification and further calibrated using Platt scaling. Model performance was evaluated using the area under the receiver operating characteristic curve (ROC-AUC), area under the precisionrecall curve (PR-AUC), Brier score, calibration intercept, and calibration slope. Given that breast cancer was a lowincidence outcome, the model’s precision, recall, and lift for the top 5% and top 10% of high-risk individuals were further calculated, so as to evaluate the model’s ability to identify and stratify risk in high-risk screening scenarios. Results In the fully adjusted model, increasing age was associated with a higher risk of breast cancer (OR=1.050, 95%CI: 1.033-1.067), whereas older age at menarche (OR=0.921, 95%CI: 0.859-0.987) and a greater number of pregnancies (OR=0.875, 95%CI: 0.770-0.986) were associated with a lower risk of breast cancer. Women with a history of estrogen use had a higher risk of breast cancer than those without such a history (OR=3.098, 95%CI: 1.481-5.728). Hyperglycemia was associated with an increased risk of breast cancer (OR=1.754, 95%CI: 1.305-2.334). Compared with women without anxiety or depression, those with mild (OR=2.239, 95% CI: 1.484-3.266) and severe anxiety or depression(OR=10.104, 95% CI: 1.106-32.798) had a higher risk of breast cancer. Restricted cubic spline analyses showed a nonlinear association between hemoglobin A1c and breast cancer (P=0.003). The logistic regression-based risk stratification model showed moderate discriminative ability (ROC-AUC=0.730, 95%CI: 0.673-0.784) and a low overall prediction error (Brier score=0.006, 95%CI: 0.006-0.006). The lift values were 4.261 for the top 5% high-risk group and 3.199 for the top 10% high-risk group, identifying 21.3% and 32.0% of breast cancer cases, respectively. Conclusion Breast cancer among adult women in suburban Shanghai was associated with older age, earlier age at menarche, fewer pregnancies, history of estrogen use, hyperglycemia, and mild and severe anxiety or depression, suggesting that greater attention should be paid to reproductive hormonerelated factors, metabolic abnormalities, and psychological health in community-based women’s health management. The logistic regression-based model using routine epidemiological indicators showed moderate discriminative ability and potential for enriching high-risk populations and may help support initial screening and stratified management of women at high risk of breast cancer in resource-limited community settings.