Determinants of breast cancer involving hematologic indices, age, and geographic residency: Insights from comparative multivariable logistic regression
نویسندگان
doi
10.48309/jmpcr.2026.561878.1914چکیده
Breast cancer remains the most common malignancy among women, with rising incidence in Southeast Asia. Evidence linking demographic, hematologic, and geodemographic factors to breast cancer risk is still limited. This study examined these associations and evaluated their predictive contributions using multiple logistic regression models. A case–control study was conducted involving women with benign and malignant breast tumors. The variables included age, hemoglobin, leukocytes, platelets, PLR, TILs, urban–rural residence, population density, and agricultural land proportion. Three alternative multivariable logistic regression models were constructed, and model performance was evaluated using the Hosmer–Lemeshow test, Nagelkerke R², overall accuracy, and AUC. Model 1 demonstrated the strongest discriminative performance (AUC = 0.903; accuracy = 86.7%), with age (OR = 35.4) and PLR (OR = 1.241) emerging as significant predictors. In Model 2, hemoglobin (OR = 0.719), TILs (OR = 2.969), and rural residence (OR = 5.516) were significant (AUC = 0.709). Model 3 yielded similar findings, with hemoglobin (OR = 0.730) and rural residence (OR = 4.537) remaining significant (AUC = 0.668). Across all models, age consistently showed the strongest association with breast cancer. Overall, age was the most robust predictor; PLR exhibited a positive association, and hemoglobin showed an inverse association. TILs and rural residency added predictive value in select models. Model 1 offered the best overall performance, indicating that combined hematologic and geodemographic variables may support risk stratification, though external validation is required before broader application .