ABSTRACT
BACKGROUND: Breast cancer is the most common malignant tumor in females. Many studies have been carried out in order to assess the reproductive risk factors. Particular attention has focused on information regarding fertility, including breastfeeding, age at first birth and number of live births. These factors are highly correlated with each other. The objective of this study was to employ latent variables to reduce the confounding effect of this correlation with a logistic regression analysis. METHODS: The investigation drew upon results from a dataset belonged to a hospital based case-control study covering 303 breast cancer patients and 303 hospital controls. Data were collected through interview and reproductive variables included age at first full-term pregnancy and live birth, number of pregnancies and live births, and total length of breast feeding. Latent variables were generated using factor analysis and principal components analysis. RESULTS: The study revealed that for both latent variable approaches the odds ratios of two latent variables significantly indicated a protective impact of number of pregnancy and live birth and breastfeeding and a prognostic relation with age at first pregnancy or live birth. CONCLUSION: The findings suggest that breastfeeding and decreasing age at first live birth have protective influences on breast cancer risk. Also using statistical model with latent variables in the presence of collinear data leads to reliable results.