Abstract
Confounding in epidemiologic studies occurs when there is a distortion in the estimate of the disease-exposure relationship due to the mixing of the true effect with that of a third factor. The traditional definition of a confounder states that a true confounder must have three characteristics: 1) it must be associated with the exposure; 2) it must be associated with the disease independent of its effect on exposure; and 3) it should not be an intervening variable on the causal pathway between exposure and disease. This investigation examined one criterion for confounding in a data set which was used to analyze the association between maternal serum alpha fetoprotein (MSAFP) and breast cancer. The criterion examined in this analysis is the relationship between the potential confounders and the exposure. The potential confounders of interest were the known predictors of breast cancer which included age, race, body mass index, age at first full-term pregnancy, age at menarche, alcohol intake and number of previous pregnancies. Evidence of an association with maternal serum alpha fetoprotein was found to exist for race, body mass index, number of previous pregnancies and age. The study findings suggest that these four factors have the potential to act as confounders in the MSAFP-breast cancer relationship. The absence of an association between MSAFP and alcohol intake, age at first full-term pregnancy and age at menarche indicates that these variables do not lead to confounding because each factor fails to meet the criterion of association with the exposure.
David, Jennifer Louise (1995). Initial identification of confounding factors in an epidemiologic study: confounding in a case-control study of maternal serum alpha fetoprotein and breast cancer. Master's thesis, Texas A&M University. Available electronically from
https : / /hdl .handle .net /1969 .1 /ETD -TAMU -1995 -THESIS -D385.