Bergkvist L, Tabàr L, Adami H O, Persson I, Bergström R
Department of Surgery, University Hospital, Uppsala, Sweden.
Am J Epidemiol. 1989 Sep;130(3):503-10. doi: 10.1093/oxfordjournals.aje.a115364.
Two population-based registers, one of women with at least one estrogen prescription, and the other of women who had undergone mammographic screening, were linked by means of the national registration number, revealing 3,436 estrogen-treated women who had attended the mammography screening program. Their mammographic parenchymal pattern was classified according to the criteria of Wolfe into two groups, one with a low risk and the other with a high risk of development of breast cancer. Estrogen replacement therapy was more common in women with a high risk than in those with a low risk mammographic parenchymal pattern (standardized morbidity ratio = 1.3). The increased probability of receiving noncontraceptive estrogen treatment in women with a high risk mammographic parenchymal pattern remained when the impact of several possible confounding factors had been taken into account in a logistic regression analysis (odds ratio = 1.36, 95% confidence interval 1.24-1.49). No evidence was found of a shift toward the unfavorable mammographic parenchymal pattern during treatment, which suggests that the treatment itself was not responsible for the high proportion of P2 and DY patterns. The mammographic parenchymal pattern must therefore be looked upon as a confounder in the context of studying breast cancer risk after estrogen treatment.
