This article is only available in the PDF format. Download the PDF to view the article, as well as its associated figures and tables.
—Drs Mendlovic and Kleerekoper suggest that we should report risk estimate for 15 years of estrogen use calculated on the basis of the US studies only. We did report an estimate for the proportional increase in risk per year for the 13 US studies (0.010, 95% confidence interval [CI], 0.002, 0.017; 15-year risk estimate, 1.2). We cannot conclude from this, however, that the increased risk is due to the European studies. Other factors, such as study quality or inclusion of premenopausal women, may have accounted for higher risk estimates. For example, when we excluded the European studies, we found that the mean proportional increase in risk for the highest-scoring tertile of the remaining studies was 0.024 (CI, 0.007, 0.040), which gave an RR of 1.4 after 15 years, while the two lower-scoring tertiles did not show increased risk.With regard to Dr Katerndahl's question about publication bias, we
Steinberg KK, Smith J, Thacker SB, Stroup DF. A Meta-analysis of Estrogen Replacement and Breast Cancer-Reply. JAMA. 1991;266(10):1359-1360. doi:10.1001/jama.1991.03470100050022