To the Editor.
—In 1991, THE JOURNAL published1 a meta-analysis on the effects of estrogen replacement therapy on the risk of breast cancer. Data for the analysis were extracted from 18 published case-control studies. Six of these reported relative risks (RRs) associated with ever-use of estrogen replacement therapy for women with and without a family history of breast cancer. Results from these studies were combined and an RR of 3.4 (95% confidence interval [CI], 2.0 to 6.0) was reported in women with a family history, which was significantly greater than in women without (RR, 1.5; CI, 1.2 to 1.7). These results raised a question as to whether estrogen replacement therapy should be given to women with a family history of breast cancer. As this has implications for our local research interests, the methods of the meta-analysis and reported results were examined.The method of calculating summary RRs used for