In Defense of Pharmacoepidemiology — Embracing the Yin and Yang of Drug Research

The past decade has not been kind to observational studies of medications. The damage began in 1998 with the publication of the Heart and Estrogen–Progestin Replacement Study, a randomized controlled trial showing that hormone replacement increased the risk of cardiac events among postmenopausal women with heart disease. Like many physicians, I had been teaching the gospel that estrogen use prevented heart disease — an idea based on observational studies1 showing that postmenopausal women who regularly took estrogen were less likely to have heart disease than apparently similar women who did not take hormones. It now appeared that this had been . . .