Cardioprotective effect of hormone replacement therapy Is not due to a selection bias

BMJ 1994; 309 doi: https://doi.org/10.1136/bmj.309.6957.808b (Published 24 September 1994) Cite this as: BMJ 1994;309:808
  1. M Stampfer,
  2. F Grodstein
  1. Harvard School of Public Health, Boston, MA, USA Channing Laboratory, Department of Medicine, Brigham and Women's Hospital, Harvard Medical School, Boston, MA 02115, USA
  2. E Merck Pharmaceuticals, Alton, Hampshire GU34 1HG
  3. Department of Clinical Epidemiology, Leiden University Hospital, PO Box 9600, 2300 RC Leiden, Netherlands.

    EDITOR, - Ward F M Posthuma and colleagues have shown that many of the studies that report reductions in the risk of coronary disease among postmenopausal women using hormones also show decreases in cancer.1 They claim that this implausible result shows a selection bias toward healthier women for hormone treatment and that the benefits in terms of heart disease must be considered to be suspect. In the exchange of correspondence they do not address the substantive arguments raised by Stevenson and Baum but instead accuse them of showing “ignorance of common medical reasoning.”2, 3 Posthuma and colleagues, however, seem themselves to have ignored some common medical reasoning.

    In their paper they talk about “total cancer within each study,” but in many instances the studies were reporting the mortality from cancer rather than the incidence of cancer. This is an important distinction because, typically, patients who die of cancer have had the disease diagnosed previously. It is well known that many physicians refrain from giving hormones to postmenopausal women who have already been diagnosed as having cancer. Hence studies of mortality from cancer will naturally tend to show a spurious protective association of postmenopausal hormones unless women with cancer at the baseline have been excluded. This spurious association has nothing to do with the selection of women according to their risk of coronary disease.

    A direct and more relevant approach would have been to assess the distribution of risk factors for …

    View Full Text

    Log in

    Log in through your institution


    * For online subscription