Study had insufficient power to investigate safetyBMJ 2012; 345 doi: http://dx.doi.org/10.1136/bmj.e8146 (Published 03 December 2012) Cite this as: BMJ 2012;345:e8146
- Jacques E Rossouw, medical officer1,
- JoAnn E Manson, professor of medicine2,
- Andrew M Kaunitz, professor of obstetrics and gynecology3,
- Marcia L Stefanick, professor of medicine4
- 1National Heart, Lung, and Blood Institute, Bethesda, MD, USA
- 2Brigham and Women’s Hospital and Harvard Medical School, Boston, MA, USA
- 3University of Florida College of Medicine-Jacksonville, FL, USA
- 4Stanford University School of Medicine, Stanford, CA, USA
There are several concerns about the robustness of Schierbeck and colleagues’ findings.1 The open label design means that the trial could not control for placebo effect, and knowledge of active treatment could have led to differential ascertainment of outcomes.
The composite outcome of mortality, myocardial infarction, and heart failure was not prespecified in the design paper, which does, however, note the lack of statistical power for effects on breast cancer and coronary heart disease.2 We are not aware of an effect of hormone therapy on heart failure, which is a curious choice of outcome.
Our concern about outcomes ascertainment is heightened by the low event rates for myocardial infarction compared with those found in the Women’s Health Initiative (WHI) data for younger women.3 The low proportion of myocardial infarctions compared with strokes and heart failure, and the high number of cardiovascular deaths compared with cases, are unexpected. Point estimates from the WHI suggests that oestrogen plus progestin increases the risk of stroke, breast cancer, and venous thromboembolism compared with placebo and might decrease total mortality risk in younger women. Non-significant hazard ratios for myocardial infarction in WHI were 1.29 for women aged 50-59 years and 0.88 for women less than 10 years into the menopause. Hence, only this study’s findings for total mortality are consistent with the much larger WHI trial. Finally, combining results from the oestrogen plus progestin and oestrogen alone intervention arms is inappropriate, especially for outcomes such as breast cancer, for which WHI intervention effects were widely divergent.
This study had insufficient power to investigate the safety of menopausal hormone therapy. The findings add further questions about the study design and results.
Cite this as: BMJ 2012;345:e8146
Competing interests: None declared.