Many of the observational studies that prompted the Women’s Health Initiative (WHI) showed a mortality and cardiovascular benefit for hormone replacement therapy (HRT) in younger peri-menopausal women. However, these same studies showed HRT increased cardiovascular morbidity and mortality in older post-menopausal women. The WHI was nonetheless undertaken as a study of secondary prevention, examining the effect of HRT on post-menopausal women with a history of cardiovascular events. This study (Effect of hormone replacement therapy on cardiovascular events in recently postmenopausal women: randomised trial) was stopped due to the results of the WHI. It is not unlikely that the benefits reported here are less than they might have been with continued treatment due to this cessation.
HRT thus represents one of the first medical interventions where its effects as secondary prevention are diametrically opposed to its effects as primary prevention. The approach taken with WHI, where a medications effect in secondary prevention is taken as a proxy or measure of its effectiveness as primary prevention, is the standard approach taken with medical interventions. This approach is thought to increase the signal and thus to increase the power of medical intervention studies. This result, in confirming the benefit of HRT perimenopausally, not only impugns the study design for WHI but raises the concerning possibility that our approach to primary prevention requires a more careful understanding of observational data than is currently commonplace. That is, because the physiology in those post-event or with a confirmed disease may be subtly different from those without the disease, it may be that use of studies showing benefit in secondary prevention may not be appropriate to guide primary prevention. In the case of HRT, this study reveals a benefit for primary prevention that was hidden using a study of secondary prevention. In other interventions, it is possible that studies showing a benefit in secondary prevention belie harm from the same intervention in primary prevention.

Competing interests: None declared