From the challenge to the reassessment of the Women's Health Initiative: A personal initiative for women's health

Abstract

In the summer of 2002, a few weeks after publicationof the results from the Women’s Health Initiative(WHI) trial [1], we published an editorial warningagainst unchallenging acceptance of the conclusionsdrawn from this study [2]. Obviously we did notreject the data, which were obtained from the largestclinical trial ever conducted in this field, but in ourmind the data interpretation and the conclusionswere questionable. The WHI results on coronaryheart disease (CHD) being in conflict with previoussolid evidence [3–5], we suggested that the age of thepopulation included in the trial, the timing of initia-tion of hormone replacement therapy (HRT) and thetype and doses of the hormones used could explainthe apparently discordant findings.We stated that the WHI trial was conducted on apopulation with characteristics totally different fromthose of women typically considered for HRT andreported on in observational trials [2]. In fact, a vastproportion of the WHI women were elderly ladies,obese or overweight, diabetics and hypertensive, withelevated cholesterol levels requiring medication. Noweverybody accepts that the WHI trial was not con-ducted in ‘healthy postmenopausal women’ as repeat-edlystatedin the originalpaper [1]. Asa golden ruleofevidence-based medicine, the results of a given trialshould be referred and applied only to populationswith characteristics that are similar to the populationstudied. Thus, the ultimate message coming from this‘gold standard’ of research (the randomized clinicaltrial) was misleading, since the population was notideal to harvest results that ‘will help inform medicalpractice and guide millions of women in making criti-cal decisions about their health as they grow older’ [2].The authors stated that the WHI results have broadapplicability. In contrast, we underlined the limita-tions of age, health status of the population andtiming of HRT initiation [2]. Now new informationcoming both from the WHI [6] and the Nurses’Health Study (NHS) [7] clearly demonstrates thatthere were and are conflicts not in the numbers, but intheir interpretation. These new evaluations are clearlydemonstrating that the apparent contradictions arebasically founded only in the reading of the results.The biological plausibility of cardiovascular eventprevention with early HRT is now substantiated. Inthe estrogen-only arm of the WHI, there were fewerCHD events in the estrogen-treated women com-pared with women in the placebo arm, while therewere more events in women older than 70 years andin those who had been menopausal for more than 20years [6]. However, these differences did not reachstatistical significance due to the quantity of womenincluded. The bottom line is that the WHI study wasnot powered to investigate the consequences of HRTin early, healthy postmenopausal women, aged lessthan 60–65 years. Therefore, the extrapolation of theglobal WHI results to young perimenopausal womenis definitely incorrect and misleading, inducing thefalse idea that HRT in standard postmenopausalwomen may cause CHD harm [2]. Conversely,re-examination of the larger NHS database demon-strated that for women starting HRT when they areperimenopausal there was a significantly reduced riskof CHD (relative risk (RR)¼0.66, 95% confidenceinterval (CI) 0.54–0.80 for estrogen alone; RR¼0.72,95% CI 0.56–0.92 for estrogen with progestogen).Considering HRT-treated women with characteristicssimilar to those included in the WHI startinghormones 10 years after the menopause, in the NHSthere was for CHD neither benefit nor harm fromHRT (RR¼0.87, 95% CI 0.69–1.10 for estrogenalone;RR¼0.90, 95% CI 0.62–1.29 forestrogen withprogestogen).These recent results of the WHI and NHS are inline with the premise that estrogen may prevent CHD