In the recently published analysis of the effects of estrogen plus progestin on health-related quality of life (QOL) in the Women’s Health Initiative (WHI), the authors concluded that this therapy did not have a clinically meaningful effect.1 However, caution is warranted in interpreting the findings of this analysis. The study design has several limitations that diminish the strength and usefulness of its conclusions.
The main finding — that QOL was not significantly improved — was to be expected in this population. Symptomatic women were discouraged from participation, and vasomotor symptoms were reported by only 12% of the participants. Furthermore, it is unlikely that many women who had bothersome menopausal symptoms would volunteer for a study in which they had a 50% chance of receiving placebo for a prolonged time. Symptom relief is the most relevant factor for improvement of QOL with hormone therapy; therefore, women without symptoms are not likely to experience large improvements in their QOL. Nevertheless, even this older population showed improvements in sleep, physical functioning, and pain. Thus, it is possible that there was an effect on the QOL, but that the study was not designed to capture this information.
The more important question of effect on QOL in symptomatic women was addressed by a subgroup analysis of 574 women ages 50 to 54 years who reported moderate to severe vasomotor symptoms at baseline. Women in the estrogen and progestin group demonstrated a significant improvement of vasomotor symptoms (p < 0.001) and sleep quality (p=0.02), as compared to placebo. Despite this, improvement in QOL was not observed. This contradicts previous well-designed prospective trials.2 There are several limitations in this study’s design that may explain the failure of this analysis to detect an improvement in QOL. First, the primary objectives of the WHI did not include evaluation of the effect on QOL. The study, and particularly the subgroup, may not have had the power to detect a meaningful effect.
Second, the main instrument used to evaluate QOL was the RAND-36.1 Although a good measure of health-related functioning, it was not designed to assess variables relevant to QOL in postmenopausal women.3 For example, important domains, such as sexual function, are not addressed. The authors attempted to evaluate this variable by asking a single question with a four-point response scale: very unsatisfied, a little unsatisfied, somewhat satisfied, and very satisfied. This approach is clearly inadequate as a method for assessing change in such a complex aspect of human function. Nevertheless, the data in this subgroup revealed an improvement in the estrogen and progestin group that had borderline statistical significance (p=0.06). Thus, the RAND-36 was not designed to assess the “sense of well-being,” particularly in the postmenopausal population. Validated instruments such as the Utian Quality of Life Scale (UQOL) are available for this purpose, and should be utilized in future studies.3
Another weakness of this study was the high rate of drop-ins. Many women randomized to placebo initiated hormones. Symptomatic women were most likely to do this, which may have introduced bias toward the null hypothesis, thereby decreasing the likelihood of detecting an improvement in QOL.
In summary, it is incorrect to conclude, on the basis of this study, that estrogen plus progestin therapy does not improve QOL in women with vasomotor symptoms. Previous randomized controlled trials have provided evidence to the contrary. Furthermore, even in asymptomatic women this conclusion is questionable, as the study was not designed to evaluate QOL in postmenopausal women.