Letter to the Editor

To the Editor: We are writing in reference to the article “Clinical Effects of a Standardized Soy Extract in Postmenopausal Women: A Pilot Study,” by Scambia et al. (Menopause 2000;7(2):105–111). We applaud the authors for trying to improve the available evidence on phytoestrogens by conducting this randomized, double-blinded trial. However, we are concerned that the authors have drawn conclusions that are not supported by their data. They conclude that “on the basis of the results obtained in this pilot short-term trial, this standardized soy extract (SOYSELECT) has a good compliance and safety and has an action on vasomotor symptoms, decreasing hot flushes and night sweats, without estrogenic effects on the reproductive tract.” In regard to reduction of hot flashes, we noted that the treatment group started with a mean of 33 hot flushes per week; the placebo group's mean was 27. The reduction in hot flash frequency, and the mean hot flash frequency at 6 weeks, is not given. We infer from Fig. 1 that the reductions were 45% for the treatment arm and 25% for the placebo, reflecting absolute means after 6 weeks of 18.15 and 20.75 hot flushes per week, respectively. We are left to wonder whether these two endpoints of 18 and 20 are actually different, because the standard deviations are not provided. Since the treatment group started with a higher number of hot flushes per week, it is not clear whether the greater reduction in hot flashes in this group reflects the treatment effect or illustrates regression to the mean. We also question the clinical importance of the difference between 18 and 20 hot flushes per week. In regard to the safety claims made about the lack of estrogenic effects on the endometrium, the authors have insufficient evidence to comment on long-term usage of this soy supplement. The rate of endometrial hyperplasia with unopposed estrogen is about 20% at 1 year;(1) 63% at 3 years. (2) Extrapolating from this data, we would expect the rate of endometrial hyperplasia at 6 weeks on standard dose estrogen to be 2.3%. This study had 12% power to detect a difference of this magnitude, assuming all subjects underwent the complete follow-up. In fact, 8 of the 20 subjects in the soy group dropped out, and their data was not included for analysis, lowering the already weak power of the study. To obtain 80% power to detect important changes in endometrial response in a 6-week time frame, this study would have needed 470 patients in each arm. Clearly we need larger and longer studies, with adequate power to evaluate long-term outcomes, such as endometrial hyperplasia and breast cancer, and pilot studies are a good start. Until then, we cannot tell our patients that long-term use of phytoestrogens is without risk. As our patients search for safer, simpler answers, we must resist the temptation to overlook the limitations of the current evidence. We cannot allow the popular demand for natural products to compromise our scientific rigor. Melanie Swift MD Jule West MD