Are the benefits of low-dose estrogens really proven?

Click to increase image sizeClick to decrease image size

Can we analyze the data more clearly by employing a different approach to estimate hot flushes?

Haines and colleagues[[1]] conducted a randomized clinical trial under the hypothesis that Dang Gui Buxue Tang would significantly reduce vasomotor symptoms compared with inert placebo capsules. The number of hot flushes, which were categorized by the symptom severity, was measured according to data from a daily hot flush diary. The authors reported that the number of hot flushes decreased in both the Dang Gui Buxue Tang and placebo groups without statistical significance, except mild hot flushes in the treatment group and moderate hot flushes in the placebo group. However, the analysis methodology employed in this trial makes the results unclear. Would it be justifiable to say that the intervention reduced mild hot flushes? Here is a different approach which might draw quite a different interpretation compared to that of this trial.

We can hypothesize that participants can report mild, moderate, and severe hot flushes according to their severity of hot flushes during 1 day[[2]]. Two possibilities could occur. First, there could be a certain participant who reported six mild hot flushes and two severe hot flushes daily at baseline, and scored 1 for mild hot flush and 5 for severe hot flushes daily after 6 months. In terms of hot flush scores, which are generally calculated by multiplying frequency and severity of the hot flushes (mild: 1, moderate: 2, severe: 3)[[2]],[[3]], the scores increased from 12 to 16. This implies aggravation of symptoms. However, the authors might report a reduction in mild hot flushes without considering the overall deterioration of hot flushes.

Second, there could be another participant who complained of two mild hot flushes and four severe hot flushes daily at baseline, and scored 5 for mild hot flushes and none for severe hot flushes daily after 6 months. In this case, the hot flush score greatly decreases from 14 to 5. However, according to the analysis of this trial, mild symptoms become worse despite the overall improvement in hot flushes. Given these interpretations, focusing on only the number of flushes, categorized by severity, might produce misunderstandings about the clinical implications. The authors should estimate hot flush scores for changes both in frequency and in severity rather than just frequency, as in this trial.

Another point concerns the possible placebo effects of the intervention. The authors mentioned that a high placebo response was usual in clinical trials of menopausal symptoms. However, they did not discuss the placebo effects in any detail. First, the authors reported the daily total dose of the intervention without mentioning the frequency. According to recent studies, the size of the placebo effect tends to rely on the frequency of administration (de Craen, cited in Ernst[[4]]). Second, there is no information about whether non-hormonal agents and over-the-counter drugs are permitted for the purpose of alleviating menopausal symptoms or not. The placebo effects of concomitant treatment, therefore, could not be ruled out. Third, it is uncertain whether the patient–treatment provider relationship is controlled for or not at each visit. Recent trials have suggested that the patient–practitioner relationship might be the most robust component of placebo effects[[5]]. Finally, if there are ‘no treatment group’ or ‘pre-treatment waiting period’ as control groups, it can exclude the effect of regression to the mean or the natural history of symptoms in both of the intervention groups. Larger placebo effects tend to be in subjective symptoms like hot (flushes) rather than objective measures of disease (van Leeuwen, cited in Ernst[[4]]). When attempting to compare the effects of treatment intervention with placebo, it is important to adequately control for, and therefore minimize, the factors which contribute to placebo effects, allowing for clearer data analysis.

Department of MedicalKun Hyung Kim ResearchMyeong Soo LeeKorea InstituteSun-Mi Choi of Oriental Medicine Daejeon, Korea

Acknowledgement

The authors specially thank Kate Boddy in Peninsula Medical School, Universities of Exeter & Plymouth, Exeter, UK for editing this manuscript.

Conflict of interest Nil.

Source of funding The authors were supported by the Acupuncture, Moxibustion and Meridian Research Project (K08010) of Korea Institute of Oriental Medicine.