Hypnosis and hot flashes: When will they ever learn?

ResearchBlogging.orgFor women undergoing menopause, hot flashes are a real problem. In my specialty, as I’ve pointed out before, women undergoing treatment for breast cancer are often forced into premature menopause by the treatments to which we subject them. It can be chemotherapy, although far more often it’s the estrogen-blocking drugs that we use to treat breast cancers that have the estrogen receptor. Estrogen stimulates such tumors to grow, and blocking estrogen is a very effective treatment for them, be it with tamoxifen or the newer aromatase inhibors like Arimidex. The utterly predictable consequence, unfortunately, is an artificially-induced menopause.

I’ve written at least twice before about this topic in the context of various poorly designed studies of acupuncture for breast cancer-induced hot flashes. There’s a reason for this. Despite studies demonstrating that hormone replacement therapy doesn’t decrease cardiovascular disease in postmenopausal women and increased the risk of breast cancer, for severe menopausal symptoms in women without breast cancer, estrogen remains the gold standard, and it’s reasonably safe to use for short periods of time. Consequently, for menopause having nothing to do with breast cancer, estrogen can be used, at least for the short term, if nonhormonal therapies don’t work. Not so in the case of women rendered menopausal by breast cancer therapy. Indeed, it defeats the purpose of antiestrogen drugs to replace the estrogen they are blocking. Not only that, but even after breast cancer therapy when a woman undergos menopause naturally, estrogen replacement increases their risk of a recurrence. Consequently, if nonhormonal methods supported by science don’t work, then there’s nothing else, and, unfortunately, most science-based nonhormonal therapies such as antidepressants do not work very well and have significant side effects.

That’s where the temptation to turn to woo comes in.

And there is a veritable panoply of woo to choose from. There are any number of herbal remedies, for example. Not surprisingly, many of the herbs that actually may work contain estrogenic compounds. Indeed, soy phytoestrogens are commonly used to relieve menopausal symptoms, as I alluded to, although did not discuss in detail, recently. Then there’s acupuncture, which does not appear to work better than placebo, but that doesn’t stop poorly designed study after poorly designed study from being done, even after better designed studies have already shown that acupuncture is no better than placebo for hot flashes. In other words, hot flashes during and after breast cancer therapy are the perfect condition for the woo-meisters to exploit. It’s also a fertile area for so-called “mind-body” modalities, some of which are woo, some of which may not be.

I haven’t made my mind up about hypnosis, but that’s the latest therapy that’s being tested for hot flashes:

NEW YORK (Reuters Health) – Hypnosis can help reduce hot flashes among breast cancer survivors, new research published in the Journal of Clinical Oncology shows.

The authors of the report note that “hot flashes are a significant problem for many breast cancer survivors.” The new findings are particularly important because the current best treatment for hot flashes, estrogen therapy, is off limits for most women who have had breast cancer.

Furthermore, many women must take estrogen-blocking drugs like tamoxifen for years after breast cancer treatment, but “hot flashes can be so severe that some women make a decision to not continue those medications,” Dr. Gary Elkins told Reuters Health.

Based on some small studies that found hypnosis benefited women suffering from hot flashes, Elkins of Baylor University in Waco, Texas, and his team randomly assigned 60 breast cancer survivors to hypnosis once a week for five weeks or no treatment.

The hypnosis sessions, which lasted about 50 minutes, involved helping the patient to reach a deeply relaxed state, and then offering suggestions for mental imagery to help her relax and feel cool. This could mean having a woman imagine herself walking down a cool mountain path, for example. Women also received instructions on how to practice hypnosis on their own.

Among the 51 women who completed the study, those who had hypnosis reported a 68% reduction in the severity and frequency of their hot flashes. This translated to 4.39 fewer hot flashes a day, on average, for women in the hypnosis group, while there was little change in the control group.

You know what it means when I see a news report like this, don’t you? It’s time to head to the study and check it out, straight from the horse’s mouth, so to speak. (Or, I could say, “Let’s go to the tape.”) The study is entitled Randomized Trial of a Hypnosis Intervention for Treatment of Hot Flashes Among Breast Cancer Survivors and comes out of Baylor University and Johns Hopkins University, headed up by Dr. Gary Elkins.

The design of the study was very simple. As mentioned above, women with breast cancer therapy-induced hot flashes were randomized to either no treatment of sessions of hypnotherapy. The specific hypnotherapy was described as follows:

Patients in the hypnosis intervention condition were scheduled for five weekly sessions, each to last approximately 50 minutes. At each session, a hypnotic induction was completed according to a standard transcript…Hypnotic suggestions for each session included the following: hypnotic induction; mental imagery and suggestions for relaxation; mental imagery for coolness; deepening hypnosis and dissociation from hot flashes; positive suggestions and imagery for the future; self-hypnosis; and the alert, “In a few moments, return to conscious alertness.”

In addition, participants were given instruction in self-hypnosis practice and were provided with an audiocassette tape recording of a hypnotic induction and of instructed in-home practice. Although the hypnotic induction followed a transcript, specific imagery for relaxation and imagery for coolness were individualized on the basis of each patient’s preference regarding such imagery.

As far as hot flash severity, investigators used hot flash severity score, the HFRDIS, the HADS anxiety sub-scale, the CES-D depression scale, and the MOS-Sleep Scale. You don’t actually need to know all that much about these tests (I don’t know their details) other than that investigators estimated hot flashes based on patient self-reporting. There are pros and cons to this. The pros are that all that patients really care about is the subjective feeling that hot flashes produce; so one could argue that objective measures of hot flash severity are unnecessary. On the other hand, if a treatment is really having an effect at a physiologic level some sort of objective measurement is highly desirable.

Be that as it may, the study is what it is, and what it is is highly, highly flawed.

I bet regular readers of this blog can guess what the problem is. It’s so glaringly obvious (again) that I’m surprised a journal as prestigious as JCO published this articel. Have you guessed yet? Of course you have! That’s right, there is no placebo control group. It’s no treatment versus hypnosis. Remember yet again, as I’ve discussed before, hot flashes are highly subject to the placebo effect. Indeed, placebo effects as high as 40% have been documented in previous studies.

Of course, the authors dance around this by saying that they saw a reduction in the hot flash severity score of 68%, which, they argue, must be greater than placebo. That’s not entirely a fallacious argument, but pretty close. It might have had validity if the number of subjects in this study were much, much larger, but at only 30 subjects per group for a total of 60 patients, of whom nine either dropped out or were lost to followup. That’s 15%. Moreover, only three dropped out of the treatment group, while six dropped out of the control group. Either way, in groups of 27 and 24, it wouldn’t take much to skew the results. Granted, such a seemingly large effect may be real, but it’s a stretch to assume that, just because it’s larger than a 50% reduction it must be larger than placebo. Using historical controls or assumptions such as that have lead more than one investigator down the garden path, so to speak, into self-deception. Heck, using such assumptions is how Nicholas Gonzalez used to justify his “Gonzalez protocol” of supplements and coffee enemas to treat pancreatic cancer. Studies such as the study under discussion require three groups ideally: no treatment control, placebo control, and hypnotherapy.

This leads me to the question of what, exactly, would serve as a good placebo against which to test hypnotherapy. While it it probably difficult, if not impossible, to double blind the study so that the therapists do not know which subjects are in which group, there are ways around that; for example, separate therapists for administering the hypnotherapy and for evaluating the subjects’ response. But what, specifically, would be the placebo? One potential type of placebo control is known as a “contact group,” where subjects receive teh same amount of contact time as the treatment group. Another possibility that I’ve heard discussed is to do all the same things in the “treatment” hypnotherapy but to use imagery that has nothing to do with the patient’s condition. In other words, no imagery of coolness would be used; an irrelevant image would be used, described as follows:

A therapeutic hypnotic intervention in general consists of the induction of the hypnotic state or trance, trance deepening, therapeutic suggestions, and finally alerting.11 To produce a placebo pill for hypnosis we should know at first that what is (are) the actual ingredient(s) of a therapeutic hypnotic intervention. At a glance, it seems that therapeutic suggestions are the actual ingredients. Indeed, disease specific hypnotic suggestions are used with the aim of targeting special issues of the disease. For example, gut-directed suggestions which are used for the treatment of IBS are all about the digestive system.12 Although induction procedure (when labeled hypnosis) and trance deepening can increase the response to suggestions,7,13 it is therapeutic (specific) suggestions that have the unique effects of hypnotherapy.14–16 In this way, one can put therapeutic suggestions into the active but not placebo pills and coat them with induction procedure, trance deepening, and termination and then labeled them as ”hypnotherapy”. This method of hypnosis, without specific suggestions is also called ”neutral hypnosis” which has been used in some experimental trials and showed to be a suitable placebo control condition.17–19 By ”neutral hypnosis” with the only trance deepening suggestion, participants reported alterations in body image, time sense, perception and meaning, sense of being in an altered state of awareness, affect, attention, and imagery,20 none of which seems to have therapeutic effect.

Shockingly, the above passage was found in the woo-ful journal Complementary Therapies in Medicine.

Studies like these are not necessarily bad science if they are treated as nothing more than highly tentative pilot studies. They are not, however, particularly good science either. After all, not even an attempt at a decent control group was made. True, the authors in essence promise to do better next time by addressing these complaints, but I still find it disturbing that a study ridden with so many flaws found its way into one of the most prestigious cancer journals out there. I find it even more disturbing that the investigators do not appear to be planning a study that would really answer the question:

In fact, Elkins said, he and his colleagues are now launching a National Institutes of Health-funded study to address this issue, which will enroll 180 postmenopausal women and will compare hypnosis to another type of mind-body intervention.

The mechanism behind hot flashes is still poorly understood, Elkins noted. “We know that they are related to decreases in estrogen, however that relationship is not direct in the sense that hot flashes lessen over time even though estrogen levels remain low,” he explained.

Hot weather, spicy food and stress can also trigger hot flashes, he added, so it’s possible that women undergoing menopause may have a more difficult time regulating their body temperature in response to these triggers. Hypnosis treatment can reduce stress by helping women to relax, Elkins explained, and may also give them a sense of control that allows them to keep their body temperature more stable.

Of course, there’s no evidence that any of this “helps women to keep their body temperature more stable,” given that only self-reporting was used to document hot flashes. We don’t know if this has any objective effect on women’s body temperatures. Worse, the study described above does not sound as though it’s designed to test anything concretely because investigators are now leaving out the no-treatment control group and instead testing against “another type of mind-body intervention,” whatever that means.

In other words, they’re testing woo against woo with no baseline untreated control to document the size of the placebo effect, and we have no idea whether the selected “mind-body” intervention is a valid placebo control. What they’ll then report is that hypnosis did better than the other “mind-body” intervention or vice versa, but what they won’t be able to report is whether any of these were better than placebo. Does hypnosis do anything for hot flashes? Who knows? It’s possible, although I always wonder whenever a therapy is billed to be able to accomplish so many different things for diseases and conditions with very different pathophysiology. What I do know is that this study is not particularly good evidence that it does, and the proposed followup study doesn’t on the surface sound all that much better.


G. Elkins, J. Marcus, V. Stearns, M. Perfect, M. H. Rajab, C. Ruud, L. Palamara, T. Keith (2008). Randomized Trial of a Hypnosis Intervention for Treatment of Hot Flashes Among Breast Cancer Survivors Journal of Clinical Oncology DOI: 10.1200/JCO.2008.16.6389