A randomised controlled trial of a cognitive behavioural intervention for women who have menopausal symptoms following breast cancer treatment (MENOS 1): Trial protocol

Non-hormonal treatments for HF/NS tend to be preferred but the efficacy and acceptability of treatments limit their use. Following a systematic review of non-hormonal medicines for HF/NS, Nelson, Vesco, Haney and others [23] concluded that SSRIs or SNRIs, clonidine and gabapentin trials provide evidence for efficacy but that adverse effects might restrict use for some women. Carpenter, Storniolo, Johns et al [19] reported modest reductions (22%) in HF/NS with Venlafaxine (37.5 mgs) but most patients discontinued treatment at 12 months follow up. These conclusions support those of a position statement by the North American Menopause Society [24]. Furthermore, possible negative interactions between some SSRIs and the effects of tamoxifen may limit prescribing options for women who have had breast cancer [25].

Non-medical treatments are popular, including acupuncture, yoga, exercise, and cognitive behavioural therapy. Studies of acupuncture often report significant reductions in HF/NS following treatment, although few studies report between group effects when compared to sham, superficial or placebo acupuncture [26]. A recent meta-analysis [27] of 4 RCTs of acupuncture for HF/NS in women who have had breast cancer found one RCT that reported significantly fewer HF/NS following acupuncture compared to sham acupuncture [28]. In well women one study out of six reviewed [29] reported a reduction in severity following acupuncture compared to placebo [30]. Both studies involved a comparatively large number of acupuncture sessions, suggesting that it may be effective over and above placebo effects if delivered intensively. Furthermore it would also be worth investigating the reasons for these consistent placebo improvements, and whether they are enduring.

Regular exercise has been associated with fewer HF/NS in cross-sectional studies, but interventions to reduce symptoms using exercise have been less successful [31]. In a systematic review of seven studies, about half reported significant within-group improvements following exercise, but only two reported improvements in HF/NS compared to a no treatment control (one was not powered to find a significant difference [32]). Although there were no studies focusing on breast cancer patients, an RCT of exercise for breast cancer patients with treatment induced HF/NS is due to publish findings in 2011 [33]. Theoretically, yoga may be more effective than exercise; as well as being low intensity exercise, yoga is used as a relaxation technique, which may also reduce HF/NS [34]. A systematic review revealed a reduction in reported HF/NS in 4 out of 7 studies [35], however most RCTs did not find an effect. The exception was an RCT of an intensive yoga programme (60 minutes a day, 5 days a week for 8 weeks) [36]. The only study to measure frequency of physiological HF/NS as well as self-reported HF/NS found no reduction in sternal skin conductance or diary reported hot flushes and night sweats following yoga [37]. There is growing evidence that relaxation therapy and paced respiration can alleviate HF/NS in well women [38‐41]. In two preliminary studies with breast cancer patients relaxation significantly reduced HF/NS after 12 sessions [42] but not after one session [43]. A treatment package including behavioural (relaxation and counselling) and pharmacological interventions for breast cancer patients has been evaluated in the US with positive results [44].

A CBT intervention for HF/NS, which includes relaxation and paced respiration, has been developed in the UK by Hunter and colleagues [45‐47]. This four-session CBT intervention was shown to be effective in an exploratory patient preference trial of well women (around 40% reduction of HF frequency and problem rating) [45]; a finding replicated by Keefer and Blanchard [48] using 8 sessions of CBT. Women who have had breast cancer tend to experience more frequent and problematic menopausal symptoms than well women going through the natural menopause [3]. In a study of 113 women, who had completed active treatment for breast cancer within the past 5 years, 80% reported HF and 72% NS, a significantly higher prevalence than found in well postmenopausal women [49]. Treatment preferences were explored and preferences were expressed for CBT (63%), complementary therapies (46%), antidepressant treatment (25%) and HT (26%). 53% expressed interest in group and/or individual CBT sessions [3].

Based on earlier work, an exploratory trial of a six-session Group CBT intervention was evaluated in a pre-post study of 24 breast cancer patients [47]. The weekly problem rating reduced on average from 5.9 (SD = 2.1) to 3.4 (SD = 1.8) at post-treatment and 2.8 (SD = 1.8) at follow-up (10 point scale); frequency of HF/NS per week reduced from 68.2 (SD = 28.7) to 46.7 (SD = 30.2) at post-treatment and to 36.9 (SD = 32.5) at 3 months follow up; a significant reduction in HF problem rating and frequency. Feedback from the women, elicited during and at the end of the groups, was very positive [47]. Relaxation/breathing and cognitive strategies used at the onset of HF/NS, as well as group support to make behavioural changes, were seen as most useful. The majority (88%) completed the 3 month follow-up assessment, suggesting good adherence to the study. These results suggest that CBT may be effective in reducing HF/NS and their impacts in breast cancer patients. However, a randomised controlled trial is needed. Furthermore, most studies of the efficacy of CBT used self-reported improvements in HF/NS, thus investigation into the impacts on physiological symptoms is warranted.

Participants are recruited from six hospital sites in South East London over an 18 month period from March 2009 to September 2010. Breast cancer patients who report HF/NS at clinic appointments are being recruited into the study by referrals from oncologists, surgeons, radiographers, breast care nurses and SELCRN research nurses working in clinics. Participants are also able to opt in to the study by responding to posters, leaflets and websites. Inclusion criteria

Women are excluded if depression or general cancer concerns are the main problem that they are seeking help for, not HF/NS, or if they are unable to commit time to attend the sessions.

Women referred from the clinics are contacted by telephone for a screening interview. A researcher conducts an eligibility check using the above criteria and explains what happens in the two arms of the trial and the randomisation process. Those eligible and interested in participating are sent a participant information sheet, a socio-demographic questionnaire and a two-week HF/NS diary to complete, and an appointment for the baseline assessment is arranged. The consent process is completed whereby the researcher summarises the information in the participant information sheet and responds to questions about the study. Participants then provide written consent. Diaries and demographic questions are not collected from those who do not consent or who withdraw consent during baseline assessment. The researcher then conducts a clinical interview covering breast cancer treatment history, medical history including mental health issues, menopause symptoms and treatments used, concomitant medication and therapy use, and current life concerns. Participants are given a questionnaire to complete, assessing HF/NS frequency and problem rating, beliefs about HF/NS, health related quality of life, stress, dispositional optimism and somatisation. After checking that the participant has no allergy to adhesive dressings (e.g. plasters) and does not wear a pacemaker, the SSC hot flush monitor is fitted. Participants are given a magnet event marker, worn on a bracelet, and a 24-hour hot flush diary. They mark the experience of a HF/NS by passing the magnet over the monitor and noting the time and severity of the HF/NS in the diary. Participants wear the monitor for 24 hours, carrying out their usual daily activities and removing the monitor only to shower or bathe.

The measures taken at each stage of the trial are summarised in table 2.

An adverse events protocol has been agreed by the DMC and TSC, adapted for a non-medical trial using the Medical Research Council/Department for Health Joint Project Guideline notes on Pharmocovigilance [64]. This excludes study outcomes, such as increases in frequency or severity of HF/NS that occur as part of normal fluctuations in menopause symptoms. Adverse events are logged in the trial master file and reported to the principle investigator and trial coordinator. In accordance with the National Patient Safety Agency's requirements for reporting of serious adverse events (SAE), any adverse event deemed to be a related or unexpected SAE will be reported to the principle investigator, the sponsor and the DMC within 48 hours, and reported to the Hospital Research Ethics Committee in accordance with their procedures. Reoccurrence of breast cancer or secondary cancer does not need to be reported in this way, as it is an expected unrelated serious adverse event (see appendix 3 of the Joint Guidelines [64]). An adverse events report of these occurrences will be included in the report to the DMC in accordance with the DMC charter.

The sample size is calculated on the HFRS problem rating subscale [55], estimating a mean of 5 (SD = 2.4) and a clinically relevant difference of 2 points [45] at two weeks post-treatment. A total sample size of 76, 38 in each group, will have 90% power to detect a difference in mean HF/NS Problem Rating of 2, controlling for baseline value, assuming a common standard deviation of 2.4, a correlation between baseline and end of treatment values of 0.4 (giving a correction factor of 0.84 [65]) and a 5% two-sided significance level. Due to clustering of outcomes within therapy groups, the sample size has been adjusted by assuming an intra-class correlation of 0.07 [66] and 8 participants per group giving a variance inflation factor of 1.49. Ninety six women will be recruited in to the trial to allow for a 20% attrition rate.

An analysis plan has been agreed (i.e. prior to completion of data collection). A trial statistician, employed by the Mental Health and Neurology Clinical Trials Unit at King's College London, carries out primary and secondary data analysis, and is blind to group allocation. Unless indicated, scales are calculated according to published scoring algorithms. All statistical tests are two sided (level of significance = 0.05). The analyses are preformed using the statistical software Stata version 11.0. An intention to treat analysis of the primary outcome is performed. Difference in problem rating between the two intervention groups is tested using a linear mixed model, utilising fixed and random effects. The regression model compares the problem rating subscale of HFRS between intervention groups, after allowing for the baseline measurements to control for pre-treatment differences, stratification factors (age: 50 years or older versus younger than 50 years), and clinically relevant covariates. Greater precision of estimates is expected within therapy groups. Therefore models are adjusted for the therapy group; this variable is able to randomly vary around the intercept (random effects). Secondary outcomes at post-treatment are analysed similarly. Follow-up data at 6 months is also explored through linear mixed models utilising repeated measures analyses, allowing simultaneous modelling of the two outcome time points. Mediation of treatment effects by hot flush beliefs and behaviours, mood, stress, quality of life and frequency of HF/NS defined by skin conductance rises will be explored using a two stage least squared method in Stata [67]. Moderation by optimism and somatisation is also tested using multiple-group modelling [68].

The design has several strengths. The treatment has been extensively developed and piloted for feasibility, acceptability and preliminary outcomes. Preliminary results found an average 2.5 point reduction in problem rating (49% reduction) as well as a significant reduction in self-reported frequency, and significant improvements in sleep quality and quality of life were also reported [47]. A treatment manual specifies the detail for each session so that it is replicable, and checks are in place to ensure adherence to protocol. Measures of treatment efficacy are robust: as well as questionnaire-based estimations of recent symptoms, the trial includes prospective monitoring of symptoms using self-report and physiological measures. Very few trials of treatments for HF/NS include both types of prospective monitoring, and no trials of CBT for HF/NS have done so. The study sample is drawn from a socially and culturally diverse area of the UK; the uptake and adherence to the treatment by women from different backgrounds will be reported.

There are some limitations that should be considered. Recruitment procedures are designed to reach all of the symptomatic breast cancer patients who attend clinics in the South East London Network (SELCRN) hospitals. Clinic and research staff approach patients directly and posters and leaflets are placed in waiting rooms so that patients can opt in themselves. Nonetheless, the sample might not be representative of symptomatic breast cancer patients in general and we cannot know how many people were made aware of the trial through these recruitment procedures.

The treatment is compared to a usual care group to see whether it improves on current practice, but usual care control groups have some drawbacks. Usual care may vary between centres, although typically treatment is minimal. Those receiving usual care may be more likely to seek complimentary therapies if they are having no other treatment. Complementary therapy use and clinic treatments are monitored in the trial and may have to be controlled for in analysis. If usual care is minimal then it will be similar to a no-treatment control group, and treatment effects might be due to the extra attention rather than the treatment per se (i.e. any intervention is better than nothing). Transitory changes and biased reporting should be apparent by comparing self-reported HF/NS with physiological measures and post-treatment with follow up data. Early in the treatment development phase it makes sense to compare with current practice, especially as some improvement in usual care would be predicted if only because symptoms do get better in time. Should treatment effects be robust, further research into the active treatment components can be tested using attention control groups.