- Email: [email protected]
Cognitive behaviour therapy for menopausal symptoms following breast cancer treatment: Who benefits and how does it work?
Maturitas 78 (2014) 56–61
Contents lists available at ScienceDirect
Maturitas journal homepage: www.elsevier.com/locate/maturitas
Cognitive behaviour therapy for menopausal symptoms following breast cancer treatment: Who benefits and how does it work? Joseph Chilcot, Sam Norton, Myra S. Hunter ∗ Institute of Psychiatry, King’s College London, UK
a r t i c l e
i n f o
Article history: Received 18 December 2013 Received in revised form 19 January 2014 Accepted 20 January 2014 Keywords: Menopause Hot flushes Cognitive behaviour therapy CBT Mediator Moderator
a b s t r a c t Objectives: Cognitive behaviour therapy (CBT) has been found to reduce the impact of menopausal symptoms, hot flushes and night sweats. This study investigates the moderators and mediators of CBT for women who had problematic menopausal symptoms following breast cancer treatment. Study design: Analysis of 96 patients with breast cancer induced menopausal symptoms recruited to the MENOS1 trial; 47 were randomly assigned to Group CBT and 49 to usual care. Questionnaires were completed at baseline, 9 and 26 weeks post randomisation. Potential moderators and mediators, including sociodemographic, clinical and psychological factors, of the treatment effect on the primary outcome were examined. Main outcome measure: Hot Flush Problem Rating. Results: CBT was effective at reducing problem rating at 9 weeks regardless of age, BMI, time since breast cancer diagnosis, menopausal status at time of diagnosis, or type of cancer treatment (radiotherapy or chemotherapy or endocrine treatment). The treatment effect was significantly greater in women not receiving chemotherapy, those with higher levels of psychological distress at baseline and for non-white women. Beliefs about control/coping with hot flushes were the main mediators of improvement in problem rating following CBT. Beliefs about hot flushes in a social context, depressed mood and sleep problems were also identified as mediators. Conclusions: These findings suggest that CBT is widely applicable for breast cancer patients who are experiencing treatment related menopausal symptoms, and that CBT works mainly by changing beliefs and improving mood and sleep. © 2014 Elsevier Ireland Ltd. All rights reserved.
1. Introduction Hot flushes and night sweats (HFNS) are commonly reported by women who have had breast cancer but are challenging to treat [1]. Between 65% and 85% of women treated for breast cancer report having HFNS, 60% rate them as severe, and these symptoms impact on quality of life, sleep, and mood [2,3]. Chemotherapy or adjuvant endocrine treatments can result in rapid reduction of oestrogen concentrations, which in turn induce or exacerbate HFNS. Hormone replacement therapy is generally contraindicated because it can increase the likelihood of recurrence, and, if left untreated, HFNS can reduce adherence to endocrine therapy [4,5]. A Cochrane review of non-hormonal medical treatments concluded that selective serotonin reuptake inhibitors (SSRIs), serotonin-norepinephrine reuptake inhibitors (SNRIs), clonidine
∗ Corresponding author at: Health Psychology Section, 5th Floor Bermondsey Wing, Guy’s Campus, King’s College London, London SE1 9RT, UK. Tel.: +44 02071880189/0. E-mail address: [email protected] (M.S. Hunter). 0378-5122/$ – see front matter © 2014 Elsevier Ireland Ltd. All rights reserved. http://dx.doi.org/10.1016/j.maturitas.2014.01.007
and gabapentin are mild to moderately effective in reducing the frequency of HFNS in women with a history of breast cancer [6] but side-effects were often reported [8]. Non-medical treatments tend to be preferred by breast cancer survivors [4] but non-pharmacological therapies, such as vitamins, herbal remedies, in general, do not have a strong evidence base [7]. There is increasing awareness that multidisciplinary approaches are needed [8], and growing evidence from three recent randomised controlled trials that cognitive behaviour therapy (CBT) can effectively reduce the impact of HFNS for women who have had breast cancer [9,10] and for well women during the menopause transition [11]. The three trials used group CBT (four to six weekly sessions of CBT; 8 h in total) developed by Hunter and colleagues. The MENOS1 trial [9,12] is an RCT of CBT (n = 47) versus treatment as usual (TAU) (n = 49) targeted at improving HFNS in breast cancer survivors. At 9 weeks after randomisation HFNS problem rating scores were significantly lower in the CBT group compared to usual care (adjusted mean difference [AMD] = −1.67, 95% CI −2.43 to −0.91, p < .001), an effect that was maintained at 26 weeks (AMD = −1.76, 95% CI −2.54 to −0.99); relating to standardised mean differences of d = 1.19 and d = 1.07, respectively [9].
J. Chilcot et al. / Maturitas 78 (2014) 56–61
A recently conducted gap analysis of UK breast cancer research highlighted the need for the development of effective theorybased interventions for treatment-related symptoms experienced by breast cancer survivors, with analysis of moderators and mediators and identified components [13]. This paper reports on planned analyses of the MENOS1 study to consider moderators and mediators of the treatment effect – that is, to identify for whom CBT works and how. The MENOS1 study included a relatively heterogeneous sample, involving women of different menopausal stages at diagnosis and on different treatments. While there is evidence that CBT is effective for women with HFNS, who were premenopausal when diagnosed with breast cancer [10], we do not know whether CBT can be confidently offered to different subgroups of women, for example those who had had chemotherapy or were having endocrine treatments. Similarly, does educational level, age or ethnicity influence CBT outcomes? In addition, while the main reports determined efficacy of CBT for breast cancer patients, neither considered the mechanisms by which CBT works [14]. Based on a cognitive model of HFNS [15] we hypothesised that CBT works by changing overly negative beliefs concerning HFNS and by helping women to use more adaptive behavioural strategies, which reduce the perceived impact of HFNS rather than their frequency. 2. Materials and methods 2.1. Study design The design of the MENOS1 RCT, and intervention procedure, is described in detail in the trial protocol [12] and main outcome paper [9]. Recruitment took place between March 2009 and August 2010 from breast cancer clinics in London, UK. Patients having at least ten problematic HFNS per week, who had completed medical treatment for breast cancer (surgery, radiotherapy, or chemotherapy), with no evidence of other cancers or metastases were included. Those taking adjuvant endocrine treatment were eligible. Sample characteristics are shown in Table 1. Following baseline assessment they were randomised to Group CBT or TAU and reassessed after 9 and 26 weeks; Group CBT involved 6 weeks of 1.5 h of CBT in groups of 6–8 women. All participants gave written, informed consent before taking part. Ethical approval was obtained from the UK NHS Research Ethics Committee. 2.2. Measures 2.2.1. HFNS measures The primary outcome was the HFNS problem rating (Hot Flush Rating Scale) [16] at 9 weeks after randomisation, which is the mean of three 10 point scales assessing the extent to which symptoms are problematic and interfere with daily life; 10 indicates most problematic HFNS. A difference of two points or more is regarded as clinically relevant. The scale had good reliability in the MENOS1 sample (Cronbach ˛ = 0.89). HFNS frequency subscale measures the total number of HFNS reported in the past week [16]. Sternal skin conductance (SSC) was included to measure physiological HFNS frequency using the Bahr SSC monitor [Simplex Scientific; MiddleYs ton, WI, USA]. A 6-cm by 6-cm monitor measured SSC every 10´ by passing an electric current across two electrodes attached to the sternal region of the chest. 2.2.2. HFNS beliefs and behaviours Hot Flush Beliefs Scale [17] is a 27-item scale comprising three subscales: (i) beliefs about HF in social context (e.g. everyone is looking at me), (ii) beliefs about coping/control of hot flushes (e.g. when I have a HF I think they will never end), and (iii) beliefs about night sweats and sleep (e.g. if I have NS I’ll never get back to
57
sleep). The HFNS Behaviour Scale [18] was developed using factor analysis and includes three subscales measuring, (i) positive coping behaviour, e.g. accepting HFNS, using breathing and calming responses; (ii) avoidance behaviour, and (iii) cooling behaviours, such as fanning oneself. 2.2.3. Stress and mood measures The Perceived Stress Scale [19] includes 14 items, on a scale from 0 never to 4 very often; items are summed to form a 0–56 scale with a high score representing high stress. Subscales of the Women’s Health Questionnaire (WHQ) [20] were used to measure depressed mood, anxiety and sleep problems. The WHQ was standardised on women aged 45–65 years and has been widely used to evaluate interventions for menopausal symptoms. 2.2.4. Personality measures The Somatosensory Amplification Scale (SSAS) [21] has 10 items rated on 5 point scales measuring respondent’s tendency to experience somatic sensation as intense, noxious and disturbing. Dispositional optimism. The Revised Life Orientation Test (LOT-R) [22] measures dispositional optimism on a 6-item scale rated on a 5 point scale. High scores indicate greater dispositional optimism. 2.2.5. Demographic and health behaviour variables Demographic and health behaviour factors were recorded at baseline including: age, height, weight, ethnicity, education, employment status, smoking, and exercise frequency. Breast cancer treatments, use of concomitant medications and therapies were also recorded. 2.3. Statistical analysis The moderator analysis extended the model used in the main study to test changes in HFNS problem rating over the study. This involved the estimation of linear mixed effects model with random intercepts for participant and cohort group. Time, treatment group, baseline HFNS problem rating score and age at randomisation were included in the model as covariates. A time by treatment group interaction term was also included to allow the calculation of adjusted means at individual time points. This model was extended to allow for the testing of potential moderators of the effect of CBT on HFNS problem rating at 9 weeks by including the main effect of the moderator variable, and two and three-way interactions of the moderator variable with time and treatment group. Inclusion of moderator by time by treatment group interaction terms allowed for the assessment of effect modification at 9 weeks. To aid interpretation effect sizes were calculated for the moderator effects. Effect sizes were standardised mean differences (Cohen’s d) for the categorical variables and standardised regression coefficients for continuous variables (beta’s). Although the study was not specifically powered to detect moderator variables, power was adequate: assuming 80% power, a medium sized moderator effect was detectable (R2 = 7.7). The original trial identified that patients receiving CBT reported significantly less depression symptoms, anxiety, stress and sleep problems at 9 weeks compared to those receiving TAU. In the present analysis, we evaluated whether HFNS beliefs and behaviours also altered over the intervention using ANCOVA to estimate the effect of treatment on the variable at 9 weeks, adjusted for the baseline level of the variable [23]. Using the variables identified as changing significantly from the original trial and analysis conducted here (HFNS beliefs and behaviours), mediation was evaluated using path models that estimated the indirect effect of treatment group on HFNS problem rating at 26-weeks through the residualised change in the potential mediator at the 9-week followup. Both the potential mediator at 9 weeks and HFNS problem
58
J. Chilcot et al. / Maturitas 78 (2014) 56–61
Table 1 Demographic and baseline clinical characteristics, with test for CBT effect moderation and adjusted mean difference in HFNS problem rating at 9 weeks or simple slope. TAU n Ethnicity 40 White Non−white 9 Education beyond 16 years Yes 33 No 16 Employed 32 Yes 15 No Menopausal status at diagnosis Pre 24 Peri/post 24 Chemotherapy Yes 37 12 No Radiotherapy 41 Yes 8 No Endocrine treatment 36 Yes 13 No Smoking status 23 Never 47 Current/ex Exercise frequency (times/week) 18 <2 2–3 19 12 >3
Age (years) Body mass index Time since diagnosis (months) Depression Anxiety
CBT
Adjusted mean difference
Effect size
Moderator
(d)
p-value
(−2.14, −0.78) (−5.95, −2.21)
−0.72 −2.02
0.01
−2.17 −1.07
(−2.95, −1.38) (−2.24, 0.10)
−1.07 −0.53
0.127
64 36
−2.06 −1.44
(−2.92, −1.21) (−2.46, −0.41)
−1.02 −0.71
0.356
24 21
51 44
−2.31 −1.37
(−3.25, −1.36) (−2.32, −0.43)
−0.67 −1.14
0.162
76 24
26 21
55 45
−1.3 −2.86
(−2.11, −0.49) (−4.01, −1.71)
−0.64 −1.42
0.029
84 16
36 11
77 23
−1.79 −1.79
(−2.53, −1.06) (−3.21, −0.37)
−0.89 −0.89
0.996
73 27
34 13
72 28
−1.53 −3.47
(−2.21, −0.84) (−5.35, −1.58)
−0.76 −1.72
0.058
47 53
17 36
30 64
−2.26 −1.41
(−3.26, −1.26) (−2.27, −0.55)
−1.12 −0.7
0.212
23 40 36
11 19 17
37 39 24
−3.39 −1.81 −0.17
(−4.61, −2.17) (−2.80, −0.82) (−1.36, 1.02)
−1.68 −0.9 −0.08
0.001
%
n
%
82 18
42 5
89 11
−1.46 −4.08
67 33
30 17
64 36
35 35
30 17
49 49
95% CI
Mean
SD
Mean
SD
Simple slope
95% CI
Effect size (beta)
Moderator p-value
54.05 27.51 31.08 0.31 0.45
7.76 6.9 30.63 0.27 0.3
53.16 27.13 47.75 0.23 0.34
8.1 5.3 53.38 0.26 0.25
−0.01 0.03 −0.01 −3.36 −2.09
(0.09, 0.08) (−0.08, 0.14) (−0.02, 0.01) (−5.28, −1.44) (−5.34, −0.45)
−0.08 0.18 −0.43 −1.01 −0.87
0.908 0.598 0.56 0.001 0.02
rating at 26-weeks were adjusted for their baseline level, treatment cohort, and age at randomisation. 3. Results 3.1. Moderators of CBT: who benefits? Demographic and clinical characteristics at baseline are shown in Table 1. At baseline HFNS were frequent and problematic with a mean of 69 per woman (SD = 39) per week for an average of 2 years, with a mean problem rating of 6.3 out of 10. Moderators of the effect of CBT on HFNS problem rating score at 9 weeks were considered (Table 1) with interaction plots shown for those observed to be significant (Fig. 1). The graphs show the adjusted problem rating scores for each group (CBT and TAU) at 9 weeks divided into categories depending on the baseline moderators described, e.g. chemotherapy versus no chemotherapy. The only demographic variable observed to be a significant moderator of treatment effect was ethnicity. HFNS problem rating score at 9 weeks was lower in the CBT group compared to usual care for both white and non-white ethnic groupings, but the difference was greater for non-white women (−4.08, 95% CI −5.95 to −2.21, p < 0.001) compared to white women (−1.46, 95% CI −2.14 to −0.78, p < 0.001). The only baseline clinical characteristic observed to moderate the CBT effect was chemotherapy, with a larger difference in HFNS problem rating score at 9 weeks for those not receiving chemotherapy (−2.86, 95% CI −4.01 to −1.71, p < 0.001) compared to those who did (−1.30, 95% CI −2.11 to −0.49, p = 0.002). These findings, for both ethnicity and chemotherapy, remained significant even after controlling for baseline levels of depression.
Concerning health behaviours, exercise frequency, but not smoking status, was observed to significantly moderate the effect of CBT. Further analysis revealed that the significant moderating effect was due to no significant difference in adjusted HFNS problem rating at 9 weeks between the CBT and usual care groups (−0.17, 95% CI −1.36 to 1.02, p = 0.779) for patients who exercised four or more times per week. However, after controlling for baseline depression the moderating effect of exercise became non-significant, suggesting that the finding is likely to be due to lower levels of distress in those who exercise more frequently. Depressed mood (−3.36, 95% CI −5.28 to −1.44, p = 0.001) and anxiety (−2.90, 95% CI −5.34 to −0.45, p = 0.020) were significant moderators of treatment effect at 9 weeks. The simple slopes relate to a difference in the adjusted mean difference between CBT and TAU of −0.83, −1.15 and −0.82 in HFNS problem rating at 9 weeks, respectively, for a one standard deviation higher score on the moderator variable. Together these observations indicate that CBT is most effective for individuals reporting high levels of psychological distress at baseline (Fig. 1). Non-significant treatment effects were indicated for individuals scoring zero on the WHQ depression and anxiety subscales. 3.2. Mediators of CBT: how does it work? There were no significant changes in sternal skin conductance – the physiological measure of HFNS frequency, between the groups over the intervention period. With regards to HFNS beliefs and behaviours those receiving CBT reported more positive social HFNS beliefs (adjusted mean difference = −0.77, 95% CI −1.12 to −0.38, p < 0.01), HF coping/control beliefs (adjusted mean difference =
J. Chilcot et al. / Maturitas 78 (2014) 56–61
59
Fig. 1. HFNS problem rating at 9 weeks post-randomisation stratified by significant moderator variables, adjusted for age and baseline HFNS problem rating.
−0.97, 95% CI −1.33 to −0.60, p < 0.01) and beliefs about NS and sleep (adjusted mean difference = −0.99, 95% CI −1.40 to −0.60, p < 0.01), and reported more positive coping behaviours (adjusted mean difference = 0.88, 95% CI 0.60 to 1.17, p < 0.01) at 9 weeks compared to those receiving TAU. Accordingly these potential process variables were selected for the mediation analysis. Table 2 summarises the mediation analysis showing the estimate of the indirect treatment effect via each potential mediator on HFNS problem rating at 26 weeks and the proportion of the total treatment effect that is mediated by the variable. Changes in depressed mood, sleep problems and social beliefs all significantly mediated the treatment effect, although, the mediation was partial since in each path model the direct effect of treatment on HFNS problem rating at 26 weeks was significant. This suggests that CBT successfully reduced depression and sleeping problems and altered social beliefs regarding the impact of hot flushes, which in turn accounted for some of the improvements in HFNS problem rating. However, the largest mediation effect was HF coping/control beliefs (standardised beta = 0.50, p < 0.01), which accounted for 60.3% of the total effect of treatment on HFNS problem rating. 4. Discussion The results of this study suggest that CBT is effective in reducing the impact of HFNS for women following breast cancer treatment regardless of age, BMI, time since breast cancer diagnosis, menopausal status at time of diagnosis, or type of cancer treatment (radiotherapy or chemotherapy or endocrine treatment). However, while CBT worked well for women who had received chemotherapy, it worked rather better for those who did not. This could not be explained by initial level of mood or hot flushes, as we adjusted for baseline scores; it is possible that there are physiological explanations given that chemotherapy can induce HFNS. Interestingly, the physiological measure of HFNS frequency (sternal skin conductance) did not significantly change in this trial, whereas in a parallel study of well women going through the menopause
transition or postmenopause (MENOS2) there were small but significant reductions following CBT [24]. Women of non-white ethnicity improved more following CBT than those of white ethnicity. This was the case after we controlled for baseline mood and HFNS, and also for the moderating effect of mood. Ethnic differences in HFNS reporting have been previously documented [25], and other factors might explain this result. The finding requires replication, as there were only 14 women in this category (nine black British and five of Asian or mixed ethnicity), but overall these findings suggest that CBT is beneficial regardless of ethnicity. Interestingly, we found that those who were more distressed at baseline, in terms of depressed mood and anxiety scores, reported greater improvement. The only sub-group of women who showed no benefit were those who on entry to the trial participated in high levels of exercise (four times a week or more), but this finding became non-significant when the moderating effect of depression was controlled. Exercise has been associated with lower prevalence rates of HFNS and with lower levels of depressed mood and was found to impact on HFNS frequency in a recent trial [26]. Therefore, in women who frequently undertake exercise and report low levels of psychological distress CBT may either be unnecessary or ineffective at reducing HFNS. When we examined how the CBT works, we found that the main mediator was beliefs about coping and control over hot flushes, and additional mediation was provided by depressed mood, sleep and changes in beliefs about hot flushes in social situations. Thus learning to control and cope with the HFNS was an important factor in reducing problem rating of HFNS. These results generally support the cognitive model of HFNS [15,27] and suggest that CBT works mainly by changing cognitions, i.e. the cognitive appraisal of HFNS, but also mood and sleep problems. The CBT targeted cognitions, behavioural reactions, stress/wellbeing and the impact of night sweats upon sleep, so these results are heartening as they show that improvement in these areas was clearly associated with clinical outcomes. It is generally not easy to find satisfactory control conditions for complex psychological interventions, in the same
60
J. Chilcot et al. / Maturitas 78 (2014) 56–61
Table 2 Summary of the indirect effects between treatment group and Hot Flush Problem Rating. Mediator
Indirect effect (95% CI) Group → mediator → HF Problem rating
% of total effect explained by the mediator
Depression Anxiety Stress Sleep problems Social beliefs Control beliefs Sleep beliefs Positive coping behaviours
0.18* (0.00 to 0.36) 0.23 (−0.04 to 0.51) 0.09 (−0.03 to 0.20) 0.24* (0.04 to 0.43) 0.24* (0.03 to 0.46) 0.50** (0.15 to 0.90) 0.32 (−0.05 to 0.70) 0.28 (−0.04 to 0.61)
24.3 31.1 11.5 30.6 29.1 60.3 41.2 33.5
Indirect estimates are the standardised change in Hot Flush Problem rating. * p < 0.05. ** p < 0.01.
way that placebos are used for medical trials [26] so mediation analysis is important to show that the treatment, such as CBT, is actually changing what it is intending to change and that the effects cannot be entirely explained by attention or ‘non-specific factors’. These results are supported by those of a qualitative study based on interviews with women who had CBT in the MENOS1 trial [28]. The women reported that CBT improved their ability to cope with their HFNS and that they ‘regained a sense of control’, which is consistent with our findings that changes in beliefs about HFNS mediate improvements in symptom experience. A sense of control over HFNS has been associated with subjective HFNS distress and overall wellbeing [17], and it has been argued that interventions to counter lack of control could be particularly beneficial for cancer patients experiencing treatment related symptoms [29]. Study limitations include the sample size; some non-significant potential moderators and mediators might not have been detected due to power considerations. While we did not correct for multiple testing because the analysis was exploratory [30], it is useful to note that even using a conservative Bonferroni correction the effects of exercise and depression would have remained significant. Further confirmatory research could include larger samples in multicentre trials in which the impact of variables such as ethnicity could be clarified. In conclusion, this study shows that CBT is beneficial to breast cancer survivors with troublesome treatment related symptoms and significantly reduces the impact of HFNS. Those who were most distressed benefited more but there were no contraindications to CBT. CBT appeared to work by changing HFNS beliefs, i.e. cognitive appraisal, and by improving mood and sleep, suggesting that the treatment might be tailored to individuals depending on their scores on these active components. Contributors J.C. wrote the first draft; J.C. and S.N. conducted the statistical analysis; M.S.H. designed the study; all authors contributed to writing, interpretation and the final manuscript. Competing interest The authors have no competing interests to declare. Funding This study (C8303/A6130).
was
supported
by
Cancer
Research
UK
Ethics Ethical approval was obtained from the UK NHS Research Ethics Committee (South East London 2 REC, ref: 08/H0802/106).
Patient consent We confirm that consent was obtained from participants in the study.
References [1] Kaplan M, Mahon S, Cope D, Keating E, Hill S, Jacobson M. Putting evidence into practice: evidence-based interventions for hot flashes resulting from cancer therapies. Clinical Journal of Oncology Nursing 2011;15(2):149–57. [2] Gupta P, Sturdee DW, Palin SL, et al. Menopausal symptoms in women treated for breast cancer: the prevalence and severity of symptoms and their perceived effects on quality of life. Climacteric: The Journal of the International Menopause Society 2006;9(1):49–58. [3] Hunter MS, Grunfeld EA, Mittal S, et al. Menopausal symptoms in women with breast cancer: prevalence and treatment preferences. Psycho-oncology 2004;13(11):769–78. [4] Biglia N, Cozzarella M, Cacciari F, et al. Menopause after breast cancer: a survey on breast cancer survivors. Maturitas 2003;45(1):29–38. [5] Cella D, Fallowfield LJ. Recognition and management of treatment-related side effects for breast cancer patients receiving adjuvant endocrine therapy. Breast Cancer Research and Treatment 2008;107(2):167–80. [6] Rada G, Capurro D, Pantoja T, et al. Non-hormonal interventions for hot flushes in women with a history of breast cancer. Cochrane Database of Systematic Reviews 2010, http://dx.doi.org/10.1002/14651858.CD004923.pub2. [7] Boekhout AH, Vincent AD, Dalesio OB, et al. Management of hot flashes in patients who have breast cancer with venlafaxine and clonidine: a randomized, double-blind, placebo-controlled trial. Journal of Clinical Oncology 2011;29(29):3862–8. [8] Hickey M, Emery LI, Gregson J, Doherty DA, Saunders CM. The multidisciplinary management of menopausal symptoms after breast cancer: a unique model of care. Menopause 2010;17(4):727–33. [9] Mann E, Smith MJ, Hellier J, et al. Cognitive behavioural treatment for women who have menopausal symptoms after breast cancer treatment (MENOS 1): a randomised controlled trial. Lancet Oncology 2012;13(3):309–18. [10] Duijts SF, van Beurden M, Oldenburg HS, et al. Efficacy of cognitive behavioral therapy and physical exercise in alleviating treatment-induced menopausal symptoms in patients with breast cancer: results of a randomized, controlled, multicenter trial. Journal of Clinical Oncology: Official Journal of the American Society of Clinical Oncology 2012;30(33):4124–33. [11] Ayers B, Smith M, Hellier J, Mann E, Hunter MS. Effectiveness of group and self-help cognitive behavior therapy in reducing problematic menopausal hot flushes and night sweats (MENOS 2): a randomized controlled trial. Menopause 2012;19(7):749–59. [12] Mann E, Smith M, Hellier J, Hunter MS. A randomised controlled trial of a cognitive behavioural intervention for women who have menopausal symptoms following breast cancer treatment (MENOS 1): trial protocol. BMC Cancer 2011;11:44, http://dx.doi.org/10.1186/1471-2407-11-44. [13] Eccles S, Aboagye E, Ali S, et al. Critical research gaps and translational priorities for the successful prevention and treatment of breast cancer. Breast Cancer Research 2013;15:R92. [14] Prigerson HG. Mind over menopausal symptoms in women breast cancer. Lancet Oncology 2012;13(3):227–9, with http://dx.doi.org/10.1016/S1470-2045(11)70381-3. [15] Hunter MS, Mann E. A cognitive model of menopausal hot flushes and night sweats. Journal of Psychosomatic Research 2010;69(5):491–501. [16] Hunter MS, Liao KL. A psychological analysis of menopausal hot flushes. British Journal of Clinical Psychology 1995;34(Pt 4):589–99. [17] Rendall MJ, Simonds LM, Hunter MS. The Hot Flush Beliefs Scale: a tool for assessing thoughts and beliefs associated with the experience of menopausal hot flushes and night sweats. Maturitas 2008;60(2):158–69. [18] Hunter MS, Ayers B, Smith M. The Hot Flush Behavior Scale: a measure of behavioral reactions to menopausal hot flushes and night sweats. Menopause 2011;18(11):1178–83.
J. Chilcot et al. / Maturitas 78 (2014) 56–61 [19] Cohen S, Kamarck T, Mermelstein R. A global measure of perceived stress. Journal of Health and Social Behavior 1983;24(4):385–96. [20] Hunter M. The women’s health questionnaire: a measure of mid-aged women’s perceptions of their emotional and physical health. Psychology & Health 1992;7(1):45–54. [21] Barsky AJ, Wyshak G, Klerman GL. The somatosensory amplification scale and its relationship to hypochondriasis. Journal of Psychiatric Research 1990;24(4):323–34. [22] Scheier MF, Carver CS, Bridges MW. Distinguishing optimism from neuroticism (and trait anxiety, self-mastery, and self-esteem): a reevaluation of the Life Orientation Test. Journal of Personality and Social Psychology 1994;67(6):1063–78. [23] Van Breukelen GJ. ANCOVA versus change from baseline: more power in randomized studies, more bias in nonrandomized studies [corrected]. Journal of Clinical Epidemiology 2006;59(9):920–5. [24] Stefanopoulou E, Hunter MS. Does pattern recognition software using the Bahr monitor improve sensitivity, specificity and concordance of
[25]
[26] [27] [28] [29]
[30]
61
ambulatory skin conductance monitoring of hot flushes. Maturitas 2013, http://dx.doi.org/10.1016/j.maturitas.2013.09.013 (in press). Gold EB, Sternfeld B, Kelsey JL, et al. Relation of demographic and lifestyle factors to symptoms in a multi-racial/ethnic population of women 40–55 years of age. American Journal of Epidemiology 2000;152(5):463–73. Avis NE. Breast cancer survivors and hot flashes: the search for nonhormonal treatments. Journal of Clinical Oncology 2008;26(31):5008–10. Hunter MS, Chilcot J. Testing a cognitive model of menopausal hot flushes and night sweats. Journal of Psychosomatic Research 2013;74(4):307–12. Fenlon DR, Rogers AE. The experience of hot flushes after breast cancer. Cancer Nursing 2007;30:E19–26. Balabanovic J, Ayers B, Hunter MS. Women’s experiences of Group Cognitive Behaviour Therapy for hot flushes and night sweats following breast cancer treatment: an interpretative phenomenological analysis. Maturitas 2012;72(3):236–42. Bender R, Lange S. Adjusting for multiple testing – when and how? Journal of Clinical Epidemiology 2001;54(4):343–9.
Contents lists available at ScienceDirect
Maturitas journal homepage: www.elsevier.com/locate/maturitas
Cognitive behaviour therapy for menopausal symptoms following breast cancer treatment: Who benefits and how does it work? Joseph Chilcot, Sam Norton, Myra S. Hunter ∗ Institute of Psychiatry, King’s College London, UK
a r t i c l e
i n f o
Article history: Received 18 December 2013 Received in revised form 19 January 2014 Accepted 20 January 2014 Keywords: Menopause Hot flushes Cognitive behaviour therapy CBT Mediator Moderator
a b s t r a c t Objectives: Cognitive behaviour therapy (CBT) has been found to reduce the impact of menopausal symptoms, hot flushes and night sweats. This study investigates the moderators and mediators of CBT for women who had problematic menopausal symptoms following breast cancer treatment. Study design: Analysis of 96 patients with breast cancer induced menopausal symptoms recruited to the MENOS1 trial; 47 were randomly assigned to Group CBT and 49 to usual care. Questionnaires were completed at baseline, 9 and 26 weeks post randomisation. Potential moderators and mediators, including sociodemographic, clinical and psychological factors, of the treatment effect on the primary outcome were examined. Main outcome measure: Hot Flush Problem Rating. Results: CBT was effective at reducing problem rating at 9 weeks regardless of age, BMI, time since breast cancer diagnosis, menopausal status at time of diagnosis, or type of cancer treatment (radiotherapy or chemotherapy or endocrine treatment). The treatment effect was significantly greater in women not receiving chemotherapy, those with higher levels of psychological distress at baseline and for non-white women. Beliefs about control/coping with hot flushes were the main mediators of improvement in problem rating following CBT. Beliefs about hot flushes in a social context, depressed mood and sleep problems were also identified as mediators. Conclusions: These findings suggest that CBT is widely applicable for breast cancer patients who are experiencing treatment related menopausal symptoms, and that CBT works mainly by changing beliefs and improving mood and sleep. © 2014 Elsevier Ireland Ltd. All rights reserved.
1. Introduction Hot flushes and night sweats (HFNS) are commonly reported by women who have had breast cancer but are challenging to treat [1]. Between 65% and 85% of women treated for breast cancer report having HFNS, 60% rate them as severe, and these symptoms impact on quality of life, sleep, and mood [2,3]. Chemotherapy or adjuvant endocrine treatments can result in rapid reduction of oestrogen concentrations, which in turn induce or exacerbate HFNS. Hormone replacement therapy is generally contraindicated because it can increase the likelihood of recurrence, and, if left untreated, HFNS can reduce adherence to endocrine therapy [4,5]. A Cochrane review of non-hormonal medical treatments concluded that selective serotonin reuptake inhibitors (SSRIs), serotonin-norepinephrine reuptake inhibitors (SNRIs), clonidine
∗ Corresponding author at: Health Psychology Section, 5th Floor Bermondsey Wing, Guy’s Campus, King’s College London, London SE1 9RT, UK. Tel.: +44 02071880189/0. E-mail address: [email protected] (M.S. Hunter). 0378-5122/$ – see front matter © 2014 Elsevier Ireland Ltd. All rights reserved. http://dx.doi.org/10.1016/j.maturitas.2014.01.007
and gabapentin are mild to moderately effective in reducing the frequency of HFNS in women with a history of breast cancer [6] but side-effects were often reported [8]. Non-medical treatments tend to be preferred by breast cancer survivors [4] but non-pharmacological therapies, such as vitamins, herbal remedies, in general, do not have a strong evidence base [7]. There is increasing awareness that multidisciplinary approaches are needed [8], and growing evidence from three recent randomised controlled trials that cognitive behaviour therapy (CBT) can effectively reduce the impact of HFNS for women who have had breast cancer [9,10] and for well women during the menopause transition [11]. The three trials used group CBT (four to six weekly sessions of CBT; 8 h in total) developed by Hunter and colleagues. The MENOS1 trial [9,12] is an RCT of CBT (n = 47) versus treatment as usual (TAU) (n = 49) targeted at improving HFNS in breast cancer survivors. At 9 weeks after randomisation HFNS problem rating scores were significantly lower in the CBT group compared to usual care (adjusted mean difference [AMD] = −1.67, 95% CI −2.43 to −0.91, p < .001), an effect that was maintained at 26 weeks (AMD = −1.76, 95% CI −2.54 to −0.99); relating to standardised mean differences of d = 1.19 and d = 1.07, respectively [9].
J. Chilcot et al. / Maturitas 78 (2014) 56–61
A recently conducted gap analysis of UK breast cancer research highlighted the need for the development of effective theorybased interventions for treatment-related symptoms experienced by breast cancer survivors, with analysis of moderators and mediators and identified components [13]. This paper reports on planned analyses of the MENOS1 study to consider moderators and mediators of the treatment effect – that is, to identify for whom CBT works and how. The MENOS1 study included a relatively heterogeneous sample, involving women of different menopausal stages at diagnosis and on different treatments. While there is evidence that CBT is effective for women with HFNS, who were premenopausal when diagnosed with breast cancer [10], we do not know whether CBT can be confidently offered to different subgroups of women, for example those who had had chemotherapy or were having endocrine treatments. Similarly, does educational level, age or ethnicity influence CBT outcomes? In addition, while the main reports determined efficacy of CBT for breast cancer patients, neither considered the mechanisms by which CBT works [14]. Based on a cognitive model of HFNS [15] we hypothesised that CBT works by changing overly negative beliefs concerning HFNS and by helping women to use more adaptive behavioural strategies, which reduce the perceived impact of HFNS rather than their frequency. 2. Materials and methods 2.1. Study design The design of the MENOS1 RCT, and intervention procedure, is described in detail in the trial protocol [12] and main outcome paper [9]. Recruitment took place between March 2009 and August 2010 from breast cancer clinics in London, UK. Patients having at least ten problematic HFNS per week, who had completed medical treatment for breast cancer (surgery, radiotherapy, or chemotherapy), with no evidence of other cancers or metastases were included. Those taking adjuvant endocrine treatment were eligible. Sample characteristics are shown in Table 1. Following baseline assessment they were randomised to Group CBT or TAU and reassessed after 9 and 26 weeks; Group CBT involved 6 weeks of 1.5 h of CBT in groups of 6–8 women. All participants gave written, informed consent before taking part. Ethical approval was obtained from the UK NHS Research Ethics Committee. 2.2. Measures 2.2.1. HFNS measures The primary outcome was the HFNS problem rating (Hot Flush Rating Scale) [16] at 9 weeks after randomisation, which is the mean of three 10 point scales assessing the extent to which symptoms are problematic and interfere with daily life; 10 indicates most problematic HFNS. A difference of two points or more is regarded as clinically relevant. The scale had good reliability in the MENOS1 sample (Cronbach ˛ = 0.89). HFNS frequency subscale measures the total number of HFNS reported in the past week [16]. Sternal skin conductance (SSC) was included to measure physiological HFNS frequency using the Bahr SSC monitor [Simplex Scientific; MiddleYs ton, WI, USA]. A 6-cm by 6-cm monitor measured SSC every 10´ by passing an electric current across two electrodes attached to the sternal region of the chest. 2.2.2. HFNS beliefs and behaviours Hot Flush Beliefs Scale [17] is a 27-item scale comprising three subscales: (i) beliefs about HF in social context (e.g. everyone is looking at me), (ii) beliefs about coping/control of hot flushes (e.g. when I have a HF I think they will never end), and (iii) beliefs about night sweats and sleep (e.g. if I have NS I’ll never get back to
57
sleep). The HFNS Behaviour Scale [18] was developed using factor analysis and includes three subscales measuring, (i) positive coping behaviour, e.g. accepting HFNS, using breathing and calming responses; (ii) avoidance behaviour, and (iii) cooling behaviours, such as fanning oneself. 2.2.3. Stress and mood measures The Perceived Stress Scale [19] includes 14 items, on a scale from 0 never to 4 very often; items are summed to form a 0–56 scale with a high score representing high stress. Subscales of the Women’s Health Questionnaire (WHQ) [20] were used to measure depressed mood, anxiety and sleep problems. The WHQ was standardised on women aged 45–65 years and has been widely used to evaluate interventions for menopausal symptoms. 2.2.4. Personality measures The Somatosensory Amplification Scale (SSAS) [21] has 10 items rated on 5 point scales measuring respondent’s tendency to experience somatic sensation as intense, noxious and disturbing. Dispositional optimism. The Revised Life Orientation Test (LOT-R) [22] measures dispositional optimism on a 6-item scale rated on a 5 point scale. High scores indicate greater dispositional optimism. 2.2.5. Demographic and health behaviour variables Demographic and health behaviour factors were recorded at baseline including: age, height, weight, ethnicity, education, employment status, smoking, and exercise frequency. Breast cancer treatments, use of concomitant medications and therapies were also recorded. 2.3. Statistical analysis The moderator analysis extended the model used in the main study to test changes in HFNS problem rating over the study. This involved the estimation of linear mixed effects model with random intercepts for participant and cohort group. Time, treatment group, baseline HFNS problem rating score and age at randomisation were included in the model as covariates. A time by treatment group interaction term was also included to allow the calculation of adjusted means at individual time points. This model was extended to allow for the testing of potential moderators of the effect of CBT on HFNS problem rating at 9 weeks by including the main effect of the moderator variable, and two and three-way interactions of the moderator variable with time and treatment group. Inclusion of moderator by time by treatment group interaction terms allowed for the assessment of effect modification at 9 weeks. To aid interpretation effect sizes were calculated for the moderator effects. Effect sizes were standardised mean differences (Cohen’s d) for the categorical variables and standardised regression coefficients for continuous variables (beta’s). Although the study was not specifically powered to detect moderator variables, power was adequate: assuming 80% power, a medium sized moderator effect was detectable (R2 = 7.7). The original trial identified that patients receiving CBT reported significantly less depression symptoms, anxiety, stress and sleep problems at 9 weeks compared to those receiving TAU. In the present analysis, we evaluated whether HFNS beliefs and behaviours also altered over the intervention using ANCOVA to estimate the effect of treatment on the variable at 9 weeks, adjusted for the baseline level of the variable [23]. Using the variables identified as changing significantly from the original trial and analysis conducted here (HFNS beliefs and behaviours), mediation was evaluated using path models that estimated the indirect effect of treatment group on HFNS problem rating at 26-weeks through the residualised change in the potential mediator at the 9-week followup. Both the potential mediator at 9 weeks and HFNS problem
58
J. Chilcot et al. / Maturitas 78 (2014) 56–61
Table 1 Demographic and baseline clinical characteristics, with test for CBT effect moderation and adjusted mean difference in HFNS problem rating at 9 weeks or simple slope. TAU n Ethnicity 40 White Non−white 9 Education beyond 16 years Yes 33 No 16 Employed 32 Yes 15 No Menopausal status at diagnosis Pre 24 Peri/post 24 Chemotherapy Yes 37 12 No Radiotherapy 41 Yes 8 No Endocrine treatment 36 Yes 13 No Smoking status 23 Never 47 Current/ex Exercise frequency (times/week) 18 <2 2–3 19 12 >3
Age (years) Body mass index Time since diagnosis (months) Depression Anxiety
CBT
Adjusted mean difference
Effect size
Moderator
(d)
p-value
(−2.14, −0.78) (−5.95, −2.21)
−0.72 −2.02
0.01
−2.17 −1.07
(−2.95, −1.38) (−2.24, 0.10)
−1.07 −0.53
0.127
64 36
−2.06 −1.44
(−2.92, −1.21) (−2.46, −0.41)
−1.02 −0.71
0.356
24 21
51 44
−2.31 −1.37
(−3.25, −1.36) (−2.32, −0.43)
−0.67 −1.14
0.162
76 24
26 21
55 45
−1.3 −2.86
(−2.11, −0.49) (−4.01, −1.71)
−0.64 −1.42
0.029
84 16
36 11
77 23
−1.79 −1.79
(−2.53, −1.06) (−3.21, −0.37)
−0.89 −0.89
0.996
73 27
34 13
72 28
−1.53 −3.47
(−2.21, −0.84) (−5.35, −1.58)
−0.76 −1.72
0.058
47 53
17 36
30 64
−2.26 −1.41
(−3.26, −1.26) (−2.27, −0.55)
−1.12 −0.7
0.212
23 40 36
11 19 17
37 39 24
−3.39 −1.81 −0.17
(−4.61, −2.17) (−2.80, −0.82) (−1.36, 1.02)
−1.68 −0.9 −0.08
0.001
%
n
%
82 18
42 5
89 11
−1.46 −4.08
67 33
30 17
64 36
35 35
30 17
49 49
95% CI
Mean
SD
Mean
SD
Simple slope
95% CI
Effect size (beta)
Moderator p-value
54.05 27.51 31.08 0.31 0.45
7.76 6.9 30.63 0.27 0.3
53.16 27.13 47.75 0.23 0.34
8.1 5.3 53.38 0.26 0.25
−0.01 0.03 −0.01 −3.36 −2.09
(0.09, 0.08) (−0.08, 0.14) (−0.02, 0.01) (−5.28, −1.44) (−5.34, −0.45)
−0.08 0.18 −0.43 −1.01 −0.87
0.908 0.598 0.56 0.001 0.02
rating at 26-weeks were adjusted for their baseline level, treatment cohort, and age at randomisation. 3. Results 3.1. Moderators of CBT: who benefits? Demographic and clinical characteristics at baseline are shown in Table 1. At baseline HFNS were frequent and problematic with a mean of 69 per woman (SD = 39) per week for an average of 2 years, with a mean problem rating of 6.3 out of 10. Moderators of the effect of CBT on HFNS problem rating score at 9 weeks were considered (Table 1) with interaction plots shown for those observed to be significant (Fig. 1). The graphs show the adjusted problem rating scores for each group (CBT and TAU) at 9 weeks divided into categories depending on the baseline moderators described, e.g. chemotherapy versus no chemotherapy. The only demographic variable observed to be a significant moderator of treatment effect was ethnicity. HFNS problem rating score at 9 weeks was lower in the CBT group compared to usual care for both white and non-white ethnic groupings, but the difference was greater for non-white women (−4.08, 95% CI −5.95 to −2.21, p < 0.001) compared to white women (−1.46, 95% CI −2.14 to −0.78, p < 0.001). The only baseline clinical characteristic observed to moderate the CBT effect was chemotherapy, with a larger difference in HFNS problem rating score at 9 weeks for those not receiving chemotherapy (−2.86, 95% CI −4.01 to −1.71, p < 0.001) compared to those who did (−1.30, 95% CI −2.11 to −0.49, p = 0.002). These findings, for both ethnicity and chemotherapy, remained significant even after controlling for baseline levels of depression.
Concerning health behaviours, exercise frequency, but not smoking status, was observed to significantly moderate the effect of CBT. Further analysis revealed that the significant moderating effect was due to no significant difference in adjusted HFNS problem rating at 9 weeks between the CBT and usual care groups (−0.17, 95% CI −1.36 to 1.02, p = 0.779) for patients who exercised four or more times per week. However, after controlling for baseline depression the moderating effect of exercise became non-significant, suggesting that the finding is likely to be due to lower levels of distress in those who exercise more frequently. Depressed mood (−3.36, 95% CI −5.28 to −1.44, p = 0.001) and anxiety (−2.90, 95% CI −5.34 to −0.45, p = 0.020) were significant moderators of treatment effect at 9 weeks. The simple slopes relate to a difference in the adjusted mean difference between CBT and TAU of −0.83, −1.15 and −0.82 in HFNS problem rating at 9 weeks, respectively, for a one standard deviation higher score on the moderator variable. Together these observations indicate that CBT is most effective for individuals reporting high levels of psychological distress at baseline (Fig. 1). Non-significant treatment effects were indicated for individuals scoring zero on the WHQ depression and anxiety subscales. 3.2. Mediators of CBT: how does it work? There were no significant changes in sternal skin conductance – the physiological measure of HFNS frequency, between the groups over the intervention period. With regards to HFNS beliefs and behaviours those receiving CBT reported more positive social HFNS beliefs (adjusted mean difference = −0.77, 95% CI −1.12 to −0.38, p < 0.01), HF coping/control beliefs (adjusted mean difference =
J. Chilcot et al. / Maturitas 78 (2014) 56–61
59
Fig. 1. HFNS problem rating at 9 weeks post-randomisation stratified by significant moderator variables, adjusted for age and baseline HFNS problem rating.
−0.97, 95% CI −1.33 to −0.60, p < 0.01) and beliefs about NS and sleep (adjusted mean difference = −0.99, 95% CI −1.40 to −0.60, p < 0.01), and reported more positive coping behaviours (adjusted mean difference = 0.88, 95% CI 0.60 to 1.17, p < 0.01) at 9 weeks compared to those receiving TAU. Accordingly these potential process variables were selected for the mediation analysis. Table 2 summarises the mediation analysis showing the estimate of the indirect treatment effect via each potential mediator on HFNS problem rating at 26 weeks and the proportion of the total treatment effect that is mediated by the variable. Changes in depressed mood, sleep problems and social beliefs all significantly mediated the treatment effect, although, the mediation was partial since in each path model the direct effect of treatment on HFNS problem rating at 26 weeks was significant. This suggests that CBT successfully reduced depression and sleeping problems and altered social beliefs regarding the impact of hot flushes, which in turn accounted for some of the improvements in HFNS problem rating. However, the largest mediation effect was HF coping/control beliefs (standardised beta = 0.50, p < 0.01), which accounted for 60.3% of the total effect of treatment on HFNS problem rating. 4. Discussion The results of this study suggest that CBT is effective in reducing the impact of HFNS for women following breast cancer treatment regardless of age, BMI, time since breast cancer diagnosis, menopausal status at time of diagnosis, or type of cancer treatment (radiotherapy or chemotherapy or endocrine treatment). However, while CBT worked well for women who had received chemotherapy, it worked rather better for those who did not. This could not be explained by initial level of mood or hot flushes, as we adjusted for baseline scores; it is possible that there are physiological explanations given that chemotherapy can induce HFNS. Interestingly, the physiological measure of HFNS frequency (sternal skin conductance) did not significantly change in this trial, whereas in a parallel study of well women going through the menopause
transition or postmenopause (MENOS2) there were small but significant reductions following CBT [24]. Women of non-white ethnicity improved more following CBT than those of white ethnicity. This was the case after we controlled for baseline mood and HFNS, and also for the moderating effect of mood. Ethnic differences in HFNS reporting have been previously documented [25], and other factors might explain this result. The finding requires replication, as there were only 14 women in this category (nine black British and five of Asian or mixed ethnicity), but overall these findings suggest that CBT is beneficial regardless of ethnicity. Interestingly, we found that those who were more distressed at baseline, in terms of depressed mood and anxiety scores, reported greater improvement. The only sub-group of women who showed no benefit were those who on entry to the trial participated in high levels of exercise (four times a week or more), but this finding became non-significant when the moderating effect of depression was controlled. Exercise has been associated with lower prevalence rates of HFNS and with lower levels of depressed mood and was found to impact on HFNS frequency in a recent trial [26]. Therefore, in women who frequently undertake exercise and report low levels of psychological distress CBT may either be unnecessary or ineffective at reducing HFNS. When we examined how the CBT works, we found that the main mediator was beliefs about coping and control over hot flushes, and additional mediation was provided by depressed mood, sleep and changes in beliefs about hot flushes in social situations. Thus learning to control and cope with the HFNS was an important factor in reducing problem rating of HFNS. These results generally support the cognitive model of HFNS [15,27] and suggest that CBT works mainly by changing cognitions, i.e. the cognitive appraisal of HFNS, but also mood and sleep problems. The CBT targeted cognitions, behavioural reactions, stress/wellbeing and the impact of night sweats upon sleep, so these results are heartening as they show that improvement in these areas was clearly associated with clinical outcomes. It is generally not easy to find satisfactory control conditions for complex psychological interventions, in the same
60
J. Chilcot et al. / Maturitas 78 (2014) 56–61
Table 2 Summary of the indirect effects between treatment group and Hot Flush Problem Rating. Mediator
Indirect effect (95% CI) Group → mediator → HF Problem rating
% of total effect explained by the mediator
Depression Anxiety Stress Sleep problems Social beliefs Control beliefs Sleep beliefs Positive coping behaviours
0.18* (0.00 to 0.36) 0.23 (−0.04 to 0.51) 0.09 (−0.03 to 0.20) 0.24* (0.04 to 0.43) 0.24* (0.03 to 0.46) 0.50** (0.15 to 0.90) 0.32 (−0.05 to 0.70) 0.28 (−0.04 to 0.61)
24.3 31.1 11.5 30.6 29.1 60.3 41.2 33.5
Indirect estimates are the standardised change in Hot Flush Problem rating. * p < 0.05. ** p < 0.01.
way that placebos are used for medical trials [26] so mediation analysis is important to show that the treatment, such as CBT, is actually changing what it is intending to change and that the effects cannot be entirely explained by attention or ‘non-specific factors’. These results are supported by those of a qualitative study based on interviews with women who had CBT in the MENOS1 trial [28]. The women reported that CBT improved their ability to cope with their HFNS and that they ‘regained a sense of control’, which is consistent with our findings that changes in beliefs about HFNS mediate improvements in symptom experience. A sense of control over HFNS has been associated with subjective HFNS distress and overall wellbeing [17], and it has been argued that interventions to counter lack of control could be particularly beneficial for cancer patients experiencing treatment related symptoms [29]. Study limitations include the sample size; some non-significant potential moderators and mediators might not have been detected due to power considerations. While we did not correct for multiple testing because the analysis was exploratory [30], it is useful to note that even using a conservative Bonferroni correction the effects of exercise and depression would have remained significant. Further confirmatory research could include larger samples in multicentre trials in which the impact of variables such as ethnicity could be clarified. In conclusion, this study shows that CBT is beneficial to breast cancer survivors with troublesome treatment related symptoms and significantly reduces the impact of HFNS. Those who were most distressed benefited more but there were no contraindications to CBT. CBT appeared to work by changing HFNS beliefs, i.e. cognitive appraisal, and by improving mood and sleep, suggesting that the treatment might be tailored to individuals depending on their scores on these active components. Contributors J.C. wrote the first draft; J.C. and S.N. conducted the statistical analysis; M.S.H. designed the study; all authors contributed to writing, interpretation and the final manuscript. Competing interest The authors have no competing interests to declare. Funding This study (C8303/A6130).
was
supported
by
Cancer
Research
UK
Ethics Ethical approval was obtained from the UK NHS Research Ethics Committee (South East London 2 REC, ref: 08/H0802/106).
Patient consent We confirm that consent was obtained from participants in the study.
References [1] Kaplan M, Mahon S, Cope D, Keating E, Hill S, Jacobson M. Putting evidence into practice: evidence-based interventions for hot flashes resulting from cancer therapies. Clinical Journal of Oncology Nursing 2011;15(2):149–57. [2] Gupta P, Sturdee DW, Palin SL, et al. Menopausal symptoms in women treated for breast cancer: the prevalence and severity of symptoms and their perceived effects on quality of life. Climacteric: The Journal of the International Menopause Society 2006;9(1):49–58. [3] Hunter MS, Grunfeld EA, Mittal S, et al. Menopausal symptoms in women with breast cancer: prevalence and treatment preferences. Psycho-oncology 2004;13(11):769–78. [4] Biglia N, Cozzarella M, Cacciari F, et al. Menopause after breast cancer: a survey on breast cancer survivors. Maturitas 2003;45(1):29–38. [5] Cella D, Fallowfield LJ. Recognition and management of treatment-related side effects for breast cancer patients receiving adjuvant endocrine therapy. Breast Cancer Research and Treatment 2008;107(2):167–80. [6] Rada G, Capurro D, Pantoja T, et al. Non-hormonal interventions for hot flushes in women with a history of breast cancer. Cochrane Database of Systematic Reviews 2010, http://dx.doi.org/10.1002/14651858.CD004923.pub2. [7] Boekhout AH, Vincent AD, Dalesio OB, et al. Management of hot flashes in patients who have breast cancer with venlafaxine and clonidine: a randomized, double-blind, placebo-controlled trial. Journal of Clinical Oncology 2011;29(29):3862–8. [8] Hickey M, Emery LI, Gregson J, Doherty DA, Saunders CM. The multidisciplinary management of menopausal symptoms after breast cancer: a unique model of care. Menopause 2010;17(4):727–33. [9] Mann E, Smith MJ, Hellier J, et al. Cognitive behavioural treatment for women who have menopausal symptoms after breast cancer treatment (MENOS 1): a randomised controlled trial. Lancet Oncology 2012;13(3):309–18. [10] Duijts SF, van Beurden M, Oldenburg HS, et al. Efficacy of cognitive behavioral therapy and physical exercise in alleviating treatment-induced menopausal symptoms in patients with breast cancer: results of a randomized, controlled, multicenter trial. Journal of Clinical Oncology: Official Journal of the American Society of Clinical Oncology 2012;30(33):4124–33. [11] Ayers B, Smith M, Hellier J, Mann E, Hunter MS. Effectiveness of group and self-help cognitive behavior therapy in reducing problematic menopausal hot flushes and night sweats (MENOS 2): a randomized controlled trial. Menopause 2012;19(7):749–59. [12] Mann E, Smith M, Hellier J, Hunter MS. A randomised controlled trial of a cognitive behavioural intervention for women who have menopausal symptoms following breast cancer treatment (MENOS 1): trial protocol. BMC Cancer 2011;11:44, http://dx.doi.org/10.1186/1471-2407-11-44. [13] Eccles S, Aboagye E, Ali S, et al. Critical research gaps and translational priorities for the successful prevention and treatment of breast cancer. Breast Cancer Research 2013;15:R92. [14] Prigerson HG. Mind over menopausal symptoms in women breast cancer. Lancet Oncology 2012;13(3):227–9, with http://dx.doi.org/10.1016/S1470-2045(11)70381-3. [15] Hunter MS, Mann E. A cognitive model of menopausal hot flushes and night sweats. Journal of Psychosomatic Research 2010;69(5):491–501. [16] Hunter MS, Liao KL. A psychological analysis of menopausal hot flushes. British Journal of Clinical Psychology 1995;34(Pt 4):589–99. [17] Rendall MJ, Simonds LM, Hunter MS. The Hot Flush Beliefs Scale: a tool for assessing thoughts and beliefs associated with the experience of menopausal hot flushes and night sweats. Maturitas 2008;60(2):158–69. [18] Hunter MS, Ayers B, Smith M. The Hot Flush Behavior Scale: a measure of behavioral reactions to menopausal hot flushes and night sweats. Menopause 2011;18(11):1178–83.
J. Chilcot et al. / Maturitas 78 (2014) 56–61 [19] Cohen S, Kamarck T, Mermelstein R. A global measure of perceived stress. Journal of Health and Social Behavior 1983;24(4):385–96. [20] Hunter M. The women’s health questionnaire: a measure of mid-aged women’s perceptions of their emotional and physical health. Psychology & Health 1992;7(1):45–54. [21] Barsky AJ, Wyshak G, Klerman GL. The somatosensory amplification scale and its relationship to hypochondriasis. Journal of Psychiatric Research 1990;24(4):323–34. [22] Scheier MF, Carver CS, Bridges MW. Distinguishing optimism from neuroticism (and trait anxiety, self-mastery, and self-esteem): a reevaluation of the Life Orientation Test. Journal of Personality and Social Psychology 1994;67(6):1063–78. [23] Van Breukelen GJ. ANCOVA versus change from baseline: more power in randomized studies, more bias in nonrandomized studies [corrected]. Journal of Clinical Epidemiology 2006;59(9):920–5. [24] Stefanopoulou E, Hunter MS. Does pattern recognition software using the Bahr monitor improve sensitivity, specificity and concordance of
[25]
[26] [27] [28] [29]
[30]
61
ambulatory skin conductance monitoring of hot flushes. Maturitas 2013, http://dx.doi.org/10.1016/j.maturitas.2013.09.013 (in press). Gold EB, Sternfeld B, Kelsey JL, et al. Relation of demographic and lifestyle factors to symptoms in a multi-racial/ethnic population of women 40–55 years of age. American Journal of Epidemiology 2000;152(5):463–73. Avis NE. Breast cancer survivors and hot flashes: the search for nonhormonal treatments. Journal of Clinical Oncology 2008;26(31):5008–10. Hunter MS, Chilcot J. Testing a cognitive model of menopausal hot flushes and night sweats. Journal of Psychosomatic Research 2013;74(4):307–12. Fenlon DR, Rogers AE. The experience of hot flushes after breast cancer. Cancer Nursing 2007;30:E19–26. Balabanovic J, Ayers B, Hunter MS. Women’s experiences of Group Cognitive Behaviour Therapy for hot flushes and night sweats following breast cancer treatment: an interpretative phenomenological analysis. Maturitas 2012;72(3):236–42. Bender R, Lange S. Adjusting for multiple testing – when and how? Journal of Clinical Epidemiology 2001;54(4):343–9.