Breast cancer survivors (BCS) represent the largest cancer survivor group in the United States, and the number is growing.1
Up to 65% of BCS experience menopausal symptoms such as sleep disturbances and hot flashes after treatment for breast cancer, with symptoms extending well over 5 years after diagnosis.2,3
These symptoms can be due to estrogen withdrawal resulting from natural menopause, abrupt withdrawal of hormone therapy at time of diagnosis, chemotherapy-induced ovarian dysfunction, and/or long-term treatment with estrogen ablation therapies. Even if women are post-menopausal (cessation of menstrual cycle for 12 months or more) prior to diagnosis, symptoms can often reoccur as a result of cancer treatment. These symptoms have been reported to negatively affect health-related quality of life in BCS.4
Sleep disturbances and hot flashes are menopausal symptoms that continue to be research priorities at the national level. Sleep disturbances are physiological, psychological, environmental, or behavioral nighttime disruptions that have an impact on daytime functioning and are defined as perceived or actual disruptions in nighttime sleep or daytime wakefulness. For BCS, the most common sleep disturbance is chronic insomnia, in which women have a hard time falling asleep and staying asleep. Hot flashes are a sudden sensation of heat over the face, neck, and chest, and may be accompanied by patches of flushed skin. Patients may experience sudden drenching perspiration with or without shivering.5
Hot flashes are frequently quite debilitating, and the best therapy (hormone therapy) is contraindicated in BCS. 6
Although these are common problems in BCS, there are few known effective interventions. Women are often offered separate pharmacological interventions for each symptom, and these typically have only minimal positive effects and often undesirable side effects. For example, medications for sleep disturbance in cancer patients have yielded variable results.7,8
Several trials of hot flash treatments such megestrol acetate, 6
and black cohosh17
have shown minimal success. Thus, these symptoms are exceptionally fertile areas for study using nontraditional therapies.
One non-traditional therapy receiving attention in the cancer and non-cancer research literature is acupuncture. It has been tested in several small cancer-related studies18–26
and a few randomized-controlled, non-cancer studies, with varying results possibly related to differences in subjective hot flash measurement and acupuncture technique.27–36
After the National Institutes of Health Consensus Conference on acupuncture,37
there were several positive prospective trials38,39
of acupuncture for relief of various symptoms such as adult postoperative- and chemotherapy-related nausea and vomiting, postoperative dental pain, and sleep disturbances.
The theoretical framework for this study is based on Traditional Chinese Medicine (TCM), defined as a holistic system of health and healing grounded in the notions of balance, harmony, moderation, and prevention.40
TCM recognizes several causes for hot flashes (e.g., excessive Heart Fire or depleted Kidney Water). We have previously documented that hot flashes of either menopausal or iatrogenic etiology may be diagnosed as Kidney Water Exhausted.41
For this unfamiliar Eastern diagnosis, there are any number of internally logical acupuncture points and treatment regimens prescribed by acupuncturists. These are determined by TCM physical examination of the pulse and the tongue. 42
Similarly, sleep disturbance is associated with particular causes, generally related to a Yin
vacuity. When associated with hot flashes this portends chong
The involvement of several different community-licensed acupuncturists was a unique factor in this trial, since most published acupuncture data have been from academic practices; community-licensed acupuncturists traditionally have not been widely involved in academic acupuncture. This is in contrast to standard acupuncture practice in which community-licensed acupuncturists treat virtually all patients.
The purpose of this study was to evaluate the feasibility of acupuncture as a treatment for concurrent complaints of sleep disturbances and hot flashes in BCS. The objectives were: 1) to report patterns of acupuncture point use; 2) evaluate outcome expectancy, credibility, and acceptability relative to the intervention; and 3) evaluate patterns of symptom change over time.
The study was approved by the institutional review board. Forty-five BCS were screened for treatment for sleep disturbances and hot flashes during a 12-month period. Of those screened for the study, 8 were self-referrals responding to a mailing and 37 from physician clinic referrals. 35 of the 45women screen were not eligible or interested for the following reasons; (1) lived to far from the university for weekly visits (n=10), (2) did not have sleep disturbances and hot flashes (n=4), (3) were not 3 months post-treatment (n=3), (4) had prior acupuncture treatment (n=3), (5) had a current psychiatric disorder (n=3), (6) were not post-menopausal (n= 1), (7) did not want treatment for sleep and/or hot flashes (n=1), (8) were too busy to commit to the 8 weekly visits (n=9), and (9) was participating in another clinical trial (n=1). 10 BCS consented to the study.
Women were eligible if they (1) were least 18 years of age, (2) reported sleep disturbances over the past month, (3) reported daily hot flashes and frequent sleep disturbances with desire for treatment, (4) were peri- or post-menopausal, (5) agreed not to change sleep or hot flash medication use or dosages during the course of this study, (6) lived within a 100-mile radius of the study site, (7) were English-speaking, (8) had a known diagnosis of non-metastatic breast cancer and no history of other cancers, and (9) were at least 3 months post-completion of surgery, radiation, and/or chemotherapy. Women were excluded if (1) they had known psychiatric or cognitive disorders, (2) had a prior personal history of acupuncture use, or (3) were participating in other clinical trials.
The study used a single group, non-randomized, quasi-experimental design.
This was an 8-week study (see ). Each subject selected one of four participating certified community acupuncture providers for three acupuncture treatments over a 2-week period. Demographic and disease and treatment questionnaires were administered during the first week of the baseline assessment. Outcome expectancy and credibility were assessed after the first treatment had been completed (treatment week 1) and acceptability at the first follow-up visit. Physiologic monitoring using wrist actigraphy and hot flash monitoring occurred for 2 weeks prior to acupuncture as a determinant of baseline, as well as during the 2 weeks of acupuncture, and during two follow-up weeks (weeks 5, 8 of study). Additionally, self-reported symptom data were obtained using validated questionnaires at baseline, during treatment, and during each follow-up week.
The optimal acupuncture point regimen is unknown for either sleep disturbances or hot flashes. Acupuncture points used during treatment were dependent on the individual subjects’ TCM diagnoses as determined by the licensed acupuncturists. However, each of the contracted licensed acupuncturists agreed to use the same tailored treatment regimen for all three treatments for each individual subject. The protocol gave subjects the option to have treatments from one of four licensed acupuncturists who were located at two separate clinics and held a National Certification Commission for Acupuncture and Oriental Medicine certification and who provided appropriate therapy based on their experience and Traditional Chinese Medicine (TCM) diagnosis. All of the acupuncturists were Caucasian-American, had at least 6 years of patient experience (M=11.33, SD=8.39), had a master’s degree in Oriental Medicine from an accredited American school, and were members of the local and national professional organization for acupuncturists. None of the acupuncturists practiced under a different health care role such as nursing,
Measures are described below. Monitoring using wrist actigraphy and sternal skin conductance was performed as described below. Each of these devices required that data be downloaded every 7 days. Women were required to visit the onsite clinic each week to return study materials and have their monitors programmed for the subsequent week. Data were downloaded by trained study staff and stored in a secure, encrypted database in de-identified format on a secure server.
A standard questionnaire previously used in research studies was used to record basic demographic and health information including age, race, ethnicity, marital status, employment status, socio-economic status, education, menopausal status, medication, height, weight, and number and type of co-morbid conditions. Medical comorbidity was assessed using a standardized checklist of categories of common medical disorders.44
Frequencies were reported for individual co-morbid conditions.
Disease and treatment information
Disease and treatment information was abstracted from medical records. Information included date of diagnosis, stage of disease, and dates and types of treatments including surgery, chemotherapy, radiation, selective estrogen receptor modulators, and aromatase inhibitors.
The subject TCM assessments and acupuncture were recorded on lined treatment notes by each acupuncturist. These forms were not standardized questionnaires but open- ended notes commonly used in TCM evaluations. Details included the specific meridian point, location (left side, right side, both sides) of point placement, and side effects/tolerance of treatment session.
Outcome expectancy-credibility and acceptability of treatment
The 6-item Devilly and Borkovec Outcome Expectancy/Credibility Questionnaire was used to assess outcome expectancy and treatment credibility at the on-treatment assessment.60
Three questions assess outcome expectancy and three assess treatment credibility. For outcome expectancy items participants are asked to indicate how much they think treatment will help improve symptoms using a 0% to 100% scale, how much they feel treatment helped improve symptoms using a 0% to 100% scale, and how much they feel treatment reduced symptoms using response options from 1 (not at all
) to 5 (somewhat
) to 9 (very much
). For treatment credibility items, participants are asked to indicate how logical the treatment was, how successfully they think treatment reduced their symptoms, and how confident they are in recommending the treatment to a friend. Each question for treatment credibility is rated on a 1 to 9 scale. Responses to items on each subscale are standardized and summed to create a total subscale score. Higher scores indicate greater belief that the treatment would provide beneficial results (outcome expectancy) or greater belief in the credibility of the treatment (credibility). In studies of men and women, for expectancy Cronbach’s alphas were .79 to .90 and one-week test-retest reliability was r
< .001). Cronbach’s alphas for credibility ranged from .81 to .86 with one-week test-retest reliability of r
For this study, Cronbach’s alpha was 0.67 for the outcome expectancy subscale and .96 for the treatment credibility subscale.
The Acceptability Scale was administered at weeks 4 and 8 of follow-up. It is a 10-item investigator-designed questionnaire that contains 10 items taken from a questionnaire used in a prior study evaluating hot flash interventions.59
Women are asked to read each of 10 statements and choose a response ranging from 5 (strongly agree
) to 1 (strongly disagree
) with 3 being neutral. For example, women rated their level of agreement or disagreement with statements such as “I enjoyed the acupuncture sessions” (see ). Frequencies of individual item responses were reported in the analysis of this study; a total score was not calculated. Since frequency of individual items are reported psychometrics were not appropriate for this questionnaire.
Sleep was assessed using the Actiwatch® (Mini Mitter, Bend, OR) device. The wrist Actiwatch® contains an accelerometer that measures sleep and wake activity through motion of the wrist. The device is worn on the non-dominant wrist and resembles a regular wristwatch. The unit measures 1” × 1” × 0.25” and weighs 0.75 ounces. At the end of each recording session, the device is downloaded into a personal computer and customized software is used to quantify sleep efficiency calculated as a percentage [time spent asleep/time in bed] x 100) and total rest in minutes calculated as total number of minutes asleep ([minutes of day naps + minutes of night sleep] - minutes awake after sleep onset). This measure has been widely used among cancer patients as a valid, objective measure of sleep. 13,45
Although wrist actigraphy is not as accurate as polysomnography, the gold standard for measurement of sleep onset, sleep latency (the number of minutes it takes to fall asleep once in bed), number of nighttime awakenings, and total sleep time,46
actigraphy is more accurate than self-completed sleep logs.47
A 7- night sleep diary was completed by participants to facilitate wrist actigraph data analysis. This record measures time to bed, time out of bed, and naps. Information was entered into the Actiwatch® software for interruption of total sleep time (TST), wakefulness after sleep onset (WASO), sleep efficiency (ratio of time spent asleep to total time in bed), and number of nighttime awakenings.
Subjective sleep quality and disturbance
The Pittsburgh Sleep Quality Index (PSQI) measures sleep quality and disturbance retrospectively over the previous two weeks or month using self-report.48,49
The PSQI consists of 19 items that produce a global sleep quality score as well as the following seven component scores: sleep quality, sleep latency, sleep duration, habitual sleep efficiency, sleep disturbance, use of sleeping medications, and daytime dysfunction. Global sleep scores greater than 5 are considered to indicate poor sleepers. The items use varying response categories including Likert-type responses (0=not in the past month
to 3=3 or more times per week
). A sample item is, How often in the past month did you wake up and have to use the bathroom
? Psychometric properties of the PSQI such as internal consistency reliability have been widely supported in a variety of populations50–52
including BCS (n=102; α=0.80.)53
Sternal Skin Conductance Monitoring
Physiological hot flash frequency was assessed using sternal skin conductance monitoring Biolog® model 3991 (UFI, Morro Bay, CA).54
The Biolog® is powered by a standard 9-volt battery and is programmed to sample 12-bit skin conductance data at 1 Hz (once per second). It is placed in a bag and worn around the waist or across the shoulders. An event marker on the monitor can be depressed by participants to signal the subjective perception of a hot flash. This event marker timestamps the sternal skin conductance data and was used to help interpret the skin conductance data during data analysis. At the end of each monitoring session, the monitor is connected to a personal computer through the Biolog Interface Box® (UFI, Morro Bay, CA) and data can be downloaded. Sternal skin conductance monitoring is more specific to detecting hot flashes than measures of core or peripheral temperature,55
and it is highly correlated with self-reported hot flashes under controlled conditions.56
Laboratory studies indicate 95% to 98% of subjective hot flashes correspond to objective hot flashes among midlife women.56,57
Perceived hot flash interference
The Hot Flash Related Daily Interference Scale (HFRDIS) is a 10-item scale measuring the degree to which hot flashes interfere with nine daily activities as well as overall quality of life using Likert-type responses (0=does not interfere
to 10=completely interferes
For example, women rate how hot flashes interfere with work, social life, and mood. Psychometric analysis supports validity and reliability of the HFRDIS for use with BCS (n= 69 BCS; Cronbach’s alpha=0.96), with strong correlations with other hot flash variables and demonstrated sensitivity to change over time.58
Due to the feasibility nature of this study, we planned to accrue 12 subjects to generate hypotheses for further study. According to Julious,61
12 subjects is roughly the minimum sample size needed for a pilot study using continuous outcomes (e.g., number of hot flashes and sleep disturbance measures).
Demographic and breast cancer treatment information was analyzed with descriptive statistics and frequencies using SPSS 17.0 (Chicago, IL). Questionnaires and objective measures were evaluated for missing data points. There were no missing questionnaire data. The two baseline assessments were averaged for all outcome variables to facilitate statistical analyses.
Because of the number and complexity of acupuncture points used for treatments, descriptive statistics were used to determine the mean number of points administered per treatment per participant and reported side effects. Frequencies were used to determine the number of different acupuncture points and placement location. Paired t tests were used to determine if the mean number of acupuncture points received per treatment per participant changed over time. The data was reviewed and the assumptions of paired t tests were met. Alpha was set at p<0.05 for all analyses. With conducting multiple t tests, the overall probability of a Type I error is typically > 5%. The authors’ note that this inherent risk for a Type I error did not justify an adjustment in alpha given the exploratory nature of study, limited power due to the small sample size, and need for confirmation in future studies. This should be considered by the reader when interpreting the findings of the study.
Descriptive statistics and frequencies were used to evaluate outcome expectancy, treatment credibility, and acceptability of treatment. Chi square tests of independence were used to compare individual treatment acceptability questionnaire responses between the two follow-up time points.
Descriptive statistics were used to report mean global sleep and mean hot flash interference scores over time. For objective measures, sleep disturbances were evaluated using the average of common sleep variables: 1) wake after sleep onset (WASO) in minutes, 2) total sleep time (TST) in minutes, 3) sleep efficiency (range from 0% to 100%), and 4) number of nighttime awakenings. Due to missing data for sleep monitoring, three consecutive days were used for analysis. The 3-day time frame could not include the first day of recording and had to have all hours of the day represented with a complete sleep log. Averages were then calculated based on the software output with additional visual inspection to ensure accuracy. Physiologic hot flash frequency was calculated as the total number of hot flashes per 24-hour day.
Lastly, to determine patterns of change in sleep disturbances and hot flashes across the 8 time points, continuous outcome variables were evaluated using paired t tests (SPSS 17.0). This test was selected since it is more powerful than other non-parametric tests for small sample sizes.