Sleep disturbance in cancer patients referred to an ambulatory integrative oncology consultation

Sleep disturbance (SD) is highly prevalent in oncology and negatively affects quality of life and mortality. Evidence supports the use of integrative oncology (IO) practices to treat SD, but there is limited published data on the characteristics of SD and factors associated with SD in IO. We determined the prevalence, severity, and factors associated with SD among cancer patients seen in an ambulatory IO consultation. Patients with cancer referred for initial outpatient IO consultation in 2017 were eligible. Patient demographics, clinical characteristics, and patient-reported outcomes (Edmonton Symptom Assessment Scale (ESAS), Measure Yourself Concerns and Wellbeing (MYCaW), PROMIS-10) were retrospectively reviewed. One thousand five hundred twenty patients were included in the analysis. The majority (70%) were women with breast cancer (42%). Nine hundred seventy-one (64%) patients reported significant SD with ESAS Sleep ≥ 4, yet only 11% expressed poor sleep as their primary or secondary concern for the IO consultation. The median SD (IQR) was 5 (3,7). ESAS scores for fatigue (adjusted OR 1.16; CI 1.07–1.26, p < 0.001), pain (adjusted OR 1.07; CI 1.00–1.15, p < 0.05), hot flashes (adjusted OR 1.14; CI 1.07–1.22, p < 0.001), well-being (adjusted OR 1.33; CI 1.22–1.46, p < 0.001), and psychological distress score (anxiety and depression) (adjusted OR 1.16; CI 1.01–1.11, p < 0.01) were independently associated with SD in multivariate analysis. Acupuncture was the most frequent intervention prescribed, 175 (35%). Other modalities included oncology massage (15%), health psychology (5%), and meditation (1%). Although 64% of patients seeking IO consultation reported clinically significant SD, only 11% were seeking integrative approaches for managing SD. ESAS fatigue, hot flashes, well-being, and psychological symptoms were significantly associated with SD.


Introduction
Sleep disturbance (SD) is common in cancer patients, with a prevalence of 23 to 61% compared to the general population (9-30%) [1][2][3][4][5]. SD is defined as "interference of normal sleep either by symptoms or condition with corresponding poor sleep efficiency [6]." Insomnia is defined as "difficulty with falling or staying asleep, which is associated with significant distress or impairment in daytime function and occurs despite an adequate opportunity for sleep [7,8]." Prevalence of SD in cancer can be attributed to emotional distress, physical pain, discomfort, the effects of cancer treatments, and adverse medication effects. Poor sleep has an impact on the immune system and inflammatory processes and plays a role in the psychoneuroimmunology axis [9]. A systematic review highlighted Santhosshi Narayanan and Akhila Reddy contributed equally as first authors.
Lorenzo Cohen and Sriram Yennurajalingam contributed equally as senior authors. that SD led to increased systemic inflammation [10]. Improved sleep not only leads to decreased emotional distress, but also attenuates pro-inflammatory and counter-regulatory cytokines and increases cell-mediated immunity suggesting the significant role of sleep in regulating key cancer biological processes [9,11]. Pharmacological management is often used by patients with severe SD. However, sleep medications are associated with adverse effects. Multiple systematic reviews conclude that cognitive behavioral therapy is an effective non-pharmacological intervention for SD [12]. Evidence suggests that other nonpharmacological therapies such as acupuncture, massage, exercise, music therapy, progressive muscle relaxation, meditation, yoga, and tai-chi also can help improve insomnia [13][14][15][16][17][18][19][20].
Cancer patients frequently use complementary and integrative medicine (CIM) therapies with recent estimates as high as 69% [21]. Most CIM modalities such as yoga, taichi, meditation, and music therapy are safe to use for most patients. The use of herbs and supplements for SD however requires health care professional supervision to prevent adverse effects, interactions, and toxicity [22][23][24]. To navigate the vast amount of information and to make an informed decision, patients often seek medical advice. IO programs are becoming more widely available in several cancer centers to address such needs and guide patients appropriately [25]. The practice of integrative oncology/medicine seeks to optimize health, quality of life, and clinical outcomes utilizing mind and body practices, natural products, and/ or lifestyle modifications from different traditions alongside conventional cancer treatment [26]. Therefore, SD can be treated using an integrative approach combining non-pharmacological and pharmacological therapies.
SD is a common symptom in cancer patients seeking our IO consultations [27]. Patients attending the Integrative Medicine Center at The University of Texas MD Anderson Cancer Center are often referred to meditation, yoga, or tai chi group classes, physical therapy, acupuncture, and health psychology as a part of the treatment plan for SD after initial consultation. There is limited literature on the frequency and severity of SD at the time of initial IO consultation, factors that are associated with SD, and IO strategies used by patients with SD. This descriptive, retrospective study examined the frequency and severity of SD at the initial IO consultation and explored factors associated with SD by analyzing data from clinical records of patient's evaluated in our outpatient IO clinic. We also describe integrative therapies commonly used by patients who reported SD.

Methods
Data was extracted from patients seen for new consults at our IO outpatient clinic at MD Anderson from January 1, 2017 to December 31, 2017. All patients presenting for a new patient consultation were asked to complete a series of assessments as part of the standard of care; only patients ≥ 18 years of age and presenting for their initial IO consultation were eligible for the study. Patients with missing ESAS score for sleep and absence of cancer diagnosis were excluded from the study. This study was approved by the Institutional Review Board of UT MD Anderson Cancer Center, and waiver of consent was obtained.

Intervention (IO consultation)
Prior to the patient-provider (physician and/or advanced practice provider) consultation, a nurse administered the patient-reported outcome (PRO) assessments. This comprehensive assessment of patient self-reported data is then entered into the medical record and made available to the physicians and advance practice providers immediately before the patient evaluation. During the IO consultation, each patient is evaluated comprehensively in terms of their overall health, goals of care, and strategies to optimize their health. Referrals are then made to other IO services according to the individual's physical, mind-body, or social needs.
Assessment for patients with self-reported SD includes evaluation of factors contributing to SD such as medical and cancer history, symptom burden, review of cancer treatments, medications, and supplements. Then, an integrative care plan is developed for SD which may include review of sleep hygiene, suggestion for medical evaluation (e.g., sleep study), and review of integrative approaches such as acupuncture for symptom control or sleep, health psychology for cognitive behavioral therapy-insomnia, meditation, music therapy for psychological distress, yoga or tai-chi/ qigong, or discussion of evidence-informed risks and benefits for herb/supplements or alternative treatments being pursued or considered by patients for SD or other reasons.

Measures
The patient demographics clinical characteristics and the following PRO assessments which are routinely collected as a part of the clinical care were retrospectively reviewed in all eligible patients from the electronic medical records. The PRO's included the Measure Yourself Concerns and Wellbeing (MYCaW), the Patient-Reported Outcomes Measurement Information System (PROMIS-10), and modified Edmonton Symptom Assessment Scale (ESAS). No information on insomnia specific scales were collected due to the retrospective nature of our study.

Measure Yourself Concerns and Wellbeing (MYCaW)
Patients completed a modified version of the MYCaW questionnaire [28]. Patients reported their top two concerns for their IO encounter from a list of symptoms and concerns.
The list includes sleep, integrative/holistic approach, herb/ supplements, diet/nutrition, pain, overall health, and stress/ anxiety, as well as an "other" category.

Edmonton Symptom Assessment Scale (ESAS)
Patient symptom burden was assessed using the modified ESAS [29]. Patients are asked to report on 16 items: 10 core symptoms (pain, fatigue, nausea, depression, anxiety, drowsiness, loss of appetite, decreased sense of well-being, shortness of breath, and sleep) and an additional 6 items (spiritual distress, financial distress, numbness/tingling, hot flashes, dry mouth, and memory)-as experienced in the last 24 h on a numeric scale of 0 to 10, with 10 being the worst. A change or difference of 1 on an individual item is viewed as a clinically significant difference in that symptom Table 1 Demographic and clinical differences of patients with sleep disturbances in patients referred for an ambulatory integrative medicine consultation 1 ESAS (SD) < or ≥ 4 = Edmonton Symptom Assessment Scale-sleep disturbance score less than 4 or equal to or greater than 4 * Data not available for all patients  [30]. Subscale scoring was as follows: global distress score (GDS, range: 0-90), sum of pain, fatigue, nausea, drowsiness, appetite, shortness of breath, anxiety, depression, and well-being scores; physical distress score (PHS, range: 0-60), sum of pain, fatigue, nausea, drowsiness, appetite, and shortness of breath; and psychological distress score (PSS, range: 0-20), sum of anxiety and depression. ESAS sleep disturbance scores of 4 or above have good sensitivity and a reasonable specificity to screen for SD and is what was used in this study to define clinically significant SD [31].

Patient-Reported Outcome Measurement Information System (PROMIS-10 scores)
PROMIS-10 scores intend to reflect the patient's own assessment of their QOL. The PROMIS-10 [32], an assessment of global health, includes 10 self-report items that can be divided into mental health and physical health subscales.
Responses are converted into T-score values, with T-score distributions standardized to the mean for the US population.
Higher scores represent better global, mental, or physical health.

Statistical analyses
The primary aim of this study was to determine the frequency and severity of SD in patients presenting for an IO consultation. The secondary aim was to describe the factors that are associated with SD in cancer patients and the IO treatments used. Descriptive statistics were used to describe the demographic and clinical characteristics of the patients. ESAS and PROMIS-10 scores were summarized with means, standard deviations, and ranges. We examined medical and demographic characteristics between those reporting clinically significant SD and those who did not use t-test for continuous characteristics and chi-squared test for categorical characteristics. We conducted univariate logistic regression analyses to examine independent factors associated with clinically significant SD. Variables significant in the univariate regressions were included in the logistic regression. A p-value of ≤ 0.05 was considered statistically significant. All statistical analyses were performed using Stata/MP v16.0 (College Station, TX).

Results
A total of 1566 patients were screened, and 1520 patients were included in the analysis. The reasons for exclusion were age below 18, no cancer diagnosis, and missing ESAS sleep. Table 1 represents the demographic and clinical characteristics of all patients, patients with ESAS sleep > / = 4 and less than 4. Seventy percent of the patients were women. Most common cancer diagnoses included breast (42%) or gastrointestinal (14%); close to one-third of the patients (32%) had metastatic disease. The median (IQR) for SD was 5 (3)(4)(5)(6)(7). Figure 1 shows the frequency and percentage for SD scores from the ESAS. Of 1520 patients, 1300 patients (86%) reported ESAS sleep score as 2 or higher, 971 patients 4 or higher (64%), and 447 patients 7 or higher (29%). However, only 164 of 1520 patients (11%) reported sleep as the primary or secondary concern on the MYCaW for seeking the IO consultation. The primary concern listed for consultation included integrative approach and overall health (316, 21%); diet advice (232, 15%); pain (221, 14%); stress/anxiety, depression, and relaxation (152,10%); neuropathy (144, 10%); natural product information (108, 7%); fatigue (78, 5%); and others such as dry mouth, exercise, hot flashes, appetite loss, memory, nausea, and spirituality. Comparing patients reporting clinically significant SD (ESAS sleep ≥ 4) to those who did not revealed that patients with SD tended to be younger than those without SD (SD = 55.6 vs. No SD = 57.9 years, p = 0.012) ( Table 1). Opioid use was also significantly higher in the cohort with clinically significant SD compared to those without clinically significant SD (45.4 vs. 35.1, p = 0.003). There were no other demographic or medical characteristic differences between those with and without clinically meaningful SD. Table 2 shows that the symptom burden was higher in the group with ESAS sleep ≥ 4. Table 3 shows univariate logistic regression analyses of factors associated with clinically significant SD, revealing that all ESAS symptoms, PROMIS, age, BMI, and opioid use were associated with SD. Multivariate analysis of the significant variables revealed ESAS fatigue (adjusted OR 1.16; CI 1.07-1.26, p < 0.01), ESAS pain (adjusted OR 1.07; CI 1.00-1.15, p < 0.05), ESAS hot flashes (adjusted OR 1.14; CI 1.07-1.22, p < 0.001), well-being (adjusted OR 1.33; CI 1.22-1.46, p < 0.001), and PSS (psychological distress -the sum of ESAS anxiety and depression) (adjusted OR 1.16; CI 1.01-1.11, p < 0.01) remained as factors independently associated with clinically significant SD. Age, BMI, opioid use, and symptoms such as nausea, numbness and tingling, and shortness of breath were not significant in multivariate model (NB: PHS, GDS, PROMIS PH, and PROMIS MH were not included in the final analysis as they are a combination of other variables or overlap significantly).

Discussion
While the majority (86%) of patients seen in our IO clinic reported SD, only 11% of patients reported SD as a primary or secondary concern for seeking the IO consultation. This could be due to a lack of awareness of integrative treatment options for SD, not prioritizing SD as a health concern, or that other symptoms or reasons were a priority over SD. Research by Degnon et al. found that patients worst symptoms on a numeric scale were not always consistent with what was causing them the most distress and what they prioritized for symptom management. Garland et al. also reported that cancer patients were reluctant to use sleep medications but felt that they were the only available choice for treating SD [33]. Evaluating SD and sleep management warrants more discussion and support.   (11,15) 14 (12,16) 12 (10,14) Our study found that clinically significant SD (ESAS ≥ 4) was correlated with other symptoms of physical or emotional distress. However, the cause-and-effect relationship cannot be determined with the current data as to whether SD aggravates some of the symptoms or if certain symptoms lead to SD. It is likely a bidirectional relationship. These findings were similar to other studies where ESAS symptoms were intercorrelated, especially sleep with anxiety and depression [34]. Our study found that opioid use; obesity; physical health symptoms such as fatigue, hot flashes, and pain; and psychological symptoms such as anxiety and depression were associated with SD in a multivariate model. This is consistent with other studies showing that anxiety, depression, fatigue, pain, and performance status are strong predictors of SD [35]. Studies have shown that insomnia often co-occurs with pain, fatigue, anxiety, and depression as a symptom cluster [36]. These symptom clusters may negatively impact the quality of life of cancer patients [37]. Table 3 Univariate and multivariate models for factors associated with sleep disturbances a PSS is the sum of depression and anxiety scores (total 0-20) b PHS equals the sum of pain, fatigue, nausea, drowsiness, appetite, and shortness of breath scores (total 0-60) c GDS equals the sum of pain, fatigue, nausea, depression, anxiety, drowsiness, appetite, sense of well-being, and shortness of breath scores (total 0-90). PROMIS10 scores include a physical health (PROMIS PH d ) scale (4)(5)(6)(7)(8)(9)(10)(11)(12)(13)(14)(15)(16)(17)(18)(19)(20) and mental health (PROMIS MH e ) subscale (4)(5)(6)(7)(8)(9)(10)(11)(12)(13)(14)(15)(16)(17)(18)(19)(20). Higher scores represent better mental, physical, or global health Note: PHS, GDS, PROMIS-PH, and PROMIS-MH were not included in the multivariate models as they are a combination of other variables (ESAS variables)  The relationship between opioid use and sleep disorders [38] and obesity and sleep disorders [39,40] has been well recognized and was noted in our findings as well. Therefore, for patients receiving IO with SD, integrative therapies such as acupuncture, massage, health psychology counseling, and meditation were common interventions prescribed by IO clinicians. A systematic review showed acupuncture to be superior to sham acupuncture for managing cancer-related insomnia [13]. However, the number of patients using acupuncture and oncology massage would have been higher if the service were more widely covered by insurance. Underutilization due to cost was shown in a cross-sectional survey of cancer patients on acupuncture use for pain [41]. Yoga and tai chi were also recommended, as well as sleep hygiene, but the yoga/meditation service use was limited secondary to cost and availability as well. We noted that CBT-I, although considered as the gold standard for management of insomnia [42], was not commonly used in our practice. IO physicians in our practice are not trained to perform CBT. Therefore, the lower CBT use could be related to the availability and focus of integrative/health psychologists versus patient-related factors such as stigma beliefs about therapy and requires further investigation. With CBT-I requiring multiple in-person sessions, other competing interventions as part of cancer care may have made this a less feasible choice for management of SD. There are several limitations to this study. We used ESAS score for sleep to measure SD in our patients. This scale is not designed specifically to assess clinically meaningful SD in cancer patients. It is also a measure of SD in the past 24 h and might not reflect a true assessment of SD in patients. However, due to prior data showing a good correlation of ESAS sleep ≥ 4 with clinically significant SD, and the retrospective nature of the study, we decided to use this as a primary measure. Also, the study is cross-sectional in nature, and does not report if patients had any improvement in SD after IO approaches. Further studies are needed. Not all IO interventions recommended to patients as part of their IO were specifically recommended for the management of SD. A prospective study design will help measure the impact of multimodal intervention in real-world IO practice. Lastly, our patient population was predominantly women with breast cancer adding to the limitation of generalizability of the study results to other cancer types and men. Despite these limitations, our data is informative to clinicians regarding the importance of screening for SD using patient-reported outcomes, although patients may not prioritize engaging in a discussion about the management of SD. Targeting modifiable risk factors such as hot flashes with IO modalities (e.g., acupuncture) may improve SD at the same time as treating hot flashes. Therefore, our preliminary data on IO interventions used in the management of SD could provide insight into and inform the development of a multi-modal approach incorporating pharmacological and integrative therapies. Also, factors associated with use of IM therapies need to be explored in future studies.

Conclusions and future directions
We found that SD is common in cancer patients presenting for IO consultation. Despite the availability of nonpharmacological interventions, only a small proportion of patients identified SD as a primary or secondary concern when seeking an IO consultation. Therefore, health care providers should consider screening cancer patients routinely for SD in their practice. ESAS fatigue, hot flashes, well-being, and psychological symptoms were significantly associated with SD. The IO clinical model, including routine comprehensive symptom assessment and multi-disciplinary management, can successfully engage patients in a discussion on addressing SD in the context of other health areas. Greater dissemination of evidence-based information on integrative approaches for managing insomnia would help patients develop a more comprehensive approach to their symptom management needs, using the best combination of pharmacologic and non-pharmacologic options. Future research should focus on multimodal interventions for managing insomnia and associated symptoms in cancer patients.