Factors associated with sleep quality during chemotherapy: An integrative review

Abstract Aim To identify the most important factors associated with sleep pattern changes in patients with cancer during chemotherapy treatment. Design An integrative review of the literature was performed between December 2017–August 2018. Methods Two independent reviewers searching the National Library of Medicine (PubMed/MEDLINE), Cumulative Index of Nursing and Allied Health Literature (CINAHL) and Latin American and Caribbean Literature in Health Sciences (LILACS), Scopus and Scielo. The process followed the recommendations of the PRISMA tool. A total of 16 articles were selected for the final study sample, including 11 cohort studies and 5 cross‐sectional studies. Results The predisposing factors for the most prevalent sleep disturbances were precipitants related to the disease and the treatment, such as fatigue, pain, depression, anxiety and distress. Predisposing factors related to lifestyle and demographic characteristics have a significant correlation with sleep disturbances.

among all patients. Such agents frequently cause multiple symptoms that occur simultaneously and are associated with a reduction in quality of life. Common symptoms include pain, nausea, vomiting, fatigue, depression and sleep pattern changes symptoms have the high prevalence (Miaskowski, 2014;Palesh et al, 2018;Ratcliff, Lam, Arun, Valero, & Cohen, 2014).
Sleep disorders are common in cancer patients undergoing chemotherapy and coexist with many other disabling symptoms (Induru & Walsh, 2014). Periods of sleep deprivation cause a decrease in immune function, increased levels of pro-inflammatory factors, a reduction in glucagon levels, dysregulation of emotional and cognitive skills, a significant drop in quality of life scores and aggravation of pre-existing health problems (Besedovsky, Lange, & Born, 2012).
Studies show that 30%-60% of oncology patients present with a sleep disorder, with insomnia being the most common (Palesh et al., 2012). The factors that influence these disorders can be classified as predisposing, precipitating and perpetuating, according to the 3Ps' model (Spielman, Caruso, & Glovinsky, 1987). Predisposing factors are enduring traits, such as female sex, advanced age or psychiatric history, that are known to increase the individual's susceptibility to insomnia. Precipitating factors include stressful events that occur during treatment, such as diagnosis, disease recurrence, effects of disease and treatment and late effects or other symptoms.
Perpetuating factors are those that maintain sleep disturbances after the resolution or adaptation of precipitating factors, such as maladaptive behaviours and sleep-related thoughts (Miaskowski et al., 2014).
The Oncology Nursing Society identified sleep pattern change as a research priority (Knobf et al., 2015). Although chemotherapy is a common treatment for oncology patients, research on disrupted sleep during this treatment is limited. A 2010 meta-synthesis of studies on this subject in women demonstrated this knowledge gap; the included studies had relatively small sample sizes, were limited to a specific type of tumour (breast cancer), did not assess predictors of the sleep disorder and assessed the problem at a limited number of time intervals (Enderlin et al., 2010;Mark et al., 2017). These limitations affected the quality of the evidence, led to inconsistent results and are indicators of the need for new studies with greater methodological rigour.
An understanding of the relationship between chemotherapy treatment and sleep pattern changes is essential because it will inform the development of interventions that health professionals can implement to control the factors that affect sleep quality in this population and will support evidence-based practice.
Understanding these aspects is essential for defining interventions that can improve quality of life during chemotherapy and prevent further damage and increased severity in patients. With this perspective, an integrative review was conducted to identify the most important factors associated with sleep pattern changes in patients with cancer during chemotherapy treatment. The guiding question for achieving this objective was "What are the factors associated with sleep pattern changes in cancer patients during chemotherapy?".

| AIM AND ME THODS
An integrative review of the existing scientific literature about the factors associated with sleep pattern changes in cancer patients during chemotherapy was conducted. The procedure followed the recommendations of the PRISMA tool (Moher, Liberati, Tetzlaff, & Altman, 2009). This research method allows the pre-existing literature on a given topic or question to be summarized in a systematic and orderly manner, which will support professionals in decision making, assist in improving clinical practice and reveal gaps in knowledge production that need to be filled with new studies (Souza, Silva, & Carvalho, 2010).
The articles were analysed in six stages: selection of the research question; selection of studies according to inclusion and exclusion criteria; presentation of the characteristics and categorization of the studies; analysis of the studies; interpretation of the results; and presentation of the knowledge synthesis (Souza et al., 2010).
The review was conducted between December 2017-August 2018 by two independent reviewers. The search of the databases was performed by one of the researchers and was based on the descriptors defined according to the research question. After this step, two reviewers independently analysed the articles in all stages.
Disagreements between the reviewers about the selected articles were resolved by a third reviewer.
The search queries used a combination of two or three descriptors or keywords in each database with the Boolean operator AND (Table S1).

| Eligibility criteria
The inclusion criteria established for article selection were full texts available in the selected databases without limits about the year of publication and published in English, Portuguese or Spanish; primary studies with quantitative designs including cancer patients undergoing chemotherapy; studies that used objective and subjective measurements for sleep evaluation; and studies that included an analysis of factors related to changes in the sleep pattern. The first step, which consisted of querying the databases according to the elaborated strategies, occurred between December 2017-January 2018. We adopted the definition of sleep disorders of the American Academy of Sleep Medicine, which includes four categories: dyssomnias, which include disorders of initiating sleep; parasomnias, which include manifestations of central nervous system activation; disorders associated with mental, neurological or medical disorders; and sleep disorders that include those without a defined aetiology

| Data extraction
The publications were examined first by reading the title and abstract, the studies that met the inclusion criteria were selected for full reading. The texts selected in full had their data related in an instrument built by the researchers with the following variables: identification of the article, objective of the study, name of the authors, year of publication, training of the authors, study design, population/ sample, location of the study, sleep assessment instrument used, level of evidence, results (risk factors) and conclusions. Allowing a synthesis and easier analysis of each study (Figure 1).

| RE SULTS
The search resulted in 454 articles ( Figure 2). Among the 454 articles found, those that did not meet the inclusion criteria were excluded, which resulted in 48 articles. Of these, all had the full text available in Portuguese, English or Spanish. About duplication, 22 articles were repeated in the selected databases; thus, 432 were subjected to abstract analysis. A total of 16 articles were selected for the final sample. Among these articles, nine (56.2%) were cohort studies, five (31.2%) were cross-sectional studies and two (12.5%) were cohort studies with a control group.
About the subjective and objective sleep evaluation instruments used to obtain the data, 10 studies (62.5%) used the Pittsburgh Sleep Quality Index (PSQI), eight (50.0%) used actigraphy and 2 (10.0%) used a sleep diary and the General Sleep Disturbance Scale. The Athens Insomnia Severity Index, Bergen Insomnia Scale, Epworth Sleepiness Scale, Insomnia Severity Index and WHI Sleep Disturbance Scale were among the instruments used in the studies.
Of these, six articles (37.5%) combined the use of the PSQI and actigraphy. Three studies (15%) used only actigraphy. The mean number of instruments per study was 1.5, and five studies (31.2%) used objective and subjective measures in combination.
The countries with the most publications on the subject were the United States with nine studies (56.2%), followed by China with two studies (12.5%), Italy with two studies (12.5%) and Brazil, Australia and Japan with one study each (18.7%) ( Table 1).
The studies included populations undergoing chemotherapy for the following cancer types: breast (25.0%), lung (12.5%), gynaecological (12.5%) and cervical (6.25%). In 37.5% of the studies, the sample was not restricted to a specific cancer type. Among all of the studies, 37.5% were conducted by doctors, 18.7% by nurses, 6.25% by psychologists and 18.75% by doctors and nurses together ( Table 1). None of the factors analysed in the studies was classified as a perpetuating risk. The analysed factors were grouped as predisposing or precipitating risks only, as shown in Table 2.
Among the studies that investigated predisposing factors, a significant association was observed between these factors and changes in sleep pattern and quality. In the study by Bischel et al. (2016), both advanced age and younger age, along with poorer physical functioning, more comorbidities, White race and social responsibilities were associated with a greater occurrence of sleep disturbances. Compromised quality of life was significantly correlated with the presence of sleep disturbances in two studies (Nishiura, Tamura, Nagai, & Matsushima, 2015;Romito et al., 2014) but was confirmed as a risk factor by logistic regression analysis only in the study of Romito et al. (2014). Altered physical and cognitive functioning also showed a significant correlation with altered sleep patterns and were identified along with precipitating factors such as distress, pain, depression, fatigue (Berger et al, 1999) and dyspnoea. Number of comorbidities and White race were significantly associated with changes in sleep patterns in two studies (Bischel et al., 2016;Phillips, Jim, Donovan, Pinder-Schenck, & Jacobsen, 2012). Other factors, including restless legs syndrome, smoking, low physical activity, previous surgery, low functional scale scores and social conditions (living alone and caring for children) were described in only one study each.
Among the studies that described precipitants of changes in sleep pattern, a significant association with fatigue, depression and pain symptoms was confirmed by logistic regression analysis.
Distress and anxiety symptoms were associated with the presence of sleep disturbances (Tian, Chen, & Zhang, 2015). F I G U R E 1 Database/portal and search strategy

Database Search strategy
PubMed "sleep wake disorders AND antineoplastic agents AND neoplasms" Scopus "sleep wake disorders AND neoplasms AND drug therapy" CINAHL "neoplasms AND chemotherapy AND "drug therapy" AND sleep wake disorders" LILACS "sono AND quimioterapia AND neoplasia"

| D ISCUSS I ON
The present review showed aspects relevant to sleep disturbances in cancer patients undergoing chemotherapy as for this popula- The factors classified as precipitants were the most commonly identified, especially in combination with one another (Nishiura et al., 2015;Romito et al., 2014;Tian et al., 2015); these factors included fatigue, depression, pain and distress (Berger, 1999;Ancoli-Israel, 2009 patients is an understanding that the symptoms are concomitant; such joint occurrence is known as symptom clusters, which are defined as groups of interrelated symptoms (Rha & Lee, 2017) with a negative impact on patient survival. Some studies included (Bischel et al., 2016;Chen, Yub, & Yanga, 2008;Nishiura et al., 2015;Phillips et al., 2012;Romito et al., 2014;Saini et al., 2013) in the present review found several combined symptoms that significantly interfered with sleep quality. There is a correlation between poor sleep quality and depression, pain, fatigue, distress and anxiety (Nishiura et al., 2015;Romito et al., 2014). The symptoms of fatigue, pain, depression and distress were the most prevalent.
Pain and sleep are interrelated and have interconnected physiological mechanisms, such as the activation of the neurons of the ventrolateral preoptic nucleus, which project into the tuberomammillary nucleus and affect the perception of pain due to its interaction with the brainstem nuclei. As the hypothalamus is connected to the locus coeruleus, pain triggers a noradrenergic reaction that has an impact on the control of the sleep-wake cycle (  The pain, the tumour itself, the treatment modalities, the patient's environment, the patient's lifestyle, the hormonal changes, the cytokine changes and the psychiatric issues related to the diagnosis or treatment of cancer have a significant effect on the worsening of the sleep pattern and the likely occurrence simultaneous symptoms (Kaplow, 2005). These data are relevant since the improvement of one symptom contributes beneficially to the others, which confirms the theory of symptom clusters. It was observed that some characteristics, such as the presence of sleep difficulties and the lifestyle before the beginning of treatment, combined with depression and fatigue that appear during treatment were responsible for a small but significant proportion of the variation in overall sleep quality (Phillips et al., 2012). These pre-existing altered sleep conditions tend to worsen with the onset of treatment. Modifiable risk factors among the predisposing factors for sleep pattern changes, such as smoking, sedentary lifestyle and social conditions, need to be better investigated in the cancer trajectory, although studies show positive correlations among these factors, especially those related to age (Beck et al., 2010;Bischel et al., 2016).
Controlling anxiety, a common symptom among oncology patients, may be an effective therapeutic approach to reducing sleep disturbance. A study of women with ovarian cancer observed that depression and anxiety were commonly identified with fatigue and sleep disturbances (Nho, Reul Kim, & Nam, 2017 A study compared healthy individuals to those with stage I and II cervical cancer undergoing chemotherapy and found altered sleep patterns in the latter group, especially among those receiving adjuvant treatment (Tian et al., 2015). Radiotherapy has an important effect on insomnia (Savard, Ivers, Savard, & Morin, 2015) and is significantly mediated by dyspnoea and night sweating, with greater effects in women with breast cancer.
In 2010, a meta-synthesis (Enderlin et al., 2010) that summa- There were also inconsistencies in the evaluation of the alteration in sleep patterns during treatment, due to the diversity of the instruments used (Enderlin et al., 2010). These results are consistent with the findings of the present review, which found a lack of clinical trials and a wide variety of instruments used for sleep assessment. This lack of homogeneity makes it difficult to compare results and conduct care planning that can be incorporated into clinical practice (Enderlin et al., 2010;Mark et al., 2017).
The decrease in physical capacity demonstrated in some studies (Romito et al., 2014;Tian et al., 2015) is a relevant and common problem, even though the nature and severity of the adverse consequences of treatment vary according to the type of cancer, the treatment regimen and the individual characteristics of the patients. Decreased physical capacity can affect a patient's ability to be fully functional at work, resulting in prolonged absences, suboptimal productivity and the total abandonment of work (Alfano, Kent, Padgett, Grimes, & de Moor, 2017). However, since most studies that have investigated physical functioning have significant biases that limit generalization, more research on this aspect is needed. The personal and social costs of physical, psychosocial and cognitive functioning problems that limit employment for individuals with cancer are modifiable impairments.
There is a need to better identify these problems early and refer patients for cancer rehabilitation and related interventions so that adverse treatment complications are managed successfully, and physical capacity is preserved. It is necessary for the treatment of cancer to evolve to help patients remain optimally healthy, functional and employed (Alfano et al., 2017). In patients with breast cancer, poor sleep quality was associated with the worsening of health-related quality of life (Sanford et al., 2013), a finding that was supported in the studies analysed in this review (Nishiura et al., 2015;Romito et al., 2014).
In the analysis of populations with other chronic diseases, the neurological disorder of restless legs syndrome has been identified as a relevant risk factor for altered sleep patterns (İn, Turgut, & Özdemir, 2016). In the present review, only one study investigated this variable; it deserves further study as the result showed a significant correlation with anxiety and worse quality of life (Saini et al., 2013).
A review study found that sleep should be evaluated in a multidimensional way; however, the gold standard for evaluating sleep-related problems needs to be better defined (Loh, Burhenn, Hurria, Zachariah, & Mohile, 2016;Ritmala-Castren, Lakanmaa, Virtanen, & Leino-Kilpi, 2014). It is believed that a combination of objective and subjective measures is the best way to evaluate sleep patterns.
Several questionnaires can be used for the rapid evaluation of insomnia symptoms and related problems (Schutte-Rodin et al., 2008).
The PSQI, which has good psychometric quality and internal consistency, is commonly used in research (Mollayeva et al., 2016). Among the 16 studies selected for the present review, 10 (62.5%) used the PSQI and three of these studies combined the PSQI with objective measures. As sleep is influenced by multidimensional factors, it can be inferred that studies that use both objective and subjective measures can demonstrate more consistent results.
Actigraphy, the second most commonly used instrument in the analysed studies, is the measure with the closest results to polysomnography, which is considered the best way of evaluating sleep quality.
Due to the practicality of actigraphy for investigating sleep disturbance, its use is recommended for patients with different clinical conditions (Riemann et al., 2017). As there is good agreement among studies that report correlations between the results of actigraphy and polysomnography, studies that use actigraphy offer consistent data for its use. Symptoms such as fatigue, menopausal stage, advanced age and changes in inflammatory markers showed the best evidence when used in combination with subjective measures and objective actigraphy results (Beck et al., 2010;Liu et al., 2012;Rissling, Liu, Natarajan, He, & Ancoli-Israel, 2011). Despite the improved readings of the sleep pattern when these two methods are combined, the agreement between the subjective and objective sleep efficiency in older people population was not similar, which suggest that even though actigraphy is recommended, it is not ideal for classifying sleep quality (Hughes et al., 2018).
This review found a large number of studies examining patients with breast cancer, as previously reported, which shows the concern with these breast cancer patients in light of the high worldwide prevalence of this disease (Brasil, 2014). However, in the meta-synthesis conducted by Enderlin et al. (2010), which analysed sleep patterns in women with breast cancer, the results did not present good external validity. In that review, the population samples from the studies were small, were predominantly Caucasian, had a very large age range, did not include comparison groups or analyses that considered age differences and did not observe exclusion criteria or covariates, such as menopausal status, comorbidities or depression, which may have influenced the findings. In addition to the need to improve the quality of studies with women with breast cancer, it is essential to investigate differences in sleep patterns in populations of patients with other types of cancer; this is strongly recommended as a current research priority (Knobf et al., 2015).
Changes in the circadian cycle (Mansano-Schlosser & Ceolim, 2017) were described in only one study; however, they represent an important risk factor that is exacerbated by chemotherapy and often associ-

| Implications for practice
The survey of the risk factors for sleep quality of patients undergoing chemotherapy indicates the need to implement interventions to control these symptoms. Such implementation is critical to improving the quality of life of affected individuals and influencing treatment efficacy and clinical outcomes. One of the limitations of the present review was that the therapeutic regimen of the treatments was not described in the studies, which could have provided a better understanding of the studied subject.
The nurses, as a reference professional engaged in care that prioritizes patient autonomy and centrality, are a determining element for the success of these goals, and they can develop and implement prevent and treatment's care.

| CON CLUS ION
This review has made important contributions to understanding the factors that influence the sleep patterns of cancer patients undergoing chemotherapy. Sleep disruptions are multifactorial, and these factors are often present simultaneously, which makes their effective management difficult. However, due to this characteristic, when an intervention benefits one symptom, it improves others.
The most prevalent factors were those related to the disease and to treatment, such as fatigue, pain, depression, anxiety and distress.
Predisposing factors related to lifestyle and demographic characteristics also have a significant correlation with sleep disturbances.

ACK N OWLED G EM ENTS
We thank the Teaching and Research Institute Sirio Libanes Hospital for their support.