Sleep Disturbance and Related Factors in Patients with Nasopharyngeal Carcinoma and Their Family Caregivers Prior to the Initiation of Treatment

Sleep disturbance is a common complaint in cancer patients. However, less is known about the parameters of sleep in patients with nasopharyngeal cancer (NPC) and their family caregivers (FCs) when they are about to begin treatment. We investigated the sleep quality in patients with NPC and their FCs before treatment and determined the related factors that predict sleep disturbance in these patients before therapy. A total of 101 patient-FC dyads were recruited. They completed the Pittsburgh Sleep Quality Index (PSQI) prior to treatment. No differences were found in sleep disturbance between patients (38.6%) and their FCs (31.7%). Patients reported significantly higher rates of short sleep duration than their FCs (P = 0.011). Logistic regression analyses showed that older patients were more prone to suffer from poor sleep quality before treatment (OR = 1.06, 95% CI = 1.01–1.10, P = 0.008), while patients with a higher BMI were less likely to experience sleep disturbance (OR = 0.83, 95% CI = 0.71–0.96, P = 0.012). Sleep disturbance is a significant problem in patients with NPC and their FCs before therapy. Older patients and those with a lower BMI appear to be more inclined to suffer from poor sleep before treatment.

between 36.7% and 76% of caregivers report mild to severe sleep disturbance 10,11 , which may negatively influence their emotional well-being 12 , quality of life 13 , and ability to provide care to the patients.
However, the majority of sleep studies have included patients with a variety of cancer diagnoses 3,4,14 . In addition, most of the subjects in studies conducted to examine sleep disturbances have been patients with breast cancer, prostate cancer, lung cancer, or brain tumours 3,4,14 . More research that focuses on a specific cancer diagnosis is needed to discern whether sleep patterns are the same in patients with different forms of cancer and to identify the factors that predict sleep disturbance in these patients. Although there has been some research evaluating sleep disorder in patients with NPC, these studies were nearly all focused on obstructive sleep apnoea syndrome 15,16 . Few studies have examined the sleep parameters of patients with NPC 17,18 . In one of the first studies of sleep disturbances in patients with NPC, Qin et al. 17 noted that 36.5% of patients reported sleep problems before the initiation of cancer treatment. However, the proportion of patients who displayed worse sleep quality significantly increased after radiotherapy (RT) and concurrent chemotherapy. In another study of 51 patients with NPC 18 , patients with poor sleep quality prior to RT comprised 37.3% of the subjects. Nearly 64.7% of patients experienced lower sleep quality after RT. Additional information on sleep parameters is warranted in patients with NPC before treatment because of the high prevalence and negative consequences of sleep disturbances in these patients. Furthermore, poor sleep quality before treatment was likely to correlate with worse sleep in patients with NPC after therapy 18 .
Certain demographic characteristics and other risk factors may correlate with subjective and objective sleep disturbances in patients. While a previous study reported that sleep dysfunction was associated with younger age 19 , in patients with NPC, age was found not to correlate with poor sleep quality 18 . Similarly, while a higher body mass index (BMI) has been correlated more with objective sleep disturbance at the initiation of RT 20 , a study of predictors of sleep disturbance found no association between BMI and sleep disturbance 21 . Along with these inconsistencies in risk factors for sleep disturbance in general, even less is known about the parameters of sleep in patients with NPC and their FCs about to begin treatment. Therefore, the objectives of this study were to describe the prevalence of clinically significant sleep disturbance in patients with NPC and FCs prior to the initiation of treatment and to compare the differences in their sleep parameters. The second aim was to determine which factors (demographic and personal characteristics, and cancer-specific variables) predict poor sleep quality in patients with NPC before therapy. We hypothesised that both patients with NPC and their FCs would report worse sleep quality prior to the initiation of treatment.

Results
Patient and FC Characteristics. A total of 120 patents with NPC were approached for enrolment in this study. In total, 19 patients (15.8%) and 3 FCs were excluded. Among the excluded patients, 7 had no FC, 5 refused to participate because they were afraid to take the test, and 7 patients were excluded because their FCs did not meet the eligibility criteria. A total of 101 patients were included in the final analysis. The 101 corresponding caregivers were interviewed. Table 1 lists a summary of the demographic characteristics of the 101 patient-FC dyads. Patients were significantly older (P = 0.010) and more likely to be male (78.2%, P = 0.000) than the FCs. Patients were diagnosed with stage II (13.9%), stage III (25.7%), or stage Iva + b (60.4%). The percentage of patients with Karnofsky performance score (KPS) = 90 was 100%, whereas most FCs had a KPS = 100 (98%, P = 0.000). The majority of the FCs were the patient's spouse or partner (49.5%) or their parent or child (32.6%). There were no significant differences between patients and FCs in terms of BMI and educational level.

Differences Between Patients and FCs in the Prevalence of PSQI Global or Subscale Scores
Prior to the Initiation of Treatment. The mean global PSQI scores of patients with NPC and FCs prior to the initiation of treatment were 7.2 (SD, 4.3) and 6.5 (SD, 4.1), respectively. A total of 39 (38.6%) patients had global PSQI scores greater than 7, meaning that they suffered from poor sleep quality (Table 2), while 32 (31.7%) FCs reported poor sleep quality. No differences were found between patients and their FCs in the prevalence of clinically significant levels of sleep disturbance on the basis of the cut-off scores for the PSQI global score (P = 0.302).
There was no difference between patients and their FCs in the occurrence rates of any of the PSQI subscale scores except for sleep duration before treatment. Patients reported significantly higher rates of short sleep duration than their FCs (P = 0.011). Of note, there was a low rate of use of sleep medications for both patients and their caregivers (6.0% and 4.0%, respectively). Additionally, no statistically significant difference between the two groups was found (P = 0.466).

Relationships Among the Demographic and Clinical Characteristics and PSQI Global Scores and
Subscale Scores of Patients Before Treatment. As shown in Table 3, the PSQI global score and the use of sleep medication of patients before treatment was significantly positively associated with clinical stage. Sleep duration was correlated with gender. Significant correlations were found between sleep efficiency and age, level of education and disease stage. Sleep disturbance was positively associated with age, but negatively correlated with education level. Daytime dysfunction was positively associated with education.

Predictors of Poor Sleep Quality in NPC Patients
Before Therapy. Logistic regression analyses showed that age and BMI were predictors of poor sleep quality in patients with NPC prior to the initiation of treatment (Table 4). Older patients were more prone to suffer from poor sleep quality before treatment (OR = 1.06, 95% CI = 1.01-1.10, P = 0.008), while patients with a higher BMI were less likely to experience sleep disturbance (OR = 0.83, 95% CI = 0.71-0.96, P = 0.012). None of the other values correlated with poor sleep quality in patients with NPC before treatment.

Discussion
This study is the first to assess the differences in the prevalence of sleep disturbance in patients with NPC and their FCs using subjective measures and to identify the predictors of sleep disturbance prior to the beginning of treatment in these patients. Consistent with our hypothesis, the results showed that the prevalence of sleep disturbance before treatment was similar for both patients and their FCs. However, the patients reported significantly higher rates of short sleep duration than their FCs. Older patients were more prone to suffer from poor sleep quality prior to the initiation of treatment, while patients with a higher BMI were less likely to experience sleep disturbance.
As the favourable outcomes of patients with NPC after treating with RT and chemotherapy, quality of life in these patients have received more and more attention. Sleep disturbance, which has been proven to affect quality of life, is a common symptom observed in oncology patients 22 , including patients with NPC. The results of this study showed that 38.6% of patients had clinically significant levels of sleep disturbance on the basis of global PSQI scores greater than 7 prior to the initiation of treatment, which is higher than the general population 5,23 . This percentage is similar to that reported in previous investigations of patients with NPC, which found sleep disturbance in approximately 37% of patients before treatment 17,18 . However, the cut-off point of sleep disturbance according to the global PSQI scores in these reports was greater than 5, which is lower than the cut-off point (>7) used in our study. Therefore, the proportion of sleep disturbance in our study is higher than those of previous investigations. The reasons for the difference in results may be that our study enrolled more older patients who were more likely to suffer from poor sleep quality before treatment.
There is an increasing body of literature suggesting that FCs are at high risk for sleep disturbance before, during, and after providing care to a family member 24,25 . Our present study showed that 31.7% of the FCs reported global PSQI scores that were higher than the cut point (>7), a prevalence rate that exceeds that of the general population 5 . To our knowledge, this study is the first to explore the prevalence of sleep disturbance in a sample of FCs of patients with NPC. Very little information is available on the level of sleep problems experienced by FCs. However, a previous study investigating FCs of patients with prostate cancer showed that the prevalence of sleep disturbance in FCs was 36.7% before initiating radiation therapy 10 , which is similar to our result. Conflicting results were reported in a study of FCs of patients with other cancer diagnoses (breast, prostate, lung, brain cancer), in which rates of sleep disturbance were between 40% and 50% at the initiation of RT 26 . Carter et al. 27 even found that 95% of the FCs of patients with advanced cancer reported severe sleep problems. Possible reasons for these inconsistencies may be related to the methods used to categorise the sleep disorder, the cancer stage, and differences in patients' cancer diagnoses. The prevalence of sleep disturbance in FCs of patients with a variety of cancer diagnoses may be different. Further investigation of specific cancer diagnoses is warranted to determine   Table 3. Correlation between PSQI scores and patients' demographic and clinical characteristics. Abbreviation: PSQI, Pittsburgh Sleep Quality Index; *P < 0.05; **P < 0.01. the sleep characteristics in patients with different forms of cancer. We found no significant differences between patients and their FCs in the occurrence rates of sleep disturbance based on the PSQI global score, which is consistent with most findings from previous studies 26, 28 .
In the current study, 58.5% of patients slept less than 7 hours (score >0) before treatment, while this rate was only 36.6% in FCs. Thus, patients reported significantly higher rates of short sleep duration than their FCs. This result is inconsistent with a study by Carney, who found no differences in sleep duration between oncology patients and their FCs 26 . According to the standard of 420 minutes proposed by Buysse et al. 29 , our result indicates insufficient duration of sleep among these populations.
It should be noted that the vast majority of patients and their caregivers in this study had never used sleep medication. Only 6.0% of patient and 4.0% of FCs had ever taken medicine, a non-significant difference. These percentages, however, are slightly lower than the 17% rate of sleep medication use reported by a study of FCs of patients with advanced cancer 30 . A possible reason for the higher rate in that study is that FCs report more caregiving burden and sleep disturbance (72%) when caring for patients with a variety of cancer diagnoses or advanced cancer. Although a previous study of oncology patients and their FCs showed that patients had higher scores for use of sleep medication than FCs 26 , most FCs did not take medicine to treat their sleep problems, which is similar to previous reports 24,31 . The reluctance to use sleeping medications may be because FCs were worried about the adverse effects of the medicine or because these medications could interfere with their ability to perform caregiving duties at night. Additional reasons can be found in the results from a previous study in which 35% of patients thought that the sleep problem was not particularly important, 30% thought that sleep disturbance may be transient, and 11% did not experience consequences of the disturbance such as daily weakness 32 .
Regarding the risk factors influencing poor sleep quality in patients with NPC prior to the initiation of treatment, the current study found that age was a predictor of sleep disturbance before treatment. Older patients were more likely to suffer from worse sleep. This finding is consistent with a previous study that reported a greater prevalence of poor sleep in older adults than in younger adults 33 . However, in contrast to prior findings with cancer patients, several studies have reported that younger age was associated with sleep disturbance 19,34,35 . Moreover, one study of patients with NPC even found that age did not correlate with poor sleep quality 18 . These inconsistent results may be due to the cancer type under investigation or the time point of the measurement. Our study evaluated the quality of sleep in patients with NPC prior to the initiation of treatment. Another reason may be that elderly patients have lower expectations related to this symptom and consider sleep disturbance part of the normal ageing process.
This study also found that patients with a lower BMI were more prone to suffer from worse sleep before treatment. However, contrary to our result, Dhruva et al. 20 reported that a higher BMI in patients with breast cancer was correlated with more objective sleep disturbance at the initiation of RT. This contradictory finding may be because 30% of the patients in their study had a BMI of more than 30, which is the cut-off score for obesity, whereas no patients' BMI score exceeded this level in our study. Obesity is associated with sleep disorders 36 . Of note, inconsistent results from studies of risk factors for sleep disturbance have not found a correlation between BMI and sleep disturbance 21,37 .
Although the current study did not identify any association between education and sleep disturbance in patients with NPC before treatment, a previous study reported that education was positively associated with an improvement in sleep quality among patients with poor sleep before breast cancer treatments 38 . Furthermore, less education was a significant predictor for poorer overall sleep in oncology patients at the initiation of chemotherapy 39 . The exact factors responsible for poor sleep in patients with less education are not clear. These inconsistent findings of the relation among age, BMI and education and sleep quality indicate that further investigation is needed in future.
Several limitations of this study need to be noted. First, the current study only evaluates the quality of sleep of patients before treatment. In addition, the cross-sectional study design does not allow an evaluation of how the sleep changes over time. Second, certain demographic characteristics of the patients and FCs do not match. The majority of the FCs were younger and female and had high levels of KPS. Therefore, differences in the results of patients and FCs should be interpreted with caution. Finally, in this study, factors that predicted poor sleep quality of patients with NPC prior to the initiation of treatment only included demographic and personal characteristics and cancer-specific variables. However, a great number of previous studies have shown that pain, depression, anxiety, fatigue, and cognitive function were all associated with sleep quality in cancer patients 38,40,41 . Furthermore, these symptoms seldom occur in isolation. Rather, they often appear as symptom clusters 42,43 , a term describing three or more concurrent symptoms that are related to each other. We did not investigate these symptoms as a group in the current study, although they should be considered in future studies.
Despite these limitations, findings from this study suggest that sleep disturbance is a significant problem in patients with NPC and their FCs prior to the initiation of treatment. Moreover, patients who are older and have a lower BMI appear to be more inclined to suffer from poor sleep quality before treatment. Longitudinal studies are warranted to determine the cause-and-effect relationships between sleep disturbance before treatment and demographics, personal characteristics, and cancer-specific variables and among symptoms of pain, depression,   44,45 . The PSQI had a Cronbach's α of 0.79 and displayed good test-retest reliability (r = 0.79) 46 . Therefore, in this study, a PSQI score >7 is defined as sleep disturbance, poor sleep quality or sleep disorder 47 .

Study Procedures.
Patients and FCs were approached by a trained nurse to discuss study participation and to administer the questionnaire. The details of the study were explained to each patient and their FCs. After patients gave their written informed consent, they were asked to complete the baseline study questionnaires. The total assessment lasted approximately ten minutes. Baseline questionnaires were completed prior to the initiation of treatment.

Statistical analyses.
Descriptive statistics and frequency distributions were generated for sample characteristics. The differences of demographic and clinical characteristics between patients and their FCs were compared by using Chi Square analyses or paired t-tests. Dyadic differences of PSQI were evaluated by using Chi Square analyses. Spearman correlation analysis was used to analyse correlations between global PSQI score and subscores and their demographic and clinical characteristics. Point-biserial correlations were used to analyze the nominal data. As mentioned in the instrument before, a PSQI score >7 was defined as sleep disturbance, poor sleep quality or sleep disorder. Logistic regression models were used to analyse the relationships between variables (age, BMI, gender, KPS, education level, marital status, and clinical stage) and poor sleep quality prior to the initiation of treatment. All tests were two-sided, and a p-value of < 0.05 was considered statistically significant. SPSS 16.0 (SPSS Inc., Chicago, IL, USA) was used for all analyses.