INTRODUCTION
Insomnia is one of the most common and crucial complications among cancer survivors and is related to an increased risk of fatigue, depression, and poor self-rated health status [
1,
2]. To manage sleep problems in cancer survivors effectively, it is important to know the size of sleep disorder and understand the associated factors.
By determining the prevalence and specific risk factors that increase the risk of sleep disorder in cancer survivors, primary care physicians can easily identify cancer survivors who are vulnerable to sleep disorder and, thus, provide timely intervention. However, the estimated prevalence of sleep problems among cancer survivors and associated factors are inconsistent between studies [
1,
3]. In a cross-sectional study using the Korea National Health and Nutrition Examination Survey (KNHANES) data, the prevalence of insufficient sleep (<5 hours) was 8.1% among cancer survivors and 3.7% among non-cancer participants [
2]. In contrast, in a cross-sectional study in the United Kingdom, 39.6% of cancer patients reported sleeping less at night, while the rate was 15.2% among age- and sex-matched non-cancer controls [
4].
Several epidemiological studies have suggested that diverse factors are associated with sleep disturbance among cancer survivors [
1,
3,
5-
8], but are controversial regarding several specific factors. A previous study based on 3,343 Danish breast cancer patients found that older age and depressive symptoms were associated with an increased risk of sleep disturbance [
5]. However, a post hoc analysis of a randomized clinical trial with 823 US cancer survivors revealed that survivors aged <58 years were significantly more likely to have insomnia than survivors >58 years [
1]. A study of 200 American breast cancer survivors showed that depression was not significantly associated with sleep disturbance [
6], which is an unexpected finding given that sleep disturbance is one of the main clinical manifestations of depression. Sleep disturbance was the most prevalent primary cancer site among breast cancer survivors (39%–45.8%) in a study of American cancer survivors [
1], whereas, in a Korean study, lung cancer survivors experienced sleep disturbance the most frequently (15.2%) [
7]. Menopausal status among breast cancer survivors [
3], and detrimental effects of treatment modalities, such as radiotherapy or bone marrow transplantation, were also suggested to be associated with insomnia [
8].
Therefore, we conducted the present study to evaluate the prevalence of sleep disorder and factors associated with sleep disorder among Korean adult cancer survivors. In addition, we assumed that the factors associated with sleep disorder in cancer survivors would differ according to sex and cancer site. Therefore, we conducted analyses by dividing our data according to sex and common primary cancer sites.
METHODS
1. Study Participants
This study employed a cross-sectional design. The study participants were adult (19 years or older) cancer survivors who visited two university-affiliated hospitals located in Seoul, South Korea from September 2014 to February 2017 for cancer survivorship care after completing active cancer treatment. This study was approved by the Institutional Review Board of two institutions (SMC2015–03–045 and H-1407-066-595). Written informed consent was obtained from all the study participants.
2. Study Variables
We collected information on sleep disturbances that occurred after cancer diagnosis as outcome variables using two questions: how often do you experience difficulty initiating or maintaining sleep? and when did your sleep problems begin? We then defined sleep disorder as ‘difficulty in sleep initiation or sleep maintenance 3 times a week or more frequently that started after cancer diagnosis’, according to the diagnostic criteria of insomnia from the Diagnostic and Statistical Manual of Mental Disorders, fifth edition [
9].
Using a self-administered questionnaire, we collected information on the demographic and socioeconomic characteristics, comorbidities, health behavioral factors, and psychological status of the study participants as predictor variables. Cancer-related information, including cancer site, cancer stage, time since cancer diagnosis, and cancer treatment modality, was also collected as predictor variables by reviewing medical records.
Cancer stage was categorized into three groups: the first group consisted of stage 0 (defined as carcinoma in situ) and stage 1, the second group included stage 2, and the third group comprised stage 3 or 4, after determining cancer stage for each cancer type following 8th edition of cancer staging manual by the American Joint Committee on Cancer [
10]. We categorized time since cancer diagnosis into four groups: ≤1 year, 2–5 years, 6–10 years, and >10 years. Monthly income was categorized into three levels (<2,000,000 Korean won [KRW], 2,000,000– <4,000,000 KRW, ≥4,000,000 KRW). Education level was categorized as follows: graduate degree, high school graduate, or middle school graduate. Marital status was categorized into two groups: individuals living with a spouse or partner, and those who were not. Smoking status was categorized as never smoked or ever smoked. Alcohol consumption was categorized into two groups: current nondrinkers and current drinkers. Regular physical exercise was defined as moderate- or highstrength physical activity for more than 30 minutes each time, with a frequency of twice or more per week. Comorbidities included diabetes, cardiovascular diseases, cerebrovascular diseases, chronic respiratory diseases, chronic liver diseases, and chronic renal diseases. We then categorized comorbidity into three groups (0, 1, or ≥2) on the basis of the number of coexisting comorbidities. Information on the family history of cancer was categorized into two groups: yes or no.
We also assessed the psychological status of the study participants as a predictor variable. We measured anxiety and depression using Hospital Anxiety and Depression Scale, a widely used psychometric instrument for screening anxiety and depression [
11], and a cut off value ≥8 was used for detecting anxiety and depression, respectively [
12]. We assessed the presence of fear of cancer recurrence (FCR) using a cut off value ≥13 for the severity subscale of the Fear of Cancer Recurrence Inventory (FCRI) [
13]. The FCRI is a multidimensional questionnaire composed of 42 items with seven subscale components of FCR; Cronbach’s α coefficient for the FCRI–Korean version was 0.85 for the total scale and 0.77 for the severity subscales [
14].
3. Statistical Analysis
We examined the prevalence of insomnia by sex. We compared demographic and socioeconomic characteristics, health behavioral factors, and clinical characteristics according to the presence of sleep disorder using the chi-square test or Fisher’s exact test. We performed a trend test using the Mantel-Haenszel chi-square test after excluding participants with missing data for each relevant variable.
We estimated odds ratios (OR) and 95% confidence intervals (95% CI) using multiple logistic regression analysis to evaluate the factors associated with sleep disorder in cancer survivors. Age, sex, education level, marital status, smoking status, alcohol intake, physical activity, comorbidities, family cancer history, cancer type, cancer stage, cancer treatment modality, and time since cancer diagnosis were included in the model as covariates. We selected covariates based on the statistical significance level (P<0.1) estimated from the univariate analysis in our study or by referring to other studies. We further estimated the OR (95% CI) for sleep disorder associated with each variable in the subgroups classified by cancer type (stomach, breast, and lung cancer). All statistical analyses were conducted using PASW SPSS Statistics ver. 18.0 (SPSS Inc., Chicago, IL, USA). Two-sided tests were performed and statistical significance was set at P<0.05.
RESULTS
Initially, 2,038 cancer survivors were recruited. A total of 145 survivors were excluded for the following reasons: missing data on sleep (N=5), missing data on psychological variables (N=2), inconsistent responses between sleep measurement variables (N=98), and cancer recurrence (N=40). Thus, 1,893 participants were included in the final analysis. Among the 1,893 participants (mean age, 58.1±9.9 years; female sex 68.0%), the prevalence of sleep disorder was 19.1% (20.3% in females and 16.5% in males).
Table 1 shows the findings of univariate analysis comparing cancerrelated characteristics according to the presence of sleep disorder. Sleep disorder were distributed differently across cancer sites, and their prevalence was highest among breast cancer survivors. Survivors who were recently diagnosed with cancer were more likely to experience sleep disorder (P=0.034) than those who were not recently diagnosed. With an increasing number of cancer treatment modalities, the proportion of sleep disorder increased (P<0.001) in female participants.
Table 2 shows the findings from the univariate analysis comparing the general characteristics of the study participants according to the presence of sleep disorder. Cancer survivors with a low household income (P=0.009); living without a spouse or partner (P=0.022); higher levels of FCR (P<0.001), anxiety (P<0.001), and depression (P<0.001); and low EuroQol Visual Analog Scale (EQ-VAS) scores (P<0.001) were significantly more likely to have sleep disorder. Higher FCR levels were consistently associated with sleep disorder in both males and females (P=0.010 and 0.001, respectively). Specifically, female cancer survivors with low education level, anxiety, depression, and low EQ-VAS scores were more likely to experience sleep disorder (P=0.006, P<0.001, P=0.004, and P<0.001, respectively), whereas only higher FCR was significantly associated with sleep disorder in male cancer survivors.
The multivariate-adjusted associations of sleep disorder with the predictor variables are shown in
Table 3. With regards to age, a 1-year increase of age was associated with higher risk of sleep disorder in males (OR, 1.04; 95% CI, 1.01–1.07), while inverse association was found in females (OR, 0.97; 95% CI, 0.95–0.99). Living with spouse or partner was associated with a low risk of sleep disorder (OR, 0.43; 95% CI, 0.20–0.95), in male cancer survivors. Female participants with higher levels of FCR and anxiety were 1.45 (95% CI, 1.06–1.98) and 1.78 (95% CI, 1.25–2.55) times more likely to have sleep disorder, whereas those with higher EQ-VAS were 0.59 (95% CI, 0.43–0.82) times less likely to have sleep disorder.
Table 4 shows the findings of the subgroup analyses according to the cancer site. In stomach cancer survivors, female sex (OR, 4.25; 95% CI, 1.44–12.58) and increased age (OR, 1.06; 95% CI, 1.03–1.09) were associated with a higher risk of sleep disorder, while post-menopausal status (OR, 0.29; 95% CI, 0.11–0.76) was associated with a lower risk of sleep disorder. This association between menopausal status and sleep disorder in stomach cancer survivors was statistically insignificant (OR, 0.63; 95% CI, 0.16–2.49) when we re-analyzed the data using only female participants (
Supplement 1). In breast cancer survivors, postmenopausal status (OR, 2.36; 95% CI, 1.30–4.28), higher level of FCR (OR, 1.58; 95% CI, 1.05–2.36), and anxiety (OR, 1.98; 95% CI, 1.23–3.19) were associated with a higher risk of sleep disorder, while increased age (OR, 0.96; 95% CI, 0.93–0.98) and higher EQ-VAS score (OR, 0.60; 95% CI, 0.40–0.91) were associated with a lower risk of sleep disorder. In lung cancer survivors, 1-year increase of age (OR, 0.86; 95% CI, 0.77–0.97), higher income (≥4,000,000 KRW per a month) (OR, 0.12; 95% CI, 0.02–0.92), and higher EQ-VAS score (OR, 0.17; 95% CI, 0.03–0.86) were associated with a lower risk of sleep disorder.
DISCUSSION
This cross-sectional study evaluated the prevalence of sleep disorder and factors associated with sleep disorder among Korean adult cancer survivors. The prevalence of sleep disorder was approximately onefifth of the study participants, and the prevalence was highest in breast cancer survivors. We found that age, menopausal status, FCR, anxiety, and quality of life in female participants were independently associated with sleep disorder, although the association varied according to sex and cancer site.
In the present study, the prevalence of sleep disorder (19.1%) was higher than that reported in other Korean studies of cancer survivors. According to a study using data from KNHANES, 8.1% cancer survivors experienced short sleep duration (5h/d) [
2]. When ascertaining nonorganic sleep disorder using F51.x code of International Classification of Disease, 10th revision (ICD-10), Lee et al. [
15] found that 2.08% of 31,579 Korean cancer patients recruited at five major sites (stomach, liver, colorectal, lung, and breast cancer) experience sleep disturbance. Park et al. [
7] assumed that 8.21% of 30,400 Korean patients with 10 common cancers suffered from sleep disturbance, based on F51 or G47.0 (insomnia) codes of ICD-10 and prescription records of typical sedatives or hypnotics (either zolpidem or triazolam). The difference in prevalence of sleep disturbance can be explained in two ways. First, the definition of sleep problems varied between studies. Insomnia is characterized by difficulty in initiating sleep, maintaining sleep, and/or early morning awakening associated with impairment or significant distress in daytime functioning, that occurs at least three nights per week for a minimum of 3 months [
9]. However, herein, we adopted a slightly less stringent approach to evaluating sleep disorder by focusing on the frequency of subjective sleep disturbances rather than adhering strictly to the overall duration criteria. This approach may have resulted in an overestimation of the prevalence of sleep disorder compared to previous studies. On the other hand, studies in the United States and Canada that used different diagnostic criteria found that up to 33–43% of cancer patients suffered from sleep disturbance [
1,
3]. In our study, sleep problems of cancer survivors with subjective or short sleep disturbance might have been underestimated. Second, our study participants were selected from the cancer survivorship clinics of university-affiliated hospitals who could be diagnosed with cancer and treated for cancer more recently than participants in other studies who were selected for general public health surveys. Therefore, participants in our study were relatively more susceptible to sleep disturbances due to the detrimental effects of recent hospitalization [
16]. The finding that survivors whose time since cancer diagnosis was less than a year reported the highest prevalence of sleep disorder seems to support this explanation. Third, we recruited study participants from a cancer survivorship clinic where regular surveillance was performed to detect cancer recurrence. Before and during periodic visits to clinics, cancer survivors may worry about undesirable surveillance results. Therefore, those survivors could be more susceptible to negative psychologic factors, so called ‘scanxiety’ [
17], which may hinder sleep quality and result in over-presentation of sleep disorder.
High FCR and anxiety were independently associated with sleep disorder in female cancer survivors. Patients with depression are known to experience increased rapid eye movement (REM) sleep duration, decreased REM sleep latency, and slow-wave sleep, leading to disturbance of sound sleep [
18]. Anxiety could affect sleep condition by regulating both neurotransmitters (norepinephrine, acetylcholine, and GABA) and specific brain area (amygdala, insula, and anterior cingulate cortex) that control sleep condition [
19]. FCR could result in negative intrusive thoughts that further exacerbate survivors’ anxiety [
20,
21]. A previous systematic review demonstrated that psychiatric diseases, such as anxiety and depression, have bidirectional relationships with sleep disturbance [
22]. In other words, sleep disturbance could be both a symptomatic result and precedent cause of psychiatric comorbidities. Similarly, a previous prospective study that followed up 3,000 individuals from the general population demonstrated that anxiety (OR, 4.27; 95% CI, 2.63–6.94) and depression at baseline (OR, 2.28; 95% CI, 1.41–3.68) were associated with new onset of insomnia after 1-year follow-up [
23]. In the same study, participants who reported insomnia at baseline were more likely to experience new occurrence of high anxiety (OR, 2.30; 95% CI, 1.10–4.82) and depression (OR, 3.51; 95% CI, 2.11–5.83) on follow-up. According to a retrospective study, there was no significant difference in sex in both objective sleep index measured by sleep duration, sleep efficiency, arousal index, and wake frequency and subjective rating scales measured by commonly used sleep questionnaires in the absence of psychiatric comorbidities [
24]. However, in the presence of anxiety or depressive disorders, which are more frequent in females than males [
25], females might experience insomnia more frequently than males due to close relationships between insomnia and anxiety or depression [
26]. Along with these findings, the significant association between sleep disorder and FCR or anxiety observed in female cancer survivors in our study may be attributed to the residual effects of coexisting psychiatric conditions, even after the adjustment for the presence of psychiatric conditions.
In line with the presence of FCR and anxiety, a low EQ-VAS score was independently associated with a higher risk of sleep disorder. The poor sleep condition of cancer survivors could be both the result and cause of poor quality of life, such as low EQ-VAS scores. People who sleep poorly often experience fatigue, which may lower their quality of life. Conversely, people with severe fatigue usually experience frequent napping, which can result in difficulties in sleep initiation and maintenance at night. In our subgroup analysis, we found a positive association between low EQ-VAS scores and sleep disorder among breast cancer survivors, consistent with the findings of previous studies on breast cancer survivors [
3,
6].
Subgroup analyses according to cancer site showed that postmenopausal status was associated with a higher risk of sleep disorder in breast cancer survivors. The positive association between postmenopausal status and sleep disorder in breast cancer survivors may be explained by the peak incidence age of breast cancer. According to recent Korean statistics (2022), breast cancer most commonly occurs in individuals aged 45–54 years. Considering that the average menopausal age is 49 years in Korean women [
27]. breast cancer survivors are likely to be diagnosed with cancer and receive active cancer treatment during the transition to menopause. Both chemotherapy and hormonal therapy using selective estrogen receptor modulators (such as tamoxifen) can induce menopause [
28]. This menopausal transition could decrease the level of melatonin, the primary hormone that regulates the normal circadian rhythm [
29]. These reasons could explain the positive relationship between post-menopausal status and sleep disorder in breast cancer survivors.
Our study had several limitations. First, this was a cross-sectional study; therefore, the temporal relationship between cancer experience and the occurrence of sleep disturbance may be unclear. To overcome this problem, we measured sleep disturbances that developed only during and after cancer treatment. Secondly, we had no information on the psychological status of the study participants before their cancer diagnosis. Therefore, it is unclear whether the study participants had suffered from depression and/or anxiety before being diagnosed with cancer or if depression and/or anxiety developed after cancer diagnosis. Third, our study participants were recruited from cancer survivorship clinics of university-affiliated hospitals and may not be representative of Korean cancer survivors in general. Therefore, caution is needed in interpreting our results. Fourth, we did not collect information on sleep patterns using the most commonly used standard screening methods for sleep disturbance, such as the Pittsburgh Sleep Quality Index and the Epworth Sleepiness Scale. In addition, we were unable to measure the actual duration of sleep disturbance, which is a crucial diagnostic criterion for sleep disorder [
9]. For example, participants with sleep disturbances of less than 3 months do not satisfy the diagnostic criteria for sleep disorder. Therefore, misclassification bias is possible. Fifth, although the sample size of our study was relatively large (n=1,893), we cannot completely eliminate the beta error issue given that the association of sleep disorder with FCR in males (P=0.057) and family history of cancer in females (P=0.097) was borderline significant. Finally, because we evaluated the association between sleep disorder and a range of possible risk factors, we could not categorize the covariates into causal risk factors, confounders, or effect modifiers. To overcome these limitations, further studies with a longitudinal design, using more precise tools for measuring sleep patterns are needed based on more specific hypotheses for risk factors, confounders, and effect modifiers.
However, we were able to evaluate the association between sleep disorder and a range of factors with consideration of detailed cancerrelated factors, including cancer stage, time since cancer diagnosis, and total number of cancer treatment modalities.
In conclusion, we found that approximately one-fifth of Korean cancer survivors suffer from sleep disorder even after the primary completion of cancer treatment. Age, menopausal status, FCR, anxiety, and quality of life were associated with a higher risk of sleep disorder. As primary family physicians who can guide long-term survivorship of cancer survivors, we need to identify survivors with risk factors that deteriorate sleep conditions and provide timely intervention to improve their sleep qualities [
30].