The prevalence of insomnia among persons with cancer is higher than in the general population (up to 50% versus 4-22%) [1–3]; its prevalence among patients with advanced cancer admitted to Palliative Care Units (PCUs) has been estimated as being between 45% and 95% [4, 5]. Nevertheless, insomnia has received little attention from the oncology community compared with other symptoms such as pain and fatigue [6].

Insomnia is defined as the subjective perception of difficulty with sleep initiation, duration, consolidation, or quality that occurs despite adequate opportunity for sleep and results in some form of daytime impairment [7]. Insomnia is strongly correlated with patient satisfaction and with quality of life in cancer [6, 8]. It brings patients discomfort and suffering, and has physical and psychological consequences such as increase of fatigue, pain intolerance, irritability and depressive mood [6, 9]. However, patients tend to underreport insomnia [10], and the routine clinical assessment of insomnia is often suboptimal.

The primary goal of the treatment of insomnia should be first to relieve any underlying disorder that may be causing the sleep disturbance. As a variety of factors may influence insomnia in hospitalised patients suffering from advanced cancer, a multimodal treatment, including both pharmacologic and non-pharmacologic therapies, should be considered. A plan combining attention to sleep hygiene and cognitive-behavioral therapy with prescription of hypnotic medications may help to relieve insomnia in cancer patients and improve their quality of life [6].

In the literature, a variety of factors that may influence sleep quality among palliative care patients have been mentioned. The factors associated with insomnia most commonly cited are pain and psychological symptoms. Uncontrolled pain, pain treatment and “interference of pain with mood” have been suggested as being associated with insomnia [11–13], as well as low or variable moods, dreams, concerns, hopelessness, post-traumatic experience, anxiety and depression [4, 11, 13–16]. Anxiety has been associated with difficulties in falling asleep, with less restoring sleep and with nightmares, while depression has been associated with early awaking, non-restorative sleep, fatigue, nightmares and insomnia according to the Pittsburgh Sleep Quality Index [5, 15]. However, only a few studies have used specific and validated tools for the assessment of psychological symptoms in relation to insomnia [5, 11].

Other factors which could influence insomnia in cancer population are age, performance status and certain pharmacological treatments [11, 14]. Female gender and non-white race, as well as excessive consumption of coffee and/or alcohol, or certain physical problems like chronic kidney disease may also increase the risk of incident insomnia in the general population [17]. Empirical evidence suggests that biological factors such as cytokines [18, 19], genetic and metabolic disorders [20] and cortisol awaking response [19, 21] can contribute to the disruption of the sleep cycle.

Environmental factors, and especially those in PCUs as for example interruptions during the night by the nursing staff, noise coming from the ward or from another patient in the same room, or even having a room for individual use, could influence the sleep quality of the patients during admission. However, most studies evaluating the factors associated with insomnia have not taken environmental factors into account.

Patients are admitted to PCUs for rapid symptom control and intensive psychosocial care, normally for relatively short periods. The length of stay may vary from two days to more than two weeks [22, 23]. Disturbing symptoms are evaluated daily by professionals in PCUs, as their intensity and characteristics can change from one day to the next. Evaluating the presence of insomnia specifically during admission to PCU, instead of evaluating persistent or chronic insomnia, may help professionals to adapt the management of insomnia, taking the changing situation of symptoms, possible acute disease and type of unit into account. Studies evaluating the prevalence of insomnia in this population have often used instruments which refer to the preceding 15 days or four weeks or which take relatively long to respond to, making their use more difficult and less specific for the evaluation of insomnia during admission in PCUs [24]. Other studies have used a single question [12, 25].

The aims of the present study were: 1) to evaluate the frequency of insomnia during hospitalisation among the patients admitted to a PCU by asking a single question and by using an assessment tool that fits the characteristics of this population (the 3-item questionnaire Sleep Disturbance Scale) and 2) to study the association between insomnia and emotional distress, considering both anxiety and depression and using a validated tool, and taking also physical, environmental and other psychological factors into account.

One hundred and seventy-six patients were consecutively admitted to the PCU during the eight-month period. Sixty-one patients (34.7%) fulfilled the inclusion criteria. Thirty-four of them were men and 27 women, and their mean age was 71.5 ± 11 years. The majority of included patients suffered from cancer (95.1%), with a high prevalence of metastatic cancer.

Ninety-five patients (54.0%) could not be included because either they did not give informed consent (n = 2) or they did not fulfil the inclusion criteria because they were too cognitively impaired (n = 34), had a Karnofsky Index < 30 (n = 44), spent less than two nights at the PCU (n = 8), or had a Karnofsky Index < 30 and spent less than two nights at the PCU (n = 7).

Twenty admitted patients (11.4%) were not assessed for inclusion either because they died (n = 5), or were discharged (n = 15) before assessment.

The frequency of insomnia among the patients admitted to the Palliative Care Unit (PCU) was considerably high both when measured with the single question (62.3%) and with the Sleep Disturbance Scale (SDS) (47.5% moderate to severe insomnia). Results of the two instruments correlated significantly with each other. These results are comparable to those on insomnia prevalence existing in the literature. Studies using the Pittsburg Sleep Quality Index (PSQI) found prevalences of 73% to 96% of “poor sleepers” using different cut-off scores [5, 11]. Other studies showed prevalences of 72% and 45% using a constructed questionnaire and one item of the major depressive episode module of the Structured Clinical Interview for DSM-IIIR (SCID), respectively [4, 25]. A prevalence of 70% was found in another study by asking the single question “Do you have trouble sleeping at night?” [12].

The single question “Have you been suffering from insomnia since your admission?” might be a sensitive method for detecting insomnia. However, the use of a single question has been considered as insufficient for the evaluation of insomnia in cancer patients, as it may have poor sensitivity and specificity for detecting insomnia [32]. In the present study, patients with higher HADS score, more anxiety, clear knowledge of the diagnosis, better performance status, nocturnal rumination, “dyspnoea” as a symptom interfering with sleep quality, and a feeling of non-restorative sleep answered more often “yes” to this single question in the univariate analysis.

The SDS assesses the presence of moderate to severe insomnia, is easy and quick to administrate, refers to a short recent period, and gives information about the three sleep items: falling asleep, waking up during the night and early waking. The use of a Likert scoring system allows participants to respond in a degree of agreement accommodating neutral or undecided feelings, and without requiring them to provide a simple and concrete yes or no answer [33]. Therefore, the assessment of insomnia of patients during admission in a PCU using SDS may have some advantages over the insomnia single question and other tools which take longer to respond to and refer to a longer preceding period of time. In the present study, patients with better performance status, nocturnal rumination, patients referring environmental disturbances and daytime sleepiness showed more moderate to severe insomnia in the univariate analysis.

The most affected sleep item was sleep maintenance, followed by difficulties in falling asleep and early waking, coinciding with data from other studies on patients with cancer [9, 14].

Clinically significant emotional distress was frequent in this study. A higher HADS score and the presence of anxiety were associated with answering “yes” to the insomnia single question in the univariate analysis, but this association was not found in the multivariate analysis, and no association was found with clinically significant emotional distress. Moderate to severe insomnia measured with the SDS was not associated with clinically significant emotional distress, higher HADS score, anxiety or depression. These results contrast with those in previous studies reporting on a correlation between insomnia and anxiety and/or depression measured with an analogue visual scale as part of the Edmonton Symptom Assessment System [34] in palliative patients [13, 16], or with a recent study reporting on a correlation between insomnia severity and depression measured using a non-validated tool [35]. Another recent study found a significant correlation between emotional distress measured with the HADS and insomnia measured with Athens Sleep Insomnia Scale [36] among patients with lung cancer [37].

Nevertheless, in the present study the psychological symptom nocturnal rumination did show a significant association with insomnia assessed with both instruments. In the unwanted mental activity at bedtime, rumination is a dysfunctional control strategy used by poor sleepers, which maintains cognitive and affective arousal instead of helping them to relax [3, 38]. The association between rumination and insomnia is consistent throughout the literature [3, 12, 14, 31, 38, 39]. Rumination has been associated with an increased and prolonged emotional distress (i.e. depression and anxiety) and with physical health problems in palliative care patients [40] and in the general population [38, 41, 42]. However, some recent studies have suggested that rumination may contribute to clinical insomnia, irrespective of worry and depressed mood states [39, 43, 44].

Different ways of assessing rumination have been proposed in the literature [31, 41, 43, 45], but only few have been used in palliative care patients [30, 40]. In the present study, rumination was evaluated by means of a simple question based on a definition obtained from the literature [30, 31]. Several methods for managing rumination have been also developed, and it has been proposed that its treatment may be imperative for a successful management of insomnia and depression [31, 45]. Strategies aimed at improving rumination could be included into the cognitive behavioral therapy of insomnia [31, 45].

Despite the association and possible role of rumination in connection with insomnia, patients in this study claimed physical factors to be the type of symptoms that interfered most with their sleep quality, coinciding with results reported in other studies [12]. This result suggests that patients may tend to underreport psychological symptoms such as rumination, but also highlights the importance of careful evaluation of physical symptoms. Among the physical factors, pain was the symptom that patients mentioned most often as interfering with their sleep quality, suggesting that pain should be especially taken into consideration when assessing and managing insomnia, as reported in other studies [11, 14]. Nevertheless, pain was not found associated with insomnia in the multivariate regression analysis in the present study.

Dyspnoea showed a significant association with insomnia measured with the single question in the multivariate regression analysis. Dyspnoea has been also reported in the literature as being the cause of sleep disturbance [12, 46]. However, this association has not been confirmed in other studies [4]. Also a higher performance status (higher Karnofsky index) has shown association with insomnia measured with the single question in the multivariate regression analysis of this study. Unlike this study, other studies found that a lower performance status might be associated with insomnia [4, 11]; however, these results were not confirmed in the multivariate analyses of these studies. Results regarding the Karnofsky index in the present study should be interpreted cautiously, as the range of performance status included was limited by the inclusion criteria.

Patients with daytime sleepiness in this study showed moderate to severe insomnia less often according to the SDS, in comparison to those patients without daytime sleepiness (37.8% vs 75.0%), suggesting that patients with moderate to severe insomnia might be more often awake in PCUs during the daytime. In a secondary analysis, daytime sleepiness was associated with a higher intake of psychotropic medication; it could be hypothesized that sleepiness due to psychotropic medication might also be happening at night. Literature on the relationship between daytime sleepiness and insomnia in palliative care is scarce. Although there is some controversy, our result is comparable with other studies [46, 47]. Nevertheless, excessive daytime sleepiness in the general adult population seems more likely to be attributed to short sleep duration [48], but the insufficient sleep can be due to insomnia or to other factors. Pharmacological treatment and biological aspects associated with advanced cancer, such as cytokines influencing the sleep-wake cycle [49], could influence the presence of daytime sleepiness. Further studies are needed in order to understand the relationship between daytime sleepiness and insomnia.

To the best of our knowledge, environmental factors have rarely been considered in the literature in relation to insomnia. Despite not being significantly associated with insomnia in the multivariate analysis, results of this study show that patients do care about environmental factors with regard to their sleep. Although PCUs are generally quieter than other conventional hospital wards, this result suggests that special care should be taken to ensure that these units are free of unnecessary noise, light or other interruptions during the night-time. As suggested by the hyperarousal theory, patients with insomnia might find themselves in a cycle of increased cognitive and physiological sustained sensory processing of environmental stimuli [50, 51].

In the present study no association between insomnia and other sociodemographic (e.g. gender, age) or clinical characteristics (e.g. diagnosis and stage of the disease, medication use) was found.

This study has limitations. The low sample size does not allow us to speak about prevalence of insomnia and limits the interpretation of the results of the multivariate analysis. This study should be seen as exploratory. Replication of the results in further studies with larger sample sizes is required. The SDS was translated into the Catalan language, but no backward translation was performed, as recommended by methodological literature [52]. However, the items were not complex in terms of language and no important cultural differences were expected after translation. Furthermore, no formal validation of the scale was carried out.

The day during admission on which interviews took place varied, with some taking place after one week of admission. Only 36% of the consecutively admitted patients were included in the study. About 10% of the admitted patients were not assessed for inclusion because they were discharged or died before, but their sociodemographic characteristics did not differ from the other patients admitted. More than half of the admitted patients were excluded because they were too confused or had a too bad performance status, similar to that experienced in other studies [4]. Results of the present study may not be representative of all patients admitted, but of those with better cognitive and performance status.

In conclusion, this study suggests a high frequency of insomnia among palliative care patients during their admission in a PCU, both when measured using a short single question and the Sleep Disturbance Scale. Furthermore, a high frequency of clinically significant emotional distress (including anxiety and depression) was shown. Pain, dyspnoea and environmental factors were often mentioned by patients as interfering with their sleep quality, and patients mentioned physical factors as the type of factor interfering most. While clinically significant emotional distress was not associated with insomnia in this study, nocturnal rumination was associated with insomnia when measured with both tools and this association remained in the multivariate analysis. The assessment of nocturnal rumination might be of particular interest in the management of insomnia. These results should be confirmed in further prospective studies with a larger sample size.