Introduction

Sleep, essential for healthy functioning, is characterized by sensory disconnection1. Buysse et al.2 defined sleep as “a multidimensional pattern of sleep-wakefulness, adapted to individual, social and environmental demands, that promotes physical and mental well-being”. Without question, sleep has a global restorative function for the organism, fundamental for maintaining neurobehavioral activity and brain health during wakefulness1,3.

Sleep quality and quantity have a critical role in promoting health4, and poor sleep quality has been associated with an increased risk of cardiovascular events5, cancer6, metabolic disorder7, and all-cause mortality8. Moreover, chronic poor sleep has been related to cognitive function impairment and neurodegenerative diseases9, a weakened immune system10,11 and a higher fall risk12. Additionally, it has been demonstrated that good sleep quality can be of therapeutic use for patients with chronic pain13, or mental disorders14.

There are numerous patient-related factors that have potential negative impact on sleep of patients such as pain, the underlying acute illness, unemployment or age, and psychological factors such as anxiety15,16,17,18. A review on sleep health concluded that sociodemographic variables are of enormous importance when evaluating sleep. Factors such as low physical activity or high fat diets can impair sleep, and an increased incidence of sleep disruptions can also be attributed to the female gender and low socioeconomic level19. Additionally, previous studies have suggested that patients’ sleep quality and quantity during hospitalisation is poor20,21,22,23, and associated sleep disturbances, such as a decrease in total sleep time, poor sleep efficiency, increased awakenings per night, or an increased sleep onset latency, in both domains of self-reported sleep quality and objective measured sleep, are shown to continue up to 12 months after hospitalisation24. During hospitalisation, symptoms such as abnormal melatonin secretion, showed by both a delay in the endogenous rhythm of plasma melatonin and 6-SMT excretion in urine25,26, or sleep beathing disorders or obstructive sleep apnea27,28 have been observed. The clinical environment may be a cause of this sleep impairment20. Factors such as light and sound disturbances, night-time care interventions, environmental temperature, and bed comfort are possible factors that can lead to poor sleep, causing sleep disruptions during the night and inducing misalignment in the sleep–wake cycles23,29. However, as hospitals are a place of healing, it could be expected that they should provide the best possible conditions to achieve this healing process. Also, hospitals could have an educational function by offering exemplary behaviour and support to patients in order to achieve sleep health in the long term, even after hospitalisation30.

To date, in Spain, no previous multicentre studies, including patients from public hospitals from several regions, have been carried out in medical and surgical units to describe sleep quality and quantity.

Therefore, we aimed:

  • To assess self-reported sleep quality and quantity of medical and surgical unit inpatients before and during their hospitalisation.

  • To describe the association of inpatients’ sociodemographic and poor health habits and sedentary behavior to their sleep quality

Methods

Study design and participants

A multicentre, observational descriptive study was carried out, recruiting patients from medical and surgical units in 12 hospitals of the Spanish National Health System located in nine Spanish regions (Fig. 1). All hospitals were participants of the SueñOn ® health awareness initiative31. Inpatients, aged 18 years or older with at least 4 consecutive nights of hospitalisation who consented to participate voluntarily in the project were included in this study. Patients with visual or auditory impairment, intellectual disability or moderate to severe cognitive impairment as reflected in the clinical patients’ record were excluded from this study. We did not include patients with a COVID-19 infection.

Figure 1
figure 1

Geographical distribution of included hospitals.

Full size image

Data were collected between January 2020 to January 2022 and patients were included on pre-established days every month. The Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) guideline was followed for reporting this study.

Data collection

Sociodemographic and clinical characteristics

Data were collected on age, gender, education level, current work situation, mean income per household (euros per month), Body Mass Index (BMI), and unhealthy habits such as smoking or any other recreational drug intake.

Patients’ clinical characteristics were collected by the nursing staff of the unit. The motive of admission was classified and data on diagnosed and self-reported comorbidities of the patient were collected. Patients were classified as having multimorbidity if multiple diseases or conditions were reported using a definition of two or more conditions32. Also, covariates such as sedative hypnotic medication use, and staying in a single or shared room were recorded.

To assess physical activity level over the last seven days before hospitalisation, the International Physical Activity Questionnaire (IPAQ)33 was used.

Sleep quality and quantity

The Spanish version of the Pittsburgh Sleep Quality Index34 (PSQI) was used to assess patients’ perception of sleep quality referring to the past 1 month, and at discharge referring to the patients’ hospital stay. The PSQI is a validated and highly reliable self-report questionnaire consisting of 19 questions divided into seven major components. These seven components are: subjective quality of sleep, sleep latency, length of sleep, sleep efficiency, sleep disturbances, use of sleep medication and daytime dysfunction. Scores range from 0 to 3 for each component and the global score ranging from 0 to 21 is obtained by the sum of all components scores. A higher global PSQI score indicates worse sleep quality2. Using a cut-of score of 5 to differentiate poor sleep with good sleep the PSQI reported a sensitivity of 89.6% and a specificity of 86.5%. The validated Spanish version of the PSQI has shown a correlation coefficient of 0.773 for the global score34. The question used to evaluate sleep quantity was: “How many hours do you estimate you actually slept each night?”.

Data analysis

Data were collected on paper and entered into an online REDCap electronic database35. Analysis of data was performed using SPSS 25.0. Descriptive data were categorized and compared with study variables. Categorical variables were summarised by percentages and continuous variables were reflected by means and SD. Kolmogorov–Smirnov test was used for checking whether sleep quality and quantity could be assumed to be normally distributed. For skewed or non-normality data non-parametric analysis were performed. Sleep quality and quantity differences at admission and during hospitalization were compared with student’s paired t test (Wilcoxon test) and differences between groups were analysed with the student’s t test (Mann–Whitney test) or ANOVA (Kruskal–Wallis test). Correlation of physical activity level measured with the IPAQ questionnaire was calculated with the Spearman Rho test. All tests were two-tailed at 0.05 significance level.

Ethical approval

Patients participated voluntarily and were included in this study after giving their written consent in the informed consent document in regulation with the law 14/2007 on biomedical research and ethical principles after the declaration of Helsinki. Patients were previously informed of the entire process, and their possibility to revoke their participation at any time. Approval of all ethical committees of all participating hospitals were obtained as well as the approval of the ethical committee of the Carlos III health institute (HIP CI CEI PI 18_2019-V3). Additionally, this study was registered with Clinicaltrials.gov (#NCT04113876).

Results

General characteristics of the participants

A total of 343 patients from 12 different Spanish public hospitals with a minimum hospital stay of 4 days were included in our study. Our study included hospitals of both rural and urban regions and covered hospitals from geographical areas all over Spain. The mean age of the participants was 63.47 ± 14.96 years. Our sample was comprised of 143 (41.69%) female, and 199 (58.02%) male participants (one patient didn’t identified their sex) and 20.41% of the participants reported being habitual smokers. Included patients were hospitalised in surgical n = 162 (47.23%) and medical n = 181 (52.77%) units. The majority of the patients were retired n = 202 (58.89%), 97 (28.28%) were employed, 37(10.79%) were unemployed, and 2 (0.58%) were students. Over half of our sample (54.52%) had been educated only to the primary level, and only 9.62% were university graduates. Household incomes per capita were mostly (38.38%) between 0 and 1000 euros, 35.86% had an average income between 1001 and 2000 euros monthly, and 9.33% of the participants disposed of > 2000 euros a month (9.33% of the participants did not want to provide data about their income). The mean BMI of the participants was 27.70 ± 5.98. Most patients were diagnosed with multimorbidity (62.09%). Main hospital admission motives were digestive diseases (27.41%), neoplasms (25.07%) and respiratory diseases (20.41%). Most of the patients stayed in double rooms n = 289 (84.26%) and the median hospital stay was 8 days (IQR = 7.00–11.00). More details about sociodemographic variables, hospital admission motive, comorbidities, and the distribution of the 343 participants and missing data are shown in Table 1.

Table 1 Sociodemographic characteristics and admission motive of participants.
Full size table

Sleep quantity and quality before and during hospitalisation

Looking at total sleep quantity, participants’ report of how many hours they actually slept before and during hospitalisation resulted in a mean total sleep time of 56 min less during hospitalisation (6.28 ± 1.64 vs 5.34 ± 1.58, p < 0.001).

Global PSQI scores measuring overall sleep quality could be calculated in only 298 patients (86.88%) because of missing data. However, because missing data was sporadic across the 19 questions, we also calculated the scores of the seven PSQI components relating to sleep quality.

Poor sleep quality (PSQI > 5) before hospitalisation was present in 77.00% and was 95.10% during hospitalisation. Mean global PSQI score before hospitalisation was 8.62 ± 4.49 and rose to 11.31 ± 4.04 during hospitalisation showing a decrease in overall sleep quality during hospitalisation compared to sleep quality before admission in the hospital (Z = −8.539, p < 0.001). Daytime dysfunction was the lowest scoring component of the PSQI (0.67 ± 0.88) and sleep latency was the highest (1.52 ± 1.09) before hospitalisation. When participants answered the PSQI at discharge, daytime dysfunction remained the lowest component (1.12 ± 1.01), and sleep latency was scored the highest (1.98 ± 1.03), along with sleep duration (1.98 ± 0.94), and sleep efficiency (1.98 ± 1.18). The greatest increases could be seen also in these components; sleep efficiency (+ 0.63, p < 0.001), sleep duration (+ 0.54, p < 0.001), and sleep latency (+ 0,46, p < 0.001) comparing PSQI scores before and during hospitalisation suggesting that these components had the greatest influence on the decline of patients’ sleep quality in the hospital. Student’s t-test and Mann–Whitney test confirmed increases in all seven components, all statistically significant except for component 1: subjective sleep quality (p = 0.056). Changes in PSQI scores and reported sleep quantity are summarized in Table 2.

Table 2 Sleep quality and quantity at admission and during hospitalisation.
Full size table

Association of patients’ sociodemographic and clinical characteristics with sleep quality and quantity

A decrease in sleep quality could be observed in all sociodemographic categories during hospitalisation, all statistically significant except for age 31–50 (p = 0.071), unemployed participants (p = 0.071), students (p = 0.655) and participants with underweight (p = 0.157). Self-reported total sleep hours also decreased significantly in all almost all groups except for students (p = 0.180) and participants that were underweight (p = 0.714), however, this could be explained by small sample size in these categories. No significant differences in sleep quantity or quality could be observed between groups for motive of admission before or during hospitalisation (all > 0.05).

It was shown that age, income, room type, length of hospital stay, smoking or recreational drug intake, or BMI had no significant influence on the patients’ sleep quality at admission or during the hospital stay (all p > 0.05). Patients’ sleep quality scores at admission were higher in patients admitted to medical units (Z = −4.144, p < 0.001) and a higher PSQI score could be demonstrated in female patients (Z = −3.284, p = 0.001). Also, a significant difference between groups at admission could be observed for patients’ working situation, with a better sleep quality for employed patients (H(3) = 9.736, p = 0.021). However, these differences between groups were not maintained during hospitalisation. Patients’ sleep quality scores during hospitalisation were higher in those with only primary education than in those with education at the secondary or university level (H(2) = 6.246, p = 0.044), whereas there was no difference found between these groups at admission. This difference could also be observed in reported sleep hours during hospitalisation (H(2) = 6.833, p = 0.033). Having 2 or more comorbidities was associated with a significant difference in sleep quality both at admission (Z = −3.441, p = 0.001) and during hospitalisation (Z = −2.731, p = 0.006). Patients taking sedative-hypnotic medication had poorer sleep quality both before (Z = −2.285, p = 0.022) and during hospitalisation (Z = −6.092, p < 0.001) compared to those who didn’t. These differences were not shown for sleep duration.

Finally, correlation with the Spearman’s Rho test showed evidence that higher physical activity measured with the IPAQ questionnaire led to better sleep quality during hospitalisation (R = −0.148, p = 0.012). However, differences were not significant at admission, nor were they significant for sleep quantity. No other differences between groups for both sleep quality (Table 3) or quantity (Table 4) were statistically significant.

Table 3 Effects PSQI global scores and sleep quantity.
Full size table
Table 4 Effects sleep quantity (in hours).
Full size table

Discussion

To our knowledge, this nationwide, multicentre study is the first to examine subjective sleep quality and quantity in patients of surgical and medical units in 12 public hospitals in Spain. Sleep is necessary for the human being to lead normal activities and maintain good health36. Hospitalised patients who are recovering from surgery or illness are supposed to have an even greater need of restorative sleep to achieve full recovery37. However, our results demonstrated that sleep quality in patients admitted to hospitals of the public national health system experienced poorer sleep quality during their hospitalisation compared to at home, and slept almost one hour less than before hospitalisation. This suggests that the experience of hospitalisation has a negative effect on both self-reported sleep quality and sleep duration. These findings are in line with numerous other studies reporting poor sleep quality and quantity in hospitalised patients in countries such as China21 , the Netherlands22, Turkey38,39,40 Canada41 and Germany42. Although full comparison with other studies could not be done because of reporting on patients from specific units, single hospital samples, patients with specific conditions, or the use of other sleep quality measurement tools. However, this may be an indication that, despite previous research showing poor sleep quality of hospitalised patients, in general there is still a problem in getting good sleep quality in a hospital. Even if hospital technology and hospital buildings have become more sophisticated43 there is still room for improvement in the way they address one of the basic needs of patients during their recovery, such as their sleep. More research is needed to identify factors that help promoting good night-time sleep in hospitals. Therefore, we believe that this study adds information to the existing literature.

Highest PSQI scores during hospitalisation could be observed in the components of sleep latency, sleep duration and sleep efficiency which may suggest a lack of the use of behavioural intervention to improve the sleep experience in hospital44. Using the cut-of score of 5 proposed by Buysse et al.2, more than three quarters of our included sample had poor sleep quality at admission to the hospital with a mean PSQI score of 8.62. This mean score is higher than in recent studies on sleep quality in the general population in Spain who reported PSQI scores of 5.1445 and 5.4546 respectively. This higher score may be due to the effects of acute illness before hospitalisation. Also, our data were collected during the COVID-19 pandemic. Although our study did not include COVID-19 patients, there is evidence that COVID-19 outbreak-associated factors correlate with a decrease in sleep quality46. These findings were confirmed by a recent study who concluded that both patients with and without an acute COVID-19 infection experienced poor sleep quality and quantity47. This would also explain why the results of a study conducted during the COVID-19 period on general Spanish population observed a similar global PSQI score5,14 with our study population48. During the COVID-19 outbreak an increase of stress, anxiety and depression affected the general population49,50. Anxiety and emotional distress have been shown to be negatively correlated to sleep quality in hospitalised41,51,52. Stress-induced activation of the hypothalamic–pituitary–adrenal (HPA) axis disrupts sleep, and variations in sleep enhance this activation, causing a vicious cycle of stress and sleeplessness53. Also, quarantine and social isolation may have caused changes in sleep–wake rhythms and a reduction of sleep quality54. These factors could have had an even greater impact on hospitalised patients during peak COVID-19 periods due to restrictions on visitors or night-companions, or a higher fear of getting infected with COVID-19 in the hospital55. Treatment delay because of the pandemic could also have caused patients to be in a more acute phase of their illness before getting admitted to the hospital.

Our sample was comprised of 83% of patients older than 50 years and 37% older than 70 years. Changes in the normal sleep-cycle and a variety of sleep problems are reported in elderly people and sleep quality is expected to deteriorate with age56. Our study found higher PSQI scores as age groups increased, but differences were not found to be significant at admission or during hospitalisation. Having higher educational level was observed to be a protective factor for poor sleep quality and quantity during hospitalisation, however, these results could be related to the age of the participants as secondary education became mandatory in 1964 in Spain57. At admission women had poorer sleep quality than men. Taking into account the mean age of our sample, it has been demonstrated that several hormonal factors, such as menopause, and physical changes in women’s life can have impact on their sleep health58.

Patients with 2 or more comorbidities were found to have poorer sleep quality which could also be related to higher age, but also because of more physical symptoms. Higher physical activity has also been shown to lead to better sleep quality during hospitalisation and no association was found for BMI which confirms the findings of prior researches59,60.

Although a decrease in the consumption of benzodiazepines is reported in Spain61, over 40% of our included population regularly took sedative medication at home or in hospital. A negative association at admission and during hospitalisation was found for sedative medication intake and sleep quality. This confirms findings of other studies suggesting that benzodiazepines do not improve sleep quality and cause a significant risk of falls, especially in elderly patients what makes that it should be regularly reviewed whether intake is necessary62,63,64 and the implementation of other non-pharmacological therapies with a long-term effect and fewer side effects should be considered65. The length of hospital stay was not associated with differences in sleep quality or quantity in our sample, although due to our inclusion criteria of a minimum length of stay of 4 nights in hospital, short-term effects of hospitalisation on the sleep wake cycle could not be fully explored.

Management of sleep in hospital is challenging as intrinsic sleep disturbances are clinically heterogeneous and complex38. Also, although it is observed that sleep is insufficient among patients, several studies reported a lack of the use of standardized sleep assessment tools and sleep promoting interventions in hospital66.

Limitations

This study has some limitations. The dependence on patient recall for their regular sleep time prior to and during hospitalisation is one of the study’s limitations. Reported sleep quality before admission may have been deviated from habitual sleep quality because of the situation of illness of the participant. We used the PSQI to measure sleep quantity and quality, we did not report on sleep dimensions with objective measurements, future research could focus on verifying and comparing our findings with non-subjective sleep evaluation methods such as actigraphy or polysomnography. Also, The PSQI was developed evaluating “usual” sleep of the past month2. PSQI scores during hospitalisation could be biased because of a short hospital stay and patients with a hospital stay shorter than 7 days could not have had the opportunity to answer correctly on the 4 answers provided in some components of the PSQI. This could have led to a sub estimation of sleep quality during hospitalisation. However, median length of stay was 8 days and before passing the test to patients, we made a clear statement that this their answers should only reflect on sleep experience during their stay in hospital. Finally, the study counted 13% missing data in the PSQI test. This can be explained by the high workload of the nurses who could not go through every questionnaire or by the acute illness which caused patients to leave some questions unanswered. Although there is no established cut-off from the literature regarding an acceptable percentage of missing data in a data set for valid statistical inference, Enders67 stated that a missing rate of 15–20% was common.

Conclusion

This study demonstrated poor sleep quality in patients admitted to hospitals of the public health system in Spain. Significant differences between sleep quality at admission and during hospitalisation in most of the components and global score of the PSQI were found, and inpatients slept a mean difference of 56 min less. A lower educational level, two or more morbidities and the intake of sedative hypnotic medication were shown to be associated with poorer sleep in hospital. However, a higher physical activity level before hospitalisation may function as a protective factor for poor sleep quality in hospital. These results may contribute to drawing patients’, and professionals’ and policymakers’ attention on the importance of sleep both before and during hospitalisation taking into account sociodemographic and clinical variables of the patient.

Future research should focus on disturbing factors causing this sleep quality and quantity decrease in order to draw attention of healthcare workers and design evidence-based interventions that improve the sleep experience in the hospital. Additionally, the positive impact of physical activity on hospitalised patients has to be studies further. Also, however the use of benzodiazepines is decreasing, a high prevalence of dispense is maintained, especially in population over 65 years, to lower this rate integral actions are needed.