Sleep Quality of Patients on a General Department During the First Days of Hospitalization

Introduction

Adequate sleep duration and structure are essential to a person’s health and well-being. Sleep problems are common and include deficits in both quantity and quality. Sleep and sleep deprivation affect virtually every system level of an individual.¹

Unfortunately, hospital rooms are not optimal environments for sleep. Hospitalized patients suffer from sleep disturbances caused by symptoms of the underlying disease (eg, pain, discomfort, comorbidities, and medications), the environment (eg, nursing and medical interventions, noise, and light), psychological factors (eg, anxiety), and social influences (eg, loss of autonomy or bedtime routines).²

Growing evidence of the importance of sleep for physical and mental health and recovery, including those of inpatients, has prompted research on this topic, including the search for appropriate techniques to assess sleep quality in a hospital setting. Sleep is commonly assessed using subjective tools such as questionnaires and scales, and objective measures such as polysomnography, actigraphy, the electroencephalogram-derived bispectral index, or endocrinological and biochemical biomarkers (eg 6-sulfatoxymelatonin, cortisol). Each method has advantages and disadvantages or shortcomings, but they are often complementary.^3,4 So far, there has been no perfect tool for nurses to use in the assessment of sleep in hospitalized patients, even though sleep is an independent factor in successful treatment.⁵

For most people, hospitalization is associated with stress, one of the factors affecting sleep. The most important general determinants of sleep disturbance due to stress are family history of insomnia, female sex, and adverse environmental influences (eg unfamiliar sleep environment).⁶ Hospitalization can also be regarded as an adverse environmental factor. Drake et al^7,8 have elaborated on the concept of sleep reactivity, which describes how sleep reacts in response to various factors. Understanding individual sleep reactivity is important for individual optimization of sleep conditions. The most widely used tool for assessing sleep reactivity is the Ford Insomnia Response to Stress Test (FIRST).⁹ The Ford Insomnia Response to Stress Test (FIRST) was selected as a frequently used screening tool due to evaluate the stress reactivity and its effect on sleep.^7–9 We proposed that patients with higher FIRST scores would be more susceptible to declines in sleep quality during hospitalization and more sensitive to stressors related to the hospital environment.

The primary hypothesis of this study was to assess subjective sleep quality in patients during the first 72 hours of hospitalization and to examine the relationship between perceived sleep quality between men and women and other commonly available hospitalization data, including age, type of department, surgery, previous hospitalization, reason for admission, pain level, and use of medications to induce or maintain sleep.

The secondary hypothesis aimed to determine whether the FIRST questionnaire could identify individuals prone to sleep disturbances during the initial days of hospitalization.

Methods

Design

A multicenter cross-sectional study

Sample

The study was conducted between September 2022 and January 2023. The sample consisted of 340 patients on general medical and surgical departments in seven hospitals in the Czech Republic. The number of beds per room mainly were two – to three. The inclusion criteria were age over 18 years, hospitalization duration of longer than 72 hours, ability to answer questions and complete the questionnaire, and informed consent to participate in the study. The exclusion criteria were a history of significant sleep disorder, chronic hypnotic use, dementia, delirium, and terminal stage of disease. For FIRST testing, patients who had received sleep-inducing medication were excluded, resulting in a final study population of 285 patients.

Sample Size Estimation

To determine the required sample size, the following formula, applicable for cross-sectional studies, was applied: (Zα/2)2 × p(1-p)/d², assuming a 95% confidence interval, prevalence rate (p) of 50%, and error rate (d) of 5%. This calculation indicated that at least 340 participants were needed.

Measurements and Data Collection

Sociodemographic information (age, sex, and previous hospitalization) was recorded together with the following basic hospitalization data from the patients’ medical records: type of department (medical, surgical), type of admission (elective, acute), surgery during hospitalization (yes, no), pain (visual analog scale 0–10), and sleep-inducing medication - Timing medication was between 8:00–10:00 p.m. (yes – administered any time throughout the study period, type and dosage; no – medication not administered on any night). Patient recruitment and administration of the questionnaires were carried out by nurses trained in the research project protocol.

Questionnaires Used

The Richards-Campbell Sleep Questionnaire (RCSQ) contains five items (sleep depth, falling asleep, awakenings, return to sleep, and sleep quality) and optional sixth item about noise during the previous night (were not included in the analysis in this study). Each item is scored on a visual analog scale ranging from 0 to 100. The RCSQ total is the average of the individual items. A total score of 0 represents the worst sleep and a score of 100 represents the best sleep. The RCSQ has been validated in the Czech Republic.¹⁰ Patients completed the RCSQ from day two to day four of hospitalization, always between 7 am and 9 am, as an assessment of the previous night’s sleep.

The Ford Insomnia Response to Stress Test (FIRST) assesses sleep disturbance caused by stressful events. It contains a total of nine items (model situations) with Likert scales (1–4), meaning that the total score ranges from nine to 36. The higher the score, the greater the likelihood of sleep disturbance.⁷ On the first day of hospitalization, patients completed the questionnaire (some with the help of nurses). The original FIRST questionnaire was requested from the author and translated. The translation of the questionnaire was carried out by the following stages: (1) Forward translation: translation from English into Czech by two independent professional translators, including the creation of a unified Czech version; (2); Backward translation: translation back into English with subsequent identification and correction of differences between the original and the backward translation; (3) Expert evaluation: consensus of an expert group on the final form of the Czech version; and (4) Pilot testing: testing the questionnaire on ten patients in a selected hospital (these patients were not included in the study population).¹¹

Ethical Aspects

The study, conducted in accordance with the Declaration of Helsinki, was approved by the Ethics Committee of the Faculty of Medicine of the University of Ostrava (No. R2/2021), the Ethics Committee of the University Hospital Ostrava (No. 524/21), and the Ethics Committee of the General University Hospital in Prague (No. 54/21; grant project AVZ VES 2022 VFN). The study is also registered in the ClinialTrials.gov database (ID: NCT05402280). All participants gave their written paper informed consent. The Participants could withdraw from the study at any time without facing any penalties. Every participant in the study was an adult, aged at least eighteen years. Participants did not receive any financial or non-financial incentives.

Data Analysis

Descriptive statistics included absolute frequency (n), relative frequency (%), arithmetic mean (mean), and general deviation (SD). Relationships between pairs of metric or ordinal variables were tested with the Mann–Whitney U-test. Relationships between dichotomous variables were assessed using Fisher’s exact test. Statistical analysis of the relationships between RCSQ items, sex, and hospital duration was performed using a three-factor analysis of variance model at a significance level of p < 0.05. Effect sizes were calculated using Cohen’s d to quantify the magnitude of differences between groups. The impact of various variables (sex, age, type of department, surgery, type of admission, previous hospitalization, sleep-inducing medication, and pain) on individual RCSQ items and the total RCSQ score (averaged over three days) was analysed using multivariate regression with dimension reduction via the orthogonal projections to latent structure (OPLS) method. OPLS effectively addresses multicollinearity by isolating predictive components from orthogonal variation, enabling the identification of key predictors and estimating their combined influence on RCSQ scores. Regarding post hoc comparisons, Bonferroni test was employed, and p <0.05 served as the cutoff level for statistical significance. The internal consistency of the FIRST was estimated using Cronbach’s alpha (NCSS 2012 software; Kaysville, UT, USA). Statgraphics Centurion XV version 12.2.06 statistical software from Statpoint, Inc. (Warrenton, VA, USA) was used to calculate analysis of variance and statistical tests. SIMCA-P v.12.0 from Umetrics AB (Umeå, Sweden) was used for statistical analysis of OPLS.

Results

The analysis included 340 patients with complete data on sleep and sleep reactivity (Table 1). Participant recruitment, exclusions and reasons for exclusion are shown in Supplementary Figure 1.

Table 1 Characteristics of Patients

The sample of 340 participants consisted of 172 females and 168 males. The mean age was 58.9 (± 14.9) for females and 57.3 (± 15.0) for males (p = 0.447). There were no statistically significant differences between males and females with respect to all items in this study except intake of alprazolam which was higher in men (Supplementary Tables 1 and 2).

The average total RCSQ score and individual item scores for the entire sample over three days were higher in males than in females, although not significantly so. This is consistent with the analysis of individual RCSQ items over the three nights studied, for which subjective perception of sleep was slightly better in males on each RCSQ item. For females, RCSQ total was found to increase over the course of their hospitalization, whereas for males, deterioration was noted on the second rated night for the following items: falling asleep, return to sleep, sleep quality, and average RCSQ total (Table 2). In order to assess the internal consistency of the (RCSQ), we calculated Cronbach’s alpha for all five items (Q1–Q5) to evaluate how consistently they measure the overall construct of sleep quality. The internal consistency of the RCSQ test (Cronbach’s alpha) was 0.93 (excellent) (Table 3). Patients who had undergone surgery scored statistically significantly better on sleep depth, awakenings, subjective sleep quality, and RCSQ total compared to patients who had not undergone surgery (Table 4).

Table 2 Changes in individual RCSQ items during the first three days of hospitalization

Table 3 Reliability of the Richards-Campbell Sleep Questionnaire (RCSQ)

Table 4 Mean RCSQ Scores for All Three Nights Combined, Surgery Vs Non-Surgery Patients

The multivariate regression method was used to analyze the relationships between the selected basic hospitalization data and RCSQ scores. RCSQ scores significantly correlated (p < 0.05) with the assessed clinical and demographic parameters (age, sex, department type, surgery performed, previous hospitalization, urgency of hospitalization, pain severity, and sleep-inducing medication administration). When assessing the level of statistical certainty (t-statistic), the following parameters had the strongest positive effect on RCSQ total: male sex (t = 7.89) and alprazolam use (t = 6.51). The strongest positive effect on RCSQ total was found for alprazolam administration (R^a = 0.604). Other clinical and demographic parameters positively affecting RCSQ scores were (in descending order of correlation coefficient R^a): hospitalization in surgical department, sleep-inducing medication (other than alprazolam) administration, surgery during hospitalization, male sex, and age. Pain had the strongest negative effect on RCSQ total (R^a = −0.378); other parameters decreasing RCSQ total were (in descending order): elective admission, hospitalization in medical department, and history of previous hospitalization. The variability associated with the factors studied accounted for 18% of the RCSQ variability (Table 5).

Table 5 Relationships Between Subjective Sleep Quality (Expressed as Mean RCSQ Scores) and Clinical and Demographic Parameters Assessed by Multivariate Regression

The mean FIRST score was 20.2 (± 6.3). When evaluating individual FIRST items, the least stressful situation likely to disturb sleep was watching a frightening movie, whereas the most stressful was receiving bad news during the day (Table 6). A sub-analysis comparing males and females confirmed significantly lower scores for males: 17.8 (± 5.5) vs 22.5 (± 6.1), respectively (p < 0.001). The FIRST achieved a Cronbach’s alpha of 0.85 (Supplementary Table 3). The dependence of sleep reactivity (FIRST score) on subjective RCSQ total was assessed using two correlation tests: Spearman’s coefficient (−0.1734; CI −0.2846, −0.0575) and Kendall’s tau (−0.1234; CI −0.1987, −0.0467). Although both tests confirmed a statistically significant relationship (p < 0.001), these negative correlations were deemed very weak (less than 0.3).

Table 6 Ford Insomnia Response to Stress Test FIRST

Discussion

The present study showed that, according to the RCSQ, the perception of good sleep quality during the first three days of a stay in a room on a hospital department was associated with several factors. These included the intake of medication inducing and maintaining sleep, hospitalization in a surgical department, undergoing surgery during hospitalization, and older age of the patient. Conversely, pain level, hospitalization in a medical department, elective hospitalization, and previous hospitalization were associated with a perception of poor sleep. The factors assessed were chosen because they are readily available during stays in all types of departments and hospitals, but their variability appeared to account for only 18% of the variability in subjective sleep ratings. Another important finding of the study was that there were no significant changes in subjective sleep quality scores between gender over the first three days when the entire sample was assessed together, although there was considerable inter-individual variability in the results for individual nights and the overall trend. Finally, the study found that the FIRST instrument was not suitable for screening inpatients at risk of poor sleep at the beginning of their hospital stay.

The findings suggested that sleep quality remained poor and consistent, with no significant improvement over the first three days. The third day subjective assessment showed only a non-significant improvement in sleep quality, although patients might be expected to be familiar with and partially adapted to the hospital environment so that their sleep would improve. In practice, therefore, it is reasonable to assume that patients’ perceptions of sleep remain the same during the first few days of their hospital stay and that a poor rating of sleep quality on the first night indicates that sleep will be similar on subsequent nights. However, our results are based on average ratings of a heterogeneous group of consecutive patients with different reasons for hospitalization. Therefore, trends in sleep perception might have been observed if more precisely defined groups of inpatients were followed.

Patients who had undergone surgery gave better ratings to the questionnaire items over three whole days (sleep depth, awakenings, and subjective and overall sleep quality – RCSQ total) than patients without surgery, but at the same time, hospitalization for elective surgery seemed to be factor that worsened sleep. Thus, we can speculate that patients with acute surgery found relief, including relief from the anxiety that is very likely to accompany a condition requiring acute surgery, whereas elective surgery led to prolonged and persistent anxiety, which did not immediately improve. Sleep and overall condition depend on the type and extent of surgery, which could not be considered in this extensive study. The sleep improvement in surgery patients may be explained by the usual administration of an anxiolytic or hypnotic the night before surgery and the postoperative use of analgesics, some of which have sedative effects. It is also reasonable to assume that, in some cases, surgery removed the problem causing pain or discomfort, resulting in improved sleep.

In our study, males gave non-significantly higher sleep ratings than females. In epidemiological studies, females generally rate their sleep quality as worse than males, independent of differences in sociodemographic and/or lifestyle factors.^12,13 There are probably several reasons for this, one of which is that females are approximately twice as likely to develop anxiety disorders, which are associated with poorer sleep quality.¹⁴ In addition, females are more likely to report more problems, including depressive experiences. At the same time, they tend to report more or more intense physical symptoms, even though these may not be objectively present.¹⁵ This is consistent with a study that found no significant difference in sleep quality between males and females when analyzing subjective data.¹⁶

The RCSQ was developed primarily for, and is used in, intensive care units (ICUs),^17,18 but has also been employed to assess sleep in non-ICU hospital environments.¹⁹ Despite the lack of normative values for general hospital departments, we followed these examples due to several advantages. In the present study, the internal consistency of the RCSQ test achieved very good reliability for the studied population of patients on general departments. Therefore, we consider the results to be reliable and the use of the RCSQ on general departments to be feasible. Not surprisingly, the RCSQ scores in our study were higher than those in ICU patients, including slightly lower scores for females. The present study can only be compared with a small study of 25 patients on regular departments that assessed sleep quality during the first three days, using the RCSQ and actigraphy, which also found no significant differences between day scores for any of the questionnaire items. The RCSQ total in this study was very similar to our results (63.0 ± 5.4).¹⁹ Higher RCSQ scores and generally better sleep of patients on regular departments compared to ICU patients are anticipated and may be explained to the more stable and less disruptive environment of general departments compared to the ICU, where noise, bright lighting, frequent interventions, and sedatives disrupt sleep. In general departments, patients experience fewer disturbances, better circadian rhythm regulation, less severe condition and lower stress, leading to improved sleep quality. Additionally, more realistic rest expectations in general departments may contribute to better perceived sleep quality. According to a review of 36 studies (with a total of 2522 patients), the mean RCSQ score during an ICU stay was 49.1 (95% CI 45.6, 52.5). Cut-off values of RCSQ total for adequate sleep in ICUs were reported to range from 50.0 to 63.4.¹⁷

In the present study, the use of sleep-inducing medication was most significantly positively correlated with subjective positive assessment of nighttime sleep from the hospitalization data analyzed. The most frequently used drugs were benzodiazepines, with alprazolam being the single most frequently used drug. A retrospective study of sleep-inducing medications administered to 642 patients in four regular departments over a two-month period showed that 168 patients (26.2%) received sleep drugs during their hospital stay. Patients were most commonly treated with trazodone (30.4%), lorazepam (24.4%), and zolpidem (17.9%).²⁰ Benzodiazepines, which were used significantly more frequently than other medications in our study, play a role in comprehensive hospital care for both their hypnotic and anxiolytic effects, alleviating the usual problems with new environments, including hospital, as well as anxiety regarding illness or elective surgery.^21,22 In the hospital setting, benzodiazepines reduce sleep latency and generally have a beneficial effect on sleep architecture²³ as has been documented with other conditions and environments in subjects with insomnia. Sleep-inducing medication is necessary and beneficial when indicated, but requires ongoing assessment of reactions and adverse effects, followed by monitoring to ensure that it is reduced and discontinued at the appropriate time. It is important to note that in some patients, newly prescribed benzodiazepines during hospitalization have resulted in ongoing use after discharge, regardless of the severity of their insomnia.²⁰

Pain intensity was found to be the most significant predictor of poor sleep in our study. This aligns with findings from a similar study by Ritmala-Castren et al¹⁶ which also identified a correlation between pain levels and sleep ratings in ICU patients. Sleep and pain interact, with pain disrupting sleep and sleep deprivation amplifying perception of pain.^24,25

Unexpectedly, the present study found a positive association between sleep quality and age. Aging is typically associated with a reduced ability to maintain uninterrupted sleep, shorter total sleep duration at night, and a decrease in the proportion of deep sleep.²⁶ However, some studies based on subjective ratings reveal surprisingly different trends in sleep quality across age groups.^13,27 This suggests that while aging brings notable changes in sleep patterns, it does not necessarily lead to complaints about sleep quality. Many healthy older individuals may adjust their perception of what constitutes “acceptable” sleep as they age. It is also likely that younger individuals are more aware of their environment and the experiences associated with hospitalization and have greater demands for comfort and privacy.

The 18% explained variability in sleep quality, as indicated by the multivariate regression model, suggests that the clinical and demographic parameters included in the analysis account for a modest but significant portion of the variance in subjective sleep quality scores (RCSQ) However, a substantial 82% of the variability remains unexplained. Despite their significance, the predictors in the model only capture a fraction of the complexity of sleep. Data used in our study suggests that sleep is influenced by factors beyond those that can be included, such as psychological factors (eg, anxiety, depression), environmental influences (eg, noise, lighting, staff activity), and intrinsic patient characteristics (eg, circadian preferences).

Hospitalization is a stressful situation that can affect sleep quality. To improve the quality of care, it is beneficial to identify patients at risk of sleep problems as early as admission to hospital. We hypothesized that patients with higher FIRST scores would be more susceptible to declines in sleep quality during hospitalization and more sensitive to stressors related to the hospital environment. However, the FIRST questionnaire did not prove useful for this purpose since it contains items related to everyday life that cannot be matched to the stressful circumstances of hospitalization and illness. Failure to include patients indicated for sleep-inducing medication at the very beginning of their hospital stay may have weakened the correlation between subjective sleep quality and FIRST scores by excluding the potentially most vulnerable individuals.

Limitations of the Study

One key limitation of our study was the low homogeneity of the sample, which was necessary to ensure that the study objectives were met, but may have impacted the generalizability of the findings. While we analyzed a large number of patients across various departments in one country, the high exclusion rate due to short hospital stays, acute conditions, or other factors may have introduced biases, as excluded patients may differ systematically from those included. Additionally, the study did not account for several potential confounding variables, such as psychological factors (eg, anxiety, depression), environmental influences (eg, noise, lighting, temperature), and intrinsic patient characteristics (eg, circadian rhythm preferences, baseline sleep habits), which could have further influenced sleep quality. The inclusion of medications not officially classified as hypnotics, though reflective of actual patient care practices, also introduced a potential bias.

Conclusion

Drugs inducing and maintaining sleep, hospitalization in a surgical department, surgery during hospitalization, male sex, and older age are factors associated with a better assessment of the subjective perception of sleep quality at the beginning of hospitalization in a regular department. Conversely, factors associated with worse sleep quality are pain, hospitalization in a medical department, elective admission, and previous hospitalization. Among the factors examined, alprazolam intake was recognized as the most influential on sleep. However, these factors accounted for only 18% of the variability in perceived sleep quality during the first days of hospitalization on a regular department. During the first three days of hospitalization, there was no significant change between men and woman in subjective ratings of sleep quality. The FIRST questionnaire does not appear to be a suitable screening tool for identifying individuals with situational sleep disturbance on general hospital departments.