Non-pharmacological interventions to improve sleep quality and quantity for hospitalized adult patients—co-produced study with surgical patient partners: systematic review

Abstract Background Hospitalized patients experience sleep disruption with consequential physiological and psychological effects. Surgical patients are particularly at risk due to surgical stress and postoperative pain. This systematic review aimed to identify non-pharmacological interventions for improving sleep and exploring their effects on sleep-related and clinical outcomes. Methods A systematic literature search was performed in accordance with PRISMA guidelines and was preregistered on the Open Science Framework (doi: 10.17605/OSF.IO/EA6BN) and last updated in November 2023. Studies that evaluated non-pharmacological interventions for hospitalized, adult patients were included. Thematic content analysis was performed to identify hypothesized mechanisms of action and modes of administration, in collaboration with a patient partner. Risk of bias assessment was performed using the Cochrane Risk Of Bias (ROB) or Risk Of Bias In Non-Randomized Studies – of Interventions (ROBINS-I) tools. Results A total of 59 eligible studies and data from 14 035 patients were included; 28 (47.5%) were randomized trials and 26 included surgical patients (10 trials). Thirteen unique non-pharmacological interventions were identified, 17 sleep measures and 7 linked health-related outcomes. Thematic analysis revealed two major themes for improving sleep in hospital inpatients: enhancing the sleep environment and utilizing relaxation and mindfulness techniques. Two methods of administration, self-administered and carer-administered, were identified. Environmental interventions, such as physical aids, and relaxation interventions, including aromatherapy, showed benefits to sleep measures. There was a lack of standardized sleep measurement and an overall moderate to high risk of bias across all studies. Conclusions This systematic review has identified several sleep interventions that are likely to benefit adult surgical patients, but there remains a lack of high-quality evidence to support their routine implementation.


Introduction
Sleep is vital for recovery from injury 1 .As well as the removal of metabolic waste, sleep is important for cellular responses in the body 2,3 .Sleep deprivation, therefore, can have significant adverse effects on normal physiological processes, including increased susceptibility to infection, overactivation of the sympathetic nervous system and increased risk of delirium [4][5][6] .Despite these risks, sleep deprivation is common in hospital, and particularly in the perioperative setting where environmental disturbances are common, pain and anxiety can affect sleep quantity and quality, and wards can be high-turnover and manage acute conditions and complications 7 .
Patients undergoing elective surgery experience a significant insult during major surgery and may take months to return to their functional and physiological baseline, if at all.During this time they are at risk of surgical complications that may delay their recovery and apply further systemic stress 8 .Enhanced Recovery After Surgery (ERAS) guidelines for postoperative care have been widely implemented across the world to improve mobility, diet, fluid status and analgesia, and reduce unnecessary interventions (for example routine nasogastric placement after colorectal resection) 9 .However, no ERAS guidelines currently include sleep quantity or quality.Given the negative effects of sleep deprivation, interventions to improve sleep after surgery have the potential to both moderate the surgical stress responses and mediate high compliance with other components of the ERAS pathway (for example by improving appetite and energy for mobilization).
Although other systematic reviews evaluating the effectiveness of sleep interventions for hospital inpatients exist, these have largely focused on critical care populations and drug therapies 10,11 .Pharmacological therapies typically have unattractive side-effect profiles that may hinder postoperative recovery 12 .Further research in non-pharmacological interventions (NPIs) for use in non-critical care areas including surgical wards is urgently needed 13 .In addition, traditional methods for sleep measurement such as highly controlled sleep studies are not feasible in a hospital environment.Future trials of NPIs in the surgical setting will have to adopt innovative, but validated methods for sleep measurement.The objectives of this systematic review were therefore three-fold: first, to identify and evaluate interventions tested out to improve sleep quality and quantity; second, to identify the approaches to measuring sleep in hospitalized adult patients; and third, to extract other sleep-associated health-related outcomes.The overall aim was to inform the co-development of a future randomized trial in patients undergoing surgery.

Study design and search strategy
A systematic review was conducted in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-analysis (PRISMA) guidelines and the protocol was preregistered on the Open Science Framework (doi: 10.17605/OSF.IO/EA6BN) 14 .The search included papers published up until 27 December 2020 and was developed through iterative preliminary searches using PubMed (Supplementary methods).The search was last updated in November 2023.Where available, search strategies from existing systematic reviews in similar topic areas were used to further inform the search strategy.In addition to this, the Cochrane Database of Systematic Reviews (CDSR) was reviewed to identify reviews of sleep NPIs, and a hand search was further carried out to ensure all studies that met the inclusion criteria were identified.As the aim of the review was to provide a thorough overview of a topic area, a search of the grey literature was also conducted using System for Information on Grey Literature in Europe (SIGLE).The following sources were searched without date restrictions: PubMed and Ovid via Medline.Endnote (Clarivate Analytics) was used to collate all references from the databases and identify duplicate studies.Reasons for exclusion of any full-text articles were recorded.

Study inclusion and exclusion criteria
Inclusion and exclusion criteria were based upon study, patient and hospital characteristics.Primary research studies including RCTs, prospective or retrospective observational studies were eligible.Studies reporting NPIs in adult, non-ventilated patients without pre-existing sleep disorders were included.Any studies with pharmacological interventions only or mixed interventions which could not be disaggregated or with mixed adult and paediatric data or ventilated and non-ventilated patients which could not be disaggregated were excluded.

Patient and public involvement
The idea for this review arose from discussion with surgical patients and was highlighted during a perioperative patient advisory group meeting with Patients and Research Together (PART) from Bowel Research UK.As this review targeted a high-priority area for patients, the study protocol and process was co-produced in partnership with a patient advocate (S.B.).The patient representative formed part of the core study steering group and was invited to participate in all aspects of the review.This included designing the search strategy, performing the thematic analysis, drafting and reviewing the study manuscript, and drafting the visual abstract.The final manuscript was sent back to members of the PART group for comment.In order to report the impact of patient and public involvement activity within this review, the Guidance for Reporting Involvement of Patients and the Public (GRIPP2) short-form reporting checklist was used 15 (Table S1).

Data analysis
Data extraction was performed by two independent reviewers (R.A. and B.H.).Any discrepancies were resolved during a study group meeting including the senior author (J.G.), until a consensus was achieved.Study characteristics, including study design, sample size and country of origin were presented.The reporting transparency was assessed by whether authors had cited the corresponding Enhancing the QUAlity and Transparency Of health Research (EQUATOR) network guideline 16 .Participant characteristics such as hospital setting, age groups, disease types and use of sedating analgesics were presented.For studies including surgical patients, included operation types were also described.
The data analysis plan was co-developed with a patient partner and structured around the three predefined research objectives.First, NPIs were extracted verbatim and then combined to reduce redundancy to form unique NPI definitions.A three-stage process of thematic analysis was undertaken based on the hypothesized mechanism of action and method of administration.Conceptual themes were extracted and underwent double coding.The themes were reviewed and refined in a patient advisory group meeting (R.A., J.G., S.B.).The final thematic groups were reviewed across the study steering group before being accepted.The frequency of reporting of each NPI (and thematic group) across different patient groups and hospital settings was explored to identify differences in their patterns of application.Next, using evidence synthesis and critical appraisal the directionality of effect of NPIs (and thematic groups) on hospitalized adult patients was described to identify early signals of patient benefit.Due to predicted heterogeneity of study populations and interventions, meta-analysis was not preplanned.Second, measures of sleep that were used in included studies were extracted and grouped into self-reported or physiological measures.Differences in timing, frequency and types of measure used across patient groups and hospital settings were explored and compared.Third, any short-term health-related outcomes relevant to sleep were extracted.To enrich this process, the Cochrane Risk of Bias (ROB) tool was used to assess for risk of bias in randomized studies, and the risk of bias in non-randomized studies of intervention (ROBINS-1) tool was used for non-randomized studies.

Results
A total of 59 full-text studies were included in data extraction and evidence synthesis (Fig. 1)  . TableS2 displays study and patient characteristics.Of 59 eligible studies, 28 studies were RCTs, ten were non-randomized interventional studies and 21 were prospective cohort studies.In total, data from 14 035 participants was included in this review.The reporting transparency of included studies was poor; only three studies reported the use of EQUATOR network guidelines 21,50,60 and one study used the Transparent Reporting of Evaluations with Nonrandomized Designs (TREND) reporting guidelines for non-randomized clinical trials 73 .No conflicts of interest or funding discrepancies were identified.

Thematic coding and analysis results
After iterative coding and refinement, two major themes were identified relating to the hypothesized mechanism of action: improving the sleep environment, and relaxation and mindfulness.Two further themes arose related to method of administration: self-administered and carer-administered.The two most common NPIs were environmental modifications to the patient's environment and light therapy.Apart from one study, all studies evaluating relaxation NPIs were clinician administered.In surgical patients, the most common theme was environmental, specifically multimodal interventions and physical aids.Key feedback points from involvement of patients in iterative coding and interpretation were two-fold: that future trials in this area must be co-produced to ensure that they are feasible, acceptable and speak to patients' true lived experience; that sleep disturbance in hospital is likely to be multifactorial and individual interventions are unlikely to succeed in isolation.

Environmental NPIs
Environmental NPIs focused on minimizing sleep disturbances created from the patient's environment, as further described in Table 1.Sleep-promotion aids were incorporated within intervention bundles, with ten studies investigating the direct effects of ear plugs and eye masks 19,37,40,47,61,66,69,72,74,75 .Nine of these studies reported statistically significant improvements in sleep domains 19,37,40,47,61,66,69,72,74 .Twenty (33.9%) studies focused on improving the sleep environment for hospitalized patients, including creating quiet time protocols for the patients and caring staff.The results of these studies were inconsistent across all studies.Improvements were reported in the duration of sleep, sleep efficiency and subjective ratings of sleep quality (Table 2).
Results from nurse-led observations showed improvements in the number of patients asleep during the intervention.As the protocols assessed across the studies involved several different interventions, a definitive cause-and-effect relationship cannot be established.Seven (12.5%) 21,23,24,34,36,50,71 studies investigated bright light exposure, and were mainly conducted on geriatric or psychiatric patients.Reported improvements following bright light exposure were in sleep duration and sleep quality.One study exploring the effects of a privacy curtain designed to increase speech privacy and reduce noise disturbances reported an increase in sleep measure score and an increase in the patient's self-reported ability to rest 43 .

Relaxation NPIs
Of the 21 studies investigating relaxation NPIs, 13 (61.9% of all relaxation NPI studies) showed statistically significant improvements in sleep-related outcomes.Acupuncture was evaluated in two studies, and both showed statistically significant improvements in sleep outcome.Garcia et al. reported a significant improvement in drowsiness and fatigue compared with baseline for patients and Tas et al. demonstrated a statistically significant decrease (P < 0.001) in insomnia amongst patients receiving acupuncture treatment 31,67 .Massage therapy showed improvements in the quantity of sleep in only one study, whereby the total sleep time for participants receiving the back massage group was 62.5 min longer than in the control group, as well as a shorter latency to sleep onset 57 .Progressive muscle relaxation therapy showed significant improvements in two studies where participants in the intervention group had greater improvements in sleep-related outcomes (P < 0.050) compared with the control group 17,18 .Relaxation therapy combined with guided imagery was found to reduce fatigue and sleep disturbances amongst participants 48 .The effects of aromatherapy on sleep were variable; however, three studies reported improvements in sleep quality and total sleep scores 22,35,53 .One study exploring the benefits of a milk-honey mixture reported significant improvements in sleep scores between the intervention and control group 27 .A study involving back rubs, warm drinks and relaxation tapes reported positive dose-wise effects on chart-abstracted and self-reported sleep 46 .

Sleep measurement
The outcomes relating to sleep quantity and quality varied significantly across all studies, with the majority involving subjective measurements ( ) studies created a study-specific questionnaire, with no pilot testing or validation 47,56,59,68,69 .

Reporting of clinical outcomes
The reporting of clinical outcomes was inconsistent across studies, with the most common physiological outcomes including vital signs (n = 7), depression (n = 11), delirium (n = 7), nausea (n = 4), pain (n = 14), anxiety (n = 14) and duration of hospitalization (n = 18).Studies conducted on surgical patients also reported changes to postoperative pain and duration of hospital stay.No studies reported the effect of sleep on appetite, mobility, infection or wound healing.The majority of NPIs showing an improvement in clinical outcomes (typically anxiety and delirium) were themed as relaxation/mindfulness.

Risk of bias assessment
In the ROB-2 assessment, most RCTs performed adequate randomization processes, commonly through a computer randomization method and drawing a random number.However, a few studies lacked specific details of the method of sequence generation.Allocation concealment was seldom reported across RCTs.Due to the nature of NPIs, most studies failed to blind participants or personnel involved.Therefore, all studies were at risk of performance and detection bias.Considerable risk of bias was present regarding the measurement of outcomes due to the use of self-reported questionnaires.In the ROBINS-I assessment, serious risk of bias was present due to the measurement of outcomes as well as the use of subjective sleep measures, which are at a high risk of performance bias.Several studies reporting dropouts or losses to follow-up were at risk of attrition bias, particularly as insufficient detail regarding the reasons for missing outcome data were documented.Most non-randomized studies failed to    76 .The patient advisory group highlighted sleep as one of the most disturbing influences on wellbeing during postoperative recovery.Among factors such as pain and medication effects which are common in the postoperative setting, interruptions caused by noise and light levels have been shown to contribute towards disrupted sleep 77 .Given the physical and cognitive limitations of hospitalized patients, most environmental interventions in the review were passive in nature and were administered to, rather than by, patients.Interventions designed to minimize environmental noise were multifactorial and included: clustering care activities, ensuring designated quiet time at night, dimming lights and closing doors if necessary.Currently, the use of NPIs during postoperative recovery is not standardized practice across all hospitals.Consideration of both self-administered and clinical-administered NPIs will be essential in maintaining an optimum and adaptable sleep environment for

Satisfaction of sleep
Self-reported questionnaire 26

Dream quality
Self-reported questionnaire 70

Use of sleep-promoting aids
Self-reported questionnaire 64

Number of participants asleep
Nurse-led observations 32,51

Number of awakenings
Nurse-led observations 23,36 , actigraphy 36 , polysomnography 57 , self-reported questionnaire 19,44,45,56,58 Sleep fragmentation index Actigraphy 36 Total sleep time or sleep quantity Actigraphy 33,34,36,40,71,73 , polysomnography 57,65 , nurse-led observations 19,20,23,36,39,51 , self-reported questionnaire [20][21][22]37,40,41,58,59,61,74  Self-reported questionnaire 17,20,21,25,[28][29][30]35,39,46,50,55,57,60 , nurse-led observation/medical records 25,46,66 patients in the future. Optimization of the spital environment is attainable with multidisciplinary support. However, due to the multifactorial nature of slp hygiene protocols, the review is unable to ascertain the specific components that benefitted sleep the most.Implementation of multifactorial environmental NPIs was variable and clinical activities, sometimes necessarily (for example to maintain safety), took precedence over the NPI.It is important that caring staff are motivated to reduce the most disruptive factors to patients' sleep.Therefore, future improvement initiatives should be co-designed with deep stakeholder engagement to both ensure that proposed NPIs are feasible and acceptable, and to increase awareness of the importance of sleep hygiene 78 .This should follow National Institute for Health and Care Research (NIHR) and Medical Research Council (MRC) complex intervention development recommendations 79 .Bright light therapy demonstrated improvements in sleep duration and efficiency.However, these studies were largely conducted amongst geriatric and psychiatric patients, which highlights a need for further research amongst surgical patients.
The use of ear plugs and eye masks to minimize sleep disruption proved to be a plausible and practical NPI across medical and surgical patients 19,37,40,47,61,66,69,72,74 .The low quality of studies and moderate to serious risk of bias is in accordance with other reviews 10,80 .Compliance with the use of physical sleeping aids was also variable.Relaxation/mindfulness interventions aim to induce anti-anxiolytic effects and restore the resting state.Aromatherapy oils resulted in statistically significant improvements in patients' self-reporting quality of sleep, consistent with prior reviews on the effect of aromatherapy.Acupuncture had consistent positive findings on subjective measures of sleep.However, the small sample sizes and poor compliance with self-reported questionnaires affected the validity of the results.
Identifying the mode of administration of sleep intervention is crucial for optimizing effectiveness and enhancing patient comfort.Eye masks and ear plugs showed benefits across sleep domains, and both are self-administered, allowing patients to be part of their care and providing an individualized method for improving sleep.Relaxation therapies including aromatherapy and massage therapy require the involvement of additional specialized individuals to deliver the intervention.This ought to be taken into account when assessing the feasibility of implementing a relaxation/mindfulness intervention in clinical practice.
Currently, a diverse range of methods of measuring sleep-related domains exist, with many studies using an unvalidated questionnaire.The use of unvalidated questionnaires can reduce the credibility of the data and outcomes may be subject to measurement error.Self-reported questionnaires had poor compliance and patient symptoms appeared to hinder the collection of self-reported measurements of sleep quality and quantity-a vital consideration for future trials.No studies evaluated the effect of sleep interventions on validated patient-reported measures of patient recovery from surgery such as Quality of Recovery-15 (QoR-15) 81 .Objective measures of sleep were seldom used, which may largely be due to the extensive technology required to record and analyse the recordings, rendering methods such as polysomnography expensive and labour-intensive 82 .Novel wearable technologies for sleep measurement pose an attractive, accessible target for objective sleep monitoring 83 .
The aetiology of sleep disruption in surgical patients is multifactorial.This review identified several NPIs that demonstrated significant improvements in sleep and related clinical measures.Patients undergoing surgery may benefit from both environmental NPIs such as sleep masks and ear plugs and relaxation/mindfulness NPIs, such as muscle relaxation therapy or aromatherapy.These are unlikely to interfere with patient treatment or affect the safety of other patients, but may significantly reduce anxiety, stress and sleep disturbance during the perioperative period 84 .The cost of implementing sleep-improving interventions is likely to vary significantly.
Healthcare providers must ensure the sustainability and accessibility of sleep-improving interventions prior to implementation.Eye masks and ear plugs are self-administered interventions and require minimal input from additional specialists, in contrast to relaxation/ mindfulness interventions, which may be associated with higher costs; it was not possible to compare the clinical or cost effectiveness of the included NPIs here.The findings from this study should guide the collaborative development of future RCTs, focusing on improving sleep quality and quantity in surgical wards, with active involvement and input from patients.
This review had several limitations.First, the aim of this review was not to estimate the exact efficacy of all sleep NPIs by assimilating the available evidence.The extensive scope of the subject area and heterogeneity between study interventions and outcomes meant meta-analysis was not possible.Second, although the search strategy used was thorough, grey literature may not have been captured.Third, the thematic analysis and co-production meetings were systematic and transparent, but the nature of qualitative synthesis is that other teams may have identified other themes or performed categorization in another way.The study team aimed to improve credibility and transferability by performing double coding and within-team discussion, and transparently reporting the methodology.Fourth, the review included studies without surgical patients in the data synthesis.Whilst this may provide opportunities for cross-disciplinary learning, potential differences in the cause of and solutions for sleep disturbance between surgical and other hospitalized patients should be recognized (for example acute postoperative pain).
Improving sleep in hospital for adult surgical patients is likely to require a multimodal strategy, which may include promising components such as environmental modifications, physical sleep adjuvants and aromatherapy.Measures of sleep adopted in published research are heterogenous, and paired objective measurement and patient-reported methods are likely to be important in parallel.There is a lack of high-quality evidence to link sleep improvement with other health-related outcomes and this warrants further exploration in future research.

FundingJ
.G. is funded through a doctoral research fellowship from the National Institute of Health and Care Research (NIHR) (NIHR300175).The funder and sponsor had no role in study design or writing of this article.The funder has approved the submission of this article for publication.The views expressed are those of the authors and not necessarily those of the National Health Service, the NIHR or the UK Department of Health and Social Care.

Table 3 )
. Twenty-four (40.1%) studies reported sleep duration or total sleep time (TST), 43 (72.9%)studies reported sleep quality and 24 (40.7%)studies reported sleep latency.Most studies used self-reported measures of sleep: 53 studies administered questionnaires to the participants and four studies reported sleep logs.Five studies employed the use of a questionnaire completed by the caring staff or designated surveyor.The three most common types of validated questionnaire employed by studies included: the Richards-Campbell Sleep Questionnaire (RCSQ), the Pittsburgh Sleep Quality Index (PSQI) and the Verran and Snyder-Halpern (VSH) sleep scale.Five (8.9%