Non‐pharmacological interventions for sleep promotion in hospitalized children

Abstract

This is a protocol for a Cochrane Review (Intervention). The objectives are as follows:

To evaluate the effect of non‐pharmacological sleep promotion interventions in the hospital setting on the:

1. sleep quality, sleep duration, length of stay, and mortality of hospitalized infants and children; and

2. delirium incidence, length of mechanical ventilation, and patient or parent satisfaction.

Background

Description of the condition

Sleep is a basic human need. The process of rest and restoration occur during the natural suspension of consciousness that occurs during sleep (Kryger 2005; Kudchadkar 2009; Kudchadkar 2014a; Kudchadkar 2014b; Murali 2003). Sleep is not a passive process, but one that is characterized by dynamic physiological changes. While the function of sleep remains elusive, the physiological effects and consequences of altered sleep can impact multiple systems throughout the body (Kudchadkar 2014a).

It is well known that sleep needs are in a constant state of change as a child matures from the neonatal to adolescent period, reflecting brain maturation (Feinberg 2013). During recovery from illness, children admitted to the hospital are exposed to a multitude of risk factors for sleep disruption, including noise, pain, anxiety, medications, interruptions for nursing care, and invasive medical interventions. Studies that used objective and subjective assessment tools have demonstrated severe sleep disruption in hospitalized children (Cureton‐Lane 1997; Hinds 2007; Kudchadkar 2014a; Kudchadkar 2016c; Meltzer 2012). Sleep disruption occurs at a time when recovery and healing are the goal, potentially interferes with fundamental physiological processes and can lead to increased energy release, impaired immunity, and delirium (Barnes 2016; Kamdar 2015). Additionally, sleep disturbances from infancy through adolescence are associated with changes in brain morphology and worsened short‐ and long‐term neurocognitive outcomes (Cheung 2017; Kocevska 2017; Saré 2016).

Although there are proven, inexpensive and non‐invasive modalities to promote sleep, such methods are rarely used for children in the hospital setting (Kudchadkar 2014b). A recent systematic review of the literature surrounding sleep in the pediatric intensive care unit (ICU) found only nine studies that investigated the quality of sleep in critically‐ill children (Kudchadkar 2014a). All of the studies demonstrated significant sleep fragmentation and decreases in slow‐wave sleep, which is the most restorative aspect of sleep and an integral component of cognitive maturation during childhood and adolescence (Kudchadkar 2014a). In the hospital floor setting, sleep disturbance due to noise, pain, vital sign checks and anxiety is associated with increased night awakenings and longer sleep latency (time to sleep onset) (Meltzer 2012). Among children recovering from major surgery, approximately 40% demonstrate no identifiable difference between night‐time and day‐time activity levels, as measured by actigraphy, suggesting significant circadian rhythm (sleep‐wake cycle) disturbance (Kudchadkar 2016b). Sleep promotion is not a priority in hospital culture, despite increasing evidence in the literature that sleep loss and fragmentation in critically‐ill adults increase the risk of delirium, which is an important risk factor for increased morbidity and mortality (Pandharipande 2006; Smith 2011). A recent Cochrane Review summarized available evidence for non‐pharmacological interventions to promote sleep in critically‐ill adults admitted to the ICU (Hu 2015). However, the role of non‐pharmacological interventions for infants and children have not been characterized.

Hospitals across the country are noisy, brightly lit environments where care providers make multiple interventions throughout the day and night to assist children’s recovery. Few hospitals employ noise reduction strategies to target World Health Organization (WHO) recommended levels (less than 30 A‐weighted decibels (dBA) for day and night time), and the current literature demonstrates that ICU noise levels are often greater than 50 dBA regardless of the time of day, with several intermittent peaks that exceed 80 dBA (Busch‐Vishniac 2005; Kudchadkar 2014b; Liu 2005). Additionally, a lack of natural sunlight and abolished patterns of normal light‐dark exposure is common in the hospital setting, with adverse effects on sleep architecture (the basic structural organization of normal sleep) and circadian rhythms (Glotzbach 1993; Kudchadkar 2016a). A 2007 study showed that 6% of all hospitalized children are prescribed medications to promote sleep, which can include opioids, benzodiazepines, and diphenhydramine (Meltzer 2007). Although these pharmacologic agents may decrease the time to sleep onset, they can significantly impact sleep architecture and lead to increased sleep fragmentation (Kudchadkar 2009).

Description of the intervention

Given the adverse effects of sedatives and analgesics on sleep physiology, the use of non‐pharmacological interventions has been explored in adult patients, primarily in the ICU environment (Hu 2015; Kamdar 2014). These interventions can be categorized as: 1) environmental; 2) behavioral (massage, music therapy, guided imagery); and 3) physical therapy interventions (mobility/exercise during the day to improve sleep at night) (Kamdar 2016; Saliski 2015). Possible environmental interventions include: earplugs, alarm modifications, headphones, white noise or unit‐based 'quiet hours' (Foreman 2015; Freedman 2001; Hu 2015; Walder 2000), cycled lighting and bright light therapy (Simons 2016; Taguchi 2007). Behavioral interventions that have been investigated in hospitalized patients include massage, acupressure and relaxation interventions, including music therapy and guided imagery (Richards 1998; Richardson 2003). All of these modalities have potential applications for the pediatric population, with the addition of kangaroo care (method of holding an infant that involves skin‐to‐skin contact) in the neonatal ICU (NICU) setting (Baley 2015). Although potential pharmacological therapies, such as melatonin and atypical antipsychotic medications, are increasingly being used in the pediatric hospital setting with a goal of sleep promotion or delirium prevention (or both), the short‐ and long‐term effects of these agents are not well studied in the developing brain. Therefore, a focus on non‐invasive, low‐risk interventions is needed to promote sleep in hospitalized children undergoing active neurocognitive development.

How the intervention might work

There is a growing body of literature supporting the effectiveness of non‐pharmacological interventions in promoting sleep in hospitalized adults (Kamdar 2014; Koo 2008; Richards 2003; Richardson 2007; Scotto 2009). Hospitalized pediatric patients also encounter disrupted sleep and are just as likely to experience side effects with pharmacological interventions that promote sleep. Sudden peaks in noise levels that occur frequently in the ICU increase arousals from sleep as demonstrated by polysomnography, the gold‐standard test for sleep measurement (Aaron 1996). In the pediatric ICU (PICU), noise is a strong predictor of sleep state, with louder noises correlating to increased awakenings from sleep (Cureton‐Lane 1997). Continuous exposure to artificial light can abolish the normal circadian rhythm and diminish the normal peak in melatonin secretion at night. Environmental interventions, such as noise reduction and lighting optimization, provide safe and effective strategies to promote sleep. Reducing the background level as well as peak levels of noise with noise reduction interventions decreases the likelihood of arousing a child from sleep in the hospital setting. Day‐night cycling of light could help normalize the circadian rhythm of hospitalized children. While behavioral interventions provide modes of relaxation to decrease stress and anxiety that is common for hospitalized infants and children, physical therapy interventions engage the hospitalized child in activities and exercise during the day to promote rest at night (Hopkins 2015; Wieczorek 2016). These interventions must be easy for caregivers to implement, and ideally, should have high compliance amongst patients to optimize efficacy and effectively change the culture for sleep promotion (Kudchadkar 2014b).

Why it is important to do this review

Despite evidence that sleep disruption in hospitalized adults has a negative impact on outcomes and inexpensive sleep promotion interventions can decrease morbidity, there is a lack of awareness in the pediatric community about modifiable risk factors for sleep disruption in hospitalized infants and children. There is a critical need for a synthesis of the current evidence to understand the potential benefits of specific sleep promotion interventions in infants and children undergoing active neurocognitive development. There are several risk factors for sleep fragmentation in acutely‐ill children and several medications used to improve sleep (i.e. benzodiazepines, diphenhydramine) that may, in fact, have a negative effect on sleep quality. Therefore, it is important to evaluate the effect of non‐pharmacologic interventions to promote sleep in these vulnerable patients.

Objectives

To evaluate the effect of non‐pharmacological sleep promotion interventions in the hospital setting on the:

1. sleep quality, sleep duration, length of stay, and mortality of hospitalized infants and children; and

2. delirium incidence, length of mechanical ventilation, and patient or parent satisfaction.

Methods

Criteria for considering studies for this review

Types of studies

Randomized controlled trials (RCTs).

Types of participants

We will include studies of infants and children from birth to 18 years of age, admitted to the hospital for more than 48 hours in any inpatient unit. Infants and children may be surgical or medical patients, or need mechanical ventilation. We will include studies with participants both above and below the age of 18 years providing we can obtain the data for the pediatric patients only.

We will exclude studies that focus solely on children with obstructive sleep apnea (OSA). If studies include children with other types of sleep disorders, we will consider a sensitivity analysis to evaluate these studies (Sensitivity analysis).

Types of interventions

Studies investigating the effects of any of the following non‐pharmacological sleep promotion interventions on the sleep quality or sleep duration (or both) of infants and children in the hospital setting.

1. Environmental interventions, including but not limited to earplugs, headphones, alarm modifications, white noise, music therapy or unit‐based 'quiet hours', lighting control/cycling, eye masks, and bright light therapy, or a combination of these

2. Behavioral Interventions, including but not limited to: kangaroo care (skin‐to‐skin contact), massage, music therapy, and guided imagery

3. Physical therapy interventions such as mobility or exercise during the day

4. Complementary and alternative therapies such as aromatherapy and acupressure or acupuncture

5. Any other non‐pharmacological intervention intended to promote the sleep of children in the hospital

Studies may include one or a combination of interventions and compare them to usual care or an alternative intervention.

Types of outcome measures

Primary outcomes

1. Changes in objective and validated measures of sleep, as assessed using a polysomnography or actigraphy (or both); specifically, total sleep time (TST, based on age‐based normative data), percentage active/rapid eye movement (REM) sleep, percentage of slow‐wave sleep, sleep efficiency (Laffan 2010), and Sleep Fragmentation Index (SFI; Haba‐Rubio 2004), during the period of intervention

2. Subjective measures of sleep, as measured by:

   1. validated parent and nurse surveys; and

   2. subjective, validated sleep assessment tools such as the Patient Sleep Behavior Observation Tool (Corser 1996; Cureton‐Lane 1997)

Secondary outcomes

1. Patient or parent satisfaction, as described by the study authors

2. Cost‐effectiveness ratios

3. Delirium incidence or delirium‐free days at time of hospital discharge

4. Length of mechanical ventilation (days)

5. Length of hospital stay (days)

6. Mortality

Search methods for identification of studies

Electronic searches

To identify relevant studies for inclusion in this review, we will search the following databases and trials registers.

1. Cochrane Central Register of Controlled Trials (CENTRAL; current issue) in the Cochrane Library, which includes the Cochrane Developmental, Psychosocial and Learning Problems Specialised Register

2. PubMed US National Library of Medicine (www.ncbi.nlm.nih.gov/pubmed; current issue)

3. Embase Elsevier (1980 onwards)

4. CINAHLPlus EBSCOhost (Cumulative Index to Nursing and Allied Health Literature; 1950 onwards)

5. Cochrane Database of Systematic Reviews (CDSR; current issue) in the Cochrane Library

6. Epistemonikos (www.epistemonikos.org/en)

7. ProQuest Digital Dissertations and Theses (current issue)

8. ClinicalTrials.gov (clinicaltrials.gov)

9. ISRCTN Registry (www.isrctn.com)

10. WHO International Clinical Trials Registry Platform (ICTRP; www.who.int/ictrp/en)

In addition, we will search the grey literature using the following websites.

1. OpenGrey (www.opengrey.eu)

2. Grey Literature Reports at the New York Academy of Medicine Library (greylit.org)

We will not limit our searches by publication date or language, or exclude studies based on the year the study was performed, if all other inclusion criteria are met (Criteria for considering studies for this review). We will search PubMed using the search strategy in Appendix 1 , which will be adapted appropriately for other electronic sources listed above.

Searching other resources

We will search the reference lists of included studies and relevant review articles as well as searching Google Scholar (1980 to present). We will also handsearch relevant conference proceedings, including the Associated Professional Sleep Societies and the American Thoracic Society Conferences, as well as key sleep journals, including the following.

1. Sleep (1990 to present)

2. Journal of Clinical Sleep Medicine (2005 to present)

3. American Journal of Respiratory and Critical Care Medicine (1995 to present)

4. Sleep Medicine Reviews (1998 to present)

Data collection and analysis

Selection of studies

After merging the results of the above literature search strategy in Endnote (Endnote 2016), we will remove any identified duplicates and export the titles and abstracts into Excel (Microsoft 2016). Each study will be randomly assigned a number and rearranged in numeric order. Two pairs of reviewers will be assigned an equal number of studies for review. The two review authors in each pair will independently review the titles and abstracts based on the inclusion criteria defined under Criteria for considering studies for this review. Each reviewer will classify the abstracts as 'yes ‐ include' or 'no ‐ exclude'. Any disagreements will be reviewed and discussed together by both review authors for consensus. If there is disagreement after discussion, the study will be included for full‐text review. The same pair of authors will review the full‐texts of the 'include' group of titles and abstracts. Disagreements will be resolved through discussion until a consensus is reached. After full‐text review, studies deemed as 'exclude from data extraction' by both review authors will be excluded. The final list of studies chosen by both pairs of review authors for data extraction will be merged again and renumbered for random assignment to two independent review authors as detailed in Data extraction and management. We will record the results of our selection process in a PRISMA diagram (Moher 2009).

Data extraction and management

Two review authors will extract data independently utilizing a standardized form developed collectively by the review authors. One review author will enter data onto this form and the other member of the pair will re‐enter the data using a double‐data strategy. All disagreements will be resolved by group consensus and third party consultation, as needed. Since it is possible that some studies may be reported in more than one article, we will extract the data from these studies separately from each article and combine them across multiple data extraction forms. We will extract the following data.

1. Study characteristics

   1. country of study

   2. year of study

   3. study design

   4. method of randomization (if applicable)

   5. unit of analysis

   6. setting (inpatient general floor, subspecialty floor [i.e. oncology, surgery, etc.], PICU or NICU)

   7. outcome measures, and

   8. conflict of interest and declaration of conflict of interest

2. Participant characteristics

   1. age

   2. sex

   3. race

   4. inclusion/exclusion criteria

   5. comorbidity (prematurity, developmental delay, traumatic brain injury, surgery)

   6. admission diagnosis, and

   7. hospital length of stay

3. Intervention

   1. type of intervention (earplugs, earmuffs, headphones, music therapy, white noise, unit‐based 'quiet‐time' protocol, alarm modifications etc.)

   2. timing of intervention

   3. duration of intervention

   4. frequency of intervention, and

   5. any other associated interventions (i.e. sleep promotion 'bundles' using multiple interventions to promote sleep)

Assessment of risk of bias in included studies

Independently, two review authors will assess the included studies for sources of systematic bias. We will assess studies using the Cochrane 'Risk of bias' tool, as recommended by the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2017). For each included study, we will classify the risk of bias as 'low', 'high' or 'unclear' (see Appendix 2), across each of the following domains: random sequence generation and allocation concealment (selection bias), blinding of outcome assessment (detection bias), incomplete outcome data (attrition bias), selective reporting (reporting bias), and other potential sources of bias (other bias). If there is disagreement, we will discuss and resolve it with collaboration of a third author and, if needed, a fourth author.

Measures of treatment effect

Dichotomous data

We will report dichotomous outcomes as risk ratios (RR), and present these with 95% confidence intervals (CI) and P values.

Continuous data

After verifying a normal distribution of continuous outcomes, we will calculate mean values for the primary outcomes of interest (TST, SFI, sleep efficiency), and present these with 95% CI. Specifically, we will calculate the mean difference (MD) where studies have assessed the same outcome using the same assessment measure, and the standardized MD (SMD) where studies have used different assessment measures.

Multiple outcome data

If there are multiple time points included in a study, we will make comparisons at the following time points when feasible: short term (within one week postintervention), medium term (within two months postintervention), and long term (any time after two months postintervention).

Time‐to‐event data

We will analyze time‐to‐event (survival) outcomes (i.e. incidence of delirium, death) using hazard ratios (HR) and present these with 95% CIs.

Unit of analysis issues

Due to the nature of the intervention, it is possible that we will encounter studies utilizing cluster‐randomization, cross‐over trials, and studies with more than two intervention groups. For these studies, we will conduct the analysis as recommended in the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2011), and outlined below.

Cluster‐randomised trials

If the cluster design was appropriately accounted for in the study (i.e. cluster‐level analyses or regression analyses of individual level data using methods for clustered data (e.g. random‐effects models, marginal modeling using generalized estimating equations); Higgins 2011), we will analyze the direct estimate of the required effect measure at the level of the individual. If not, we will conduct the analysis at the same level as the allocation, where each cluster is treated as an individual. We will extract adjusted data only and pool it using the generic inverse variance method.

Cross‐over trials

We will analyze continuous data from a two‐period, two‐intervention cross‐over trial using a paired t‐test. A paired analysis will be possible if any of the following are available:

1. individual participant data;

2. means and standard deviations or errors of participant‐specific differences between measurements from the experimental and control groups;

3. MD and either a t statistic from paired t‐test or CI from a paired analysis; or

4. a graph of measurements on interventions and controls where individual matched data values can be identified and extracted.

We will note cross‐over trials with problematic periods or carry‐over effects in the 'Risk of bias' assessment.

Multiple groups

We will manage unit‐of‐analysis issues with studies including multiple, correlated comparisons by combining all relevant experimental groups of the study into one group and combining all relevant control groups into one group, to create a single pair‐wise comparison, where possible. If it is not possible to combine all relevant experimental groups of the study into one group, we will split the control group to ensure participants are not double counted.

Dealing with missing data

When possible, we will contact the original investigators of the included studies to provide missing participant data, including reasons for drop out as well as missing information regarding study design, to assist with 'Risk of bias' assessments. If it is not possible to obtain these data from the authors, we will calculate missing statistics (e.g. correlation coefficients or standard deviations) from other available statistics (e.g. standard error or CI). We will impute data using all available information, and for missing participant data, we will impute the data using the last observation carried forward (LOCF) (Higgins 2011). If we make assumptions during imputation, these will be documented on the data extraction sheet and reported in the ‘Risk of bias’ tables. If we impute data, we will conduct a sensitivity analysis to assess the impact on the results (see Sensitivity analysis). In summarizing aggregate data where LOCF is not possible, we will address the potential impact of the missing data on the findings of the review in the Discussion section.

Assessment of heterogeneity

Through assessment of variations in participant, intervention and outcome characteristics, we will examine clinical as well as methodological heterogeneity. These will include the following.

1. Participant characteristics: age, gender, comorbidities, inclusion/exclusion criteria

2. Intervention characteristics: type of noise reduction intervention, timing, duration

3. Outcome characteristics: method of measurement, timing

We will use the I² statistic (%) to determine the proportion of variation between included studies that is due to heterogeneity; a value above 50% will define substantial statistical heterogeneity (Deeks 2017). In addition, we will examine the result of the Chi² test (P value < 0.10 = significant heterogeneity) and CI overlap of included studies, with visual inspection of the forest plot, where poor overlap is suggestive of heterogeneity. The likelihood of high variability in participants, interventions, and outcomes between studies is great (see Data synthesis).

We will report Tau² — an estimate of between‐study variance — with use of a random‐effects model.

Assessment of reporting biases

To reduce the risk of publication bias, we will attempt to obtain and include data from unpublished trials through a search of the grey literature. If a meta‐analysis includes more than 10 studies, we will draw a funnel plot and inspect it visually for asymmetry that may be caused by publication bias or other small study effects attributable to, for example, poor methodological quality, true heterogeneity, or chance.

Data synthesis

We will use Review Manager 5 (RevMan 5) to conduct the meta‐analyses (Review Manager 2014). We will make separate comparisons between each category of intervention (i.e. environmental interventions, behavioral interventions, physical therapy, complementary and alternative therapies, and any other non‐pharmacological intervention) and usual care or alternative interventions (see Types of interventions) in each of the following three settings: NICU, PICU, and inpatient floor.

We will use a random‐effects model to combine the results of trials included in the review for meta‐analysis, as there will likely be clinical and methodological heterogeneity. Using RevMan 5, we will calculate the RR for dichotomous outcomes using the Mantel‐Haenszel method. When continuous outcomes are measured in the same way across studies, we will summarize across‐study MD estimates using the inverse variance method, with 95% CI and P values; we will use the SMD when studies use different methods to measure the same outcome. If variance data are not available, we will not estimate the mean between‐group difference with 95% CI and P value.

We will not perform a meta‐analysis if there is significant clinical and methodological heterogeneity (see Assessment of heterogeneity). Instead, we will present a narrative synthesis of our review with relevant tables (of study characteristics and results), along with a comprehensive discussion of each study’s methodology or quality (or both) that may affect the quantitative result obtained. We will consider all potential sources of bias in the review, in addition to our methodology for assessment and control of bias.

'Summary of findings' tables

We will summarise the results of each category of comparisons described in Data synthesis and present the primary outcomes in a 'Summary of findings' table. We will create this table with software developed by the GRADE Working Group (GRADEpro 2015). Using the GRADE approach, two independent review authors will assess the quality of the body of evidence for each outcome as 'high', 'moderate', 'low', or 'very low' according to the presence of the following criteria:

1. limitations in design and implementation;

2. indirectness of evidence;

3. unexplained heterogeneity or inconsistent results;

4. imprecision of results; and

5. high risk of publication bias.

We will provide a narrative description of important, clinically‐relevant findings.

Subgroup analysis and investigation of heterogeneity

We will conduct the subgroup analyses listed below using the statistical test for subgroup differences.

1. Age (neonates, 1 month to < 2 years, ≥ 2 years to < 6 years, ≥ 6 years to < 13 years, and ≥ 13 years to 18 years): age is an important effect modifier to consider given developmental differences in sleep between infants, toddlers, children of school age and adolescents.

2. Diagnosis or subspecialty unit (oncology, cardiology, surgical, etc.): diagnosis is an important effect modifier to consider given the potential effect of different organ system pathology on sleep. Additionally, surgical pain can impact sleep significantly.

Sensitivity analysis

We will perform sensitivity analyses to determine the impact of the following issues, if they are present among the included studies.

1. Reanalysis excluding studies of lower methodological quality (higher risk of bias)

2. Reanalysis excluding studies published prior to 1990

3. Reanalysis including cluster‐RCTs

4. Reanalysis considering decisions made regarding cross‐over studies

Appendices

Appendix 1. PubMed search strategy

1. "Sleep"[Mesh] 2. "Sleep Stages"[Mesh] 3. "Sleep, REM"[Mesh] 4. "Sleep Deprivation"[Mesh] 5. (sleep[tiab] OR sleeping[tiab] OR slept[tiab] OR sleeps[tiab] OR sleepless[tiab] OR sleeplessness[tiab] OR sleepy[tiab] OR asleep[tiab] OR insomnia[tiab] OR insomniac[tiab] OR somnolent[tiab] OR somnolence[tiab])) 6. "Circadian Rhythm"[Mesh] OR "circadian"[tiab] 7. rest[tiab] OR resting[tiab] OR nap[tiab] OR naps[tiab] OR napping[tiab] 8. #1 OR #2 OR #3 OR #4 OR #5 OR #6 OR #7 9. "Noise"[Mesh] OR "noise"[tiab] OR "noises"[tiab] OR "noisy"[tiab] 10. "Sound"[Mesh:noexp] OR "sound"[tiab] OR "sounds"[tiab] OR "sound masking"[tiab] 11. "Ear Protective Devices"[Mesh] OR "ear protective device"[tiab] OR "ear protective devices"[tiab] OR "earplug"[tiab] OR "earplugs"[tiab] OR "ear plug"[tiab] OR "ear plugs"[tiab] OR "earmuff"[tiab] OR "earmuffs"[tiab] OR "ear muff"[tiab] OR "ear muffs"[tiab] OR "headphone"[tiab] OR "headphones"[tiab] OR "head phone"[tiab] OR "head phones"[tiab] OR "hat"[tiab] OR "hats"[tiab] OR "music"[tiab] 12. "Light"[Mesh] OR light*[tiab] OR "Eye Protective Devices"[Mesh] OR "eye protective device"[tiab] OR "eye protective devices"[tiab] OR "eye masks"[tiab] OR "eye mask"[tiab] 13. "Polysomnography"[Mesh] OR "polysomnography"[tiab] OR "polysomnographic"[tiab] OR "polysomnogram"[tiab] OR "polysomnograms"[tiab] OR "Actigraphy"[Mesh] OR "actigraphy"[tiab] OR "actigraphic"[tiab] OR "actigram"[tiab] OR “actigrams"[tiab] 14. "Music Therapy"[MeSH Terms] OR "musical"[tiab] OR musicotherap*[tiab] OR "song"[tiab] OR "songs"[tiab] OR lullab*[tiab] OR sing[tiab] OR sings[tiab] OR "singing"[tiab] OR "sang"[tiab] OR sung[tiab] 15. "Complementary Therapies"[Mesh:noexp] OR "Mind‐Body Therapies"[Mesh:noexp] OR "Aromatherapy"[Mesh] OR "Relaxation Therapy"[Mesh] OR "Biofeedback, Psychology"[Mesh] OR "Imagery (Psychotherapy)"[Mesh] OR "Therapeutic Touch"[Mesh] OR "Breathing Exercises"[Mesh] OR "Massage"[Mesh] OR "Acupressure"[Mesh] 16. "non‐pharmacological"[tiab] OR massag*[tiab] OR "acupressure"[tiab] OR aromatherapy[tiab] OR complementary therap*[tiab] OR alternative therap*[tiab] OR mind‐body therap*[tiab] OR biofeedback[tiab] OR breathing exercise*[tiab] OR imagery[tiab] OR muscle relaxation[tiab] OR relaxation therap*[tiab] OR therapeutic touch[tiab] 17. environmental intervention*[tiab] OR "quiet time"[tiab] OR "alarm modifications"[tiab] OR "alarm modification"[tiab] 18. "Social Support"[Mesh] OR social support[tiab] 19. "Physical Therapy Modalities"[Mesh] OR "physical therapy"[tiab] OR "physical therapies"[tiab] OR "rehabilitation"[tiab] OR "rehab"[tiab] 20. "Kangaroo Care"[All Fields] OR "Kangaroo Mother Care"[All Fields] 21. #9 OR #10 OR #11 OR #12 OR #13 OR #14 OR #15 OR #16 OR #17 OR #18 OR #19 OR #20 22. #8 OR #21 23. "Intensive Care Units"[Mesh] OR "ICU"[tiab] 24. "Intensive Care Units, Pediatric"[MeSH] OR "PICU"[tiab] OR "pediatric unit"[tiab] OR "pediatric units"[tiab] OR "paediatric unit"[tiab] OR "paediatric units"[tiab] 25. "Intensive Care Units, Neonatal"[MeSH] OR "NICU"[tiab] OR "neonatal unit"[tiab] OR "neonatal units"[tiab] 26. "intensive care"[tiab] OR "intensive therapy"[tiab] OR "high dependency"[tiab] OR "Critical Care"[Mesh] OR "critical care"[tiab] 27. "Burn Units"[Mesh] OR "burn units"[tiab] OR "burn unit"[tiab] 28. "Recovery Room"[MeSH] OR "recovery room"[tiab] OR "recovery rooms"[tiab] 29. "Coronary Care Units"[MeSH] OR "coronary care unit"[tiab] OR "coronary care units"[tiab] OR "cardiac care unit"[tiab] OR "cardiac care units"[tiab] OR "CCU"[tiab] 30. "Respiratory Care Units"[MeSH] OR "respiratory care unit"[tiab] OR "respiratory care units"[tiab] 31. hospitali*[tiab] OR "Inpatients"[Mesh] OR "inpatients"[tiab] OR "inpatient"[tiab] OR "in hospital"[tiab] OR "hospital ward"[tiab] OR "hospital wards"[tiab] 32. "pediatric wards"[tiab] OR "pediatric ward"[tiab] OR "paediatric wards"[tiab] OR "paediatric ward"[tiab] OR "patient rooms"[tiab] OR "patient room"[tiab] 33. "critically ill"[tiab] OR "Critical Illness"[Mesh] OR "critical illness"[tiab] OR "critical illnesses"[tiab] OR "acute illness"[tiab] OR "acute illnesses"[tiab] OR "acute care"[tiab] 34. "Hemodialysis Units, Hospital"[Mesh] OR "hemodialysis units"[tiab] OR "hemodialysis unit"[tiab] OR "haemodialysis units"[tiab] OR "haemodialysis unit"[tiab] 35. "post acute care unit"[tiab] OR "post anesthesia care unit"[tiab] OR "post anesthesia care units"[tiab] OR "post anaesthesia care unit"[tiab] OR "post anaesthesia care units" OR “PACU"[tiab] 36. #23 OR #24 OR #25 OR #26 OR #27 OR #28 OR #29 OR #30 OR #31 OR #32 OR #33 #34 OR #35 37. #22 AND #36 38. randomized controlled trial[pt] OR controlled clinical trial[pt] OR randomized[tiab] OR placebo[tiab] OR "drug therapy"[Subheading] OR randomly[tiab] OR trial[tiab] OR groups[tiab] NOT ("animals"[MeSH Terms] NOT "humans"[MeSH Terms]) 39. #37 AND #38

Appendix 2. Criteria for judging risk of bias in RCTs

Random sequence generation (selection bias)

1. Low risk of bias: sufficient use of a random component for adequate sequence generation (i.e. computer random number generator or random number tables)

2. High risk of bias: insufficient component of randomization (i.e. participants are assigned by date of birth or date of presentation)

3. Unclear risk of bias: authors have not sufficiently explained the randomization and how it was performed in order to permit a judgment of low or high risk of bias

Allocation concealment (selection bias)

1. Low risk of bias: adequate concealment of the allocation (i.e. use of consecutively‐numbered, sealed, opaque envelopes or telephone randomization)

2. High risk of bias: the allocation was definitely not adequately concealed (i.e. open random number lists, odd or even date of birth)

3. Unclear risk of bias: uncertainty about whether the allocation was adequately concealed (i.e. concealment method not known)

Blinding of participants and personnel (performance bias)

We will not assess blinding of participants and personnel, as study participants, families, and personnel will be aware of the intervention they are receiving or implementing.

Blinding of outcome assessment (detection bias)

1. Low risk of bias: a. no blinding of outcome assessment but the review authors judge that the outcome measurement is not likely to be influenced by the lack of blinding; or b. blinding of outcome assessment is ensured

2. High risk of bias: blinding did not occur and the outcome measurement is likely to be influenced by the lack of blinding

3. Unclear risk of bias: blinding of outcome assessor is not reported and cannot be confirmed through contact with study authors

Incomplete outcome data (attrition bias)

1. Low risk of bias: a. no missing outcome data; b. reasons for missing outcome data are unlikely to be related to the true outcome (for survival data, censoring is unlikely to introduce bias); c. missing outcome data are balanced in numbers across intervention groups, with similar reasons for missing data across groups; d. for dichotomous outcome data, the proportion of missing outcomes compared with the observed event risk is not enough to have a clinically relevant impact on the intervention effect estimate; e. for continuous outcome data, plausible effect size (difference in means or standardized difference in means) among missing outcomes is not enough to have a clinically relevant impact on observed effect size; or f. missing data have been imputed using appropriate methods

2. High risk of bias: missing data are not accounted, not imputed using appropriate methods, or not evenly distributed between groups

3. Unclear risk of bias: missing data or losses to follow‐up, or both, are not reported

Selective reporting (reporting bias)

1. Low risk of bias: a. the study protocol is available and all of the study’s pre‐specified (primary and secondary) outcomes that are of interest in the review have been reported in the pre‐specified way; or b. the study protocol is not available but it is clear that the published reports include all expected outcomes, including those that were pre‐specified (convincing text of this nature may be uncommon)

2. High risk of bias: not all pre‐specified primary outcomes have been reported

3. Unclear risk of bias: not enough information to determine if the study is at low or high risk of reporting bias

Other potential sources of bias

1. Low risk of bias: study is free from other apparent sources of bias

2. High risk of bias: there is potential, additional source(s) of bias that is not captured by the other domains above (i.e. bias related to study design)

3. Unclear risk of bias: not enough information to determine if the study is at low or high risk of other bias

Contributions of authors

Sapna R Kudchadkar conceived the review, coordinated the writing of the protocol, authored and revised the protocol, contributed to the formulation of the proposed search strategy, and has overall responsibility for the review. Sean Barnes authored and revised the protocol. Blair Anton authored and revised the protocol, and contributed to the formulation of the proposed search strategy. Daniel J Gergen authored and revised the protocol. Naresh M Punjabi authored and revised the protocol.

Sources of support

Internal sources

• The Johns Hopkins University School of Medicine, Department of Anesthesiology and Critical Care Medicine, Baltimore, USA.

  Provided salary support for Dr Sapna R Kudchadkar

• Johns Hopkins Clinical and Translational Science Award (CTSA) Number 5KL2RR025006 from the National Center for Advancing Translational Sciences of the National Institutes of Health, Baltimore, USA.

  Provided funding for Dr Sapna R Kudchadkar

External sources

• None, Other.

Declarations of interest

Sapna R Kudchadkar is employed at The Johns Hopkins University, School of Medicine. She and her institution have received grants from the following non‐profit organizations: Society for Anesthesia and Sleep Medicine, The American Thoracic Society, the Thomas Wilson Sanitarium, and the National Institutes of Health. Sean Barnes ‐ none known. Blair Anton ‐ none known. Daniel J Gergen ‐ none known. Naresh M Punjabi has received research grant support to his institution from the for‐profit organizations Resmed and Philips‐Respironics, which are unrelated to the current publication, with no implications for potential bias.

Disclaimer: The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health. ​

New