Non‐pharmacological interventions for sleep disturbances in people with dementia
Abstract
This is a protocol for a Cochrane Review (Intervention). The objectives are as follows:
To identify, appraise and summarise the best evidence for the efficacy of non‐pharmacological interventions for sleep disturbances in people with dementia. To describe the components and processes of any complex intervention used.
Background
Description of the condition
Dementia is a clinical syndrome characterised by cognitive, neuropsychiatric, and functional symptoms. It involves difficulties in memory, disturbances in language, psychological and psychiatric changes as well as impairments in activities of daily living (Burns 2009).
Worldwide about 24 million people are affected by dementia. Because of the age profile of the population, numbers are especially high in western European and northern American countries with approximately 6% of people over 60 affected by dementia. Worldwide the prevalence rate in people over 60 has been estimated between 5% and 8% (Prince 2015). Predictions that the prevalence will increase steadily, with numbers likely to double every 20 years (Burns 2009; Ferri 2005) have been challenged by the results of newer studies indicating a later onset of dementia, possibly leading to a less marked increase in incidence (Larson 2013). There are different presentations of dementia. About 50% of people with dementia have Alzheimer’s Disease (AD), 25% vascular dementia (VaD), 15% dementia with Lewy bodies (DLB), and 5% other forms of dementia. At least 25% of people with dementia have more than one pathology (Burns 2009).
Sleep disturbance and insomnia occur frequently in people with dementia. Common problems are an increase in the duration and number of awakenings and an increased percentage of time spent in stage 1 sleep (Colton 2006). Prevalences of sleep disturbances of up to 40% have been reported in different settings (Dauvilliers 2007; McCurry 1999; Ritchie 1996). Progressive dementia (for example Alzheimer’s disease) has effects on sleep that can be distinguished from normal aging, particularly fragmentation of the sleep/wake cycle and disruption of the circadian regulation of sleep (Song 2010). These changes in sleep regulation and architecture have been related to the deterioration of brain structures and the supply of neurotransmitters relevant for sleep, as well as psychosocial and behavioural changes occurring in people with dementia (Ancoli‐Israel 2006; McCleery 2014).
Sleep disturbances are associated with a number of problems for the affected persons, their relatives and caregivers. In people with dementia it may lead to worsening of cognitive symptoms, challenging behaviours, such as restlessness and wandering, and further harms, such as accidental falls. Sleep disturbances can be associated with significant caregiver distress and have often been reported as a factor contributing to the decision to admit a person with dementia to institutional care (Ancoli‐Israel 2006; Gibson 2014; Lee 2011).
Increases in costs attributable to dementia have been shown for impairment in activities of daily living and cognitive deficits (Hurd 2013; Leicht 2013), but there are insufficient data on which to base reliable estimates of the costs associated with sleep disturbances in people with dementia. Apart from drug therapies (McCleery 2014), a number of non‐pharmacological interventions have been proposed for behavioural symptoms, including sleep disturbances in people with dementia (Lee 2011; Livingston 2014; O'Neil 2011; Salami 2011).
Description of the intervention
Non‐pharmacological interventions include all treatment options that are not medication or drug therapies (Brown 2013; Livingston 2014; O'Neil 2011). A number of classifications of non‐pharmacological interventions for behavioural and psychological symptoms of dementia have been used in earlier systematic reviews (Livingston 2014; O'Neil 2011). However, a pre‐planned categorisation seems inappropriate here as non‐pharmacological interventions which have been proposed to improve sleep in people with dementia are frequently multi‐faceted. The components may include modifications to the environment (e.g. increased exposure to natural light, decreased night‐time noise and light), changes to care routines (e.g. decreased daytime in‐bed time, increased daytime physical activity, structured bedtime routines), behavioural interventions (increased daytime physical activity and exercise), or other sleep hygiene measures (e.g. decreased night‐time noise and light, avoidance of caffeinated drinks). Other interventions may include sensory stimulation (e.g. aromatherapy, touch and massage, transcutaneous electrical nerve stimulation (TENS) (O'Neil 2011)), individual relaxation therapies, and complementary therapies (e.g. acupuncture). Bright light therapy has been suggested as an intervention of specific benefit to sleep and may be implemented alone (Forbes 2014; Montgomery 2003).
How the intervention might work
Non‐pharmacological interventions use different mechanisms to improve the management of sleep disorders in people with dementia. For example, environmental interventions aim to improve sleep by providing conditions that allow for physiological sleep, while sensory interventions target the lack of sensory input in people with dementia that could cause disruptions in peoples’ internal circadian rhythms. There seem to be several advantages over pharmacological interventions. Depending on the intervention, compliance tends to be good and adverse events are usually minimal. Furthermore, treatment efficacy may last longer compared to pharmacological treatments, where positive effects tend to stop with treatment cessation (e.g. in behavioural interventions). Because of the increased frequency of side effects of drug treatments in older people with dementia, non‐pharmacological management of sleep problems has been proposed as a first treatment option (David 2010).
Why it is important to do this review
Sleep disturbances are associated with a number of problems for people with dementia as well as significant caregiver distress, and have often been reported as a contributing factor to institutionalisation. Therefore, there is a clear need to rigorously synthesise the research evidence on strategies to improve sleep in people with dementia.
A recently published Cochrane review (McCleery 2014) shows a distinct lack of evidence regarding successful pharmacological interventions to manage sleep problems in people with dementia. Therefore, a Cochrane review on non‐pharmacological interventions would complement the results of McCleery 2014. A number of non‐pharmacological interventions as outlined above have been proposed for sleep disturbances in people with dementia. Most published systematic reviews analysed specific interventions in specific settings, but not always for people with dementia. For example Brown 2013; Livingston 2014, and Montgomery 2003 included all people older than 50 or 65 with and without dementia.
There are two recent systematic reviews of non‐pharmacological interventions for sleep disturbances in people with dementia including 13 (Brown 2013) and nine (Salami 2011) randomised controlled trials. These reviews also included a range of other study designs and use suboptimal tools for quality appraisal (e.g. the outdated Jadad score in Brown 2013). In addition, there are two recent, more general systematic reviews on non‐pharmacological interventions for people with dementia (Livingston 2014; O'Neil 2011) that do not specifically focus on sleep disturbances.
The results of this review may overlap with two Cochrane reviews on light therapy for people with dementia (Forbes 2014) and for adults aged 60+ (Montgomery 2003), but studies on bright light therapy will be included in this review if appropriate as we aim to be comprehensive in reviewing the evidence on non‐pharmacological interventions for sleep problems in dementia.
In contrast to already existing reviews, the planned systematic review aims to give an overview of any kind of non‐pharmacological interventions irrespective of setting and type of dementia. Importantly, none of the available reviews has adequately considered the challenges of synthesising complex interventions (Anderson 2013). Recently, we (Möhler 2015) and others (Datta 2013) have highlighted the need to adequately describe and summarise important factors concerning the development, evaluation and implementation of interventions used in systematic reviews of complex interventions. This seem warranted in order to identify effective intervention approaches and in addition lead to the development of new interventions on the basis of the current best evidence.
At present, there is no rigorous systematic review focusing on non‐pharmacological interventions for sleep disturbances. We believe this is needed, because of the importance of the clinical question and the widespread use of drug treatments of questionable effectiveness which may cause significant harm.
Objectives
To identify, appraise and summarise the best evidence for the efficacy of non‐pharmacological interventions for sleep disturbances in people with dementia. To describe the components and processes of any complex intervention used.
Methods
Criteria for considering studies for this review
Types of studies
We will include all randomised controlled trials (RCTs), randomising either individuals or clusters, investigating the effects of interventions to improve physiological sleep in people with dementia. Cross‐over designs will also be included. To be included, studies must have a sleep‐related outcome measure as a primary outcome. Studies may be published in any language.
Types of participants
We will include all people with a diagnosis of dementia, irrespective of age, type of dementia, severity of cognitive impairment and setting. Diagnoses of dementia may be made using any established diagnostic criteria. In studies also including people without dementia, we will use results for the subgroup of people with dementia. If these data are not available, studies will only be included if at least 80% of participants are people with dementia. If necessary, authors will be contacted to determine rates of people with dementia. Participants must also have a sleep problem at baseline, diagnosed on the basis of any subjective or objective measures.
We will exclude studies of people with dementia and sleep apnoea, as this is primarily a respiratory problem requiring different treatment strategies (McCleery 2014).
Types of interventions
We will include any non‐pharmacological intervention aiming to improve physiological sleep in people with dementia. Studies where patients receive medication but no other type of intervention will be excluded. As described above, we will not group interventions into specific intervention categories.
It is highly likely that there will be a number of interventions designed as complex interventions (Craig 2008), where it may not be possible to extract the effective components of the interventions. Therefore, components of included programs will be analysed in detail using suitable guidelines e.g. the TIDieR (Hoffmann 2014) or the CReDECI 2‐criteria (Möhler 2015), or both.
Comparator interventions may be of any type, including usual care (may be described as 'no treatment') and optimised usual care, any other non‐pharmacological intervention or any drug treatment intended to improve sleep.
Types of outcome measures
Where possible, outcomes will be categorised as short‐term (following treatment of up to six weeks) and long‐term (following treatment of more than six weeks).
Primary outcomes
The main outcomes of interest are objective sleep‐related outcomes as used in the primary studies. We expect a number of different outcome measures such as sleep onset latency (the length of time between wakefulness and sleep), waking after sleep onset, total waking time, total sleep duration, ratio of day‐time sleep to night‐time sleep or of night‐time sleep to total sleep over 24 hours, number of nocturnal awakenings and sleep‐efficiency. The outcomes will be assessed by objective measurements e.g. (wrist) actigraphy, polysomnography or observation e.g. using the Observational Sleep Assessment Instrument (OSAI) or other sleep‐related rating scales. As additional primary endpoint, we will assess adverse events (e.g. use of physical restraints or psychotropic medication) as reported in primary studies.
Secondary outcomes
-
All subjective sleep‐related outcomes, e.g. quality of sleep, patient‐ or caregiver‐reported sleep satisfaction assessed using sleep‐related rating scales (e.g. Pittsburgh sleep quality index), sleep logs, diaries, surveys or sleep charts.
-
Behavioural and psychological symptoms of dementia, including agitation and the 'sundown' phenomenon.
-
Quality of life.
-
Functional status.
-
Institutionalisation
-
Compliance with the intervention.
-
Attrition rates (as indicator for intervention acceptability).
For caregivers we will consider relevant endpoints such as distress and quality of life.
Search methods for identification of studies
Electronic searches
We will search ALOIS, the register of the Cochrane Dementia and Cognitive Improvement Group (CDCIG) using the following search terms: non‐pharmacological [using the intervention type filter] AND sleep disturbance (“sleep disturbance” OR “insomnia” OR “sleep disorder” OR “circadian rhythm” etc.) AND treatment dementia [using the study aim filter].
ALOIS is maintained by the trials search co‐ordinator and contains dementia and cognitive improvement studies identified from:
-
Monthly searches of a number of major healthcare databases: MEDLINE, EMBASE, CINAHL, PsycINFO and Lilacs;
-
Monthly searches of a number of trial registers: metaRegister of Controlled Trials; UMIN Clinical Trials Register (Japan); the World Health Organization (WHO) portal (which covers ClinicalTrials.gov; International Standard Randomised Controlled Trial Number (ISRCTN); Chinese Clinical Trials Register; German Clinical Trials Register; Iranian Registry of Clinical Trials and the Netherlands National Trials Register, plus others);
-
Quarterly searches of the Cochrane Central Register of Controlled Trials (CENTRAL);
-
Six‐monthly searches of a number of grey literature sources: ISI Web of Knowledge Conference Proceedings; Index to Theses; Australasian Digital Theses.
To view a list of all sources searched for ALOIS see About ALOIS (http://www.medicine.ox.ac.uk/alois) on the ALOIS website.
Additional separate searches will be run in many of the above sources to ensure the most up‐to‐date results are retrieved. See Appendix 1 for the MEDLINE (via Ovid SP) search strategy.
Searching other resources
We will review reference lists of included studies and relevant reviews as well as other potentially relevant trials identified through the search. We will contact study authors and experts in the field for unpublished and ongoing studies.
Data collection and analysis
Selection of studies
We will obtain lists of references from different sources and merge these to check for duplicates. Independently, two review authors will assess titles and abstracts from all search results to identify eligible studies. After selection of potentially relevant articles, full reports will be obtained and assessed for inclusion and exclusion criteria. When necessary, we will resolve any disagreement on the eligibility of studies through discussion to reach consensus or, if required, by involving a third experienced review author.
We will access full texts that are not in English or German, if necessary using a language translation service.
Data extraction and management
Two authors will independently read and extract the data from each study included. In case of disagreement or discrepancies, we will involve a third review author to reach consensus. We will use a standardised data extraction form, including source, study characteristics, methods, participants, interventions, comparators, outcomes, results, and adverse events according to the Cochrane Handbook for Systematic Reviews of Interventions (Chapter 7.3; Higgins 2011a).
Assessment of risk of bias in included studies
Assessment of risk of bias of included studies will follow the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2011b). Two authors will independently assess and score included studies’ methodological quality in order to identify any potential sources of systematic bias. Criteria for appraisal of studies will be internal validity and low risk of bias through selection bias, performance bias, attrition bias, and detection bias. Study validity will be determined by categorising individual studies into low, high or unclear risk of bias.
As recommended by the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2011b), we will use a two‐part tool, addressing six domains (sequence generation, allocation concealment, blinding, incomplete outcome data, selective outcome reporting, and other issues), the first part describing what has been reported in the study and the second part assessing the related risk of bias (low, high, or unclear).
The domains of sequence generation, allocation concealment (avoidance of selection bias) and selective outcome reporting (avoidance of reporting bias) will be addressed in the tool by a single entry for each study. Blinding of participants, staff and outcome assessors (avoidance of performance bias and detection bias) will be considered separately for objective outcomes and subjective outcomes. Incomplete outcome data (avoidance of attrition bias) will be considered separately for different lengths of follow up (shorter and longer follow up).
Measures of treatment effect
For continuous outcome data we will use mean differences with 95% confidence intervals (CIs). If we intend to pool data from studies using different measurement scales for the same outcome, we will calculate standardised mean differences (SMD). For the analysis of dichotomous outcome data risk ratios (RR) with 95% CIs will be calculated. Should any relevant ordinal outcome be encountered, we will only consider this if it can justifiably be treated as a continuous variable or can be sensibly dichotomised. We will perform all statistical analyses using Review Manager 5.3 (RevMan 2014).
Unit of analysis issues
Cluster randomised studies
Cluster‐randomised studies will be analysed at the level of individuals while accounting for cluster effects. If individual‐level data are not available, we will use the direct estimate of effect measure (OR with CI) from cluster randomised controlled trials. If study authors failed to control for a clustering effect, we will request individual patient data to calculate an estimate of the intracluster correlation coefficient (ICC). If there are no data available, we will obtain an external estimate of the ICC from similar studies and calculate “effective sample sizes” as decribed in the Cochrane Handbook for Systematic Reviews of Interventions (16.3.4) to approximate analyses of cluster‐randomized trials for meta‐analyses. Meta‐analyses will be performed using the generic inverse variance method in RevMan 5.3 (RevMan 2014).
Cross‐over studies
For cross‐over studies we will only use first period data up to the first point of cross over to rule out carry‐over effects.
Studies with multiple treatment groups
If a study compares two or more eligible intervention groups to one eligible control group, we will split the sample size for the shared comparator group as outlined in chapter 16.5 of the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2011c).
Dealing with missing data
We will report any missing data. Where necessary, we will contact the authors for additional information about missing data. We will use intention to treat (ITT) data where these are available, reporting on any imputation methods used in the primary studies. We will use completer‐only data if no other data are available.
Assessment of heterogeneity
For the assessment of clinical heterogeneity we will examine extracted data for between‐study variability with respect to participants, interventions, technology and outcome measurements. We will use RevMan software (RevMan 2014) to assess statistical heterogeneity using I2 and Chi2 statistics and visual inspections of forest plots (Higgins 2003). If no substantial heterogeneity is found (I2 ≤ 50%, P value > 0.05), we will perform meta‐analyses using random‐effects models.
Assessment of reporting biases
We will include all studies in any language to minimise language bias. If there are at least ten studies included in the meta‐analysis, we will prepare funnel plots to visually estimate small study effects which may reflect reporting bias (Sterne 2011).
Data synthesis
We will only perform meta‐analyses when we consider that studies are sufficiently clinically homogeneous in terms of participants, interventions and outcomes. Meta‐analyses will be performed using the generic inverse variance method in RevMan 5.3 (RevMan 2014). We expect frequent clinical heterogeneity between studies due to mostly diverse or complex interventions, or both. Therefore only random‐effects models will be applied.
Subgroup analysis and investigation of heterogeneity
If possible, subgroup analyses will be performed for:
-
different settings (e.g. nursing homes and home care); as well as
-
type and severity of dementia (e.g. Alzheimer’s disease and vascular dementia);
-
short (0‐ 6 weeks) and long term (> 6 weeks follow up);
-
type of comparator intervention (e.g. (optimised) usual care, non‐pharmacological intervention, drug treatment, or mixed).
Sensitivity analysis
Sensitivity analyses will be used to examine the effect of inclusion or exclusion of low quality studies as well as studies using or not using validated outcome instruments.
Presentation of results – 'Summary of findings' tables
For each comparison, we will assess the overall quality of the evidence for all primary outcomes (i.e. objective sleep‐related outcomes and adverse events) using the GRADE approach (Schünemann 2011a). The GRADE approach defines the quality of a body of evidence as the extent to which one can be confident that an estimate of effect is close to the true quantity of interest. It takes into account risk of bias, imprecision, inconsistency between studies, indirectness of the evidence and publication bias. We will present the main results of the review in ‘Summary of findings’ tables, which provide key information concerning the best estimate of effect of the interventions examined, and the quantity and the quality of the evidence behind each estimate (Schünemann 2011b).
