Cognitive behavioural interventions for sleep problems in adults aged 60+ (Review)

Transcription

1 Cognitive behavioural interventions for sleep problems in adults aged 60+ (Review) Montgomery P, Dennis JA This is a reprint of a Cochrane review, prepared and maintained by The Cochrane Collaboration and published in The Cochrane Library 2003, Issue 1

2 T A B L E O F C O N T E N T S HEADER ABSTRACT PLAIN LANGUAGE SUMMARY BACKGROUND OBJECTIVES METHODS RESULTS DISCUSSION AUTHORS CONCLUSIONS ACKNOWLEDGEMENTS REFERENCES CHARACTERISTICS OF STUDIES DATA AND ANALYSES Analysis 1.1. Comparison 1 Sleep quality: cognitive-behavioural therapy vs. no-treatment control or placebo, Outcome 1 Sleep onset latency (SOL) as reported in participants diaries Analysis 1.2. Comparison 1 Sleep quality: cognitive-behavioural therapy vs. no-treatment control or placebo, Outcome 2 Sleep onset latency (SOL) as measured by polysomnography Analysis 1.3. Comparison 1 Sleep quality: cognitive-behavioural therapy vs. no-treatment control or placebo, Outcome 3 Wake after sleep onset (WASO) as reported in participants diaries Analysis 1.4. Comparison 1 Sleep quality: cognitive-behavioural therapy vs. no-treatment control or placebo, Outcome 4 Wake after sleep onset (WASO) as measured by polysomnography Analysis 1.5. Comparison 1 Sleep quality: cognitive-behavioural therapy vs. no-treatment control or placebo, Outcome 5 Total wake time (TWT) as reported in participants diaries Analysis 1.6. Comparison 1 Sleep quality: cognitive-behavioural therapy vs. no-treatment control or placebo, Outcome 6 Total wake time (TWT) as measured by polysomnography Analysis 1.7. Comparison 1 Sleep quality: cognitive-behavioural therapy vs. no-treatment control or placebo, Outcome 7 Sleep duration (total, in minutes) as reported in participants diaries Analysis 1.8. Comparison 1 Sleep quality: cognitive-behavioural therapy vs. no-treatment control or placebo, Outcome 8 Sleep duration (total, in minutes) as measured by polysomnography Analysis 1.9. Comparison 1 Sleep quality: cognitive-behavioural therapy vs. no-treatment control or placebo, Outcome 9 Early morning wakening (as defined by trialist) as reported in participants diaries Analysis Comparison 1 Sleep quality: cognitive-behavioural therapy vs. no-treatment control or placebo, Outcome 10 Early morning wakening (as defined by trialist) as measured by polysomnography Analysis Comparison 1 Sleep quality: cognitive-behavioural therapy vs. no-treatment control or placebo, Outcome 11 Sleep efficiency (ratio of time asleep/ time in bed) as reported in participants diaries Analysis Comparison 1 Sleep quality: cognitive-behavioural therapy vs. no-treatment control or placebo, Outcome 12 Sleep efficiency (ratio of time asleep/ time in bed) as measured by polysomnography Analysis Comparison 1 Sleep quality: cognitive-behavioural therapy vs. no-treatment control or placebo, Outcome 13 Pittsburgh Sleep Quality Index (PSQI) -- global score APPENDICES WHAT S NEW HISTORY CONTRIBUTIONS OF AUTHORS DECLARATIONS OF INTEREST NOTES INDEX TERMS i

3 [Intervention Review] Cognitive behavioural interventions for sleep problems in adults aged 60+ Paul Montgomery 1, Jane A Dennis 2 1 The Centre for Evidence-Based Intervention, University of Oxford, Oxford, UK. 2 School for Policy Studies, University of Bristol, Bristol, UK Contact address: Paul Montgomery, The Centre for Evidence-Based Intervention, University of Oxford, Barnett House, 32 Wellington Square, Oxford, OX1 2ER, UK. paul.montgomery@socres.ox.ac.uk. Editorial group: Cochrane Developmental, Psychosocial and Learning Problems Group. Publication status and date: Edited (no change to conclusions), published in Issue 1, Review content assessed as up-to-date: 2 February Citation: Montgomery P, Dennis JA. Cognitive behavioural interventions for sleep problems in adults aged 60+. Cochrane Database of Systematic Reviews 2003, Issue 1. Art. No.: CD DOI: / CD Background A B S T R A C T The prevalence of sleep problems in adulthood increases with age. While not all sleep changes are pathological in later life, severe disturbances may lead to depression, cognitive impairments, deterioration of quality of life, significant stresses for carers and increased healthcare costs. The most common treatment for sleep disorders (particularly insomnia) is pharmacological. The efficacy of non-drug interventions has been suggested to be slower than pharmacological methods, but with no risk of drug-related tolerance or dependency. Cognitive and behavioural treatments for sleep problems aim to improve sleep by changing poor sleep habits, promoting better sleep hygiene practices and by challenging negative thoughts, attitudes and beliefs about sleep. Objectives To assess the efficacy of cognitive-behavioural interventions in improving sleep quality, duration and efficiency amongst older adults (aged 60 and above). Search methods The following databases were searched: The Cochrane Library (Issue 1, 2002); MEDLINE ( January 2002); EMBASE ( January 2002), CINAHL ( January 2002); PsycINFO (1887 to 2002); National Research Register (NRR) (2002, Issue 1_. Bibliographies of existing reviews in the area, as well as of all trial reports obtained, were searched. Experts in the field were consulted. Selection criteria Randomised controlled trials of cognitive behavioural treatments for primary insomnia where 80% or more of participants were over 60. Participants must have been screened to exclude those with dementia and/or depression. Data collection and analysis Abstracts of studies identified in searches of electronic databases were read and assessed to determine whether they might meet the inclusion criteria. Data were analysed separately depending on whether results had been obtained subjectively or objectively. 1

4 Main results Six trials, including 282 participants with insomnia, examined the effectiveness of cognitive-behavioural treatments (CBT) for sleep problems in this population. The final total of participants included in the meta-analysis was 224. The data suggest a mild effect of CBT for sleep problems in older adults, best demonstrated for sleep maintenance insomnia. Authors conclusions When the possible side-effects of standard treatment (hypnotics) are considered, there is an argument to be made for clinical use of cognitive-behavioural treatments. Research is needed to establish the likely predictors of success with such treatments. As it may well be the case that the treatment efficacy of cognitive-behavioural therapy itself is not durable, the provision of top-up ( refresher sessions of CBT training to improve durability of effect are worthy of investigation. P L A I N L A N G U A G E S U M M A R Y Cognitive behavioural therapy for older adults (aged 60+) with sleep problems Sleep problems become more common with age, affect quality of life for individuals and their families, and can increase healthcare costs. Older people are often prescribed a range of drugs for their health problems (including with sleep) many of which have side effects. This review considered the effectiveness of cognitive and behavioural treatments (CBT). These aim to improve sleep by changing poor sleep habits and by challenging negative thoughts, attitudes and beliefs about sleep. Reviewers report that there is only limited evidence available, and what data there is suggests a mild effect of CBT. B A C K G R O U N D Description of the condition The prevalence of sleep problems in adulthood increases with age (Brabbins 1993; National Commission on Sleep Disorders Research (NCSDR 1993); Bliwise 1993; Foley 1995; Ford 1989). In the general population the most common types of sleep problems reported are insomnia (both difficulties in initiating and maintaining sleep) and early morning waking with an inability to return to sleep. Older adults primarily report difficulty in maintaining sleep and, while not all sleep changes are pathological in later life (Morin 1989; Bliwise 1993), severe sleep disturbances may lead to depression and cognitive impairments (Ford 1989). Night waking produces significant stresses for carers and is a common cause for demands that institutional living arrangements be made (Pollak 1990). Prevalence rates of insomnia in people aged 65 and over range between 12 and 40% (Morin 1999b). There are reports that the impact of chronic sleep disturbance impairs waking functions (e.g. mood, energy, performance) and life quality (Borkovec 1982; Morin 1989). There is evidence that sleep disturbances contribute significantly to healthcare costs (Stoller 1994; Simon 1997). Overall, the aetiology of sleep problems in the elderly remains uncertain. There may be a developmental perspective to sleep in this population as the question of whether older adults need less sleep has not yet been answered (Bliwise 1993). Prevalence rates of insomnia are even higher when co-existing medical or psychiatric illness is taken into account (Ford 1989; Mellinger 1995). Lifestyle changes related to retirement, the increased incidence of health problems, and the use of medication, all place older people at increased risk of disrupted sleep (Morgan 1988). The relationship between sleep problems and depression in the elderly is particularly strong, but difficult to disentangle. It has been reported in a large study by Ford and Kamerow that depression can predict future sleep disturbance, and that unremitting insomnia can itself cause depression (Ford 1989). However, amongst the eight symptoms of major depressive disorder which may predict development of the full syndrome, sleep disturbance is not the most predictive. Sleep disturbances may also be comorbid with impending dementia, but that does not mean they are the cause. Alzheimer-related deterioration of suprachiasmatic nucleus neurons could cause comorbid sleep disturbance in sleep-wake cycle disorders in particular (Kripke 2001). Despite the high prevalence of sleep disorders and their negative impact, it is estimated that fewer than 15% of patients with chronic insomnia receive treatment (Mellinger 1995). This may be due to a lack of knowledge 2

5 about sleep and its disorders amongst health professionals. It is reported that the median amount of time spent on sleep issues in medical training in the UK is five minutes (Stores 1998) and that in clinical psychology it is no better (Wiggs 1996). The most common treatments for sleep disorders are pharmacological, particularly for insomnia (Hohagen 1994; Kupfer 1997; Morin 1999b). Lack of knowledge about non-drug treatment and limited access to other forms of professional help are cited by physicians as the main reason for prescribing sleeping pills (Baillargeon 1996); however, the long-term efficacy of this approach, which usually involves the administration of hypnotics (typically benzodiazepines) is not certain. Two consensus conferences sponsored by the National Institute of Health (NIH 1983; NIH 1990) concluded that short-term use of hypnotic medications may be useful for acute and situational insomnia across all age groups, but that long-term use remains controversial because of the potential risk of tolerance and dependency. The same NIH studies indicate that the drug of choice for the symptomatic treatment of insomnia is a benzodiazepine receptor agonist (e.g. temazepam, zolpidem etc). One study by Nowell et al found that these drugs improve sleep latency (the time between going to bed and going to sleep), number of awakenings, total sleep time and total sleep quality (Nowell 1997). Post-treatment problems were not adequately investigated, as follow-up in this study only extended to one to two nights following discontinuation of the drug s administration. Other drugs, e.g. zaleplon, one of the most popular current benzodiazepine agonists, do not even significantly increase total sleep time. Ultimately, both consensus conferences clearly recommended against long term use of hypnotics. It has been reported that older people are more likely to be affected by daytime residual effects of these types of drugs (Morgan 1988; Prinz 1990; Kripke 2000); that these drugs may increase the likelihood of patients developing sleep apnoea (Kripke 1983), as well as increasing the risk for falls and fractures (Wettstein 1992; Meyer 1998). Constipation has been correlated positively with hypnotic use (Campbell 1993). Despite this, data have suggested that persons over 60 years of age in the USA are prescribed sedative-hypnotic drugs at more than twice the rate of people years of age (Baum et al 1986). Survey data suggest that older adults in France, Italy, Germany and Canada are even more likely to use hypnotics than Americans (Morin 1999b). The effects of non-drug interventions has been suggested to be slower but more durable than pharmacological methods (McClusky 1991;Milby 1993; Hauri 1997). In view of the potential risks of tolerance and dependency and the frequently high numbers of other drugs that older people may be taking, an evidence-based non-drug approach would be of interest. Description of the intervention This is the first of four interrelated reviews investigating non-drug treatments for sleep problems in the older adult. In addition to the present review on cognitive behavioural treatments, subsequent reviews will cover physical treatments (Montgomery 2002b), bright light therapy (Montgomery 2002a) and hypnosis. A paper summary review will set out the evidence for the full range of nondrug treatments in an effort to answer the clinical question What alternatives to medication exist for sleep problems in the older adult? Cognitive and behavioural treatments for sleep problems aim to improve sleep by: 1. changing poor sleep habits 2. challenging negative thoughts attitudes and beliefs about sleep Cognitive-behavioural interventions include a broad range of treatments, from educational packages to purely behavioural strategies. Those included within this review are: Sleep Hygiene Education Sleep hygiene education aims to teach individuals about the impact of lifestyle habits such as diet, exercise and drug use and the influence of environmental factors e.g. light, noise and temperature (Hauri 1991). While these issues are unlikely to cause insomnia (Reynolds 1991) they may well exacerbate it. Studies generally advocate (1) the avoidance of caffeine and nicotine (both stimulants) in the 6 hours before bed; (2) the avoidance of alcohol around bedtime (alcohol may facilitate sleep onset but it tends to cause fragmentation of sleep and nightmares); (3) the avoidance of a heavy meal before sleep (although a light meal may be helpful); (4) the avoidance of exercise close to bed-time (even though in general exercise is helpful for sleep); and (5) the minimisation of noise, light and excessive heat during the sleep period. Sleep hygiene also includes information concerning age-appropriate sleep duration to ensure realistic expectations. Stimulus Control Stimulus control (Bootzin 1991) involves a set of instructions aimed at helping the individual to re-associate the bed, bedtime and bedtime stimuli with sleep rather than with the frustration or anxiety resulting from lying in bed trying to sleep. Participants are instructed to (1) only go to bed when sleepy; (2) only to use the bed for sleeping and sex; (3) to leave the bed if they have not gone to sleep within minutes and to go back only when feeling sleepy again, to be repeated as often as necessary through the night; (4) to get up at the same time each morning regardless of the amount of sleep achieved in the previous night; and (5) not to sleep during the day. Countercontrol treatment (Zwart 1979) can be seen as a version of stimulus control. It was designed to disrupt sleep-incompatible activities but omitted introducing other features of classical stimulus control, e.g. leaving the bed. Muscle Relaxation Therapy 3

6 Relaxation therapy consists of alternating tension and relaxation of various muscle groups in sequence (Woolfolk 1983). The Progressive Muscle Relaxation technique as devised by Edmund Jacobson involved systematic training in tensing and relaxing groups of muscles, first separately, then with the aim of relaxing the whole body at once (Jacobson 1974). Some studies have added imagery to the relaxation (Borkovec 1978). To assess the efficacy of cognitive-behavioural interventions in improving sleep quality, duration and efficiency amongst older adults (aged 60 and above). M E T H O D S Sleep Restriction Therapy A common treatment is sleep restriction therapy (Spielman 1987). This method limits the time spent in bed at night and restricts sleep during the day. Participants estimate an allowable time in bed from the diaries kept over the previous two weeks. For example if a person spends eight hours in bed and only actually sleeps for six of them, their allowed time in bed would be six hours. Weekly adjustments are made to this amount by looking at the individual s sleep efficiency (ratio of total sleep time to time spent in bed). When sleep efficiency reaches 90%, the time allowed in bed increases by minutes. These adjustments continue until the expected optimal amount of sleep time is reached. The urge to sleep will be increased during each stage of the treatment and in this way it is thought to increase the homeostatic drive for sleep. This type of treatment is considered likely to be effective with older people, many of whom may have tried to compensate for their poor sleep by spending more time in bed (Miles 1980). How the intervention might work Cognitive therapy for insomnia (Morin 1993) consists of identifying, challenging and altering a set of dysfunctional belief and attitudes about sleep and its impact on day-to-day life. It encompasses many aspects of the above treatments. The object is to break the vicious cycle of insomnia; dysfunctional cognitions and emotional distress that leads to further sleep disturbance. For example, commonly misplaced beliefs include I cannot function unless I sleep for 8 hours per night and if I try harder I will eventually fall asleep. Why it is important to do this review Meta-analyses of these treatments with younger people (Morin 1994; Murtagh 1995) have suggested that these interventions improve sleep in 70-80% of such people with insomnia and restoring sleep latency and time awake after sleep onset to near normative values. O B J E C T I V E S Criteria for considering studies for this review Types of studies All relevant randomised controlled trials in which participants had been randomly allocated to an intervention group and a control group. The control groups were either waiting-list control groups or placebo. Types of participants In determining a cut-off point in age for this review, the age of 60 was chosen as being most clinically relevant, following consultation (DPOA 2000). Trials whose focus was explicitly on the older adult were included where 80% or more of participants were recorded as being over the age of 60. Participants must have been diagnosed with sleep problems via standardised measures (e.g. the PSQI [PSQI 1989]), objective measures in sleep laboratory (e.g. polysomnography, actigraphy) or by participants own sleep diaries or reports/diaries kept by partners or nursing staff. Participants must also have been screened to exclude those with dementia and/ or depression by the use of psychometrically sound measures such as the Mini Mental State Examination (MMSE) (Folstein 1975 ), Beck Depression Inventory (Groth-Marnat 1990) or comparable instrument(s). This was to avoid the confounding effects of these conditions. Sleep problems addressed in this, and related reviews include: Primary sleep problems: difficulties in initiating and maintaining sleep sleep efficiency sleep latency delayed or advanced sleep phase problems parasomnias impaired daytime functioning As sleep apnoea is primarily treated as a respiratory condition, trials whose participants who had been diagnosed as having sleep apnoea were excluded. Those with secondary insomnia or sleep disturbance caused by a psychiatric or medical disorder were also excluded. 4

7 Types of interventions All forms of cognitive-behavioural therapy, including sleep hygiene, stimulus control, muscle relaxation, sleep restriction and cognitive therapies, were included. Types of outcome measures Outcomes measures of interest to the review question include: Sleep onset latency (time taken to fall asleep) Wake after sleep onset (WASO) Total wake-time (TWT) Sleep duration (total) Early morning wakening (defined by the trialist) Sleep efficiency (ratio of time asleep / over time in bed) Self-report of sleep satisfaction Scales related to sleep, e.g.. the Pittsburgh Sleep Quality Index (PSQI (PSQI 1989)); the Sleep Impairment Index (Morin 1993a). Daytime functioning (as measured by attentional tasks tests, self-report using a standardised measure, e.g. the Stanford Sleepiness Scale (Hoddes 1973), the Epworth Sleepiness Scale (Johns 1991)). Quality of life, as measured by validated scales Outcomes were divided, where possible, into immediate posttreatment, medium term (3-12 months), and long term (more than 12 months). Search methods for identification of studies Electronic searches The following electronic databases were searched: The Cochrane Controlled Trials Register (The Cochrane Library, Issue 1, 2002), MEDLINE ( January 2002); EMBASE ( January 2002); CINAHL ( January 2002); PsycINFO Journal Articles and Chapter/Books (1887 to 2002); National Research Register (NRR) (2002, Issue1) ; and the sleep bibliography available at ( ). The search terms used to isolate controlled trials are shown in Appendix 1 and were used to search the Cochrane Library. They were modified as necessary for the other databases. Searching other resources Reference lists of articles identified through database searches were examined to identify further relevant studies. Bibliographies of systematic and non-systematic review articles were also examined to identify relevant studies and experts in the field were consulted. Data collection and analysis Selection of studies All reports of studies identified as above were inspected independently by the two reviewers. Disagreements regarding relevance were resolved by acquisition and reading of the full article and discussion between the reviewers. All selected articles were independently assessed to determine if they met inclusion criteria including limits on age, diagnosis and screening for comorbid conditions. The reviewers were not blinded to the names of the authors, institutions or journal of publication. Provision was made for arbitration by a third reviewer although this was not required. Data extraction and management Data were extracted independently by each reviewer, and compared using data extraction sheets and the double entry feature in RevMan 4.1. Where it was not possible to extract any data because they were not available or further information was needed, the first author of the trial was contacted for clarification. Where it was possible to extract relevant data, comments on the methods, participants, interventions and outcomes were presented in the Included Studies table. Individual patient data were requested from authors of all trials identified from our searches in which participants had a wide spread of ages amongst their participants and a mean age over 50 (e.g., Davies 1986; Engle-Friedman 1992). The authors of the former trial have reported that lost data from their trial has recently been located and individual data from nine participants over 60 in the treatment arm and seven in the wait-list control group have been supplied and these data were then entered into the outcome (WASO, immediate post-treatment) for which they were appropriate. If the authors of the latter trial respond and it proves possible to isolate data for participants over the age of 60, such data will be entered. Assessment of risk of bias in included studies Assessment of methodological quality Quality assessment was made of all included studies, to consider the following questions: Was the assignment to treatment groups truly random? Was allocation adequately concealed? How complete was follow-up? How were the outcomes considered for people who withdrew? Were they included in the analysis? Were those assessing outcomes blind to the treatment allocation? 5

8 The Cochrane Collaboration Handbook criteria are based on the evidence of a strong relationship between the potential for bias in the results and allocation concealment and are defined as below: A. Low risk of bias (adequate allocation concealment) B. Moderate risk of bias (some doubt about the results) C. High risk of bias (inadequate allocation concealment) We contacted authors of all the included studies to acquire details both of the method of randomisation and that of allocation concealment. For the purpose of the analysis in this review, it was decided that trials would only be included if they met criteria A (adequate) or B (unclear) of the Handbook. Measures of treatment effect an entire trial (see Puder 1983) which has been used elsewhere to support claims for the efficacy of non-pharmacological treatment of late-life insomnia (Morin 1999b; Pallesen 1998). Assessment of heterogeneity Statistical heterogeneity was assessed using the chi-squared test for heterogeneity along with visual inspection of the graph. A significance level of less than 0.10 was interpreted as evidence of heterogeneity. For data where heterogeneity was found the reviewers looked for an explanation. When studies with heterogeneous results were found to be comparable, the statistical synthesis of the results was performed using a random effects model; where they were not comparable, no meta-analysis was undertaken. Continuous (including scale) data Rating scales: a range of instruments are available to measure sleep quality and the aspects of mental health which are associated with it (see for example Hoddes 1973; PSQI 1989; Johns 1991). For outcome instruments some minimum standards were required: (i) the psychometric properties of the instrument should have been described in a book or peer-reviewed journal; (ii) the instrument should either have been: (a) a self report, or (b) a report completed by an independent rater, bed-partner or relative/carer (not the therapist); and (iii) the instrument should be either a global assessment of an area of functioning or a specific feature of sleep quality, duration or timing. Combining mean treatment effects is straightforward when all measurements are comparable and on the same scale. The fixed effect estimate of the overall treatment effect can be computed as the weighted mean of the individual study effects,...where the weights are equal to the individual study specific variance estimates. On other occasions it is necessary to transform the mean effect from each study to a standardised value by dividing by the sample standard deviation within each study. This was not necessary in the current version of this review. Normal data: to avoid the pitfall of applying parametric tests to non-normally distributed data the following standards were applied to all data before inclusion: (i) standard deviations and means had to be reported in the paper or obtained from the authors; (ii) when a scale starts from a finite number (such as 0), the standard deviation had to be less than the mean (otherwise the mean was considered unlikely to be an appropriate measure of the centre of the distribution). Data which did not meet the second standard were not entered on RevMan software (which assumes a normal distribution). Dealing with missing data With the exception of the outcome of loss to follow up, if attrition rates were greater than 30%, these data were not used as they were considered to be too prone to bias. This caused the exclusion of Assessment of reporting biases Data from all identified and selected trials were entered into a funnel graph where appropriate (trial effect vs. variance) in an attempt to investigate the likelihood of overt publication bias. Data synthesis General In all cases the data were entered into RevMan in such a way that the area to the left of the line of no effect indicates a favourable outcome for the relevant behavioural intervention. In outcomes where a higher number indicated a benefit to the participant when compared to a lower number (e.g., total sleep duration as measured in minutes) data were entered as negative numbers.multiple treatment armswhere studies contained more than one eligible CBT therapy versus a control group, the Ns, means and standard deviations were pooled for use in the analysis. This was required in the case of McCurry 1998, where data for CBT delivered individually and by group therapy were combined, and in Lichstein 2001, where treatment arms included sleep restriction, relaxation and control, although only data from the relaxation arm were used in the outcome total sleep time. This decision was taken because sleep restriction is a paradoxical intervention where, during treatment, total sleep-time is limited. In another trial, data were combined from both intervention arms in a trial comparing stimulus control, imagery training and wait-list control (Morin 1988). R E S U L T S Description of studies 6

9 See: Characteristics of included studies; Characteristics of excluded studies; Characteristics of ongoing studies. Results of the search Following searches in the Cochrane Library, MEDLINE, EM- BASE, CINAHL and PsychINFO, 2677 references were located. Titles and abstracts were examined by both reviewers and 45 papers acquired. Six trials met the inclusion criteria (Davies 1986; Morin 1988; Morin 1993; McCurry 1998; Morin 1999a; Lichstein 2001) and data are being sought from three unpublished dissertations (Espie 1987; Epstein 1994; Creti 1998) as well as a trial in which the mix of ages necessitates that individual patient data be obtained (Engle-Friedman 1992). Included studies Specific details of each study are reported in the Characteristics of included studies table. The six studies included in this review were published between 1986 and 2001 (Davies 1986; Morin 1988; Morin 1993; McCurry 1998; Morin 1999a; Lichstein 2001). Five trials were conducted in the United States (Davies 1986; Morin 1993; McCurry 1998; Morin 1999a; Lichstein 2001) and one in Canada (Morin 1988). All participants were recruited from media advertisements, which suggests a highly-motivated (because selfselected) participant population. Included studies were relatively small, with a total of 282 participants, of whom 224 were used in the analysis in this review. Data from a further trial (Engle- Friedman 1992) and three unpublished dissertations (Espie 1987; Epstein 1994; Creti 1998) are still being sought. The stated purpose of each included study was to improve sleep in the older adult with sleep problems through behavioural or cognitive techniques. In two studies (McCurry 1998 and Morin 1988) trialists permitted participants to remain on hypnotics during the trials, but only if the participants had stabilised on their courses of medication for at least six weeks prior to the start of the trial. Four trials positively excluded those participants who had an inability or unwillingness to discontinue medication (Davies 1986; Morin 1993; Morin 1999a; Lichstein 2001). Severity of sleep problems at baseline was measured by participants sleep diaries (used for two weeks before the trials) in all cases (Davies 1986; Morin 1988; Morin 1993; Morin 1999a; McCurry 1998; Lichstein 2001). Two trials also used pre-test polysomnography (Morin 1993; Lichstein 2001;) to help establish baseline severity. Trialists from all studies claimed to have screened participants for dementia, depression, sleep apnoea and other potential causes of secondary insomnia, via interviews and administration of psychiatric tests, although one trial (McCurry 1998) permitted one participant in its control group to remain in the trial despite being prescribed a tricyclic anti-depressant. Three trials randomised participants in severity blocks (Morin 1988; Morin 1999a; Lichstein 2001). All trialists attempted to screen for comorbid conditions such as Alzheimer disease, psychoses and dementia, the effects of which have a known impact on sleep. Morin 1999a used the Mini Mental State Examination (MMSE), a structured clinical interview and screening tests for sleep apnoea and periodic limb movement problems (Folstein 1975). McCurry 1998 excluded potential participants if they reported or were suspected to be at risk of sleep apnoea and emphysema and were also screened for depression using the Clinical Center for Epidemiologic Studies Depression scale (CES-D) (Radloff 1977). Morin 1988 screened for depression using the Beck Depression Inventory (BDI) (Groth-Marnat 1990). Davies 1986 used the Minnesota Multiphasic Personality Inventory (MMPI 1943) and the Zung Self-Rating Depression Scale (Zung 1965), as well as personal interviews calculated to gain information on serious, painful medical conditions (e.g. arthritis), psychopathology, sleep apnoea, or nocturnal myoclonus (Davies 1986). Symptomatic evidence of sleep apnoea and restless leg syndrome or periodic limb movements was also sought. Lichstein 2001 screened for sleep apnoea (using polysomnography ), as well as for depression, anxiety and dementia, using the MMSE, the Cornell Medical Index, the State-Trait Anxiety inventory and the Geriatric Depression Scale (Folstein 1975; Spielberger 1983; Yesavage 1983; Brodman 1986). Morin 1993 screened for cognitive impairments using the MMSE (Folstein 1975), periodic limb movements (using polysomnography) and also reported having screened for depression, alcohol abuse and multiple medical problems, without describing the tests used. The oldest trial we included (Davies 1986) compared countercontrol behavioural therapy to wait-list control. The most recent trial included (Lichstein 2001) compared relaxation with sleep compression against a placebo described as quasi-densensitisation. As relaxation training and sleep compression are forms of CBT, data from both groups have been pooled within the analysis. McCurry used a combination of information and sleep hygiene together with sleep compression techniques (McCurry 1998). Morin and Azrin chose to compare stimulus control with imagery training against a no-treatment control group and as with Lichstein et al, data from both arms have been pooled for analysis in this review (Morin 1988; Lichstein 2001;). Morin 1993 used a combined treatment of sleep hygiene, sleep compression, stimulus control and cognitive therapy, and compared this with a wait-list control group. Morin 1999a compared a combination of sleep hygiene, sleep compression, stimulus control and cognitive therapy, versus temazepam, versus both the combined cognitive-behavioural treatments with temazepam, versus a placebo hypnotic (only data from the first and last arms described are used within this review, i.e. cognitive-behavioural treatments and placebo). Three studies used sleep latency as an outcome measure (Morin 1988; Morin 1993; McCurry 1998) although data from one study were unusable as no data were reported for the control group 7

10 (McCurry 1998). Four studies recorded data on the amount of time participants managed to remain asleep before first waking, known as wake after sleep onset (WASO) (Davies 1986; Morin 1993; Morin 1999a; Lichstein 2001); one on total wake-time (Morin 1993); four on sleep duration (Morin 1988; Morin 1993; Morin 1999a; Lichstein 2001); one on early morning wakening (Morin 1993); three on sleep efficiency (Morin 1993; Morin 1999a; Lichstein 2001), one on validated scales related to sleep, i.e. the Pittsburgh Sleep Quality Index (PSQI 1989) and the patients and clinicians versions of the Sleep Impairment Index (Morin 1993a; McCurry 1998; Morin 1999a) although data supplied for the latter outcome were inadequate for inclusion in the analyses for this review. Various non-validated scales developed by individual trialists were reported but not included in this review s analysis. Triallists used a five-point scale for previous night s sleep quality and a five-point scale for how I felt upon awakening (McCurry 1998), a five-point scale for patient s outcomes ratings, including severity, interference and noticeability of sleeping problems (Morin 1988; Morin 1993), a five-point scale for the views of significant others on participants sleep (Morin 1988; Morin 1993) and a five-point rating scale measuring perceived quality of sleep (Lichstein 2001). Morin 1988 included a treatment reliability measure which was constituted of anonymous ratings by treated participants of expectancies for success, treatment plausibility and confidence in recommending treatment to an insomniac friend; similarly, Lichstein 2001 assessed treatment credibility including dimensions such as reasonableness of treatment, opinion of the therapist and willingness [of participant] to recommend treatment to a friend. Follow-up data were sought after four weeks in one study (Davies 1986) (though data has not been presented in published or unpublished form); after three months in four of the six studies (McCurry 1998, Morin 1988; Morin 1993; Morin 1999a ), after one year in five studies (Davies 1986; McCurry 1998; Morin 1993; Morin 1999a; Lichstein 2001) and after two years in one study (Morin 1999a). In all but two studies (Morin 1999a and Lichstein 2001) the use of a wait-list control group rendered follow-up data unusable for meta-analyses, as by time of follow-up the control groups were also receiving a CBT intervention. Excluded studies Please see Characteristics of excluded studies for details. the published papers. Two responses have been received (McCurry 1998; Lichstein 2001). The latter trials both merit A s for randomisation and Bs for allocation concealment, due to the use of random numbers tables. The remainder of the studies are rated at B for both randomisation and allocation concealment until further information is supplied by the trialists. Blinding Only one trial reported that those who assessed outcomes were blind to treatment allocation (McCurry 1998). Incomplete outcome data Drop-out rates varied from 0-29% between studies (for details see Table of Included Studies). Follow-up times tended to group at similar intervals across studies and included post-treatment, in one study (Davies 1986) to three months in four of the five studies (Morin 1988; Morin 1993; McCurry 1998; Morin 1999a) to one year in four studies (McCurry 1998; Morin 1993; Morin 1999a; Lichstein 2001) and after two years in one study (Morin 1999a). As to how outcomes of people who withdrew from trials were considered by trialists, reports varied enormously. Lichstein 2001 and Morin 1999a considered the outcomes of participants who withdrew from the study between post-treatment and follow-up separately and no information is available concerning McCurry As Morin 1988 had no drop-out rate at all, the issue is not applicable. Other potential sources of bias Three groups of trialists attempted to distribute potential confounders equally by randomising in blocks according to severity of sleep problems (Morin 1988; Morin 1993; Lichstein 2001). Samples need to be of sufficient size for differences between groups to become statistically significant. Small samples can obscure treatment effects and result in an unequal distribution of confounders. Critical appraisal of study sizes was made. Sample sizes varied from 16 to 83 patients per trial that could be included in this analysis. There was no reason to think that there was a particular problem with publication bias even in view of the asymmetry of the funnel plot, which is likely to be due to the small numbers of studies in this field of research. Risk of bias in included studies Effects of interventions Allocation We contacted authors of all the six included studies to acquire details both of the method of randomisation and that of allocation concealment, because no such information was given within Sleep onset latency Sleep onset latency was used as an outcome measure in three studies included in the post-treatment analysis (Morin 1988 and Morin 1993; Lichstein 2001, total n = 135), and in one study at one 8

11 year + follow-up (Lichstein 2001, total n = 74 ). The overall effect estimate immediately at post-treatment was very mild as measured by sleep diaries (estimate of reduction of time to sleep onset = -3 minutes, 95% CI = -8.92, 2.92; Analysis 1.1) (Morin 1988; Morin 1993; Lichstein 2001;). When measured by PSG (Morin 1993, n = 24) the effect was also slight (estimate reduction of time to sleep onset, -4.4 minutes, 95% CI = , 4.55; Analysis 1.2). After a year or more, the effect is reported in one trial has improved slightly (estimate favours treatment group, who fall asleep 11.5 minutes faster than control, 95% CI = , 0.58; Analysis 1.1) (Lichstein 2001). This was not supported by PSG, where the improvement was only 2.5 minutes (95% CI = -3.24, 8.22; Analysis 1.2). Wake after sleep onset (WASO) Wake after sleep onset (WASO) as measured by sleep diaries was used as an outcome measure in four studies included in the post-treatment analysis (Davies 1986; Morin 1993; Morin 1999a; Lichstein 2001, total n = 159), in two studies at three-month follow-up (Morin 1993; Morin 1999a, total n = 50) and in two studies at one year + follow-up (Morin 1993; Lichstein 2001, total n = 98). The overall effect immediately post-treatment was modest, as measured by sleep diaries, with patients in the treatment group decreasing their time of being awake after the onset of sleep by 21.9 minutes (95% CI = , -6.38; Analysis 1.3). This effect was more marked when measured by PSG in the two trials which used this instrument (Morin 1993; Morin 1999a, total n = 59) where participants showed a decrease in WASO of 24.4 minutes (95% CI = , -7.57; Analysis 1.4). At three-month follow-up, the improvement is good when measured by sleep diaries, with the treatment group now 33 minutes better off than control (95% CI = , -8.35; Analysis 1.3) (Morin 1999a). After a year or more (in a different trial) the effect appears more modest, with WASO decreased by 13 minutes the treated group (95% CI = -28.8, 3.42; Analysis 1.3) (Lichstein 2001). Lichstein 2001 (n = 74) also used PSG twelve months after treatment and in this trial the effect appears to support the diary results (decrease of WASO by 10.1 minutes, [95% CI = , 14.17; Analysis 1.4). Total wake-time (in minutes) Total wake-time was used as an outcome measure in only one (small) study (Morin 1993, n = 24). Total wake-time was measured at all of the three outcome periods (immediate post-treatment, three-month, and one year+). The overall effect immediately posttreatment was encouraging as measured by sleep diaries, with participants in the control group spending an hour less awake during the night than control (decrease of minutes, (95% CI = , ; Analysis 1.5), but the very wide confidence intervals mean this should be interpreted cautiously. When measured by PSG, this effect appeared weaker, but still significant (decrease of 38 minutes, (95% CI = , -7.76; Analysis 1.6). Sleep duration (in minutes) Sleep duration, or total sleep time was used as an outcome measure in four studies included in the post-treatment analysis (Morin 1988; Morin 1993; Morin 1999a; Lichstein 2001, n = 143), in one study at three-month follow-up (Morin 1999a ) and in one study at one year + follow-up (Lichstein 2001, n = 54). For this outcome, only data from one of the two treatment arms involved in Lichstein 2001 has been used. This is because relaxation techniques (in the first arm) were used to increase total sleep time, while the second arm (sleep compression) involves a paradoxical technique in which participants time in bed is limited. Thus a lower N is shown for Lichstein 2001 for all data measurement points. The overall effect immediately post-treatment was mild when measured by sleep diaries, with participants in the treatment group having 14.6 minutes more sleep a night than control (95% CI = , -7.01; Analysis 1.7). When measured by PSG, this effect was not supported with overall effect reversing to favour control (participants in the control group slept 19 minutes more a night than treatment group, 95% CI = , 68.5; Analysis 1.7). It should be noted that PSG was conducted at post-treatment in only two of the four studies which measured sleep duration. At three-month follow-up, participants in the treatment groups appear 14.8 minutes worse off (sleeping less) than control group (95% CI = , -68.5; Analysis 1.7). After a year or more, the effect appears very positive according to diary reports (treatment group reporting an increase in sleep per night of 32 minutes, 95% CI = , -8.13; Analysis 1.7); however this is contradicted by the PSG measurement used in one trial (Lichstein 2001, n = 50) and in this trial the effect had reported that the control group was sleeping almost 7 minutes more a night than the treatment group (95% CI = , 47.37; Analysis 1.8). Early morning wakening Early morning wakening was used as an outcome measure in only one study (Morin 1993, n = 24). Early morning wakening was measured at all of the three outcome periods. The overall effect immediately post-treatment was modest as measured by sleep diaries, with the treatment group waking 17 minutes later than the control group ( 95% CI = , 0.46; Analysis 1.9). When measured by PSG, this effect was supported (overall effect reporting an increase of 14.9 minutes (95% CI = , 2.55; Analysis 1.10). Sleep efficiency Sleep efficiency was used as an outcome measure in three studies included in the post-treatment analysis (Morin 1993; Morin 9

12 1999a; Lichstein 2001, n = 143), in one study at three-month follow-up (Morin 1999a) and in one study at one year + follow-up (Lichstein 2001). The overall effect immediately post-treatment was modest (albeit with wide confidence intervals) as measured by sleep diaries, with participants showing an improvement of 7.5% (95% CI = , 0.47; Analysis 1.11). When measured by PSG, this effect was almost identical, and showed narrower confidence intervals (6.25% improvement 95% CI = 10.18, 2.31; Analysis 1.12) (Morin 1993; Morin 1999a). At three-month follow-up, the effect had improved slightly as measured by sleep diaries in the one included trial (Morin 1999a) (9.6%, 95% CI = 18.21, 0.87; Analysis 1.11 ). After a year or more, the effect was reduced further (4.4% improvement, 95% CI = 9.85, -1.13; Analysis 1.13) (Lichstein 2001). Lichstein 2001 s use of PSG did not support even this modest finding and the treatment group appeared 2.4% worse, 95% CI = -3.36, -8.14; Analysis 1.12). Standardised scales Only one study (McCurry 1998, n = 36) used the Pittsburgh Sleep Quality Index (PSQI 1989) at post-treatment and at threemonth follow-up. This is a subjective measure of sleep quality, ranging from 0 to 21. Any score above five is considered indicative of sleep disturbance. Initial effect of CBT in this trial was good, with participants scoring 7.8 in the treatment group as compared with 10.6 in the control, overall effect -2.80, (95% CI = -5.44, ; Analysis 1.13). At three month follow-up, the effect of CBT had improved further, although all participants were still above the threshold of pathological sleep disturbance (treatment group = 6.20, control group = 10.20). The overall effect at follow-up was (95% CI = -6.62, -1.38; Analysis 1.13). D I S C U S S I O N Summary of main results Overall, the results of this review suggest that cognitive-behavioural treatments for sleep problems in people aged 60 and over are mildly effective for some aspects of sleep according to the patient diaries in the short term, but that the effect of these treatments is not always durable. However, in general, beneficial effects are not as great as for these treatments when used with younger adults (see Pallesen 1998 for a review making this comparison) which is consistent with other research into the efficacy of CBT with older adults. The objective reports of sleep by polysomnography do not always agree with the patient reports, which was particularly notable in the data for sleep duration. It should be recognised that there is a large variation in the interpretation of many sleep variables (McGhie 1962) and so, both objective and subjective measures have been included in this review. The results of this review of the effects of cognitive-behavioural treatments on sleep are mixed. Total sleep duration appears to show a modest improvement at post-treatment, which declines with time. Night waking (WASO) shows clinically important improvements, although these appear to diminish somewhat over time. Sleep onset latency (SOL, or settling ) seems to change very little as a result of the interventions investigated, while early morning waking reduces slightly. Results for sleep efficiency (the percentage of time in bed in which participants are actually asleep) suggests a modest gain initially, which erodes over time. Participant reports of total sleep duration are not supported by polysomnography (PSG); in contrast, the substantial improvements reported by participants for night waking (WASO) are objectively supported. Neither sleep onset latency, early morning waking nor sleep efficiency have objective confirmation of participant reports. Results from the one trial which used the Pittsburgh Index of Sleep Quality suggested that all participants in the trial remained at pathological levels of sleep disturbance throughout the trial (a finding which may at least partially be explained by the fact that all were full-time partners or carers of people with Alzheimers disease). Nevertheless, whilst participants from the treatment group improved their scores from baseline to follow-up (10.8 down to 6.20), the control group over the same time period improved only from 11.9 to Overall completeness and applicability of evidence While some of the improvements reported here appear modest, they might well be considered clinically useful and patient reports indicate that they are both worthwhile and valued, particularly when compared with drug-induced sleep (Giblin 1983). Some studies have reported considerable reductions in hypnotic usage following CBT for sleep problems and in a population whose usage of drugs is generally high this may be particularly important. A prospective study examining 5-year mortality among hypnotic drug users and respondents with subjective insomnia identified in a longitudinal study of health, activity, and lifestyle (Nottingham Longitudinal Study of Activity and Ageing) which involved 1042 survey respondents, aged over 65 years, concluded that the mortality rate of participants was significantly greater among those taking some form of medication for sleep than for those not taking sleep medication (Rumble 1992). Generally, research supports the idea that when informed of options for the treatment of chronic insomnia, CBT treatments are welcomed (see Vincent 2001, wherein the concept of treatment preference and patient satisfaction is investigated). Researchers reported that cognitive-behavioural therapy was significantly preferred over pharmacological therapy at pre-treatment... The study further reported that, amongst the 10

13 treatment techniques used, participants least liked sleep restriction and most liked sleep hygiene. Use of the model for PPA (prospective preference assessment) described by Halpern might also enhance participant satisfaction and adherence to treatment (Halpern 2002). Homogeneity Specific cognitive-behavioural interventions and the mode and quality of therapist delivery vary somewhat. It may be that these differences can explain some of the heterogeneity in these results. Moreover, research has suggested that the efficacy of CBT declines with age (Morin 1994; McCurry 1998). Morin 1994 also reported that sleep restriction (reducing time in bed to as near as possible to time asleep) resulted in a worsening of sleep duration at least in the short-term, or alternatively that sleep efficiency had improved (the ratio of time asleep to time in bed) at the expense of sleep duration. It should also be considered that there may be a developmental perspective to sleep in this population as the question of whether older adults need less sleep or cannot get more sleep has not yet been answered (Bliwise 1993). Limitations There is considerable overlap between many cognitive behavioural interventions such as aspects of sleep hygiene which may restrict sleep. It is therefore not possible to determine which parts of these therapies are effective. Studies did not refer to standard AASM criteria for diagnosing patients sleep problems which may limit the internal reliability of this meta-analysis (AASM 1990). Agreements and disagreements with other studies or reviews All treatment gains reported in this review are modest when compared with those reported by others (Morin 1994; Pallesen 1998). This can be explained partly by the rigorous inclusion criteria applied within this review; for example, our requirements that 80% of the sample be over the age of 60; that dropout be less <30%; that clear measures of the sleep problem at baseline be present and that screening for psychiatric co-morbidities by standardised measures take place. Excluded studies almost invariably reported more significant treatment gains (see Characteristics of excluded studies ). The lack of effect at follow-up in this review (even for those sleep variables for which CBT did appear modestly effective) also contrasts with the findings of other reviews, possibly for the same reasons. It may be that using written materials or videotapes would help patients maintain or enhance the changes they had achieved as shown in adults (Marrs 1995) in children (Montgomery 2001) and more specifically in adult insomnia (Mimeault see below, and Giblin 1983). Research into the effectiveness of booster sessions for maintaining the benefits of a cognitive-behavioural programme for insomnia for adults aged reported no difference at follow-up between groups which received post-treatment booster sessions and those which did not (Cook 1986). However, it is highly likely that Cook s timing (a mere two months after cessation of the CBT treatment programme) as well as the fact that both groups were still showing significant improvement in their sleep problems, suggests that booster sessions were not yet required. Further testing is needed to show whether or not booster sessions at six months or a year from initial treatment (when, as this review shows, benefit has markedly declined) might maintain earlier improvements. A U T H O R S C O N C L U S I O N S Implications for practice The data suggest a mild effect of CBT for different aspects of the sleep problems in older adults. The small number of studies meeting the stringent criteria for inclusion in this review may also influence the apparent efficacy. It is notable that a large number of excluded studies reported positive results and that the main reason for their exclusion was that their mean age was younger than 60. While it may well be the case that the treatment efficacy of cognitive-behavioural interventions seems to decline with the age of participants, they may still be worthy of consideration by clinicians, especially in a population for whom alternative treatments may be limited. As some studies have reported that older adults are more likely to be taking a wide range of medications for other health problems and that many hypnotics have side-effects, it is likely that reducing additional medication for sleep problems would be a positive benefit to the health of older adults. Implications for research Cognitive-behavioural therapy for insomnia has several related dimensions, as has been discussed in this review. It might be the case that some aspects of it are of greater importance than others and that research into the active ingredient(s) would be useful in refining the treatment. In addition, cognitive-behavioural treatments involve a considerable commitment on the parts of both the patient and therapist since the apparent paradoxical intervention (especially in sleep restriction) can make compliance difficult to achieve. Research is needed to establish the likely predictors of success with these treatments. In view of the high prevalence of sleep problems and the common co-morbidity of them with other disorders, notably anxiety and depression, particular attention must be paid to these conditions as potential confounders in the older adult population. Where such interventions are reported to have an effect, durability must be considered. To increase durability of effect the provision of either top-up sessions of CBT training at intervals of

14 months following initial treatment and/or the provision of written materials and/or videocassettes for home use following treatment should be tested. Cost-benefit analysis as regards these treatments versus pharmacological treatments would be a particularly useful addition to this area of study. Finally, the role of cognitive-behavioural therapy in preventative education for sleep disorders may also be worthy of investigation. Given that the efficacy of CBT is known to reduce with age, it may be that sound prevention programmes delivered in the middle years might benefit adults as they age and become more prone to sleep problems. A C K N O W L E D G E M E N T S The authors would like to thank Jo Abbott, Margaret Burke, Esther Coren, Annemarie Courtiour, Lindsay Dow, Julian Higgins, Daniel Kripke, Liz Lloyd, Stuart Logan, Geraldine Macdonald, Gregory Stores, Katrina Williams and Philp Wilkinson for their helpful comments and advice on this review. Thanks also to Brant Riedel, Susan McCurry and Ruth Davies-Sulser for supplying data. R E F E R E N C E S References to studies included in this review Davies 1986 {published data only} Davies R, Lacks P, Storandt M, Bertelson AD. Countercontrol treatment of sleep-maintenance insomnia in relation to age. Psychology and Aging 1986;1(3): Lichstein 2001 {published data only} Lichstein KL, Riedel BW, Wilson NM, Lester KW, Aguillard RN. Relaxation and sleep compression for late-life insomnia: a placebo-controlled trial. Journal of Consulting and Clinical Psychology 2001;69(2): McCurry 1998 {published data only} McCurry SM, Logsdon RG, Vitiello MV, Teri L. Successful behavioral treatment for reported sleep problems in elderly caregivers of dementia patients: a controlled study. Journal of Gerontology 1998;53B(2): Morin 1988 {published data only} Morin CM, Azrin NH. Behavioral and cognitive treatments of geriatric insomnia. Journal of Consulting and Clinical Psychology 1988;56(5): Morin 1993 {published data only} Morin C, Kowatch RA, Barry T, Walton E. Cognitivebehavior therapy for late-life insomnia. Journal of Consulting Clinical Psychology 1993;61(1): Morin 1999a {published data only} Morin CM, Colecchi C, Stone J, Sood R, Brink D. Behavioral and pharmacological therapies for late-life insomnia: a randomized controlled trial. JAMA 1999;281 (11): References to studies excluded from this review Alperson 1979 {published data only} Alperson J, Biglan A. Self-administered treatment of sleep onset insomnia and the importance of age. Behavior Therapy 1979;10: Anderson 1988 {published data only} Anderson MW, Zendell SM, Rosa DP, Rubinstein ML, Herera CO, Simons O, Caruso L, Spielman AJ. Comparison of sleep restriction therapy and stimulus control in older insomniacs: an update. Sleep Research 1988;17:141. Friedman 1991 {published data only} Friedman L, Bliwise DL, Yesavage JA, Salom SR. A preliminary study comparing sleep restriction and relaxation treatments for insomnia in older adults. Journal of Gerontology 1991;46(1):1 8. Giblin 1983 {published data only} Giblin MJ, Clift AD. Sleep without drugs. Journal of the Royal College of General Practitioners 1983;33(255): Hoch 2001 {published data only} Hoch CC, Reynolds CF, Buysse DJ, Monk TH, Nowell P, Begley AE, Hall F, Dew MA. Protecting sleep quality in later life: A pilot study of bed restriction and sleep hygiene. Journals of Gerontology: Series B: Psychological Sciences and Social Sciences 2001;56B(1):52 P59. Puder 1983 {published data only} Puder R, Lacks P, Bertelson AD, Storandt M. Short-term stimulus control treatment of insomnia in older adults. Behavior Therapy 1983;14: Riedel 1995 {published data only} Riedel BW, Lichstein KL, Dwyer WO. Sleep compression and sleep education for older insomniacs: self-help versus therapist guidance. Psychology of Aging 1995;10(1): Schnelle 1999 {published data only} Schnelle JF, Alessi CA, Nahla R, Al-Samurrai NR, Fricker RD, Ouslander JG. The nursing home at night: effects of an intervention on noise, light, and sleep. Journal of the American Geriatric Society 1999;47(4): References to studies awaiting assessment Creti 1998 {unpublished data only} Creti L. An evaluation of a new cognitive-behavioral technique for the treatment of insomnia in older adults [dissertation]. McGill University, Engle-Friedman 1992 {published data only} Engle-Friedman M, Bootzin RR, Hazlewood L, Tsao C. An evaluation of behavioral treatments for insomnia in 12

15 the older adult. Journal of Clinical Psychology 1992;48(1): Epstein 1994 {unpublished data only} Epstein DR. A behavioral intervention to enhance the sleepwake patterns of older adults with insomnia [dissertation]. University of Arizona, Espie 1987 {unpublished data only} Espie CA. A controlled comparative investigation of psychological treatments for chronic sleep-onset insomnia [dissertation]. Glasgow: University of Glasgow, References to ongoing studies Morgan 2002 {unpublished data only} Psychological treatment in the management of hypnotic dependent chronic insomnia in primary care. Ongoing study Unknown. Additional references AASM 1990 American Academy of Sleep Medicine (AASM). International Classification of Sleep Disorders: Diagnostic and Coding Manual. Rochester MN, USA: American Academy of Sleep Medicine, Ancoli-Israel 1997 Ancoli-Israel S, Poceta JS, Stepnowsky C, Martin J, Gehrman P. Identification and treatment of sleep problems in the elderly. Sleep Medicine Reviews 1997;1(1):3 17. Bachman 1992 Bachman DL. Sleep disorders with aging: Evaluation and treatment. Geriatrics 1992;47(9): [MEDLINE: ] Baillargeon 1996 Baillargeon L, Demers M, Gregoire JP, Pepin M. [Study on insomnia treatment by family physicians] [French] [Enquete sur le traitement de l insomnie par les omnipraticiens]. Canadian Family Physician 1996;42: [MEDLINE: ] Baum et al 1986 Baum C, Kennedy DL, Knapp DE, Faich GE. Drug utilization in the U.S : seventh annual review. Rockville, Maryland USA: Food and Drug Administration, Bliwise 1993 Bliwise DL. Sleep in normal aging and dementia. Sleep 1993;16(1): [MEDLINE: ] Bootzin 1991 Bootzin RR, Epstein D, Wood JM. Stimulus control instructions. In: Hauri P editor(s). Case Studies in Insomnia. New York, NY: Plenum Medical, 1991: Borkovec 1978 Borkovec TD, Hennings BL. The role of physiological attention-focusing in the relaxation treatment of sleep disturbance, general tension, and specific stress reaction. Behaviour Research and Therapy 1978;16(1):7 19. [MEDLINE: ] Borkovec 1982 Borkovec TD. Insomnia. Journal of Consulting and Clinical Psychology 1982;50(6): [MEDLINE: ] Brabbins 1993 Brabbins CJ, Dewey ME, Copeland JR, Davidson IA, McWilliam C, Sunders P, Sharma VK, Sullivan C. Insomnia in the elderly: prevalence, gender differences and relationships with morbidity and mortality. International Journal of Geriatric Psychiatry 1993;8(6): Brodman 1986 Brodman K, Erdmann AJ Jr, Wolff HG, Miskovitz PF. Manual: Cornell Medical Index. New York: Cornell University Medical College, Campbell 1993 Campbell AJ, Busby WJ, Horwath CC. Factors associated with constipation in a community based sample of people aged 70 years and over. Journal of Epidemiology and Community Health 1993;47(1): Cook 1986 Cook MA. Effectiveness of booster sessions inthe treatment of sleep maintenance insomnia [PhD dissertation]. St. Louis, Missouri: Washington University, Missouri, DPOA 2000 Wilkinson P (Consultant Psychiatrist at the Dept of the Psychiatry of Old Age, Oxfordshire Mental Healthcare Trust, UK). Personal communication Foley 1995 Foley DJ, Monjan AA, Brown SL, Simonsick EM, Wallace RB, Blazer DG. Sleep complaints among elderly persons: an epidemiologic study of three communities. Sleep 1995; 18(6): [MEDLINE: ] Folstein 1975 Folstein MF, Folstein, SE, McHugh PR. Mini-Mental State: A practical method for grading the state of patients for the clinician. Journal of Psychiatric Research 1975;12: Ford 1989 Ford DE, Kamerow DB. Epidemiologic study of sleep disturbances and psychiatric disorders. JAMA 1989;262: [MEDLINE: ] Groth-Marnat 1990 Groth-Marnat G. The handbook of psychological assessment. 2nd Edition. New York: John Wiley & Sons, Halpern 2002 Halpern SD. Prospective preference assessment: a method to enhance the ethics and efficiency of randomized controlled trials. Controlled Clinical Trials 2002;23(3): Hauri 1991 Hauri PJ (Editor). Case studies in insomnia. New York: Plenum Medical Book Company, [: ISBN ] Hauri 1997 Hauri PJ. Can we mix behavioral therapy with hypnotics when treating insomniacs?. Sleep 1997;20(12): [MEDLINE: ] 13

16 Hoddes 1973 Hoddes E, Zarconi V, Smythe H, Phillips R, Dement WC. Quantification of sleepiness: a new approach. Psychophysiology 1973;10: Hohagen 1994 Hohagen F, Kaeppler C, Schramm E, Rink K, et al.prevalence of insomnia in elderly general practice attenders and the current treatment modalities. Acta Psychiatrica Scandinavica 1994;90(2): Iler 1997 Iler JA. Efficacy of a nonpharmacological intervention for insomnia: An empirical investigation [PhD dissertation]. Pasadena, California: Fuller Theological Seminary, Jacobson 1974 Jacobson E. Progressive Relaxation. Chicago: University of Chicago Press, Johns 1991 Johns MW. A new method of measuring daytime sleepiness: the Epworth sleepiness scale. Sleep 1991;14(6): Kripke 1983 Kripke DF. Why we need a tax on sleeping pills. Southern Medical Journal 1983;76(5): [MEDLINE: ] Kripke 2000 Kripke DF. Chronic hypnotic use: deadly risks, doubtful benefit. Sleep Medicine Reviews 2000;4(1):5 20. Kripke 2001 Kripke DF. Personal communication ( ) November Kupfer 1997 Kupfer DJ, Reynolds CF 3rd. Management of insomnia. New England Journal of Medicine 1997;336(5): [MEDLINE: ] Lacks 1989 Lacks P, Powlishta K. Improvement following behavioral treatment for insomnia: clinical significance, long-term maintenance and predictors of outcome. Behavior Therapy 1989;20: Lichstein 1993 Lichstein KL, Johnson RS. Relaxation for insomnia and hypnotic medication use in older women. Psychology and Ageing 1993;8(1): Marrs 1995 Marrs RW. A meta-analysis of bibliotherapy studies. American Journal of Community Psychology 1995;23(6): McClusky 1991 McClusky HY, Milby JB, Switzer PK, Williams V, Wooten V. Efficacy of behavioral versus triazolam treatment in persistent sleep-onset insomnia. American Journal of Psychiatry. 1991;148(1): [MEDLINE: ] McGhie 1962 McGhie A, Russell S. The subjective assessment of normal sleep patterns. Journal of Mental Science 1962;108: Mellinger 1995 Mellinger GD, Balter MB, Uhlenhuth EH. Insomnia and its treatment. Archives of General Psychiatry 1995;42: Meyer 1998 Meyer TJ. Evaluation and management of insomnia. Hospital Practice (Office Edition) 1998;33(12):75-8; Milby 1993 Milby JB, Williams V, Hall JN, Khuder S, McGill T, Wooten V. Effectiveness of combined triazolam-behavioral therapy for primary insomnia. American Journal of Psychiatry 1993;150(8): [MEDLINE: ] Miles 1980 Miles LE, Dement WC. Sleep and Aging. New York: Raven, Mimeault 1999 Mimeault V, Morin CM. Self-help treatment for insomnia: bibliotherapy with and without professional guidance. Journal of Consulting and Clinical Psychology 1999;67(4): MMPI 1943 Hathaway SR, McKinley JC. Minnesota Multiphasic Personality Inventory. Minneapolis: University of Minnesota Press, Montgomery 2001 Montgomery P, Bjornstad G, Dennis J. Media-based behavioural treatments for behavioural disorders in children. Cochrane Database of Systematic Reviews 2006, Issue Issue 1. [DOI: / CD pub3; : ] Montgomery 2002a Montgomery P, Dennis J. Bright light therapy for sleep problems in adults aged 60+. Cochrane Database of Systematic Reviews 2002, Issue 2 Art. No.: CD [DOI: / CD003403] Montgomery 2002b Montgomery P, Dennis J. Physical exercise for sleep problems in adults aged 60+. Cochrane Database of Systematic Reviews 2002, Issue 4 Art. No.: CD [DOI: / CD003404] Morgan 1988 Morgan K, Dallosso H, Ebrahim S, Arie T, Fentem PH. Prevalence, frequency, and duration of hypnotic drug use among elderly living at home. BMJ 1988;26: Morin 1989 Morin CM, Gramling SE. Sleep patterns and aging: comparison of older adults with and without insomnia complaints. Psychology and Aging 1989;4(3): [MEDLINE: ] Morin 1993a Morin CM. Insomnia: psychological assessment and management. New York, NY USA: Guilford Press, Morin 1993b Morin CM. Dysfunctional beliefs and attitudes about sleep among older adults with and without insomnia complaints. Psychology and Aging 1993;8(3): [MEDLINE: ] 14

17 Morin 1994 Morin CM, Culbert JP, Schwartz SM. Nonpharmacological interventions for insomnia: a meta-analysis of treatment efficacy. American Journal of Psychiatry 1994;151(8): [MEDLINE: ] Morin 1999b Morin CM, Hauri PJ, Espie CA, Spielman AJ, Buysse DJ, Bootzin RR. Nonpharmacologic treatment of chronic insomnia. Sleep 1999;22(8): Morin 1999c Morin CM, Mimeault V, Gagne A. Nonpharmacological treatment of late-life insomnia. Journal of Psychosomatic Research 1999;46(2): Murtagh 1995 Murtagh DRR, Greenwood KM. Identifying effective psychological treatments for insomnia: a meta-analysis. Journal of Consulting and Clinical Psychology 1995;63(1): [MEDLINE: ] NCSDR 1993 National Commission on Sleep Disorders Research (WC Dement, Chairman, National Commission). Wake Up America: A National Sleep Alert. Report of the National Commission on Sleep Disorders Research. Washington DC: U.S. Department of Health and Human Services, NIH 1983 National Institute of Health. Drugs and Insomnia: The Use of Medications to Promote Sleep. NIH Consensus Statement 1983 (Nov 15 17); Vol. 4, issue 10. NIH 1990 National Institute of Health. The Treatment of Sleep Disorders of Older People. NIH Consensus Statement 1990 (Mar 26 28); Vol. 8, issue 3:1 22. Nowell 1997 Nowell PD, Mazumdar S, Buysse DJ, Dew MA, Reynolds CF 3rd, Kupfer DJ. Benzodiazepines and zolpidem for chronic insomnia: a meta-analysis of treatment efficacy. JAMA 1997;278(24): [MEDLINE: ] Pallesen 1998 Pallesen S, Nordhus IH, Kvale G. Non-pharmacological interventions for insomnia in older adults: a meta-analysis of treatment efficacy. Psychotherapy: theory, research, practice, training 1998;35(4): Phillips 2001 Phillips B, Ancoli-Israel S. Sleep disorders in the elderly. Sleep Medicine 2001;2: Pollak 1990 Pollak CP, Perlick D, Linsner JP, Wenston J, Hsieh F. Sleep problems in the community elderly as predictors of death and nursing home placement. Journal of Community Health 1990;15(2): [MEDLINE: ] Pollak 1991 Pollak CP, Perlick D. Sleep problems and institutionalization of the elderly. Journal of Geriatric Psychiatry and Neurology 1991;4(4): [MEDLINE: ] Prinz 1990 Prinz PN, Vitiello MV, Raskind MA, Thorpy MJ. Geriatrics: sleep disorders and aging. New England Journal of Medicine 1990;323: Prinz 1995 Prinz PN. Sleep and sleep disorders in older adults. Journal of Clinical Neurophysiology 1995;12(2): [MEDLINE: ] PSQI 1989 Buysse DJ, Reynolds III CF, Monk TH, Berman SB, Kupfer DJ. The Pittsburgh Sleep Quality Index: a new instrument for psychiatric practice and research. Psychiatry Research 1989;28: Radloff 1977 Radloff LS. The CES-D scale: A self-report depression scale for research in the general population. Applied Psychological Measurement 1977;1: Reynolds 1991 Reynolds CF, Kupfer DJ, Buysse DJ, Coble P, Yeager A. Subtyping DSM III-R primary insomnia: a literature review by the DSM IV work group on sleep disorders. American Journal of Psychiatry 1991;148: Rumble 1992 Rumble R, Morgan K. Hypnotics, sleep, and mortality in elderly people. Journal of the American Geriatrics Society 1992;40(8): Schulz 1959 Schulz JH, Luther W. Autogenic training: psychophysiological approach in psychotherapy. New York: Grune and Stratton, Simon 1997 Simon G, Von Korff M. Prevalence, burden and treatment of insomnia in primary care. American Journal of Psychiatry 1997;154: Spielberger 1983 Spielberger CD, Gorsuch RL, Lushene R, Vagg PR, Jacobs GA. State-Trait Anxiety Inventory (Form Y). Palo Alto, California: Consulting Psychologists Press, Spielman 1987 Spielman AJ, Saskin P, Thorpy MJ. Treatment of chronic insomnia by restriction of time in bed. Sleep 1987;10: Stoller 1994 Stoller MK. The economic effects of insomnia. Clinical Therapeutics 1994;16: Stores 1998 Stores G, Crawford C. Medical student education in sleep and its disorders. Journal of the Royal College of Physicians of London 1998;32(2): [MEDLINE: ] Vincent 2001 Vincent N, Lionberg C. Treatment preference and patient satisfaction in chronic insomnia. Sleep 2001;24(4):

18 Wettstein 1992 Wettstein A. Falls as a side-effect of drugs [Sturz als Nebenwirkung von Medikamentan]. Schweizerische Rundschau fur Medizin Praxis 1992;81(46): Wiggs 1996 Wiggs L, Stores R. Sleep education in undergraduate psychology degree courses in the UK. Psychology Teaching Review 1996;5: Woolfolk 1983 Woolfolk RL, McNulty TF. Relaxation treatment for insomnia: a component analysis. Journal of Consulting and Clinical Psychology 1983;51(4): [MEDLINE: ] Yesavage 1983 Yesavage JA, Brink TL, Rose TL, Lum O, Huang V, Adey M, Leirer VO. Development and validation of a Geriatric Depression Screening Scale: A preliminary report. Journal of Psychiatric Research 1982/1983;17: Zung 1965 Zung WWK. A self-rating depression scale. Archives of General Psychiatry 1965;12: Zwart 1979 Zwart CA, Lisman SA. An analysis of stimulus control treatment of sleep-onset insomnia. Journal of Consulting and Clinical Psychology 1979;47: Indicates the major publication for the study 16

19 C H A R A C T E R I S T I C S O F S T U D I E S Characteristics of included studies [ordered by study ID] Davies 1986 Methods Participants Interventions Outcomes Notes Randomised controlled trial 34 participants (mean age of [SD = 10.98]). Only data on the over-60s used in this review (9 in treatment arm; 7 from control arm). Participants were judged to be insomniac following two weeks of self-report data using sleep diaries, spouse reports (in some cases) and a history of sleep maintenance insomnia of at least 6 months duration. Sleep maintenance insomnia was defined as WASO of at least 30. min. per night at least one night per week Treatment 1: countercountrol therapy; Treatment 2: wait-list control WASO, average number of awakenings per night, average number of awakenings of 10 minutes or more, were the primary outcomes and the only ones for which data were reported. Sleep latency and posttreatment use of hypnotics were also investigated, but no data reported Initial intake for randomisation was 43; 34 finished trial; 16 remained at one-year follow-up; however, no data on the over-60s alone were available Risk of bias Item Authors judgement Description Allocation concealment? Unclear B - Unclear Lichstein 2001 Methods Participants Interventions Outcomes Notes Randomised controlled trial (participants randomised in 8 blocks following noting of gender and severity of condition) 83 participants (mean age of 68.1 [SD= 8.3, 67.9 [SD = 6.8] and 68.0 [SD = 5.9] for the three arms of the trial). Participants were judged to be insomniac following two weeks of self-report data using sleep diaries Treatment 1: relaxation; Treatment 2: sleep compression; Treatment 3: placebo, defined as quasi-desensitisation Sleep latency, no. of awakenings, WASO, total sleep time, sleep efficiency Initial intake for randomisation was 89; 83 finished trial; 74 remained at follow-up Risk of bias 17

20 Lichstein 2001 (Continued) Item Authors judgement Description Allocation concealment? Unclear B - Unclear McCurry 1998 Methods Participants Interventions Outcomes Notes Randomised controlled trial 36 participants (mean age of 68.7 [SD = 10.6]) who were caregivers of those with dementia and who had endorsed at least one sleep problem on a sleep problems screening questionnaire Treatment 1: Group behavioural treatment (including sleep hygiene, stimulus control, sleep compression, relaxation and education) Treatment 2: Individual behaviour treatment (including sleep hygiene, stimulus control, sleep compression, relaxation and education) Treatment 3: wait list control Sleep participant diaries (including onset and waking times) ; Pittsburgh Sleep Quality Index Initial intake at randomisation was 36; 35 finished trial; 29 completed 3-month follow-up. Difficulties in this study include varying pharmacological treatments some participants (6/36) were on throughout study. No data given on control group for any outcome except the PSQI Risk of bias Item Authors judgement Description Allocation concealment? Yes A - Adequate Morin 1988 Methods Participants Interventions Outcomes Notes Randomised controlled trial (participants randomised within severity blocks) 27 participants, non-institutionalised (mean age 67.4 [SD = 5.6]) with sleep-maintenance insomnia (average duration 19 years) Treatment 1: Stimulus control Treatment 2: Imagery training Treatment 3: Wait-list control Sleep diaries (including onset and waking times), hand-held switch-activated clock, reports from bed partners Initial intake for randomisation was 27; 19 finished trial Risk of bias 18