Psychological And Behavioral Treatment Of Insomnia:Update Of The Recent Evidence (1998-2004)

There is also a need to disseminate more effectively the available evidence in support of psychological and behavioral interventions to health-care practitioners working on the front line.

Background: Recognition that psychological and behavioral factors play an important role in insomnia has led to increased interest in therapies targeting these factors. A review paper published in 1999 summarized the evidence regarding the efficacy of psychological and behavioral treatments for persistent insomnia. The present review provides an update of the evidence published since the original paper. As with the original paper, this review was conducted by a task force commissioned by the American Academy of Sleep Medicine in order to update its practice parameters on psychological and behavioral therapies for insomnia. Methods: A systematic review was conducted on 37 treatment studies (N = 2246 subjects/patients) published between 1998 and 2004 inclusively and identified through PsycInfo and Medline searches. Each study was systematically reviewed with a standard coding sheet and the following information was extracted: Study design, sample (number of participants, age, gender), diagnosis, type of treatments and controls, primary and secondary outcome measures, and main findings. Criteria for inclusion of a study were as follows: (a) the main sleep diagnosis was insomnia (primary or comorbid), (b) at least 1 treatment condition was psychological or behavioral in content, (c) the study design was a randomized controlled trial, a nonrandomized group design, a clinical case series or a single subject experimental design with a minimum of 10 subjects, and (d) the study included at least 1 of the following as dependent variables: sleep onset latency, number and/or duration of awakenings, total sleep time, sleep efficiency, or sleep quality. Results: Psychological and behavioral therapies produced reliable changes in several sleep parameters of individuals with either primary insomnia or insomnia associated with medical and psychiatric disorders. Nine studies documented the benefits of insomnia treatment in older adults or for facilitating discontinuation of medication among chronic hypnotic users. Sleep improvements achieved with treatment were well sustained over time; however, with the exception of reduced psychological symptoms/ distress, there was limited evidence that improved sleep led to clinically meaningful changes in other indices of morbidity (e.g., daytime fatigue). Five treatments met criteria for empirically-supported psychological treatments for insomnia: Stimulus control therapy, relaxation, paradoxical intention, sleep restriction, and cognitive-behavior therapy. Discussion: These updated findings provide additional evidence in support of the original review's conclusions as to the efficacy and generalizability of psychological and behavioral therapies for persistent insomnia. Nonetheless, further research is needed to develop therapies that would optimize outcomes and reduce morbidity, as would studies of treatment mechanisms, mediators, and moderators of outcomes. Effectiveness studies are also needed to validate those therapies when implemented in clinical settings (primary care), by non-sleep specialists. There is also a need to disseminate more effectively the available evidence in support of psychological and behavioral interventions to health-care practitioners working on the front line. SLEEP, Vol. 29, No. 11, 2006 1399 month and often several years. Insomnia complaints are typically associated with reports of daytime fatigue, problems with memory and concentration, and mood disturbances, impairments that may be the primary concerns prompting patients to seek treatment. Insomnia can be a symptom of several other conditions including medical, psychiatric, substance abuse or another sleep disorder; or, it can be a disorder in itself as in primary insomnia. [6][7][8] There are several treatment options available for insomnia, including psychological/ behavioral approaches, various classes of medications, and a host of complementary and alternative therapies (e.g., herbal/dietary supplement, acupuncture). The present paper focuses on psychological and behavioral approaches to treating insomnia. These procedures have received increasing research attention in the past 2 decades, and were noted as effective therapies at a recent National Institutes of Health State-of-the-Science Conference 9 on the manifestations and management of chronic insomnia. These methods include stimulus control therapy, sleep restriction, relaxation-based interventions, paradoxical intention, cognitive therapy, and combined cognitive-behavioral therapy. A brief summary of the nature of these interventions is presented in Table 1; more extensive descriptions are available in other sources. 10,11 2.0 PURPOSE The objective of this paper is to provide an update of the evidence regarding the efficacy, effectiveness, durability, and generalizability of psychological and behavioral interventions for persistent insomnia. The evidence is reviewed for the treatment of both primary insomnia and insomnia associated with other medical, psychiatric, or substance abuse disorders. As with the initial review paper 12 this updated review was commissioned by the Standards of Practice Committee of the American Academy of Sleep Medicine.

Search Methods, Keywords, and Databases
Treatment studies selected for review in this paper were identified through PsycInfo and Medline searches for research conducted from 1998 through 2004 inclusively. The following key words were used: nonpharmacologic, behavior therapy, cognitive therapy, psychotherapy, alternative medicine, stimulus control, progressive relaxation therapy or progressive muscle relaxation, paradoxical techniques or paradoxical intention, behavior modification, cognitive behavior therapy, psychological therapy, treatment, intervention, behavioral intervention, treatment, cognitive treatment, alternative treatment, therapy, biofeedback, sleep restriction, sleep deprivation, complementary therapies, mind-body and relaxation techniques, aromatherapy, biofeedback, hypnosis, imagery, or meditation, relaxation, relaxation techniques, yoga, massage. These terms were combined with sleep disorders or sleep initiation and maintenance disorders, or insomnia, or dyssomnia. The search was limited to humans, adults (18 and older), English or French language.

Selection Criteria of Treatment Studies
The initial PsycInfo and Medline searches yielded a total of 312 titles of potential interest; an additional 34 titles were identified by members of the task force through their own reading of the literature, for a total of 346 titles of potential interest. Of these, 102 abstracts were read by the task force chair for initial screening and 53 articles were selected for full review by 2 independent members of the task force. Only peer-reviewed published articles were retained at this phase. Each rater used a standard extraction sheet to summarize information about the study including experimental design, sample (number of participants, age, gender), diagnosis, type of treatments and controls, primary and secondary outcome measures, and main findings. Data extraction was completed independently and discrepancies between 2 members of a pair of raters were resolved through discussion with the chair and other members of the task force. The criteria for inclusion of a study were: (a) the main sleep diagnosis was insomnia (primary or comorbid), (b) at least 1 treatment condition was psychological or behavioral in content, (c) the study design was a randomized controlled trial, a nonrandomized group design, a clinical case series or a single subject experimental design with a minimum of 10 subjects, (d) the dependent measure included 1 or more of the following variables (as measured by daily sleep diaries, polysomnography (PSG), or actigraphy): Sleep onset latency (SOL), number of awakenings (NA), time awake after sleep onset (WASO), total sleep time (TST), sleep efficiency (SE), or sleep quality Psychological And Behavioral Treatment Of Insomnia-Morin et al (SQ). Studies using global measures of treatment outcome were included if those measures had been validated. Studies of circadian disorders (e.g., phase delay or phase advance syndromes) and those using other nonpharmacological therapies (e.g., light therapy, electro-sleep therapy) or complementary and alternative therapies (e.g., acupuncture) were excluded from this review. The majority of the initial 346 titles of potential interest were excluded either because they were not treatment studies for insomnia or because the main intervention was pharmacological. Of the 53 full articles reviewed and rated, 16 were rejected. The main reasons for article rejection were: no documentation of an insomnia diagnosis, sample size smaller than 10, results were secondary analyses of other databases (e.g., predictors of treatment response) and used global and non-validated measures of outcome. A list of excluded studies with reasons for exclusion is provided in an, Appendix A. The 37 studies that met inclusion criteria are listed in Table 2. For each study, we report the study design and the evidence level (using the Sackett System, 1993), the sample size (enrolled/completed), age, gender, insomnia diagnosis (primary or secondary), types of treatment and control conditions, therapy dosage, format, setting, therapist training, treatment duration and longest follow-up, primary and secondary outcome measures, and summary of main findings. Although some studies not meeting inclusion criteria are discussed in the text they are not included in the evidence table. Based on the Sackett sytem, 13 the criteria for grading evidence level of each study were: Randomized well-designed trials with low alpha and beta error (Grade I), randomized trials with high alpha and beta error (Grade II), nonrandomized concurrently controlled studies (Grade III), nonrandomized historically controlled studies (Grade IV), case series (Grade V). Table 2 summarizes the main features of the 37 studies selected for this review paper. The number of studies meeting Sackett System standards for grading evidence levels 13 was as follows: 11 studies were graded I, 13 studies graded II, 2 studies graded III, 5 studies graded IV, and 6 studies graded V. A total of 2246 patients with insomnia were enrolled in the 37 studies and approximately 2029 of those completed treatment, for an attrition rate of less than 10% overall. With a few exceptions (studies of Veterans, substance abusers, internet-based treatment), there was a larger representation of women than men enrolled in most studies, with a typical ratio of 2:1, which is representative of insomnia prevalence estimates. 1 Nine of the 37 studies focused specifically on older adults (average age > 60 years old).

Descriptive Features of the Studies and Samples
The main sleep diagnoses of patients enrolled in the studies were primary or psychophysiological insomnia (28 studies), insomnia associated with medical (4 studies) or psychiatric disorders (3 studies) or a mix of both conditions (6 studies), and hypnotic-dependent insomnia (5 studies). Eight studies examined treatment efficacy in patients with different subgroups of insomnia diagnoses, leading to a total number of studies greater than 37 studies.
The majority of reviewed studies (n = 33) relied on prospective daily sleep diaries to document treatment outcome. Participants were typically required to complete a daily diary for a minimum of a 1 or 2 week baseline period, for the duration of treatment, and for an additional 1 or 2 week period at post treatment and follow-ups. We retained 3 studies using the Pittsburgh Sleep Quality Index (PSQI) as the primary outcome. [14][15][16] A few studies have also included polysomnography (n = 7) 17-23 and actigraphy (n = 6) 18,19,[24][25][26][27] to complement subjective reports from daily sleep diaries. Those studies are identified in Table 2 and in the appropriate subsections of the results. Primary dependent variables derived from these assessment methods were sleep onset latency (SOL), wake time after sleep onset (WASO), total sleep time (TST), sleep efficiency (SE), and sleep quality (SQ). Secondary outcomes included measures of insomnia severity (Insomnia Severity Index; ISI, 28 sleep quality (PSQI), 29 psychological symptoms (Beck Depression Inventory, BDI, 30 State-Trait Anxiety Inventory STAI 31 , and fatigue (Multidimensional Fatigue Inventory, MFI 32 ).
The following sections summarize the evidence regarding the efficacy of treatment for primary insomnia, the generalizability of the evidence to different forms of insomnia (primary, secondary, hypnotic-dependent), insomnia in older adults, and the clinical significance and durability of sleep improvements over time. Comparative findings of single and multifaceted therapies and of different treatment implementation models are also summarized.

Treatment of Primary Insomnia
Seventeen studies evaluated the effects of treatment for primary insomnia (see Table 2). Five of those studies were randomized clinical trials (RCT; with grade I, 17,20,21,33,34 4 of which used CBT as the main intervention. In a comparative study of CBT (without relaxation), relaxation, and a psychological placebo (i.e., quasidesensitization) with a sample of 75 primary insomniacs, 17 CBT produced greater improvements on the main diary and PSG-defined sleep measures (e.g., SE, WASO) relative to relaxation and control. More CBT patients (64%) achieved clinically significant outcomes compared to relaxation (12%) and placebo (8%). In an effectiveness trial 33 conducted in primary care that evaluated CBT against a wait-list control group, active treatment was found superior to the control condition on most primary and secondary outcome measures. SOL was reduced from 61 to 28 min following active treatment compared to a change from 74 to 70 min for the control condition. a Smaller improvements were noted on WASO. There was no significant change on TST during treatment, but an increase of about one-half hour over baseline was obtained at follow up. Of those patients using hypnotic medications at baseline, 76% were medication-free at the end of treatment and 80% at the 12-month follow up. In a comparative study 21 of CBT, medication (temazepam), and combined CBT plus medication, all 3 active treatments improved more than pill placebo on the main outcomes of WASO and SE, with a trend for the combined intervention to yield the greatest benefits. PSG data produced similar outcomes, although of smaller magnitude, but only the combined condition was significantly superior to placebo on the main outcome variables. According to PSG, more patients in the CBT (56%) and combined (68%) conditions achieved clinically significant changes (i.e., SE > 85%) relative to medication alone (47%) or placebo groups (22%). In another clinical trial of primary insomnia in older adults, 20 23,37 or treatment implementation models, 38,39 or as part of case series. 40,41 Some of these studies will be discussed in later sections of this paper.

Treatment of Insomnia Associated with other Medical or Psychiatric Disorders
Twelve investigations have evaluated the efficacy of psychological and behavioral treatments for insomnia associated with another medical or psychiatric disorder. These studies have focused, for example, on patients with chronic pain, 25 cancer, 16,42 alcohol dependence 26 and older adults with various medical illnesses. 27,43 Only 4 of those 12 studies were RCT (Grade I or II) [25][26][27]43 and the remaining were nonrandomized studies or clinical replication series.
In a study of 60 patients with insomnia associated with chronic pain, 25 CBT was significantly more effective than control on measures of SOL, WASO, and SE, but not on number of awakenings and TST. SOL was reduced from 55 min to 28 min and SE increased from 72% to 85%. Nocturnal motor activity (as measured by actigraphy) was reduced in the treated group but not in the control group; there were no group differences on pain ratings, depressive symptoms, or medication use. In a study of 51 older adults with insomnia associated with medical illness, 27 CBT and relaxation conditions were more effective than control on diary measures of WASO and SE, as well as on a measure of overall sleep quality (PSQI); the relaxation group had a greater increase in TST than CBT and controls. A higher proportion of treated patients relative to controls achieved clinically significant improvements. There were no differential group effects on actigraphy, medication use, or other secondary measures of anxiety, depression, and quality of life. In a study of 49 older adults with insomnia associated with medical and psychiatric conditions, 43 a combined intervention of stimulus control, relaxation, and education reduced WASO 25 min and increased SE 11% at post treatment. Fifty-seven percent (57%) of treated patients achieved clinically significant improvements on SE relative to 19% of control patients; there was no significant change on secondary measures of anxiety, depression, and impact of insomnia. Outcomes were similar for individuals with insomnia associated with a medical condition and those with insomnia related to a psychiatric disorder. A controlled study conducted with recovered alcoholics 26 showed modest but significant improvements of SOL (-18 min) and SE (+10%) among insomnia-treated patients. At the 6 month follow-up, 15% of treated participants had relapses with alcohol and this proportion was not different between treated and control patients.
Several additional studies (clinical replication series or uncontrolled group studies) have also provided evidence showing that patients with medical and psychiatric disorders could also benefit from sleep/insomnia specific interventions. Two investigations with cancer patients 16,42 showed that CBT was associated with improvements of sleep and of daytime functioning (e.g., fatigue, energy). One case series study of 67 patients with psychiatric disorders and insomnia 44 reported significant improvements of sleep, mood, fatigue, and reduced use of sleep medication, while a smaller study 45 found no change on specific sleep parameters (SOL and WASO) but reported significant reductions of global insomnia severity as measured by the ISI. Additional evidence supporting the efficacy of CBT was reported in 3 clinical replications series [46][47][48] conducted with heterogeneous samples of patients presenting to sleep disorders clinics with a variety of primary and secondary insomnia diagnoses. Although conclusions drawn from these uncontrolled studies should be treated with caution because of the studies' high attrition rate, the evidence suggests that among those who received an adequate treatment exposure (average of 6 -8 therapy sessions) outcomes were comparable to those patients with primary insomnia enrolled in controlled clinical trials. Furthermore, treatment response appeared comparable between patients with medical or psychiatric comorbidity and those with primary insomnia in one study. Baseline anxiety, depression, and insomnia severity did not differ among treatment responders and nonresponders. 48

Treatment of Insomnia in Older Adults
Nearly 25% (9 studies out of 37) of reviewed studies were conducted with older adults (average age > 60 years old). This is in sharp contrast to our previous review 12 that included only a handful of studies with older subjects. Three studies focused on older adults with primary insomnia, 18,20,21 2 on insomnia associated with medical or psychiatric illnesses, 27,43 1 included a mix of patients with primary and comorbid insomnia, 49 2 evaluated the impact of psychological and behavioral interventions specifically in older adults who were chronic users of hypnotic medications, 15,22 and 1 study 19 examined the moderating role of upper airway resistance syndrome in the treatment of postmenopausal insomnia. With the exception of the study conducted by Pallesen et al, 49 all these investigations were RCT.
In a study of 89 older adults with primary insomnia, 20 sleep restriction and relaxation were both more effective than a psychological placebo for reducing WASO; changes were identical (67 min to 43 min) for the 2 treatment groups at the end of the 6-week treatment phase, but sleep restriction produced the best outcome at the 1-year follow up. No significant changes were obtained on PSG measures. All 3 conditions, including placebo control, showed improvements on secondary measures of fatigue and a measure of insomnia impact. In another placebo-controlled study, 21 76 older adults treated with CBT, medication (temazepam), or combined CBT + medication improved more than those receiving placebo on the main outcome measures of WASO and SE. PSG comparisons yielded improvements in the same direction, albeit of smaller magnitude, than those reported on sleep diaries. A greater proportion of patients treated with CBT, alone or combined with medication, achieved clinically significant improvements (i.e., SE > 85%) compared to those receiving medication alone or placebo.
In a comparison of sleep restriction, with and without an optional daytime nap, to sleep education alone, 18 both sleep restriction conditions produced greater SE increase, with reduced time spent in bed, relative to the control condition. There was no sig-nificant group difference on actigraphy or PSG measures; TST was reduced for the sleep restriction conditions at post treatment and returned towards baseline values at the 3-month follow-up. There was a mild increase of physiological sleepiness (as measured by the MSLT) but no change on subjective sleepiness. In another investigation, 49 a combination of sleep education plus stimulus control was as effective as sleep education plus relaxation, and more effective than a wait-list control, in older adults with mixed primary and secondary insomnia; there were modest improvements of daytime measures for both active conditions. Two additional studies (reviewed in section 4.3) provided evidence that older adults with insomnia and comorbid medical disorders also benefitted from sleep-specific interventions. 20,27

Treatment of Insomnia Among Chronic Hypnotic Users
Four investigations examined the efficacy of psychological and behavioral interventions for insomnia in the context of chronic hypnotic usage, including 2 that were conducted with older adults. 15,22 In a study of 209 chronic hypnotic users, 15 CBT (with an optional medication taper) was associated with improved PSQI scores and reductions of hypnotic use at 3-and 6-month follow ups. A greater percentage of patients treated with CBT for insomnia (39%) relative to no treatment controls (11%) achieved at least a 50% reduction of hypnotic use relative to baseline at the 6-month follow up. A cost-offset analysis revealed that while CBT added to the initial treatment cost, there was a significant cost offset at follow up resulting from a reduction of sleep medication usage. In another study comparing a supervised medication withdrawal program, alone and combined with CBT for insomnia, to CBT alone, 22 all 3 interventions produced significant reductions in both the quantity (90%) and the frequency (80%) of benzodiazepine use, and more patients in the combined approach (85%) were medication free at post-treatment than for those receiving the taper schedule alone (48%) or CBT alone (54%). There were modest changes in sleep patterns during the initial 10-week withdrawal phase, but CBT-treated patients reported greater sleep improvements relative to those receiving the medication withdrawal alone. Several improvements were also reported on secondary measures of insomnia severity (ISI), anxiety (BAI) and depressive (BDI) symptoms.
Two additional studies have examined the impact of chronic hypnotic use on outcome with middle-aged adults. One study 50 found a significant worsening of sleep parameters during medication withdrawal and the addition of relaxation therapy did not attenuate this effect. In a similar study using stimulus control as the main behavioral intervention, 51 stimulus control produced significant improvements on most sleep parameters relative to no additional treatment. There was no difference in sleep outcomes between medicated and nonmedicated patients.

Validation and Comparative Efficacy of Single and Multifaceted Therapies
Although there are several distinct psychological and behavioral therapies for insomnia, there was a clear trend/preference for investigators to combine 2 or more of these methods when treating insomnia. The most common combination involves an educational (sleep hygiene), behavioral (stimulus control, sleep restriction, relaxation), and a cognitive therapy component, usually referred to as cognitive-behavior therapy. Indeed, 21 stud-ies have evaluated the efficacy of CBT, either with (12 studies) or without relaxation (9 studies), and 5 more studies have used a similar multi-component interventions but without cognitive therapy (see Table 2).
There has been no complete dismantling of CBT to isolate the relative efficacy of each component within the same study. However, comparisons of some components revealed that CBT was superior to relaxation alone in 1 study of primary insomnia 17 and sleep restriction was superior to relaxation at follow-up in another study with older adults. 20 In another study, 52 relaxation was more effective for sleep initiation problems relative to sleep hygiene education alone and a combination of stimulus control plus sleep restriction, whereas the latter combination had greater effects on sleep maintenance variables.
Twelve studies have isolated in a controlled trial at least 1 therapy component such as relaxation, sleep restriction, stimulus control, or paradoxical intention. All 6 studies contrasting relaxation-based interventions (either progressive muscle relaxation or similar procedures) to a control condition, have reported that this single therapy was more effective than wait-list, 27 placebo, 17,20 no treatment, 37,50 and minimal sleep hygiene education control. 52 Two studies showed that sleep restriction was superior to either placebo 20 or sleep hygiene education alone, 18 1 study found that stimulus control was more effective than a wait-list control, 51 and 1 additional investigation reported that paradoxical intention was superior to a wait-list control for sleep onset insomnia. 24 In spite of the inclusion of a cognitive therapy component in numerous studies, no study has yet evaluated its unique contribution to outcomes.
As in the earlier review paper, 12 criteria developed by the American Psychological Association 53 for defining empirically-validated psychological treatments, were used to determine whether additional evidence was available for each psychological and behavioral interventions (See Table 3). Based on the criteria outlined in our previous review, 12 stimulus control therapy, relaxation training, and paradoxical intention met criteria for wellestablished psychological treatment for insomnia. With the additional evidence from the present review (indicated by studies in boldface in Table 3), sleep restriction 18,20 and CBT 17,21 would also meet criteria for well-established treatments. Furthermore, additional studies strengthened the level of evidence supporting stimulus control, 51 relaxation, 17,20,27,37 and paradoxical intention. 24

Comparisons of Psychological/Behavioral Therapies and Medication
Five controlled studies conducted with primary/psychophysiological insomnia patients have evaluated the impact of psychological/behavioral interventions in comparison to or as an adjunct to hypnotic medications. Two studies evaluated the efficacy of CBT, singly and combined with medication, 21,34 1 used a medication alone condition as a comparator to psychological treatments, 52 and 2 examined the incremental benefits of adding 1 treatment to the other. 54,55 In a placebo-controlled comparison of CBT and medication (temazepam) singly and combined 21 (study described in Sections 4.2 and 4.4), all 3 active treatments were more effective than placebo on sleep continuity variables, with a trend for the combined approach to yield better outcomes. These results were corroborated with PSG measures, although the magnitudes of sleep improve-ments were smaller on PSG than on diary measures. Long-term follow-up data showed that subjects treated with CBT sustained their clinical gains over time, whereas those treated with medication alone did not. The combined approach showed some loss of therapeutic benefits over the follow-up periods, although there was more variability across subjects in that condition. In a similar study design with 63 young and middle-aged adults with sleep onset insomnia 34 (study described in Section 4.2), CBT was shown more effective than medication (zolpidem) and placebo on measures of SOL and SE. All 4 conditions, including placebo, increased their TST, with medication yielding the largest increase in TST (69 min), though this finding was not significantly different from the other groups. There was no significant difference between CBT alone and CBT combined with medication. Sleep changes were well maintained at the 12-month follow up for patients treated with CBT, singly or combined with medication, but no follow up data was available for those treated with medication alone. Data obtained from a Nightcap device showed sleep changes in the same direction as those from diaries, except that no improvement was obtained on any of the measures for the placebo condition.
An investigation of 41 patients treated with estazolam examined the added benefits of muscle relaxation, imagery training, and sleep hygiene education to medication. 55 There was no group difference on any outcome. Significant improvements from baseline to post treatment were obtained on WASO and SE for the relaxation (-17 min and + 9.7% respectively) and imagery training groups (-33 min and + 7.4%). TST was increased by 34 min (education), 40 min (imagery), and 65 min (muscle relaxation) for the same period but there was no significant group difference. SOL was not changed in any of the groups. Significant changes were obtained from baseline to follow up in all 3 groups on sleep measures and on secondary measures of arousal, self-efficacy, and depressive (but not anxiety) symptoms.
Another study of 30 patients 54 examined the added benefits of modafinil when combined with CBT in the management of primary insomnia. Although there was no significant gain from modafinil in terms of sleep continuity parameters, there were trends suggesting that the addition of modafinil to CBT reduced daytime sleepiness and enhanced compliance with prescribed bedtime. Finally, data from 53 patients with primary insomnia 52 showed that while relaxation was more effective for sleep onset problems and a combination of stimulus control plus sleep restriction had more benefits for sleep maintenance variables, medication (flurazepam) produced the largest improvements on all sleep variables during the initial 2-week intervention.

Treatment Implementation Methods: Individual, Group, and Self-Help
Treatment was implemented on an individual basis in 22 studies (54%), in a group format in 11 studies (29.7%), and a few additional studies relied on self-help materials with or without additional telephone consultations. An average of 5.7 consultation visits was conducted over a mean treatment period of 6.5 weeks. 1 study directly compared the relative efficacy of CBT implemented in group or individual sessions, or through self-help written materials combined with brief telephone consultations. 56 All 3 groups produced significant improvements of sleep and secondary measures and there was no between group difference on any  70 SC > Pla   71 SC > Pla Morin & Azrin (1987; 72 75 Rel > Pla Nicassio et al. (1982) 76 Rel > No treatment Turner & Asher (1979) 74 Rel > Pla Woolfolk & McNulty (1983) 77 Rel  38 CBT > WL Morin et al. (1993) 82 CBT > WL These studies provided evidence supporting more than 1 treatment. *Well established treatments according to APA criteria for empirically supported treatments. Criteria for well-established treatments require at least 2 between-group design studies demonstrating efficacy in 1 or more of the following ways: I. superior to pill or psychological placebo or to another treatment; or equivalent to an already established treatment in a study with adequate statistical power; II. a large series of single case design experiments (n > 9) demonstrating efficacy as in I; III. the studies must be conducted with treatment manuals; IV. the characteristics of the sample must be well-described; V. the effects must have been demonstrated by at least 2 different investigators or investigatory teams. †Probably efficacious treatments according to APA criteria for empirically supported treatments. Criteria for probably efficacious treatments are: I. 2 studies showing the treatment is more effective than a waiting-list control group, or II. 1 or more studies meeting the wellestablished treatment criteria I, III, and IV, but not V, or III. a small series of single case design studies (n > 3) otherwise meeting wellestablished treatment criteria II, III, and IV. measure. A similar study of insomnia in recovered alcoholics also found equivalent outcome between individual CBT and self-help CBT plus telephone consultation. 26 In contrast, 1 study 38 found that the addition of telephone consultation to self-help written material enhanced outcomes at post treatment, but those initial gains tended to disappear at follow up. An internet-based intervention produced greater improvement in several sleep parameters relative to controls, 39 but the attrition rate (24%) was higher than in studies using face-to-face consultation visits. While the majority of reviewed studies have used psychologists or psychology trainees as therapist (with treatment manual), 2 studies examined treatment efficacy as implemented by primary care physicians 36 or by nurse practitioners 33 who had also been trained before the studies; treatment benefits were generally equivalent to those obtained with therapists who had mental-health training.

Durability of Sleep Improvements
Twenty-six of the 37 reviewed studies reported follow up data of at least 1 month duration after completing treatment (mean duration = 7.7 months; range 1-36 months). The remaining 11 studies reported no follow-up data. As was the case in the initial review, a very robust finding across studies is that treatmentproduced changes in sleep parameters are well maintained at short (1-3 month), intermediate (6-month), and long-term (> 12 months) follow-ups. One interesting finding from studies using sleep restriction is that total sleep time may be reduced during the initial intervention, but it is significantly increased at follow up evaluation. 20 Despite fairly robust long-term outcomes, follow-up data must be interpreted cautiously as there are relatively few studies reporting long-term (> 1 year) follow-ups (see Table  2) and, among those that do, attrition rates increase substantially over time.

CONCLUSIONS
This updated review of treatment studies conducted between 1998 and 2004 provides additional evidence that psychological and behavioral interventions represent an effective treatment option for the management of persistent insomnia. In addition to studies further documenting treatment efficacy for primary insomnia, recent studies indicate that treatment is also effective for insomnia associated with some medical conditions and, to a lesser extent, with psychiatric conditions. Treatment benefits are well sustained over time. There is still limited evidence of clinically meaningful changes beyond reductions of insomnia symptoms (i.e., improved daytime functioning, quality of life).
These findings are consistent with our previous systematic review 12 as well as with other meta-analyses of the efficacy of psychological and behavioral interventions for insomnia. [57][58][59] For instance, of the 17 most recent treatment studies of primary insomnia, 5 were randomized controlled trials, 17,20,21,33,34 and all 5 yielded additional evidence of significant sleep improvements with psychological and behavioral interventions. Although most of this additional evidence is based on daily sleep diaries, 3 of the 5 key studies included PSG measures and 2 of them reported outcomes that paralleled findings from diary measures. Actigraphy, on the other hand, was not very sensitive for detecting changes in sleep/wake variables in the few studies using this device.
The treatment of comorbid insomnia has received limited attention until recently, perhaps owing to the traditional notion that it would not respond to treatment unless the associated condition was treated first. The present review, as well as recent findings, [60][61][62] challenge this traditional notion and indicate that insomnia-specific treatment is of benefit even among those whose insomnia is associated with comorbid conditions such as cancer, pain, alcohol abuse, and some psychiatric conditions. Nonetheless, there is a need for additional prospective and randomized controlled studies of comorbid insomnia contrasting outcomes when sleep is or is not directly targeted in treatment.
The treatment of insomnia in older adults is another area previously neglected and for which there was limited evidence to guide practitioners. Nearly 25% of the studies reviewed in this paper focused on older adults. The findings from those studies indicate that older adults with primary insomnia respond to treatment as well as younger and middle-aged adults, although the presence of comorbid medical or psychiatric condition may moderate outcomes. 27,43 A recent meta-analysis 63 also confirmed that treatment effect sizes are comparable for middle-aged and older adults. There is additional evidence that psychological treatment can facilitate hypnotic discontinuation in older adults who are chronic users of hypnotics. 15,22 This is an important finding as older adults are more likely to be long-term hypnotic users which, in some cases, may perpetuate sleep disturbances. Thus, although heterogeneity in diagnosis makes it more difficult to compare studies, 64 such heterogeneity of insomnia samples also enhances generalizability of outcomes.
This review highlights an emerging trend among investigators for combining multiple treatments, which contrasts with the earlier review describing numerous studies comparing treatment efficacy among 2 or more single therapies. Indeed, 26 of the 37 reviewed studies evaluated the efficacy a multi component approaches, including 21 studies using multicomponent therapy, with or without relaxation, and 5 more combining behavioral interventions (e.g., stimulus control and sleep restriction) but without cognitive therapy. Although the reason for this shift in paradigm is not entirely clear, the use of multi-component approaches is more likely, at least on a clinical basis, to address the different facets/perpetuating factors of insomnia. 11 Whether CBT produces outcome that is superior to single therapies remains largely unexplored. The few comparative studies available show that outcome is superior for CBT, stimulus control, and for sleep restriction relative to relaxation alone, 17,20,27,49 however, there has been no complete dismantling of CBT to isolate the relative efficacy of each component. Furthermore, although some findings 65 suggest that change in beliefs and attitudes is an important mediator of long-term outcomes, there has been no direct controlled evaluation to isolate the relative contribution of cognitive therapy.
Our previous review had identified 3 treatments as well-validated and 2 more as probably efficacious according to criteria set by the American Psychological Association. 53 This updated review provides further evidence supporting stimulus control, relaxation, paradoxical intention as well validated therapies, and new evidence to upgrade sleep restriction and CBT from probably efficacious to well-established treatments.
Although there is evidence supporting the efficacy of psychological and behavioral treatment for insomnia, there is still little information about the specificity of this treatment modality and the active therapeutic mechanisms responsible for sleep improve-ments. With a few notable exceptions using attention-placebo conditions, 17,20 most CBT trials have used wait-list control groups, precluding the unequivocal attribution of treatment effects to any specific ingredient of psychological and behavioral treatment. The lack of a pill-placebo control equivalent in psychological outcome research makes it difficult to determine what percentage of the variance in outcomes is due to specific therapeutic ingredients (i.e., restriction of time in bed, cognitive restructuring), the measurement process (i.e., self-monitoring), or to non-specific factors (e.g., therapist attention, patients' expectations).
An important limitation noted in the previous review that is still evident in recent studies is the limited evidence documenting the clinical significance of outcomes beyond insomnia symptom reductions (i.e., reduced morbidity, improved quality of life). There is a need for broadening the scope of outcome measures 66 and for standardizing assessment methodology in insomnia research, 64,67 Furthermore, even for patients meeting criteria for what might be considered a clinically meaningful change, many such treatment responders reach a plateau and continue showing residual sleep disturbances after treatment and may remain at risk for relapse. There is a need to develop and validate more potent interventions that would increase the rate of patients reaching full remission. 9 Ongoing studies are currently examining optimal treatment dosage, treatment combination involving medication, and maintenance therapies.
A related issue is that most of the outcome evidence currently available is about improving sleep initiatiation and sleep continuity parameters, with essentially no information about the impact of treatment on more qualitative aspects of sleep, i.e., non-restorative sleep. Although this qualitative feature is part of the standard insomnia definition, no study has of yet examined the impact of psychological treatment on this variable.
Proper implementation of psychological and behavioral therapies usually requires more time than prescribing a hypnotic medication, which may represent an important barrier to using such interventions in clinical practice. Nonetheless, several studies have documented the benefits of cost-effective implementation models using nurse practitioners, group therapy, or self-help materials to complement therapist-guided intervention. Whereas such implementation models are likely to make treatment more readily available, adequate therapist training remains an important consideration in using CBT effectively to optimize outcome. Additional studies examining the relative cost-effectiveness of different insomnia interventions would be warranted 68 In summary, this updated review provides additional evidence supporting the use of psychological and behavioral interventions for primary insomnia, for insomnia associated with medical or psychiatric conditions, and for insomnia in older adults. Additional research is still needed to develop and validate treatment algorithms that would optimize outcomes and reduce morbidity; clinical research that would examine treatment mechanisms, mediators and moderators of outcomes is also warranted; and, additional effectiveness trials are particularly needed to document outcomes in unselected patients seeking treatment in various clinical settings (e.g., primary care). Finally, an important challenge for the future will be to disseminate more efficiently the available evidence to health-care providers and translate that evidence into meaningful clinical guidelines in order to ensure a more widespread use of validated therapies. SLEEP, Vol. 29, No. 11, 2006