Comparative Effectiveness of Cognitive Behavioral Therapy for Insomnia: A Systematic Review

The prevalence of insomnia in primary care patients is as high as 69% [1,2] compared to 33% in the general population [3]. Insomnia can exist as a primary disorder or co-morbid with other conditions including depression [4] and chronic pain. In the past insomnia was considered to be a symptom of these conditions with the assumption that treatment of these ‘primary’ conditions would lead to the resolution of insomnia, eliminating the need for targeted insomnia treatment. There is now evidence to suggest that insomnia often persists following resolution of these ‘primary’ conditions, and that it generally does not spontaneously resolve over time if left untreated [5]. Insomnia is independently associated with significant morbidity including fatigue, impaired concentration and memory, irritability, difficulty in interpersonal relationships, decreased quality of life, and increased risk of new-onset psychiatric illness [1,6–9]. In addition, there is evidence that insomnia may confer risk for medical illness including hypertension, heart disease, and diabetes [10,11], and is associated with increased overall health care costs [9].

The most common approach to the management of insomnia is medication treatment. Numerous trials have documented moderate efficacy with benzodiazepine receptor agonists [12,13]. The advantages of medications are that they are widely available and, when effective, lead to clinical improvement rapidly. The disadvantages are the potential for side-effects, dependence, and tolerance over time. Perhaps the most important disadvantage is that medications are usually not curative, leading to long-term treatment over many years despite a lack of safety and efficacy data for their long-term use beyond 1–2 years.

An alternative treatment approach is cognitive behavioral therapy for insomnia (CBT-I). CBT-I is a non-pharmacological approach to treatment comprised of several strategies. The goal of CBT-I is to target those factors that may maintain insomnia over time, such as dysregulation of sleep drive, sleep-related anxiety, and sleep-interfering behaviors. This is accomplished by establishing a learned association between the bed and sleeping through stimulus control, restoring homeostatic regulation of sleep through sleep restriction, and altering anxious sleep-related thoughts through cognitive restructuring. By changing sleep-related behaviors and thoughts, CBT-I may target those factors that cause insomnia to persist over time. CBT-I is delivered over the course of 4–8 sessions that occur weekly or every other week for 30–60 minutes each. There are two main disadvantages to CBT-I. First, during the first few weeks of treatment there is often an acute reduction in total sleep time that can lead to the side effect of increased daytime sleepiness which, for some, is enough to lead them to drop out of treatment. Second, improvements from CBT-I are typically not seen until 3–4 weeks into treatment. While a few research studies have examined the efficacy of nurseled CBT-I in primary care settings [14], in current clinical practice it is usually necessary to refer out to individuals with specialized training in this treatment. It should be noted that the core therapies for CBT-I substantially differ from other forms of CBT and it is for this reason that the abbreviation CBT-I denotes this form of CBT specifically for insomnia [15].

Treating insomnia with CBT-I, as opposed to medication, has a number of potential advantages, including fewer known side effects, and an explicit focus on treating the factors that may be responsible for perpetuating chronic insomnia in an effort to produce more durable effects. Some patients also prefer non-medication treatments [12,16,17]. Providers often have negative attitudes towards hypnotics as well, and prefer to reduce their prescriptions of such medications [18]. These advantages of CBT-I suggest that it might be a viable option for the treatment of insomnia [12,17].

Evidence suggests that CBT-I is effective when compared with placebos [19,20]. A new systematic review [21] meta-analyzes evidence from 14 randomized trials comparing groups of patients given CBT-I to control groups. A variety of controls were combined in that analysis, such as patients on a waiting list to receive CBT-I and patients offered a sleep hygiene program which has been shown to be of little or no benefit [17]. The meta-analysis found that CBT-I had a medium to large effect in reducing insomnia, and that the effects were maintained even after conclusion of the treatment period. But this review does not tell us whether CBT-I is as effective as sleep medications. The AHRQ Evidence-based Practice review [20] addressed that question, but it was published in 2005 and lacks evidence published since September 2004.

The primary objective of this review is to examine the most up-to-date evidence comparing CBT-I to medications in patients with primary and comorbid insomnia and assess the comparative effectiveness of these treatments.

Our search found a total of five RCTs that met the inclusion criteria and compared CBT-I to medications. The five trials all involved patients with primary insomnia. The trials compared CBT-I to zopiclone [29], zolpidem [30], temazepam [31,32], and triazolam [33]. The design and quality of these studies is reported in Table 1 and their short and long term results are summarized in Table 2 and Table 3 respectively. Only one trial compared CBT-I to drug therapy in patients with comorbid insomnia [34]: the drug used was nefazodone (now withdrawn from the market), so the study was excluded from further analysis. The evidence base was too small and heterogeneous for meta-analysis.

Overall, we found that CBT-I is at least as effective for treating insomnia when compared with sleep medications, and its effects may be more durable than medications. The strength of these claims needs to be tempered with a critical evaluation of the available evidence. There have only been a handful of studies that have included both cognitive behavioral and pharmacotherapy treatment arms for inclusion in this systematic review. A significant limitation of the studies is that they all focused on patients with primary insomnia who, while more straightforward for inclusion in clinical studies, may not be representative of the majority of patients who have comorbid medical or psychiatric disorders. RCTs of CBT-I only, with no medication comparator condition, generally find that treatment is as efficacious in those with comorbid disorders compared to those with primary insomnia (for example see [38]). While this suggests that the results of this review may also hold for comorbid insomnia, studies including both CBT-I and medication arms are needed in this patient population. The studies also were diverse in terms of sleep-related outcome measures (subjective versus objective) and demographic composition. In terms of outcome measures, insomnia RCTs generally find larger effects of treatment on subjective measures such as questionnaires than on objective measures of actigraphy and polysomnography. However, this does not limit the meaning of the findings since the diagnosis of insomnia relies on patient report and not on a diagnostic laboratory test. In terms of demographic composition, three of the five studies we included were in older adults. Future studies will need to be conducted in both older and younger adult samples, but it is noteworthy that the findings were generally consistent across studies despite the diversity of subject ages.

The results of this analysis indicate that more research is warranted in this area to better understand the relative efficacy of CBT-I and hypnotics in order to make decisions about best practices.

Even with these limitations, these results have a number of important treatment implications for the primary care setting. First, providers should consider CBT-I as a treatment option for their patients with insomnia. Second, CBT-I may have advantages when compared to medications, most notably more durable treatment gains that reduce or eliminate the need for long-term pharmacologic treatment. In addition, those significant clinical gains can be made in a relatively short number of treatment sessions (most studies were 6–8 weekly or bi-weekly sessions). Future research comparing the efficacy of CBT-I and hypnotics may find that treatment efficacy is moderated by patient characteristics such as age or comorbid diagnoses. These data can lead to the development of treatment algorithms that would allow primary care providers to make more informed treatment decisions. Of note, three of the studies [29,31,32] included a treatment arm that was a combination of CBT-I and medication. Outcomes were generally comparable to those receiving CBT-I alone, but future studies should examine additional strategies for combination treatment that capitalize on the relative advantages of medication (rapid onset on therapeutic response) and CBT-I (long-term durability of treatment gains).

Despite the evidence supporting the effectiveness of CBT-I, many providers may not be aware of its existence or know how to refer patients for treatment [39]. Referral can be accomplished in several ways. First, there is a growing registry of providers who have sought specialized certification in the delivery of CBT-I and other behavioral sleep medicine interventions to whom patients can be referred (http://www.absm.org/BSMSpecialists.aspx). These providers are primarily clinical psychologists and are located in a variety of settings including sleep disorders centers and private practice [14,40]. Second, primary care practices can partner with local psychologists or others to provide CBT-I. Lastly, a few studies have yielded promising results by implementing CBT-I in the primary care setting by nurses provided the necessary training and supervision [41,42]. By utilizing these various approaches, primary care providers can make CBT-I accessible. Yet, even with these options, getting a patient access to CBT-I may present challenges not encountered when providing a patient with a simple prescription for a sleep medication. First, there can be greater difficulty in getting insurance coverage for CBT-I than for medications. Second, therapists to provide CBT-I may not be available in some parts of the country or the world. However, the evidence reviewed in this manuscript suggests that the additional effort in obtaining access to a therapist if available may be worth the more long-term improvements patients may achieve with CBT-I without the extended use of sleep medications.

The GRADE approach used to grade the evidence base for the comparisons examined in this study can help identify areas where the evidence is least robust and where additional studies can most impact our conclusions about CBT-I. GRADE can also help identify causes for the weakness of the evidence base. Data on long-term outcomes of CBT-I compared to non-benzodiazepines is still sparse, and inconsistencies in the data comparing CBT-I to benzodiazepines in the short term need to be resolved. Data on CBT-I for patients in critical subgroups of interest such as those with comorbid insomnia is sparse as well, and there have been no trials comparing CBT-I to approved drugs in these patients. This is an important limitation of the findings in our review as many patients presenting with insomnia to primary care practices will have secondary or co-morbid insomnia, and thus the findings here may not be as generalizable to those settings. In addition, comparisons of CBT-I to drugs commonly used off-label for the treatment of insomnia, including antidepressants and antihistamines, are not available. Lastly, all studies in this field suffer an unavoidable risk of bias because double-blinding is not possible.

CBT-I is an effective treatment for insomnia that can produce durable results in a relatively brief number of visits. Low- to moderate quality evidence suggests it has greater effectiveness than medications for treatment of insomnia six months or more after therapy is completed. Additional research is needed to validate its effectiveness in long-term studies beyond 1–2 years and in patients with comorbid insomnia. Additional research is also needed to establish a benefit for CBT-I with respect to important psychological outcomes including quality of life.