Cognitive behavioral therapy for insomnia: A meta-analysis of long-term eff ects in controlled studies

Van Der Zweerde, T., Bisdounis, L., Kyle, S. D., Lancee, J., & Van Straten, A. (2019). Cognitive behavioral therapy for insomnia: A meta-analysis of long-term effects in controlled studies. Sleep Medicine Reviews, 48, 101208. https://doi.org/10.1016/j.smrv.2019.08.002

Summary by: Isabel Nemeth

According to the DSM-IV, insomnia can be defined as, “a persistent difficulty initiating or maintaining sleep, for three months or longer and for at least three nights a week, resulting in impaired daytime functioning and signifi cant distress” (van der Zweerde et al., 2019). Only 6% of the general population suff ers from insomnia, but around 30% experience the symptoms of insomnia without meeting the criteria for diagnosis in the DSM-IV (Morin et al., 2006; Ohayon, 2002). If untreated, insomnia may persist for years and have an eff ect on daily life in aspects such as, mood, fatigue, cognitive ability, physical well-being, and social relationships. (Kyle et al., 2010; Morin et al., 2009).

Cognitive behavioral therapy (CBT) has already proven to be an effective treatment for insomnia in the short term. In fact, oftentimes it is the preferred treatment method over pharmacotherapy according to the American and European guidelines (Qaseem et al., 2016; Riemann et al., 2017). This is because of the potential risks that sleep medication constitutes for the patient long-term for instance, dizziness, drowsiness, addiction, and relapse when the medication is discontinued (Buscemi et al., 2007; Riemann et al., 2017). Furthermore, the lack of suffi cient evidence on the long-term eff ects of pharmacotherapy poses additional problems when it comes to prescribing. Following this information, cognitive behavioral therapy for insomnia (CBT-I) is assumed to be the better treatment option for long-term (Riemann et al., 2017).

With that being said, the researchers of the present meta-analysis don’t know of any meta-analyses published that included controlled studies on the long-term eff ects of CBT-I on patients. van der Zweerde and her colleagues aim to address this gap in the literature and include all available randomized control trials (RCT) reporting on the controlled long-term eff ects of CBT-I at 3, 6, and 12 months and quantifying these eff ects (van der Zweerde et al., 2019).

The focus of the present meta-analysis was on subjective sleep outcomes according to accounts of sleep diaries and self-reported symptoms. The researchers used their previous meta-analysis as a starting point and checked whether any of those studies had published follow-up measurements or data since then. Subsequently, they performed a new search that covered the period of time from the end of the previous search (December 2015) to May 2018 (van der Zweerde et al., 2019). The inclusion criteria were the following: RCT design, investigation of CBT-I or its components, adult participants, self-reported or formally diagnosed insomnia, the comparison to a non-active control group, inclusion of sleep diary outcomes,

reporting of follow-up data for 12 or more weeks post test, and the provision of suitable data for eff ect size calculation (van der Zweerde et al., 2019). The CBT-I components were defi ned as relaxation therapy (RE), sleep restriction therapy (SRT), stimulus control therapy (SC), paradoxical intention (PI), and cognitive therapy (CT) (van der Zweerde et al., 2019).

This study was focused primarily on insomnia severity which was measured through questionnaires (ISI), and secondarily on sleep onset latency (SOL) and sleep effi ciency (SE) measured through sleep diaries. The data extraction process involved coding characteristics such as publication year, recruitment setting, insomnia defi nition, comorbidity, age group, treatment format, number of sessions, control group type, and intervention type. Lastly, the statistical analysis involved computing Hedges’ g for eff ect sizes, using a random eff ects model to account for heterogeneity, and checking for outliers and publication bias (van der Zweerde et al., 2019).

The fi nal meta-analysis included 29 studies after excluding one outlier. The results showed that CBT-I had signifi cant eff ects on insomnia severity at 3, 6, and 12 months with eff ect sizes declining over time (large at 3 months, moderate at 6 months, small at 12 months). The researchers found that three months after treatment, the severity of insomnia complaints (primary outcome) was better for patients treated with CBT-I than for patients without the active treatment. Additionally, signifi cant eff ects were observed for SE and SOL at 3, 6, and 12 months (van der Zweerde et al., 2019).

In general, there was a steady decline overtime of the long-term eff ects of CBT-I. This indicates that the long-term eff ects of CBT-I are smaller than the eff ects seen in the short-term. There are several factors we have to consider when analyzing these results. The analyses at the diff erent time points include diff erent studies which could lead to biased eff ect size estimates. Furthermore, patients in the control conditions could start sleeping better over time either because they have sought treatment elsewhere or purely as a result of time passing. As the time of follow-up increases, so does the time frame of opportunity to seek treatment elsewhere. Finally, when looking at the data for the individual studies included in the present meta-analysis, participants reported a return of symptoms over time. This could be because the interventions are relatively short and focus on behavioral changes such as lifestyle, bedtimes, and sleep hygiene (van der Zweerde et al., 2019).