Cognitive behavioral therapy for insomnia (CBT-I) is both efficacious and effective, supported by well over 100 randomized controlled trials (RCTs) and recommended as first-line insomnia treatment by the US National Institutes of Health,¹ the American College of Physicians,² the American Academy of Sleep Medicine,³ and the British Psychopharmacological Society.⁴ Despite these robust clinical outcomes, access to CBT-I has been limited for a variety of reasons, including a well-documented shortage of trained behavioral sleep specialists⁵ as well as uncertain financial return on investment for insomnia treatment services.⁶ To increase access to care and leverage economy of scale, researchers and others have delivered multi-component CBT-I utilizing a variety of formats, including group face-to-face care, telephone consultation, bibliotherapy, and digital/mobile interventions.

The earliest known publication of a digital CBT-I intervention was a RCT conducted by Strom and colleagues⁷ in Sweden. Since that time, no fewer than 11 different digital CBT-I technologies have been developed in academic centers and published in dozens of peer-reviewed articles.⁸ Outside of the research domain, literally hundreds of commercial sleep apps are available. Ample evidence suggests that rigorously developed digital CBT-I interventions can reduce insomnia severity with effect sizes comparable to face to face care. Further, digital CBT-I approaches have been found to improve multiple aspects of daytime function including symptoms of anxiety, depression, fatigue, and health-related quality of life.^9,10 In general, longer treatment duration, inclusion of more CBT-I components, and greater support have been associated with more favorable outcomes.⁸ At the same time, relative to face to face care digital CBT-I results in smaller improvements in sleep parameters, and attrition also remains a major challenge for digital CBT-I. Greater personalization is required to reduce elevated dropout rates.

In the current issue of the Journal of Clinical Sleep Medicine, Vedaa and colleagues¹¹ report 18-month follow-up data from a previously published RCT of digital CBT-I in Norway.¹² Eighteen months following the conclusion of treatment, 66 (70%) participants who completed the clinical trial completed follow-up assessment as well as 10 days of sleep diaries. Based on these results, reductions in insomnia severity (as measured by the Insomnia Severity Index [ISI] and Bergen Insomnia Scale) were maintained, and dysfunctional beliefs about sleep (as measured by the Dysfunctional Beliefs and Attitudes about Sleep scale – 16 item version) were further reduced between post-treatment and 18 months. Conversely, excepting total sleep time, which improved slightly, gains in sleep diary variables (sleep onset latency, wake after sleep onset, and early morning awakening) that had been observed posttreatment were reduced slightly but significantly by 18 months. Forty-six percent of participants were considered treatment responders (ie, experienced a reduction in insomnia severity ≥ 8 on the ISI), and 37% of participants were considered in insomnia remission at 18 months. Finally, although some posttreatment reductions in fatigue were lost by 18-month follow-up, improvements in anxious and depressive symptomatology (as measured by the Hamilton and Anxiety Rating Scale) remained stable throughout this period. Overall, this study was well-conducted, and these results can be viewed favorably: relative to web-based education, digital CBT-I resulted in improvements in sleep parameters and reductions in clinical insomnia severity, with most gains being maintained for at least 18 months.

These findings are consistent with previous studies over the past twenty years that demonstrate the long-term gains associated with face-to-face CBTI (eg, Morin et al.¹³) and more recent results supporting the stability of gains resulting from digital CBT-I.^14–16 This study also provides important insight into the positive and sustained impact of digital CBT-I on daytime function. Eighty-five percent of insomnia is comorbid, and CBT-I has been shown to have positive impact on comorbid medical and psychiatric conditions including anxiety,¹⁷ depression,¹⁸ posttraumatic stress disorder,¹⁹ alcohol and substance use disorders,²⁰ and chronic pain.²¹ In the present study, the authors found a reduction in anxious and depressive symptom-atology that was maintained for a minimum of 18 months.

Despite these strengths and valuable insight into the long-term effects of digital CBT-I, this study missed opportunities to provide insight into three essential areas of digital CBT-I research (and sleep disorders research more broadly). First, digital CBT-I studies to date have primarily employed traditional RCT methodologies. Although traditional RCT designs such as that employed by Vedaa and colleagues can provide needed insight into efficacy and effectiveness, these studies have thus far provided limited insight into factors influencing personalized medicine approaches for digital CBT-I. Given the elevated dropout rates evident in digital CBT-I trials, greater understanding is required for which patients are likely to benefit (and not to benefit) from digital CBT-I. This will require researchers to incorporate patient-centered factors such as treatment preferences into study design and assignment of treatment conditions, in addition to sophisticated moderation and mediation analyses to identify patient-centered barriers and facilitators to digital CBT-I. Fortunately, investigators are currently conducting studies of digital CBT-I in primary care settings to employ stepped-care approaches and to improve matching of patients and treatments based on patient preference and disease severity.

Second, the study by Vedaaa and colleagues provides minimal insight into factors influencing implementation and dissemination of digital CBT-I. Although scalability is a key benefit of digital CBT-I, regrettably few studies have employed implementation or dissemination research methods. For example, studies to date have generally failed to incorporate stake-holder perspectives, such as how referring providers perceive the clinical utility of digital CBT-I and how these technologies might be incorporated into real-world clinical practice. As health care shifts toward value-based care and systems-level population health,²² health care decision makers require evidence-based guidance how to incorporate digital CBT-I technologies into clinical armamentaria on a large scale.

Third, this study fails to provide any insight into the health economic aspects of digital CBT-I. Although longer follow-up periods such as that reported in the study by Vedaa and colleagues will enable analysis of longer economic time horizons, in this study only outcomes from the patient perspective were included. A health economic perspective requires assessment of key outcomes from multiple perspectives, including the payer perspective and the employer perspective.⁶ In the present study, the authors failed to report results concerning any traditional health economic endpoints, such as quality adjusted life years (which often can be gleaned from measures of health-related quality of life, which was not assessed), health care utilization, or workplace productivity.^23–25 Thus, despite the contribution of providing additional data regarding long-term clinical outcomes of digital CBT-I, this study fails to move the needle regarding three of the largest issues facing the field.

Digital CBT-I is here to stay. Given the promise of digital CBT-I²⁶ to increase access to insomnia care for tens of millions of individuals worldwide, the past decade has seen an explosion of interest in digital CBT-I from researchers and clinicians alike. From a scientific perspective, evidence supporting digital CBT-I will continue to accrue. Yet the scientific needs outlined above remain to be addressed. From a commercial perspective, there have been flurries of activity over the years, some visible and some less so, which can be viewed as important indicators of broader interest in digital CBT-I. At the same time, stand-alone digital CBT-I technologies are largely commoditized, and there are few barriers to entering the marketplace. Thus, the scientific advancement and commercial potential of digital CBT-I share common needs: enhanced personalization, implementation in health systems, and greater attention to health economic outcomes. We and others are actively pursuing such efforts.

In summary, the well-conducted study by Vedaa and colleagues builds upon the growing body of evidence demonstrating the long-term clinical effectiveness of digital CBT-I technologies. Although much work remains, these data provide important confirmation of the maintenance of gains associated with unguided digital CBT-I as well as the positive impact on daytime functioning.