CBT-I Coach: A Description and Clinician Perceptions of a Mobile App for Cognitive Behavioral Therapy for Insomnia

Abstract

Study Objectives:

This paper describes CBT-I Coach, a patient-facing smartphone app designed to enhance cognitive behavioral therapy for insomnia (CBT-I). It presents findings of two surveys of U.S. Department of Veterans Affairs (VA) CBT-I trained clinicians regarding their perceptions of CBT-I Coach before it was released (n = 138) and use of it two years after it was released (n = 176).

Methods:

VA-trained CBT-I clinicians completed web-based surveys before and two years after CBT-I Coach was publicly released.

Results:

Prior to CBT-I Coach release, clinicians reported that it was moderately to very likely that the app could improve care and a majority (87.0%) intended to use it if it were available. Intention to use the app was predicted by smartphone ownership (β = 0.116, p < 0.05) and perceptions of relative advantage to existing CBT-I practices (β = 0.286, p < 0.01), compatibility with their own needs and values (β = 0.307, p < 0.01), and expectations about the complexity of the app (β = 0.245, p < 0.05). Two years after CBT-I Coach became available, 59.9% of participants reported using it with patients and had favorable impressions of its impact on homework adherence and outcomes.

Conclusions:

Findings suggest that before release, CBT-I Coach was perceived to have potential to enhance CBT-I and address common adherence issues and clinicians would use it. These results are reinforced by findings two years after it was released suggesting robust uptake and favorable perceptions of its value.

Citation:

Kuhn E, Weiss BJ, Taylor KL, Hoffman JE, Ramsey KM, Manber R, Gehrman P, Crowley JJ, Ruzek JI, Trockel M. CBT-I Coach: a description and clinician perceptions of a mobile app for cognitive behavioral therapy for insomnia. J Clin Sleep Med 2016;12(4):597–606.

INTRODUCTION

Cognitive behavioral therapy for insomnia (CBT-I) is an evidence-based psychotherapy shown to effectively treat chronic insomnia. Numerous randomized controlled trials, summarized in several meta-analyses, have demonstrated that CBT-I leads to significant decreases in symptoms of insomnia and improvement in sleep efficiency, sleep fragmentation, and sleep onset latency.^1–5 This body of research has shown that CBT-I results in significantly better outcomes compared to placebo therapy,⁶ sleep hygiene alone,⁷ and relaxation therapy,⁶ and improvements are comparable to sleep medication at the end of treatment but are more durable after study treatments are discontinued.^8,9

Given the high rate of insomnia found in veterans treated at U.S. Department of Veterans Affairs (VA) clinics,¹⁰ the demonstrated effectiveness of CBT-I, and its recommended use as the first-line treatment for insomnia,¹¹ the VA initiated a national dissemination effort to train its licensed mental health clinicians (i.e., non-sleep specialists) in CBT-I that is currently ongoing.¹² Program evaluation data on patient outcomes from this effort show very large reductions in insomnia (pre-post effect size d = 2.2–2.3),^13,14 which are comparable to those of published trials.^15,16

Despite the success of the training program, there is significant room for improvement. For example, nearly a quarter of patients drop out of treatment prematurely or cannot regularly attend sessions.^13,14 For those who do complete the protocol, 40% fail to show marked improvement and over a quarter continue to have moderate to severe clinical insomnia, while another 38% have sub-threshold insomnia. Because patient adherence to the protocol as rated by the therapist (e.g., going to bed only when sleepy, getting out of bed at the prescribed time) is associated with better outcomes,¹⁴ efforts to improve patient adherence may increase CBT-I effectiveness.

BRIEF SUMMARY

Current Knowledge/Study Rationale: Cognitive behavioral therapy for insomnia (CBT-I) is an effective treatment for chronic insomnia, but patient adherence issues (e.g., dropout, homework non-compliance) can limit its benefits. A smartphone app could help address these issues by making CBT-I procedures more convenient for patients to complete and easier for clinicians to deliver.

Study Impact: CBT-I Coach is a free, publicly available patient-facing smartphone app intended to augment clinician-delivered CBT-I by facilitating the delivery of major CBT-I treatment components, including sleep educational materials, daily sleep diary completion, stimulus control guidelines, sleep restriction procedures, and anxiety management and cognitive therapy tools. Surveys of CBT-I clinicians in the U.S. Department of Veterans Affairs suggest that CBT-I Coach is favorably received in terms of its potential to improve practice and address common adherence issues and that it is being widely used by CBT-I patients.

A particularly promising approach that could help address adherence issues is the use of patient-facing mobile applications (apps) on smart devices (including smartphones and tablets) in care.^17,18 Emerging research suggests that such mobile interventions could potentially improve care for depression, stress, and substance use.¹⁹ Integrating patient-facing mobile apps into existing evidence-based protocols is feasible as smartphones and other smart mobile devices are now being used by a majority of adults in the U.S.,²⁰ including a majority of VA patients being treated for posttraumatic stress disorder.²¹

Given this promise, VA's National Center for PTSD partnered with Stanford University's School of Medicine and the Department of Defense's (DoD) National Center for Telehealth and Technology to build CBT-I Coach, a mobile app designed to enhance the delivery of CBT-I and mitigate common adherence issues. This paper briefly reviews the CBT-I protocol that has been broadly disseminated throughout the VA healthcare system¹⁹ and describes CBT-I Coach, including ways in which its key features and functions can enhance care and possibly improve adherence. A second aim of the paper is to report results of two surveys of VA CBT-I protocol-trained clinicians. The initial survey assessed clinicians' perceptions of the potential value and likelihood of adoption of CBT-I Coach before it was released. The second survey briefly assessed the uptake of CBT-I Coach and impressions of its value two years after it became available.

The VA CBT-I Protocol

The VA CBT-I protocol is designed to be delivered by a licensed mental health clinician over approximately 6 hour-long, individual, weekly sessions.¹⁹ Treatment can be completed in fewer sessions if the patient makes sufficient improvement and relapse prevention strategies are discussed. The first session is a comprehensive insomnia assessment that guides an individualized case conceptualization, which informs what and in what order CBT-I treatment components are delivered. Daily monitoring of sleep takes place throughout treatment using the Consensus Sleep Diary.²³ The sleep diary is used for assessing baseline sleep patterns, tracking change over time, and setting and adjusting the time spent in bed. The Insomnia Severity Index (ISI²⁴) is also part of the continued assessment. It is completed by patients each week to gauge insomnia severity at baseline and monitor treatment progress. As described below, the 4 major therapy components included in the VA CBTI protocol are stimulus control,²⁵ sleep restriction,²⁶ methods to reduce arousal in bed, and sleep-focused cognitive therapy. Clinicians also provide tailored sleep education to support engagement in the above components, as well as education about sleep hygiene (e.g., how substances and medications, exercise and diet, and aging affect sleep).

Stimulus control consists of 5 recommendations aimed at strengthening the association between the sleep environment and sleeping. The 5 guidelines include: not going to bed unless sleepy, waking up at the same time every day, getting out of bed if unable to sleep, minimizing non-sleep activities in the bed (e.g., watching TV), and limiting napping.

Sleep restriction involves initially reducing time spent in bed and then gradually increasing it. Clinicians use standardized algorithms to generate the initial personalized time in bed window, which in most cases is the average reported total sleep time over the previous week, and to determine if and how to alter the recommended time in bed based on sleep efficiency (i.e., the percent of time in bed spent asleep) and the individual's sleep need.

Methods to reduce arousal in bed that are included in the VA CBT-I protocol include: relaxation techniques (e.g., progressive muscle relaxation [PMR], diaphragmatic breathing, meditation, and positive imagery), creating a pre-sleep buffer to allow for unwinding before going to bed, and techniques to address intrusive thoughts in bed (e.g., “worry time,” a technique used in evidence-based therapies for generalized anxiety disorder²⁷).

Cognitive therapy techniques are used to address thoughts and beliefs that interfere with sleep or adherence to the behavioral recommendations. An example of such thoughts and beliefs is: “If I don't get enough sleep, I won't be able to function.” For a more detailed description of the techniques used, see Manber et al.²²

Sleep hygiene involves recommendations about consumption of contraindicated substances, diet, exercise, and changing the sleep environment. For example, patients are encouraged to make their bedroom as conducive to sleeping as possible by setting a comfortable cool room temperature, minimizing noise and light, turning the clock around to prevent “clock watching,” and closing the door to minimize distractions.

CBT-I Coach

CBT-I Coach is the product of a partnership between VA, Stanford University, and the DoD. The development team included experts in insomnia, CBT-I, clinical intervention development, technology, and implementation science. CBT-I Coach is primarily intended to be used by patients undergoing CBT-I with a healthcare professional and targets military veterans and service members as well as civilians with insomnia. It is not designed to replace clinician-delivered CBT-I; nonetheless, the app can be used as an educational resource. While CBT-I Coach does not automatically transmit data from the mobile device, users (of the iOS version only) can email their sleep diary and ISI data to themselves so they can print it out and share it with their CBT-I clinician.

CBT-I Coach for iOS was released in May 2013 and for Android in September 2013.^28,29 Both versions of the app are licensed software owned by the VA and are available at no cost to the general public (Figure 1). Since its release, CBT-I Coach has been downloaded over 80,000 times in 86 countries (through February 2016) and has received favorable user ratings in the app marketplaces (mean user rating of 4 stars for both versions). The following describes how VA CBT-I protocol components are presented in CBT-I Coach and highlights ways in which the app can facilitate and enhance adherence to CBT-I.

Figure 1. Home screen (iOS version).

The sleep diary in CBT-I Coach (Figure 2) includes 10 drop-down and text entry fields that assess sleep behaviors for an entered date, including napping (along with total duration of naps), bedtime, the time when first tried to fall asleep, latency to fall asleep, number of awakenings (not counting final awakening), total duration of all awakenings, final wake time, whether final wake time was earlier than desired, the time got out of bed, sleep quality rating, and a comments field that allows patients to keep notes about their sleep. To provide reinforcing feedback and enhance adherence, a progress bar shows sleep diary completion for the past 7 days. For each date a sleep diary was completed, total sleep time in hours and minutes are presented. Accumulated sleep diaries are displayed in a list and when a diary entry is selected, calculated total time in bed, total time asleep, and sleep efficiency are displayed, allowing for easy review. Likewise, graphical representations of sleep diary data provide visual summaries of changes in sleep parameters over the past 7 days (Figure 3).

Figure 2. Sleep diary entry (iOS version).

Figure 3. Graphical summary of sleep diary data (iOS version).

While recent projects have explored collecting sleep diaries via telephone,³⁰ the standard VA CBT-I protocol uses paper diaries. Unfortunately, paper entry can be difficult for some patients to remember to complete, fill out accurately, and bring reliably to sessions. In addition, the protocol requires calculations of variables derived from the information collected (e.g., sleep duration and sleep efficiency), which are time-consuming for the therapist. CBT-I Coach promotes daily completion and accuracy of sleep diaries and uses the data to automatically calculate derived sleep variables. Users are prompted to set a daily sleep diary reminder that is close to the time of getting out of bed and are required to answer all questions. Back-end algorithmic data checks help to ensure accurate data entry and calculation by alerting users to possible data entry errors. Two examples of common entry mistakes include: (1) entering an earlier sleep initiation time than time in bed, resulting in a negative calculated sleep onset latency time; and (2) AM/PM confusion, which may result in unusually short or long time in bed and/or total sleep time. CBT-I Coach calculates time in bed and total sleep time in real time, prompting users to confirm or correct entries when the calculated variables are unusual.

To facilitate insomnia symptom monitoring, CBT-I Coach includes the ISI,²⁴ with each of the 7 items being presented on individual screens. After completing the ISI, the user is prompted to set a reminder for completion of the next ISI and recommended to complete it no more frequently than once per week. A graphical display of changes in ISI scores over time (Figure 4) and of changes in individual item scores provides easy-to-digest feedback about treatment progress.

Figure 4. Insomnia severity index assessment history (iOS version).

CBT-I Coach offers comprehensive sleep educational materials from the VA CBT-I protocol. It provides an overview of CBT-I and related topics, including functions of sleep, stages of sleep, sleep regulators (i.e., sleep drive, circadian process, arousal), sleep apnea, and substances and medications that impact sleep. It also includes a section with information on napping, winding down, eating, caffeine use, exercise, and alcohol use. Lastly, a glossary of CBT-I terms allows for quick reference to definitions of CBT-I-specific language (e.g., total sleep time).

Stimulus control guidelines and sleep hygiene recommendations are included in the “Create New Sleep Habits” section of the app. Users are provided with dynamic checklists that can be tailored to their interests, with ideas for encouraging adherence to stimulus control guidelines and sleep hygiene recommendations. For example, in the “Go to Bed Only When Sleepy” section, users can scroll through a list of suggestions to help keep them awake until their prescribed bedtime and choose those they find most helpful, which will then appear as primary suggestions when the user returns to this section. Additional sections also provide dynamic checklists to encourage users to get out of bed when they cannot sleep, get out of bed at their prescribed time, set up an optimal sleep environment (e.g., low noise level and comfortable temperature), and reduce caffeine intake.

In the implementation of sleep restriction therapy, CBTI Coach provides the option of automated, app-generated or manual entry of the time in bed period. The automated option is based on back-end algorithms that take into consideration the average total sleep time and scores on the Sleep Need Questionnaire.³¹ To improve adherence to the recommended time in bed window, in addition to the tool mentioned above, CBT-I Coach allows patients to set reminders for their prescribed bedtime and wake time.

CBT-I Coach includes a number of audio-guided relaxation exercises, including PMR, diaphragmatic breathing, guided imagery, and mindfulness exercises. Suggestions for “winding down” activities to do before going to bed (e.g., taking a bath or preparing dinner for the next day) are also provided. Users can select activities they find most useful and these are then displayed first when the user returns to this section of the app.

Cognitive therapy-related skills are addressed in CBT-I Coach via a set of tools that are designed to facilitate cognitive restructuring. The “Changing Your Perspective” section of the app focuses specifically on cognitive restructuring, substituting an alternative more accurate or helpful thought for common sleep-related thoughts that interfere with sleep. Users can select to focus on thoughts related to their sleep or trauma-related thoughts, in the case of PTSD. The app presents common helpful alternative thoughts (e.g., “I will survive even if I don't sleep at all tonight”). The users can review as many of these as they like until they find one that fits their situation. Tools to facilitate “worry time” are also listed in this section. Users can log the content of their worry thoughts in the app and schedule a time to worry about them outside of time allocated for sleep. The user can set a reminder to ensure they do not forget to address the worry thoughts on the list they create, which may help further ease the mind of anxious patients.

CBT-I Coach also includes a relapse prevention section (“Prevent Insomnia in the Future”), in the form of a checklist. This tool helps patients identify CBT-I tools they can use should insomnia re-emerge. This tool also provides guidelines for when patients should seek professional help should sleep deteriorate. In addition to guidelines concerning re-emergence of insomnia, the tool provides information about symptoms of common sleep problems other than insomnia (i.e., sleep apnea and restless leg syndrome) so that patients know when to seek help.

Aside from its potential to improve delivery of and adherence to specific CBT-I components, CBT-I Coach may have additional benefits. For example, as smartphones are typically carried throughout the day, they offer convenient, private access to CBT-I materials, allowing patients to take advantage of opportunities to engage in therapy content when they might not otherwise do so (e.g., while on public transportation), thereby increasing exposure to this content relative to paper handouts. Using smartphone apps might be particularly helpful for promoting adherence among younger individuals, previously identified as having poorer adherence to CBT-I³² as they are more likely to have smartphones.³³ CBT-I Coach might improve motivation to change sleep behaviors, treatment expectancies, self-efficacy, satisfaction with CBT-I, and reduce perceived barriers to changing sleep behaviors, factors all shown to relate to adherence to CBT-I.^32,34–37

VA-Trained CBT-I Clinicians' Perceptions of CBT-I Coach and Report of Use

While the development process for CBT-I Coach considered the realities of integrating the app into the VA CBT-I protocol, feedback was obtained from VA-trained CBT-I clinicians to validate design decisions intended to improve care. Preliminary goals were to assess the adoption rate as well as factors that might influence the uptake process among clinicians. The approach was grounded in Diffusion of Innovations Theory,³⁸ which posits the centrality of perceptions of an innovation that can influence the adoption decision of potential adopters. Perceptions of innovations that have been shown to be important in this process include the innovation's relative advantage over existing practices, compatibility with the adopter's needs and values, and perceived complexity or ease of use of the new practice.³⁹ We also assessed clinician characteristics (age, smart-phone ownership, having already used apps in care) that might relate to perceptions of the app and intention to use CBT-I Coach when it became available. Lastly, two years after CBT-I Coach was released, we assessed the adoption of CBT-I Coach and impressions of its value within this clinician community.

METHODS

Procedures

Study procedures were approved by the local VA hospital research and development committee and affiliated university IRB. In February 2013, all VA-trained CBT-I clinicians (n = 366) at that time were sent an email describing the study with an invitation to participate by accessing the web survey through an embedded hyperlink in the message. One week later, a follow-up email invitation was sent to those who had not accessed the survey from the initial invitation. This resulted in an overall response rate of 37.7% (n = 138) of the clinician population. Similar procedures were used to conduct a very brief survey of all CBT-I clinicians (n = 613) 2 years after both version of CBT-I Coach were released (i.e., September 2015) to assess uptake of the app and perceptions of its benefit for homework adherence and outcomes. The survey was intentionally kept very brief in attempt to maximize participation in order to obtain an accurate estimate of use. The email invitation noted that the survey would only take a few minutes and required answering just 8 questions. Overall, 28.7% (n = 176) of these clinicians completed the survey.

Participants

VA-trained CBT-I clinicians (n = 138) completed a web-based survey assessing their perceptions of the CBT-I Coach app prior to its public release. The majority of participants (87.7%, n = 121) were smartphone owners, and many (42.8%, n = 59) reported that they had already used or were currently using a smartphone app with patients as an adjunct to treatment (see Table 1), here referred to as early adopters. Two years after CBT-I Coach was released, 176 clinicians completed a very brief follow up web-based survey.

Table 1.

Participant demographic and professional characteristics.

Measures

Initial items on the pre-release survey assessed participant demographics (age, gender) and professional characteristics (e.g., profession, CBT-I caseload), as well as smartphone ownership and prior or current app usage with patients (i.e., early adopters). Participants were then asked to read a brief objective description of the core features and functions of the CBT-I Coach (see Appendix). Ten items assessed if clinicians believed that CBT-I Coach could enhance CBT-I (e.g., patient adherence and outcomes; see Table 2 for items). Lastly, as recommended,³⁸ a study-specific measure (based on one developed for evaluating diffusion of computer therapy)⁴⁰ was created with 17 items assessing participants' perceptions of the app in terms of its relative advantage, compatibility, and complexity (see Table 3 for items). As behavioral intentions have been shown to relate to changes in clinician behavior (e.g., Perkins et al.⁴¹), 2 additional items focused on participants' intention to use the app (i.e., future use intention) if it were available. Items were rated using a 7-point Likert-type scale that ranged from 1 = Strongly Disagree to 7 = Strongly Agree. Negatively scaled items were reverse-scored prior to aggregation. Alphas ranged from 0.75 to 0.84 for the 3 subscales demonstrating adequate internal consistency.

Table 2.

Potential CBT-I protocol-specific improvements of CBT-I Coach.

Table 3.

Clinician perceptions of CBT-I Coach.

For the brief survey 2 years after CBT-I Coach became available, participants reported how many CBT-I patients they treated in the last 12 months, how many of these patients used CBT-I Coach, how many patients they were currently treating using CBT-I, and how many of these current patients were using CBT-I Coach. They were also asked to rate the extent to which they agreed that CBT -I Coach improves homework adherence and improves patient outcomes using a 7-point Likert-type scale that ranged from 1 = Strongly Disagree to 7 = Strongly Agree.

Statistical Analyses

Data were analyzed using PAWS Statistics 21. Descriptive statistics (frequency, mean, and standard deviation [SD]) were used to report summary outcomes. To test if age, smartphone ownership, and early adopter status accounted for differences in mean ratings of perceptions of the app, independent samples t-tests were employed. To test if clinician characteristics and perceptions of CBT-I Coach predicted intention to use it (while controlling for number of CBT-I patients treated per week), a multiple regression analysis using simultaneous entry was conducted. For the data from the survey 2 years after CBT-I Coach became available, descriptive statistics were used to report summary outcomes and t-tests were conducted to assess mean differences between participants who had and had not used CBT-I Coach on impressions of its impact on homework adherence and outcomes.

RESULTS

As presented in Table 2, participants reported that they believed that it was moderately to very likely that CBT-I Coach could improve care in a number of ways. In general, participants also had positive perceptions of the app in terms of Diffusion of Innovation constructs, including agreeing that CBT-I Coach would provide a relative advantage to existing treatment, would be compatible with their clinical practice style, and would not be overly complicated to use (see Table 3). Participants also endorsed strong agreement that they would consider using (mean = 6.43, SD = 0.73) or intended to use the app (mean = 6.01, SD = 1.04) if it were available, with 97.1% (n = 134) and 87.0% (n = 120), respectively, agreeing to some degree with these items.

Regarding clinician characteristics relating to perceptions of the app, participants who were smartphone owners and those who were early adopters of using apps in care generally held more favorable perceptions across the diffusion of innovation constructs than their counterparts (see Table 4). Participant age (< 40 versus ≥ 40) on the other hand was not associated with significantly different ratings. As detailed in Table 5, a multiple regression analysis demonstrated that smartphone ownership and perceptions of relative advantage, compatibility, and complexity of the app were all significant predictors of future use intention, while age, number of CBT-I patients treated per week, and using or having previously used apps in care did not significantly predict future use intention.

Table 4.

Perceptions of CBT-I Coach based on clinician characteristics.

Table 5.

Summary of multiple regression analysis for variables predicting future use intention of CBT-I Coach (n = 138).

Regarding results of the survey 2 years after CBT-I Coach became available, among participants who saw at least 1 CBT-I patient in the past year (n = 172), 59.9% (n = 103) reported using CBT-I Coach with at least 1 patient during that time period and reported using it with an average of 35.4% (SD = 23.17) of their CBT-I caseload. Of those who reported currently seeing CBT-I patients (n = 138), 44.9% (n = 62) reported using CBT-I Coach with at least 1 current patient and reported using it with an average of 54.9% (SD = 28.50) of their current caseload. Regarding perceptions of CBT-I Coach's impact on homework adherence, participants who had used the app in the past year reported statistically significantly stronger agreement that it improved homework adherence (mean = 5.41, SD = 1.20) relative to those who had not used the app (mean = 4.13, SD = 0.62), t₁₇₀ = 8.15, p < 0.001. Likewise, these participants also had statistically significantly stronger agreement that CBT-I Coach improved outcomes (mean = 5.14, SD = 1.26) than did those who had not used it (mean = 4.07, SD = 0.58), t₁₇₀ = 6.56, p < 0.001.

DISCUSSION

The purpose of this manuscript was to describe CBT-I Coach, a mobile app for patients undergoing CBT-I, recently developed by a multidisciplinary development team for use by VA-trained CBT-I clinicians. Specific features of the app, such as automated sleep diary calculations, sleep education, tools to practice CBT-I skills, and reminders were included to promote adherence with completing sleep diaries, following the recommended time in bed, and potentially reducing treatment dropout and improving insomnia outcomes. Before it was publically available, VA-trained CBT-I clinicians were surveyed regarding their perceptions of the app in terms of its potential to improve care and adoption. Clinicians agreed that it was likely that the app could improve care in a number of ways and had generally favorable perceptions of the app's relative advantage, compatibility, and complexity. Clinicians owning smartphones and having already used apps in care had more favorable perceptions then did their counterparts. A majority agreed to some extent that they intended to use the app when it became available. Smartphone ownership and perceptions of the app's relative advantage, compatibility, and complexity were significant predictors of future use intention. After CBT-I Coach was available for about two years, a majority of survey respondents (59.9%) reported they had used it with a patient in the past year, and those who had used it had stronger agreement that it improved both homework adherence and outcomes compared to those who had not used it. Thus, these findings reinforce the findings of the pre-release survey.

Overall, these findings suggest that clinicians see the potential of CBT-I Coach to improve CBT-I. Diffusion of Innovations theory considers perceived relative advantage as an essential initial consideration in the adoption decision.³⁸ Thus, it bodes well that clinicians believed that CBT-I Coach would offer a relative advantage compared to current practices and that clinicians who reported using it agreed that it provided benefits in terms of improved patient adherence and outcomes. Our findings also provide insight into other factors that may affect adoption of CBT-I Coach by VA-trained CBT-I clinicians. Such insights could inform development of targeted dissemination and educational efforts that may increase adoption by clinicians who do not own smartphones or those who perceive the app less favorably (e.g., those who believe it would be too difficult to use). Based on findings of the survey at two years after CBT-I Coach became available, it may be helpful according to the Diffusion of Innovation theory's construct of observability (or the capacity to observe results of the innovation) to have clinicians who have used CBT-I Coach and found it valuable to share their successes with non-users to increase the adoption and spread of CBT-I Coach.

To our knowledge, this is the first paper to describe a patient-facing app for CBT-I, explore its potential to improve adherence to CBT-I, examine factors that could influence adoption of the app, and report on early uptake in a provider community. We had a relatively good response to our pre-release (37.7%) and two year post-release (28.7%) surveys. The pre-release survey was guided by a well-established theory and is a modification of a previously used and tested measure.

Despite these strengths, conclusions drawn from our findings are tempered by several notable limitations. Foremost among these is that participant pre-release perceptions of CBTI Coach were based on review of a brief objective description of the app, as it was not yet publically available. Although this is a recommended method of estimating the adoption potential of an innovation,³⁸ it is unknown if exposure to the app itself would have led to different responses. Unfortunately, our survey methodology was not longitudinal, so we were unable to link pre- and post-release data to evaluate if clinician factors and perceptions of the app predicted adoption. Similarly, across both surveys clinician beliefs about the potential of the app were assessed, so actual benefits in terms of enhanced adherence and outcomes related to use remain unknown. Although the overall response rate was good, a majority of VA trained CBT-I clinicians did not choose to participate in our surveys. Therefore, it is unknown if these non-responders systematically differed in important ways from those who did participate. For example, although all VA CBT-I clinicians must regularly use computers as part of their jobs (for email, electronic medical records, staff trainings), survey non-responders' choice not to participate may reflect less favorable attitudes about using technology. Despite efforts to maximize participation in the follow-up survey by making it very brief, the number of respondents was only 28% higher than the number who completed the initial survey even though the number of VA trained CBT-I clinicians increased by 67% between surveys. It cannot be ascertained if this lack of increased participation is reflective of a bias in participation toward those who hold more favorable attitudes about technology. If this is the case, the true estimate of clinicians who have used CBT-I Coach in this population could be closer to the lower limit of 17% (103 of 613). Alternative survey completion methodology (e.g., paper-or telephone-based surveys) should be considered in future studies to possibly enhance participation and minimize the potential of such biases. In addition, by making the follow-up survey very brief, information about the sample was not gathered that could have provided insight into whether smartphone ownership and age of participants has changed over time. Finally, participants were VA clinicians who may not be representative of CBT-I clinicians practicing outside of VA.

The CBT-I Coach app is currently limited by its inability to automatically transmit raw and calculated data directly to clinicians. Having such capacity, via a clinician-facing app or a web-based dashboard, would be a significant step towards increasing the efficiency of this data-intensive treatment protocol. Dashboards could additionally provide advanced analytics, aimed at an even richer understanding of sleep behaviors than is currently available through the basic calculations required in standard treatment.

The impact of connecting data from CBT-I Coach to clinician dashboards and examining objective measures of adherence, using data from the app, will be important to fully examine the potential benefit of CBT-I Coach. Our hope is that this paper will spur further investigations to answer such questions in order to aid in successful implementation of CBT-I and, thus, decrease the burden and disability associated with chronic insomnia. Further, while user reviews in the app market places suggest that CBT-I Coach is well received, examining actual patient perceptions would have provided valuable insight into factors influencing adoption of the app and should be examined in future studies. Recently, Babson and colleagues⁴² reported findings from a very small-scale pilot feasibility study showing that CBT-I Coach was perceived as user-friendly and helpful by the two veterans who used it for two weeks. Other factors, such as the expense of mobile devices and cellular or wireless plans, as well as patient demographic characteristics (e.g., age, education), lack of familiarity/comfort with technology, and sharing of data between patient and clinician should also be examined.

CBT-I Coach is an early entrant into a field that promises continued growth and adaptation over time. Personal technology is so ubiquitous, feature-rich, and supported by significant and relatively consistent nationwide infrastructure that its opportunities to enhance existing evidence-based psychotherapy practices are numerous. However, care must be taken at all stages of innovation to ensure that novelty does not overtake the need to ensure that products are appropriate, effective, and adequately integrated into existing systems and processes in order to ensure maximal consistency with theory, science, and necessities of practice. Therefore, it is incumbent to move beyond feasibility and acceptability, and to ensure that programs of research continue to evaluate CBT-I Coach for its ability to improve adherence to the standard CBT-I protocol and clinical outcomes in terms of efficiency, symptom reduction, or both. It will also be crucial to examine which features of the app are most effective, both from a patient perspective and via comparative studies, in order to maximize the effectiveness of the intervention. Evaluating whether CBT-I Coach improves outcomes relative to CBT-I without the app will also be an important next step.

DISCLOSURE STATEMENT

This was not an industry supported study. Julia E. Hoffman is a consultant to Otsuka Pharmaceuticals. Rachel Manber receives royalties from New Harbinger Press. Philip Gehrman has received grant/research support from Merck Inc. The other authors have indicated no financial conflicts of interest. The contents do not represent the views of the U.S. Department of Veterans Affairs or the United States Government.

ABBREVIATIONS

app

application

CBT-I

cognitive behavioral therapy for insomnia

DoD

Department of Defense

ISI

Insomnia Severity Index

VA

U.S. Department of Veterans Affairs

APPENDIX

Description of CBT-I Coach

CBT-I Coach is a smartphone app that is based on the therapy manual, Cognitive Behavioral Therapy for Insomnia in Veterans, by Rachel Manber, PhD, Leah Friedman, PhD, Colleen Carney, PhD, Jack Edinger, PhD, Dana Epstein, PhD, Patricia Haynes, PhD, Wilfred Pigeon, PhD and Allison Siebern, PhD.²² CBT-I Coach is meant to be used with face-to-face treatment for sleep difficulties. It can be used on its own with ease, but is not intended to replace therapy for those who need it.

The key features of CBT-I Coach include:

• An interactive sleep diary for the daily logging of sleep habits

• Automatic calculation of the sleep prescription, with options for the therapist to adjust it as needed, which can be informed by feedback from the Sleep Need Questionnaire included in the app.

• A validated sleep assessment (i.e., the Insomnia Severity Index) for diagnosing sleep problems, with a graph of scores to view monthly progress

• Dynamic tools to improve sleep, including relaxation exercises (e.g., diaphragmatic breathing, progressive muscle relaxation), coping self-statements, and a checklist for setting up the sleeping area

• Comprehensive educational materials about sleep, healthy sleep habits, barriers to sleep, and CBT-I

• Customizable reminders to alert users when to prepare for bed, go to sleep, get out of bed, record sleep habits, and take sleep assessments

• Relapse prevention information and tools